Table of contents
Extended
Abstract
Introduction.
Concepts
of intelligence
Formally
defining intelligence.
Constraints
on intelligent systems.
Intelligence
and apparent teleology
Intelligence
and Information Processing
Intelligence
in tomic systems
Intelligence
and communication
Intelligence
and subsystems
Intercommunication
between component modules of intelligent systems.
Learning
and intelligence
Coupling
between subsystems or components of intelligent systems
In
loosely coupled systems, coupling may be behavioural, even casual.
Examples of tighter coupling include
compulsory parasitism or symbiosis
Parasitic
or symbiotic vector relationships
Physiologically
profound symbiotic relationships
Colonial
sociable communities
Familial
communities
Colonial
eusocial communities
The
colonial nature of multicellular or physically crowded communities.
Language,
abstraction; dances and semiotics
Democracy
and didactics in communal intelligence
Intelligence
and subjective consciousness
Stages
and aspects of sociality
Hypothetically
independent organisms.
Sexually
reproducing species
Mutually
interdependent species
Rudimentary
intraspecific tolerance
Interspecific
and intraspecific flocking
Aggression
inhibitors (Tolerance releasers?)
Episodic
assembly
Rearing
care and defensive behaviour releasers
The
evolution of matching signal conventions
Eusocial
colonies and familial groups
Systematic
reaction to the environment
Establishment
of the colony
Maintenance
and management of the colony
Conflicts
of interests
Behavioural
pathology in communities
Intelligence
of colonies, and within colonies
Intelligence
of individuals or castes within colonies
Intelligence
of colonies
Culture
and intelligence: Elephants, apes,
orchestras and Selenites
Evolution
of intelligence
. . .wenn du lange in einen Abgrund blickst, blickt der Abgrund
auch in dich hinein.
(. . .if you gaze long enough into an abyss, the abyss will gaze back into
you.)
Friedrich Nietzsche
As compared
to the behaviour of their individual members, behaviour patterns of communities
raise questions concerning the very mechanisms and nature of intelligence. If the concept of intelligence comprises more
than one type of component then it is simplistic to reduce one’s view of
intelligence to limited aspects such as, at one extreme, information processing
and storage, or at the other, subjective consciousness and purposiveness. Currently it is not clear how many effects
should be lumped together as evidence for “intelligence” in general. It remains
inescapable however, that the intelligence of any entity is an emergent effect
of that entity's component entities' interaction.
The concept
of intelligence of communities, as opposed to that of individuals, is of
particular interest in that sense, because communities present a range of
outgroups that we may contrast to intelligence in individuals with variously
modular, but unitary, brains. To object
that one cannot compare intelligence in a unitary brain, with behaviour of a
community of communicating individuals, is not cogent; the brain’s intelligence
depends after all on the behaviour of communicating individual neurons.
That is not
in itself a definition of intelligence, but it certainly is an essential aspect
of the nature of intelligence. The subject is vividly explored in "The Mind's I" by Hofstadter &
Dennett, and I enthusiastically commend it to anyone as seminal
material.
The most
striking examples of putatively intelligent communities are eusocial insects,
but other types of communities suggest wider ranges of underlying principles
and lines of development. Concentration
on integrated brains or too narrow a range of types of community, may obscure
seminal questions.
Brains are
fuzzily modular, largely in ways that reflect the components of their
behaviour, and it is not clear how many sub-modules are the sites of
intelligences in themselves. When
modules communicate, does a new functional intelligence (a new “mind”)
emerge? If so, then how? To what extent and in what way might brain
modules retain identities when they form a team of minds in the inclusive
brain? If we unified two brains, would
the new entity have two minds or one? Or four? And how could we quantify them?
Our main difficulty in resolving such puzzles is that we lack physical measures
by which to detect or quantify what constitutes a mind, or to distinguish
between single minds, structures of minds, or fractions of minds. Without measurable criteria, discussion of
the concept of intelligence reduces to simplistic information theory,
observation, introspection, and thought experiment.
Other
information processing entities than unitary brains comprise recursive
sub-modules. Examples of such entities
include communities and possibly entire species throughout their existence,
with modules comprising individuals, groups, and super-groups. Unlike brain modules, sub-communities are not
connected physically by synapse or neuron, but they might behave intelligently
in ways reminiscent of Searle’s Chinese room, where no component recognises the
operation in which it plays a role, any more than a neuron in a human brain
might do. Conversely other types of
communities might show little intelligence, but include at least some
intelligent members who exploit their communities to achieve personal
objectives.
Such
examples show how community intelligence need not obviously be commensurable
with that of particular members. Even
the intelligence of individual members, or classes of members, such as
different castes in insects, or different professions in humans, might not be
usefully commensurable. Intelligence has
too many dimensions for adequate one-dimensional measurement.
That either
communities or brains are modular does not imply that their respective modules
are functionally analogous. It is not
logically necessary that every community must have either anatomical or
functional analogues to a brain or brains.
To faster, but not necessarily more complex minds, the intelligence of
long-lived communities or species may be imperceptible or qualitatively
different.
Apart from
community intelligence, aspects of the intelligence of insects differ from
intelligence in birds and mammals,
including eusocial mammals. Play activity
seems practically non-existent in any community regarded as an entity, and
mental generalisation is of a very low degree.
Insects, whether sociable or eusocial tend to be poor at passing on
information except of special types, such as information concerning alarm,
repeated arousal, or food. In some ways the stereotyped behaviour
patterns of individual social insects are reminiscent of the limited behaviour
repertoire of neurons. In contrast, sociable communities of intelligent
individuals such as gregarious species of Corvidae, commonly are quick to
acquire memes, and may propagate a meme for generations.
How does
one measure the relative intelligence of a system of components with
stereotyped behaviour, as opposed to systems with putatively subjectively
conscious components? It is a messy
business to compare the intelligence of a termite nest with that of a human
community.
Politics
offers one class of comparison between, if not actual measurement of, the
intelligence of the individual and the community when the intelligence of
individual members permits the establishment of rulers or mentors. Political structures are of particular
importance in the present context in that the field of politics is where individual
intelligence and whatever passes for the intelligence of the community, affect
each other directly, and the intelligence of the community simultaneously
exceeds or is exceeded by that of particular individuals.
Read not to contradict and confute; nor to
believe and take for granted;
nor to find talk and discourse; but to weigh and consider.
Some books are to be tasted, others to be swallowed,
and some few to be chewed and digested:
that is, some books are to be read only in parts,
others to be read, but not curiously,
and some few to be read wholly, and with diligence and attention.
Francis Bacon
This essay
is not a balanced treatise on any aspects of intelligence in any useful sense
of the word. At most I present
preliminary informal reading material to support a discussion on the subject of
apparent intelligence as manifested in communities of organisms. However, my only apologies are that the essay
is too short by an order of magnitude, that this adversely affects its coherence,
and that its construction has been so informal that I have produced hardly any
supporting bibliography. My biggest
disappointment is that I have been unable to structure all the subject matter
into a coherent conclusion or unifying principle. As I point out, I am doubtful that the concept
of intelligence is monolithic, any more than that of consciousness is. I am certain that we are not anywhere near
being able to demonstrate the truth or error of this opinion as yet.
Another
point of irritation to some readers might be the way in which I speak as though
colonies were comparable to individuals.
I neither do so purely accidentally, nor am I regenerate in the face of
accusations of argument by false analogy; in my view the human body – any metazoan
body - is a colony of cells. For our
purposes it generally is a clone of cells, though that is a far looser
definition than some might think; it also is arguably inaccurate.
Some of the
attributes of a colony of separate organisms are fundamentally different from
the attributes of a single, distinct body; others are logically analogous. Both those attributes that are logically
analogous and those that are alien, are relevant to various points made in this
essay. However, to assume that that a
community or colony is "nothing but" a body, or vice versa, would be
a gross example of the reductionistic fallacy, and this essay espouses no such
delusion.
Please note
that I am not being simplistic in this. For example, I am well aware that to
speak of an evolutionary strategy is not the same as speaking of a conscious
strategy in say, a chess player. I similarly realise that a decision made by a
swarm, or perhaps a mob, or the effect of such a decision, is not clearly the
same as the decision of a brain. Swarm
behaviour however, certainly includes episodes that functionally correspond to
decisions, such as where to select a new site the colony..
Similarly,
I do urge that readers remain alert for functionally valid analogies between
the intelligence of species as one type of unit, communities and colonies as
another, and unitary brains as a third. All these are systems or structures
that use information and apply its processing in meeting the opportunities or
challenges of the environment in various forms of coordinated behaviour. Even
positivistically we are far from any cogent argument against dealing with each
of the three as being intelligent in its own terms.
Given the
disputable nature of much of the subject matter, I expect and welcome
discussion, argument and criticism. As
for the document content, anyone is free to use it in any good faith connection
as they see fit.
The
question of what it means to speak of intelligence in systems other than
individual, possibly unitary, brains (or machines) is of key importance if we
are to make sense of the subject of intelligence at all. The very idea of the intelligence of a
community raises questions as fundamental as those raised by the scripting or
computer modelling of intelligent behaviour.
The intelligence of an insect colony is just one such dimension of
study.
It is
tempting to argue that it is meaningless to speak of the intelligence of a
group as anything but an analogy, and a foggy analogy at that, but the argument
is treacherous at best, and in my
opinion is simply invalid. Our brains
are groups of neurons, but I have never heard any cogent argument to
demonstrate that each neuron has a subjective consciousness. It is hard even to argue that it has any
intelligence at all, as opposed to raw information processing capacity.
To argue
that in the brain neurons are not a group, but are physically attached to each
other, is neither valid, nor unconditionally relevant. Much of the brain’s connectivity between
either different parts of itself, or between itself and the rest of the body,
is not at all by contact, nor even via neurons, but by chemical signals or even
by other media. It is theoretically
possible (though a good deal more tricky than science fiction authors might
suggest) to replace a synaptic connection with an artificial (say, electronic)
intermediary, and without obviously changing the effect on the receiving
neuron.
After all,
in spite of the belated current recognition of the ubiquitous and indefinitely
various importance of emergence as a concept and principle, it has been a
commonplace ever since the writings of Aristotle, that the whole is more than the sum of its parts. We do not assume that because a car is made
partly of bolts, partly of pistons, and partly of paint and the like, that it
should behave partly like a piston, partly like paint and so on. In fact, it behaves very like a car, even
though within it there are pistons behaving like pistons and bolts behaving
like bolts. What is more, even though
the headrests and windscreen wipers have little to do with the engine or the
wheels, we would not easily be satisfied with a car in which the headrests and
windscreen wipers did not work.
The
relationships between the parts may be as important as the parts themselves;
often even more important, because adjusting the relationships can not merely
change, but even invert the functions or effects of the entire structure.
To labour
the analogy still further, any organisational man will know that a number of
ships is not the same as a fleet, nor is a number of people the same as a
company, even if no two elements in fleet or company ever make physical
contact. The organised aggregate has
abilities that the individual lacks and it generates liabilities that the
individuals escape. To treat the
organisation in the same way as one treats a component individual, or even to
treat a small organisation in the same way as one treats a large organisation,
or the other way round, is to tempt disaster.
In this connection I strongly recommend a stimulating and entertaining
book by the late John Gall: Systemantics.
Furthermore,
simply by rearranging the parts, it is possible to change the nature of such a
structure radically, whether it is a device like a car, or an organisation like
a company or a fleet. The change might
be favourable, irrelevant or dreadful, but the fact is that a dismantled or
reassembled car can be a scrap heap, a stock of spares, a lounge suite, or a
power plant, without there being a single component unaccounted for, and
without there being any significant argument about whether what one now is
looking at still is a car. This
crushingly demonstrates that Aristotle had a point about wholes and sums of
parts, and as far as we know, he never even saw a car. When one deals with kits of interchangeable
parts, like Meccano or Lego, the concept becomes even more striking.
Exactly
analogously, a bee is not just a number of living cells, and a colony is not
just a number of bees, not even if one is a queen and all the others are
workers or larvae. And as with Meccano
or Lego, when one does have an insect colony, it is not just anyone who can
tell the components apart. Even in a
large human company, not every member could put a name to every colleague, nor
even recognise each one, and yet, not many people would argue that each time a
member dies, leaves, or joins the company, it is a new and different company. A
single change might sometimes metamorphose the entire company, but not
often.
Other
dimensions of intelligence include the intelligence of an entire species, as
opposed to any member or community of that species, or of ecological
interactions between species. One even might distinguish the population of the
species at a given moment, from its identity throughout its history. There is a lot to be said for the value of
charting the direction of information flow and the nature of the information
processing, to characterise the nature of the respective types of
intelligence. Such flow is far from the
whole story, but it is better than hand-waving about what impresses one person
and fails to impress another as defining intelligence.
In this
essay I do not explore the
ramifications of the Gaia concept, interesting though they may be, but I remark
in passing that I do not regard it as practically or intellectually cogent,
certainly not on this planet as things are at present. And the concept of
information flow is part of the reason for my scepticism.
All the
populations discussed here, whether higher taxons, species, colonies or
individuals, behave in ways that from our point of view amount to algorithmic
or at least systematic reaction to environmental situations. How is one to discriminate between such
behaviour and intelligence in any narrower sense? How is one to distinguish between human
cells and independently moving colonies of driver ants or of myxamoebae in a
slime mould grex? Is the distinction the
fact that our body cells stay together?
That holds no water. We shed skin
cells by the million daily, far more than the driver ant colony sheds non-reproductive
members; some types of leukocytes move in and out of our mucosa all the time;
blood and lymph cells range freely almost throughout our bodies.
That same
question becomes even trickier when we reflect that every distinct, non-trivial
brain that we have been in a position to investigate, has different regions
that deal, in more or less specialist ways, with more or less distinct
functions. In fact, there are good
grounds for suspicion that brains are necessarily
strongly modular in their architecture and function. The analogy to eusocial castes is hard to
refute.
Brain
modularity is of at least four major types, all of them of anatomical and
functional importance. The most
fundamental type arises from the nature of the creatures with the most advanced
brains: the vertebrates, arthropods, and more debatably, certain molluscs. This is the segmented nature of the central
nervous system. Ancestral organisms
originally had nodes of the central nervous system in each segment of the
body. Each module, each bodily segment,
had its own segment of the central nervous system. Their interconnection was necessary for
overall function and control, but they had local function and control as well. Their interaction in such coordinated
functions as the walking of millipedes has impressed generations of
biologists.
The second
type of modularity originated from a sort of reversal of the original
segmentation process, the combination of small, repetitive modules into fewer,
larger modules. The brains of both
humans and bees consist of multiple segments of their central nervous systems
assembled into a more closely integrated structure.
Vertebrates
accordingly have a cranial brain and a spinal cord. The two components are not cleanly separate,
but each does have its own architecture and range of functions. The brain evolved originally as a local
enlargement of the spinal cord, associated with the location of particular
sense organs, mainly sight, hearing, and certain classes of chemoception.
The third
type of modularity is elaboration of the functional enlargements of primitive
segments of the central nervous system.
The major regions of vertebrate and arthropod brains are outgrowths of
the segments that constituted the development of the front end of the spinal
cord into a brain. In the vertebrate brain the most conspicuous enlargements
are the cerebrum, the tectum, and the cerebellum.
Furthermore,
palaeontological evidence suggests that among some of the larger dinosaurs,
most notoriously in Diplodocus, there was a secondary enlargement, a node of
the spinal cord in the lumbar region.
This has since been criticised as a misinterpretation of the fossils,
but whatever the true facts might be, the very concept is consistent with the
idea of local enlargements of the central nervous system as adaptations to meet
local needs for control. A possibly
more significant extant example of distributed control is in the tentacles of
cephalopods. The necessary muscular and
sensory control are so elaborate that if the brain of the octopus had to
control the tentacles directly, the nervous connections would exceed the mass
of the rest of the tentacle. Instead,
local modules control simple functions and sensory data, communicating with the
brain only in terms of summarised data.
At first
sight this might seem alien, but it is not qualitatively different from the way
the nervous system in the vertebrate retina digests visual data before passing
it back for secondary processing in other areas of the brain.
However,
note the difference between the function of the modularity of local nervous
control in the cephalopod tentacle, and the spinal node of Diplodocus. The function of the latter almost certainly
would have been to overcome the problem of controlling hindquarters at a
distance from the brain that would have been so great as to cause disabling
transmission delay. In modern
cephalopods, especially the largest squids, there is a similar problem, in that
co-ordination of the control of the mantle is important, but that the animal
may be so large that nervous control signals from the brain to the various
parts would arrive out of synchronisation.
In this
case the correction is applied by varying the thickness of the axons that carry
the control signals to the various parts of the mantle. The speed of the impulses varies accordingly,
re-establishing the necessary synchronisation.
These
examples of contrasting or analogous problems and solutions focus attention on
the relevance or otherwise, of the costs of communication between processing
units. They illustrate how the requirements might affect the unity of the
mental system, or whatever it is that must deal with, react to, manage, or
disseminate, information.
The fourth
form of modularity is most familiar in the cerebral hemispheres of
mammals. Here we find that there are
many functions that according to received wisdom are located in particular
regions of the brain. Arguably the most
spectacular examples are various speech and vision processing functions. However, we now know that such locations are
far less precise than once was thought, and from person to person they vary
noticeably in placement and extent. In
short, such modules are not as clearly anatomically defined as the first three
types.
Insect
brains too are segmentally modular. In
fact, they are more clearly divided into parts above and below the digestive
tract, than any division in vertebrate brains.
Various parts of their brains produce outgrowths that may be more or
less conspicuous depending on their respective functions. In this they resemble vertebrates, and it
seems clear that in every type of nervous system we know, the preferential
growth of particular brain regions reflect the demands of storage and
processing functions. It also seems to
me that in all the more elaborate brains, there are minor functional modules
within large nodes, modules that are not clearly reflected in the gross
anatomy. The way in which zones of the
brains of say bees, swell or subside at various stages of their lives, suggests
that such anatomically indistinct regions have functionally modular roles, much
as in human brains.
There also
is room for analogy between brains that differ in such ways, and communities
that have larger or smaller populations of various specialist members. Such differences affect both the personality
of such communities and their skill bases.
Apart from
the problems of defining intelligence or intelligent systems, there are
problems concerning constraints on the validity of the venerable analogy
between colonies and organisms. As
people in this field use the various terms, there not only are problems of
fuzziness such as may be dealt with in fuzzy logic, but also problems arising
from false analogy and conflation of disparate, poorly defined concepts. Informal writers for example have compared
the queen of a colony of bees or ants with the brain, an analogy that is
neither functionally valid nor conceptually useful.
Other
problems arise where we lack the basic techniques for scientific investigation,
let alone quantification, of particular effects (for example, subjective
consciousness). Here we still lack
fundamental formulae corresponding for example to F=ma in physics.
Long before
we could claim to have solved such puzzles we would have to deal with the
questions arising from for instance, John. R. Searle’s 1980 “Chinese Room”
challenge, as described in his essay “Minds, brains, and programs”. Personally I reject Searle’s views of the
impossibility of machine intelligence equivalent to biological intelligence,
but at the same time I firmly reject the naïve AI claim (I do not like the term
“strong AI” in this sense) that related examples of scripting are essentially
equivalent to, much less that they accordingly explain, the activity of human
minds in dealing with, understanding, and solving such challenges. At the time
of writing this essay, advances in
sophistication in AI machine have been dramatic, but support, rather
than contradicting, the degree of difference between minds and currently
understood algorithmic processes. Where this will go in future is not yet
known, nor even understood, but we shall see.
One of the
major ways in which eusocial communities present assistance and challenges of
their own in understanding such systems, is that they are models strongly
reminiscent of the Chinese room, but contrast with the human brain in that
their modules are physically far less closely interdependent, and that none of
their members has the independent intelligence to substitute for the entire
system. (Remember Diplodocus and the
cephalopod innervation as similarly internally different, but interdependent
examples!) In contrast, either a single
human speaker of Chinese and English, or a small team, could substitute for the
Chinese room.
Such points
deserve emphasis because many people are over-reliant on simplistic
analogy. They will compare a particular
part of a colony, or even the entire population of the colony, with a brain,
commonly even with a human brain in particular.
In fact many metazoa exhibiting rudimentary but effective intelligent
behaviour do not have brains at all; some of them barely have nerves worthy of
the term, let alone definite central nervous systems. And yet, they do exhibit some sort of purposive
behaviour.
To argue
that a termite colony is not intelligent in any meaningful sense because the
colony has no definite anatomical analogue say, to the cerebrum or brain stem,
would be to miss the point. One cannot
in anticipation cogently dismiss the possibility of intelligence by defining it
in terms of familiar structures and pointing out, however validly, that the
subject under inspection has no structure that conforms to the definition of
those terms.
In fact
apart from pre-judgement on anatomical
grounds one also needs to be very cautious about arguments along the lines that
such a population has no functional
analogue of particular mental or brain structures. Apart from the trap of confusing anatomical
structures with structures of intelligence, there is the trap of limiting our
definition of intelligence to formal structures directly equivalent to those we
recognise in our own mental makeup. Here
too, observations of discrepancies between human and current artificial
intelligence, sound warning signals.
For one
thing, we know only a limited amount about our own mental makeup and it is
philosophically questionable in principle how comprehensively we ever can know it, or even assess it. This makes it difficult for us to recognise
equivalent mental formalisms in corresponding alien mentalities. What is more, there is no reason to suppose
that our mental modelling of our selves and our world must be or could be
comprehensive in certain important senses.
For example, we cannot show that any deeply different information
handling or control mechanisms for functionally dealing with the world must be
invalid or meaningless or not somehow mental.
Even if such chauvinism were justified to some extent, it would not
follow that the alien structures were not of similar importance in their own
terms and of value as outgroups in the study of our own mental structures.
For purposes of
recognising and characterising intelligence in taxonomically alien systems, we
are reduced to searching for intelligence that we can recognise empirically,
that is to say, recognising by its behaviour instead of by its physical
nature. As a rule it would make most
sense to look for behaviour that:
· Has selective value,
evolutionarily speaking, or could reasonably be argued to have had selective
value at some time. (The fact that it
might be currently counter-selective, say because of climatic change or
conflict with recent immigrants, such as humans, is at best contingently relevant.)
· Has homeostatic function. This is closely related to the previous
point, but is more specifically relevant to such things as food and
physiological control.
· Effectively meets the challenges
of contingency. Such challenges need not
be external circumstances, though those are the most obvious; contingencies
within the organism or colony are just as important, and often more
complex. Examples include hormonal and
neurosecretory feedback control. The
fact that there is no recognisable consciousness dealing with the challenges
does not mean that the behaviour lacks any component of intelligence.
These
criteria certainly are far too limited, but they do exclude a lot of processes
that otherwise might confuse the issue.
Apart from
examples that in our experience are generally biological, we are familiar with
programmed behaviour in inorganic systems, such that if we observed it in a
biological system, we should readily class it as intelligent to some degree or
in some sense. In strict terms programming
a machine to pass the Turing test still is beyond us, but at the same time it
also is too limited a test in the context of this discussion. Human intelligence is not the only form of
intelligence we might or should recognise.
Such
examples of empirically intelligent behaviour in programmed systems that are
not biological organisms, may or may not in specific cases explain aspects of
algorithms that the biological brain might use, but even if they do, it is by
no means clear that they describe the nature of the mind that solves the problem
and implements the solution. They do not
even prove that the brain uses the same algorithm. In fact, even within the same brain,
investigations of human brain processes suggest that the original solution of a
problem, as developed by a naïve brain, might not closely resemble the
practised or stereotyped solution of an experienced brain.
Conversely,
if humans achieved some sort of brain to brain communication as direct as the
internal neuron to neuron communication, what sort of group intelligence would emerge? And would the individual brains lose their
individual awareness of being? Whether
they did or not, would they individually be aware of any global consciousness
of which they were components? One
assumes that they could not be fully
aware of it; it is certain that we, with our currently single brains are not at
any time aware of everything that goes on within our skulls. For good discussions of the minds within, I
recommend works of Oliver Sacks, such as "Musicophilia".
Possibly
however, such a single brain might be aware of its attached meta-brain in
something like the way that a blind man feeling the foot of an elephant, may be
well aware that the foot is not the elephant, nor perhaps, the only foot of the
elephant, even if he could not conceive the full splendour of the elephant,
tail behind and trunk in front.
And if the
entire structure of connected brains developed a subjective awareness, would it
be aware of the awareness of individual brains within the system, or could
component brains have their private thoughts? Or might they struggle
desperately to get their individual thoughts to the attention of each
other? And what degree of
interconnection is necessary to generate any such shared consciousness? Must it be?
If not, then tactile, pheromonal, or vocal communication might suffice.
And even if physical neuron-to-neuron connection were not strictly necessary,
we do not yet know the extent to which insect, or even human, communities could
have awarenesses outside those of their individual community members.
If such a
connection were possible, in which individual members had individual
consciousnesses, and yet the mass had its own consciousness too, then it would
be a somewhat unimpressive sort of mass mind that did not have the capacity and
competence to deal with volumes and complexity of data and concepts beyond what
any individual brain could deal with.
Conversely,
it is conceivable that the group mind would have no use for whole classes of
concept that were of no immediate value to the group, but that occupied
individual brains. If so, it is quite possible that the intellects of
individuals could be on an altogether higher mental plane. By way of
illustration a human political body might include individuals and cliques of
intellects far beyond those of the political leaders, even if the leaders were
competent as leaders, and the intellectual population members, though competent
to perform research analysis or synthesis, would have been disastrous as
leaders. Not every president is an Einstein, and not every Einstein is a Thomas Jefferson.
These
realities are non-trivial, but not easy to establish definitively.
Another
difficulty is so great and so poorly defined that I hardly mention it and
certainly cannot discuss it cogently, or even comprehensibly: the nature or
function of subjective consciousness, its relationship to functional
intelligence, or even whether or how subjective consciousness is necessary or
relevant to functional intelligence. Such examples of intelligent behaviour as
we see in programmed systems, in eusocial colonies, or even in humans as
perceived by their fellows. They get nowhere near the problem of explaining
subjective consciousness, much less explaining it away.
We do not
know what the essential nature of consciousness in this sense may be, nor
whether is functionally important, or why, or whether it is no more than a
passive by-product of the way some part of our conscious brain works, or even
whether other information processing systems generate similar internal
subjective effects. Consider such
examples of intelligent behaviour as we see in programmed systems, in eusocial
colonies, or even in humans as perceived by their fellows. Our observations and
perceptions get nowhere near the problem of explaining, much less explaining
away, the function or reality of subjective consciousness.
Conversely
we cannot argue that, as I at one time suspected, subjective consciousness is
mentally irrelevant, just because, as I still suspect, it is a passive emergent attribute of a working
brain. A friend of mine once pointed out
that if we have such a consciousness and are aware of it, then that datum of
the awareness constitutes a material difference, an item of information,
distinguishing between two brains that otherwise would be functionally
identical, particularly in their external behaviour. This is more important than the academic
quibble it might seem. Such internal
awareness might be vital to certain classes of maintenance and modification of
the human mind, performed by operation on itself.
These
problems of definition go beyond the problems of meaningful characterisation of
a set, such as were discussed by Wittgenstein in his “Philosophical
Investigations” (roughly in paragraphs 59 to 100) or Hofstadter in “Metamagical
Themas”, topic “Variations on a Theme as the Crux of Creativity”. Not only would it require Wittgenstein’s
“family resemblances” to define some of the sets we discuss, but some of the
sets would justify subdivision on the basis of the disparate natures of their
elements, which of course, we know hardly, if at all.
Firstly, no
one has yet managed to devise even a prospect of a test for subjective
consciousness, qualia, or what Searle calls “intentionality”, though other
authors use the term apparently as interchangeable with intensionality. Some prominent writers even claim that such
things as subjective consciousness are obvious or logically necessary, but that
certainly is not obvious to everyone, nor has its logical necessity been shown
to be logically necessary. Some
obsessive empiricists stridently deny the existence of subjective
consciousness, on the basis that we cannot define it in any way that is
independently empirically verifiable or falsifiable. This view I reject, pending proof, not that
we cannot define it at present, but that we cannot in principle ever define
it. The fact that stone-age man could
not define magnetism did not mean that nothing existed that caused two
lodestones to move together or apart, even though not all stones attracted or
repelled each other all the time.
Some
respected authorities (not all of them authorities on information theory or
biology) go even further and dismiss the very concept of subjective
consciousness as an illusion. I shall
not in this essay waste space on the question of what my subjective
consciousness might be an illusion of, and to what it is an illusion, and what
it might mean for something to appear an illusion to something that is not
subjectively conscious, and why it should matter that it is an illusion, given
that the illusion is concrete enough to affect the brain’s activity physically,
as I mentioned my friend pointing out.
Personally I find the apparent intrinsic self-reference far less acceptable
than the idea that most people choke on, namely the perceived indignity that it
is conceivable that we might not really exist as aware minds, but only seem to
our unaware minds to do so.
Nor do I
personally remember ever writing a computer program that gave any evidence of
experiencing such an illusion.
Some insist
that the effects of subjective consciousness certainly are real, but are
limited to biological structures; others contemptuously dismiss the idea that
inanimate systems might lack similar internal processes, but as I see it, none
of these groups have achieved anything beyond proof by assertion.
Most people
assume implicitly that for our purposes effectively all humans have both
subjective consciousness and intentionality, but we base the assumption on
subjective analogy from introspection, plus generalisation on behavioural and
anatomical grounds, neither of which has been demonstrated to be inseparable
from subjective consciousness. The
existentialist problem of our inability to demonstrate personality in other
parties than ourselves remains intact, even without invoking sterile or
meaningless solipsism. We have no strong
evidence that behind the eyes of a human, there are more dimensions of
consciousness than behind a fine work of art or the screen of a computer that
passes the Turing test.
After all,
though that is not the way the authors typically put it, a favourite theme in
human literature, is the inability of people to demonstrate their intelligence
to each other. In essence they retail
failures of inverted Turing tests.
Joseph Conrad’s short story, “Amy Foster”, is a particularly direct
example, but the theme is ubiquitous and for example it recurs in the story
"I only came to use the phone" in "Strange Pilgrims", an
anthology of short stories by Gabriel Garcia Marquez.
Far more
importantly, as I have hinted already, we do not even know how many individual
subjective consciousnesses exist in any intact, functional human brain, nor can
we count them even in our own respective brains, nor can we tell how it feels
when they split or combine.
Investigations of split brains and stroke victims have demonstrated the
relevance of such questions, but the questions go a lot further than that. We have little idea of whether the apparently
“unconscious” or “subconscious” functions of our brains are in fact without
subjective consciousnesses of their own, and I have no idea of how we can find
out. But it seems to me that such
questions of whether we have swarms of consciousnesses in our skulls are very
relevant to the idea of the swarm or sub-swarm consciousness among colonial
animals.
Conversely,
for what it is worth, the Turing criterion stands firm. How many kinds of (or aspects of)
intelligence are involved in it, I cannot say, but if, on objective, existing
criteria, the behaviour of a system is logically indistinguishable from
intelligence, then we must find new, objective, criteria before we deny that
that behaviour comprises intelligence in any relevant sense. Searlely denial of intelligence in a
particular system, just on the grounds that one has not identified its nature
or its roots, ranks with Bouillaud, the prominent nineteenth century French
doctor who embarrassed his colleagues by suspecting that the phonograph might
be a ventriloquist’s trick. He had failed to imagine how it could be possible.
Later he humiliatingly had to rationalise his original mistake.
We who have
not yet denied the concept of conscious intelligence are in a more comfortable
position, whereas those who attempt proof by denial fail cogency until anyone
is able to define what is to be denied.
Now in this
discussion, when we examine the putative intelligence of communities, I shall
not ignore the presumable differences between “subjectively conscious
intelligence” and objectively intelligent behaviour, but, for lack of a means
of distinction, I often shall treat them in the same way except where the
putative difference turns out to be relevant.
Conversely,
when on the Turing principle one facilely accepts every arguably intelligent
behaviour of a system as evidence of intelligence, serious difficulties arise
on every hand. If a colony reacts
intelligently to deal with internal and external challenges, even though the
behaviour of its members is demonstrably stereotyped, then why is a computer
not intelligent when it controls a building or a supercritically configured
aircraft? It controls things in much the
way that a human would, sometimes far better than any human, even though it has
not even one clearly intelligent component.
And if we apply this criterion to a computer, then why not to a
thermostat or hygrostat, or even a proximity-fused bomb or a spring-loaded
mousetrap?
Why not in
general to any system that displays feedback control?
Accordingly
I both decline to define the concept of intelligence rigorously, and insist on
using the term largely as though it were meaningful and monolithic. I realise that this is unacceptable in the
long term, but it is no worse than doctors who discussed the nature of disease
in the days before human physiology or the germ theory were generally
understood, and hence before there could be any deeply meaningful
classification of disease. Had they
refused to discuss such matters before they achieved a fairly advanced
understanding of underlying mechanisms, they would have avoided talking a lot
of nonsense, but they also would have made no progress in definition or
treatment of disease, and we probably still would have had no proper
understanding of disease today.
So I insist
on continuing to speak of intelligence with no more fundamental knowledge or
justification than a medieval doctor basing his theories of disease on the four
humours.
. . . On a huge
hill,
cragged and steep, Truth stands, and he that will
Reach her, about must, and about must go;
And what the hill’s suddenness resists, win so.
John Donne
We are most
used to thinking of intelligence in terms of our own largely conscious
thought. We think of internal
representation and memory as essential for symbolic thought. We also think of external symbolic
communication. We think of abstraction,
explicit conception and manipulation of conception. Imagination, motivation and conditional
thought also are essential to intelligence as we see it in ourselves. Both to deal with internal conceptions and
external situations by reasoning with abstractions and analogies, we need
pattern recognition, abduction, induction, and deduction.
And yet,
how many of those are essential to intelligence in broader senses? Certainly not many of them occur in non-human
communities, and those that do, seem to fail to rival the development of the
human versions by several orders of magnitude.
But there
are difficulties all the same. We still
cannot recognise intelligence as such.
Suppose that a tree, or a species (as a community), truly thinks, but a
hundred thousand times more slowly than a human, and perhaps in some radically
different ways. Do we criticise it as
unintelligent because it cannot think fast, and perhaps cannot solve crossword
puzzles? Speed and literacy may not be
relevant as long as the tree can orient its leaves toward the light, avoid
contact with neighbouring trees and adapt to the prevailing wind and the
location of soil nutrients.
In
fact, it is not a simple matter to establish the nature of thought and of
intelligence as either essentially identical or distinct, though patently both
can occur in the same entity.
One can’t proceed from the informal to the formal by formal
means.
Alan Perlis
In this
discussion I simply do not formally define “intelligence”. The concept is not understood well enough for
either general agreement or cogent argument, and consequently such definitions
as one reads tend to be inconsistent and uncogent. As I point out elsewhere, we still lack any
way to recognise subjective consciousness, much less express its relationship
to subjective intelligence. This frustrates
useful definition. In the definition of
intelligence it is fairly sure that there are whole ranges of concepts that get
confused and lead to question begging and cross purposes in discussion. Rival points of view, valid or otherwise,
commonly depend on tacit assumptions that lead to conflation of ideas and to
disputants talking past each other.
As I
instance later on, comparison of the intelligence of individuals with that of
communities is fraught with traps.
Examples that we shall see include questions of complexity, redundancy,
efficiency and speed. Not one of these
turns out to be an unambiguous measure of superiority of intelligence.
The
favourite positivistic criterion for intelligence is empirical evidence for
problem solving and communication. In
many contexts it arguably is the most practically useful, but it is not
formally cogent. For one thing
empiricism without an adequate formal basis can hardly improve on (formally
fallacious) inductive argument unless there is adequate statistical argument to
support it. There cannot be cogent
statistical argument unless there is a cogent definition that supports cogent
criteria for diagnosis of intelligence.
To a great extent I do in fact rely on such empiricism, but it is for
lack of anything better, not because I mistake it for proof.
For the
present, pending major breakthroughs in the field, I personally am pessimistic
about defining intelligence compellingly.
Instead I cannot do more than propose a few constraints on what we
usefully might call intelligent systems.
Because of the poverty of our understanding of the field, it does not
follow that whatever falls within these constraints meaningfully qualifies as
intelligence. However, the construction
of such lists of constraints is as powerful an aid as we have at present, for
developing useful concepts and terminology.
The scale, properly speaking, does not permit the
measure of the intelligence,
because intellectual qualities are not superposable,
and therefore cannot be measured as linear surfaces are measured.
Alfred Binet
The
following proposals are neither exhaustive nor compelling; I intend them as
useful bases for informal discussion of constraints on what we might regard as
evidence for intelligence.
Notice that
I do not include the concept of personality as a constraint. It neither is clear that personality
(individualism or any form of unique behaviour or attribute) guarantees
intelligence, nor that it is necessary to intelligence. In practice, sufficient complexity in a
system almost guarantees individual behaviour patterns, simply as a consequence
of the nature of complexity and chaotic systems. All the same, it would require some very
peculiar arguments to justify regarding say, a hurricane as intelligent, even
though people who deal with them often personify them, and even though
hurricanes are immensely complex and no two storms have the same
behaviour.
The concept
of subjective consciousness also is not logically necessary for an entity to be
intelligent to all other intents and purposes, passing the Turing test, for
example, so I exclude it from the list of constraints.
I find you want me to furnish you with argument and
intellects too.
No, Sir, there I protest you are too hard for me.
Oliver Goldsmith
It is not clear
on what grounds to deny intelligence in any system that exhibits apparent
teleology, even when the same system simultaneously exhibits several versions
of possibly independent, sometimes conflicting, teleology. In this essay I unapologetically refer to
such behaviour as intelligent in certain contexts.
Such terminology is not to be confused with anthropomorphism. It is more like the empiric pragmatism of the
Turing test, which intrinsically does not even assume consciousness, let alone
anthropomorphic attributes, in the subject.
It also is difficult to decide at which level apparent purpose grades
into non-trivial teleology. A gin trap
might be seen as behaving with intent when it snaps shut on anything that
triggers it, but we might reasonably reject it as a candidate for teleological
behaviour on the part of the trap, though the person who set the trap
presumably acted teleologically. After
all, a door in the wind might behave very similarly to the trap, even though
there certainly would be nothing teleological about it. In any such system, how one defines the
intention, as distinct from noise in the system, is an arbitrary point. The engineer might be interested in a sprung
ratchet as a means of enforcing a single direction of movement, and any noise
it makes is well . . . no more than
noise. To the football fan with a
rattle, that self-same noise is the very point of the device.
We surely however must accept that the trapper who directs a pathway to
pass over his gin trap, does demonstrate significant teleology.
Contrast
that human trapper with the larva of many species of ant lion (Myrmeleontidae)
or of a worm lion fly (of the unrelated family Vermileonidae). The two species
have convergently evolved almost indistinguishable predatory behaviour. Their
respective larvae excavate pits in sandy soil, use those pits as traps for prey
organisms, throw up sand to keep the prey from escaping, and snap shut on any
prey as soon as it gets into the right zone of the trap. Now, you know and I know that there is a
difference between the proximal mental processes in the two types of system:
human and insect, but as we shall see in examining social systems, there may be
more for us to think about than just proximal mental processes. It also is reasonable to speculate whether a
given species of ant lion, seen as a population in a given region and time
span, exhibits teleology, even if the individual ant lion patently is not
equipped for teleology.
I think it
should be clear that it is quite possible to imagine behaviour that seems very
persuasively teleological in a system, and that in some such systems one is
very unwilling to grant intelligence once one sees how the programming of a
simple mechanism works. In cases of the
type I have in mind the apparent teleology evaporates on detailed inspection,
as in mousetraps or perhaps Skinner-type conditioning. Sometimes apparently teleological behaviour
may be very impressive, but be based on simplistic mechanisms (e.g. in scripted
systems in computers, or simple physics such as in the lifelike behaviour of
heated oil droplets or soap-bubble membranes.)
Highly sophisticated behaviour in invertebrates, whether as individuals
or in colonies, often turns out to be stereotyped and, on close investigation,
looks suspiciously mechanistic.
Similarly,
many systems that demonstrably engage in information processing are not
intelligent in any persuasive sense.
Examples include effectively target-seeking, apparently teleological
behaviour, which is necessarily controlled by information processing in simple
mechanistic programs. Common examples
are stereotypic, innate behaviour. J.H.
Fabre wrote impressive early discussions on stereotypic behaviour in
insects. In birds and mammals too, later
ethologists saw many examples, that were triggered by releasers.
Humans are
by no means free of apparently similar behaviour. John Crompton criticised
Fabre’s views on stereotypic behaviour.
He compared certain instances of behaviour in humans with stereotypic
behaviour in insects and pointed out that it is very hard to distinguish them
by simple external inspection. For
instance, Fabre spoke of the “Ignorance of instinct” that caused a hunting wasp
to seal a disturbed cell in which there was neither egg nor prey. John Crompton in his book “The Hunting Wasp”
compared this with his experience of having seen a heartbroken woman in the
Blitz, her house having been practically levelled by a bomb, but with the front
door still hanging in a fragment of standing wall. She “entered” and examined what she could,
then “exited” again, and carefully closed the door. As a deservedly popular lay writer on mainly
entomological topics Crompton cut no ice in the annals of ethology, but his
story did clearly illustrate the philosophical problems that arise from such
inter-specific comparisons.
Furthermore,
it is now well known that artificial stimulation of parts of the human brain
can elicit particular behaviour patterns.
In some cases the behaviour is of a visceral, unconscious, involuntary
type, though often coordinated, but in other cases it is fully conscious, such
as the hearing of a tune or the recall of a scene. These events are suspiciously evocative of
“earworms”, the irritating repetition of a tune or a phrase “running through one’s
head”. It also recalls the (almost?)
reflexive elicitation of appetitive behaviour by very stylised stimuli. For example, even in humans the merest smell,
or a few lines of suggestive sketch can stimulate hunger, sexual desire, or
fear.
Some such
stimuli occur extremely widely in the animal kingdom, so much so that they
occur in aposematic signals. The
eye-spots in the wings of moths or mantids might appear unpersuasive to humans,
but they have been observed to alarm small birds when suddenly revealed. I personally have repeatedly observed large
Calliphorid, but not moderately-sized muscid flies assume a pose of alertness
as soon as a chameleon brings both its eyes to bear. The flies then react so effectively as to
avoid the tongue impressively often. Is
such a response any less “intelligent” than the startle reaction of a human who
helicopters when suddenly encountering a representation of a snake or spider?
Given the
survival value of such reactions in the wild, are we to regard them as unintelligent? If they are not intelligent in the
individual, are they unintelligent in the species? There most certainly are bases for objections
to such views, but are the bases perhaps not too-narrow definitions of what
comprises intelligence? Would they not
exclude certain classes of behaviour that deserve to be called intelligence in
entities that differ from say, the higher primates and molluscs? For example, Protoctista, Cnidaria, plants,
species, programs?
For as long
as we cannot cogently define, characterise, and measure intelligence, it is
dangerous to exclude certain categories of behaviour from our definitions and
characterisations. Even in humans, the main virtue of IQ tests is that they are
numerically convenient. They also are
notoriously non-cogent, however useful they sometimes may be.
It also is
dangerous to argue that teleology depends on an understanding of the mechanisms
necessary for the success of one’s purposive actions. Consider a dog that has learnt to drop a ball
into the receiver of a machine that delivers a reward. If we then move the machine to another part
of the room, many such a dog will drop the ball in the original part of the
room, ignoring the fact that the machine is now a few metres away. If we argue that such a dog, in whom we had
originally thought we had recognised intelligently teleological behaviour, now
demonstrates a lack of intelligent teleology, then what are we to say of the
vast majority of human drivers of vehicles, or computer users? Most are incuriously,
blackly, ignorant, not only of how their apparatus works, but of the relevance
of many of the learnt actions with which they control the system.
In fact, it
is a safe bet that many experimental ethologists who use slot machines in such
studies, themselves have little idea of how the slot machines work. Also, concerning many aspects of such an
operation that they are confident that they do understand, the beliefs of such
people often are flatly wrong.
Nor is such
superstition and ignorance limited to humans low of intelligence. I have seen a whole class of gifted, lively
children, who, when I challenged them to explain how a pocket calculator
worked, eagerly informed me that one pressed the buttons. It took me an extended exchange to awaken
them to the idea that button pressing, seen as a computational primitive, left
something to be desired. If that
is teleology, then why are we sneering at those silly dogs?
I
personally have seen Argentine ant workers, Linepithema humile (formerly
Iridomyrmex humilis) walking across a leaf, where they were tending
scale insects. Perhaps 40 mm away from
such an ant, there sat a Hymenopteran parasitoid of the scale insects. When an ant came within some 10 mm of the
wasp, it apparently saw it, abruptly changed course, and charged at it at a
much higher speed than its erstwhile relaxed walk across the leaf. The wasp flew off, and the scale insects
remained unharmed.
Nothing
could be more deliberate, more clearly teleological. The wasp wanted to avoid being killed by the
ant, especially before it had finished laying its eggs in the scale
insects. The ant “wanted” to protect
its important food resource, and to eat the wasp that threatened the scale. And though I did not specifically see this at
the time, the scale “wanted” the ant to keep protecting it and removing the
troublesome honeydew that otherwise would attract fungi. Simultaneously, from a different perspective,
the communities of ants, scale and wasps intelligently, teleologically, pursued
their ends. Very likely the injured
plant also was emitting chemical signals to attract the wasp, so that it might
attack the scale insects.
Or so we
might reasonably gather from the observations.
Nothing I saw argued the contrary.
If similarly I had been in a position to watch, say a sunbird chasing a
rival from a favoured Tecomaria bush,
an orang utan chasing a monkey out of a disputed fig tree, a soldier
defending a magazine from an insurgent, or a shepherd chasing a jackal from his
flock, I should have interpreted what I had seen in just such terms. How am I to justify a different
interpretation in the case of the ant?
Indirectly
I could base counter arguments on published studies of the brains involved, or
on publications on comparative ethology, but even after that, assuming that I
accepted the arguments of those publications, I am left with nothing better
than arguments from analogy to convince me simultaneously of direct conscious
purpose in the case of the vertebrate minds and its absence in minute insects
or their communities: “I have a
subjectively conscious mind with its associated conscious purposes and I have a
brain that supports all those things, and therefore it causes me to behave in a
similarly purposive manner; they have brains of related structure and
physiology, and similarly purposive actions, therefore they have similarly
related conscious purposes. QED.”
Persuasive? Perhaps.
Cogent? Certainly not.
What is
more, various learnt behaviours, especially those that for their highest perfection
of performance rely on high degrees of coordination and speed, such as in
competitive sport, seem to rely largely on parts of the brain other than those
that support conscious behaviour.
In learning
such activity the consciously intelligent beginner relies on the cortex, and
usually performs poorly at first. After
some training, often after a necessary interval of non-participation, there may
be a discontinuity in the standard of performance. Reactions become partly automatic, with
details of the control seemingly relegated to the cerebellum; the entire
performance becomes smoother without the baggage of conscious control. And yet, some of those subjectively automatic
activities may be fairly complex, not just single, reflexive movements. Here again such modularity in the brain
suggestively resembles eusocial colonies with their relegation of particular
functions to specialist castes, or to limited stages in the maturity of the
castes.
It also
raises the question of the site of teleology in a given system. Is it the trap that has the teleology, or the
trapper? The jaw or the ant lion? The goalkeeper, or his cerebrum, or his cerebellum?
Or the
whole system?
Consider a
human practising a physical action mentally instead of physically (an effective
and important form of exercise). More
commonly he might practise partly mentally and partly physically, for example
shadow boxing in the absence of an opponent.
To a naïve observer such behaviour might not be obviously teleological
at all, because the effort is expended on no obvious objective. Compare this with someone performing much the
same mental processes with a phantom limb after amputation. Some of the distinctions raise quite
troublesome philosophical questions in terms of empirical and subjective
teleology.
Other
empirical examples that we defensibly could describe as “mindless” behaviour,
functionally seem to amount to teleology in the ontogeny of multicellular
organisms. There the growth of exquisite
organ structures is controlled by chemical and physical gradients that affect
the growth, multiplication, shape, proliferation, migration, and not least,
selective death of cells in ways that look stunningly teleological. Less precise, but none the less amazing, are
the formation of the reproductive phases, “grexes” of slime moulds, the
Myxomycetes. Their behaviour
irresistibly suggests, practically screams, teleology. By what criterion do we deny the teleology of
the system? That it is brainless? That it is stereotyped? That the individual amoebae are brainless, as
brainless as neurons? That they follow
chemical gradients?
There
might be something to
such claims, but on cross examination we find ourselves reduced to awkward
special pleading and arbitrary assumptions.
If our individual neurons are not conscious of our intentions, and it
takes a whole brain to be conscious in that sense, then until we develop a
working teleolometer, our arguments will neither embarrass the empiricists, nor
demonstrate that inorganic computers are not subjectively conscious.
Then once
more, what about the intelligence of communities? More particularly, what about the
intelligence of communities (such as the population of a species) in which some
members never meet or interact with others?
How do we define intelligence that does not rely directly on the
transmission of information? How do we
deny intelligence in systems that display apparently purposive behaviour? Do we argue that in communities with temporal
dimensions in which one member can affect the behaviour of another, but not
receive feedback, there is no control?
That would be nonsense. Not all
control has to be closed loop control, neither in engineering, nor in
physiology, nor even in communities.
Conversely,
we have the problem of arguing whether any community is a tool user or tool
creator. This is an arguable point. Consider bees’ use of wax and propolis, or
their creation of honeycombs; do those constitute tool use? Like the use of a pebble for tamping down the
cover of its tunnel by some Ammophiline wasps, all this is of course
stereotyped behaviour, but what fundamental difference does that make, apart
from the fact that the creation of new stereotyped tools, or advances on old
tools, becomes a slow, slow business?
We accept
the idea of intelligence in brains whose activities are made up of the
activities of their unintelligent cells, even though such intelligence is not
additive; when we contemplate a brain of 1010 neurons and an IQ of
102, we cannot attribute an IQ of 10-8, to each
cell. Nor do we attribute an IQ of
nearly 1012 to the entire human race as an entity. The components of an intelligent community
neither have to be individually intelligent, nor unintelligent; nor does the
individual necessarily have the same skills or the same kind of intelligence as
the compound organism or community; it all depends on the kind of relationship
that exists between unit and unit, and between unit and community.
I have no doubt that in
reality the future will be
vastly more surprising than anything I can imagine.
Now my own suspicion is that the universe is not only
queerer than we suppose, but queerer than we can suppose.
J. B. S.
Haldane
Intelligence
is not the same as information processing, as far as we can tell, but it is
hard even to imagine any system that is undebatably intelligent, but does not
include information processing as a key aspect of apparent intelligence.
One type of
information processing is vital to every decidedly intelligent creature. It has not been clearly recognised in any
colonies other than metazoan brains, or possibly in colonies composed of
metazoa with such brains: this is explicit information abstraction, particularly
abstraction in the form of generalisation from pattern recognition. Every vertebrate that recognises danger or
food from experience uses generalisation by pattern recognition; the stimuli
that the threats or prey evoke are not every time identical.
In
invertebrates, such abstraction and generalisation certainly occur, but not
always from personal experience. The
wasp that stings a spider will recognise the spider in any attitude or by its
smell, and will know how to apply the sting to the critical spot, but her
actions are innate. The experience she
draws on is the “experience” of her species, not her own. It is encoded in her genes, not in what she
has learned since her emergence from her pupa, or even her egg.
However,
the bee that learns which flowers are open where, and when, certainly does so
partly by learnt pattern recognition.
This has been demonstrated elaborately and repeatedly. She passes on the information to a clique of
workers in the hive. This too is well
known. Could one say that the hive
has now learnt to find and recognise the resource by pattern recognition?
From some
points of view one certainly could say something of the type, but one must
allow for the fact that, if so, the hive has a memory no longer than the
working life span of that clique (though not necessarily as short as the life
span of individual members of the clique): as long as members of the clique
keep finding such a resource, they will recognise its pattern if it is a good
resource. They might even remember how
to dodge the flower if it is one of a species that keeps swiping at them when
they land. But when the last members of
the clique die before the information has been passed on, any learnt skills die
with them and must be re-learnt if not to be lost permanently. In fact it is worse than that, because not
many learnt skills do get passed on in invertebrate populations.
An
interesting problem is best exemplified by the dancing language of the
bees. It is undebatable that in this
language a worker bee conveys certain variables abstractly and symbolically. The information includes distance, bearing,
and the quality and nature of the food.
Furthermore,
unless the source is particularly rich and stable, commonly only a clique of
bees, a minority of the available workers, will follow the discoverer, and go
looking for the food. If it is a
particularly fine and stable source, say the nectar flow of a large stand of
alfalfa or eucalyptus, their activity progressively recruits effectively the
entire nectar collecting population of the hive. “But it is an innate, stereotyped program,”
one might object. Certainly, but what
does that mean except that the skill is the skill of the species or the hive
rather than of the worker? It does not
mean that it is not a skill. There are
elements of the Chinese room situation here too.
At a higher
level, only humanity has demonstrated the ability to reason explicitly by
analogy. But I am always moved to
caution by reflection on a paraphrase of Orgel’s truism: “Natural selection is
cleverer than you”. Argument from
analogy has a bad name, but in the empirical universe all argument is in
essence from analogy, and some of it is functional. It remains functional as long as one does not
exceed the isomorphisms between the object of the argument, and the notation
and representation in one’s mind.
Argument from analogy comes naturally to humans — commonly such
argument is invalid, we must grant, but valid or invalid, it is a sophisticated
process. It depends on abstraction of a
putative isomorphism from one situation and application of that isomorphism to
a patently different situation. This is
not something that one commonly sees in animal behaviour, still less in
non-human community behaviour.
If we could
see such a thing in a hive or colony, we should certainly find it easier to
recognise it as intelligence of a high order, and more importantly,
intelligence that makes sense in terms of our own outlook on the world.
And he asked him, What is thy name?
And he answered, saying,
My name is Legion: for we are many.
Mark 5:9
Every
system that one might reasonably call intelligent is tomic (or is non-atomic,
if you prefer) in the sense that it has distinguishable components that so
interact as to process information for necessary activities. In fact I have failed to find any convincing
information processing system, biological or otherwise, that does not operate
by interaction between its components.
This trivial observation becomes very relevant in dealing with the
intelligence of colonies.
Conversely,
it is not clear whether, or in what sense, a complex system must necessarily
have some form of intelligence, nor that the respective roles of the components
of a clearly intelligent system will be obvious.
I confined one of these Eciton hamata under a piece of
clay
at a little distance from the line, with his [sic] head projecting.
Several ants passed it, but at last one discovered it and tried to pull it out,
but could not. It immediately set off at a great rate,
and I thought it had deserted its comrade, but it had only gone for assistance,
for in a short time about a dozen ants came hurrying up,
evidently fully informed of the circumstances of the case,
for they made directly for their imprisoned comrade, and soon set him [sic]
free.
I do not see how this action could be instinctive.
It was sympathetic help, such as man only among the higher mammalia shows.
Thomas Belt The Naturalist
in Nicaragua
It is not
cogent for an outsider to insist that if an entity does not communicate
recognisably it is not intelligent.
There are whole classes of reasons why it might not communicate
detectably; for reasons of its own it might not want to. Even if it did want to, it might not use a
channel of communication, such as a particular band of frequencies of light,
scent, or sound, that the observer might recognise. It might not use the same language as the
observer; many animals simply do not share the same signalling systems. For example, the growl of a cat simply means
nothing to most dogs, whose growls are much deeper. Conversely, a surprisingly wide range of
mammalian species, though by no means all, share innate play signals,
especially while immature. However, some
classes of signals come to invert their meaning in the course of evolution; in
context, bared teeth in humans more often mean friendliness than threat, while
in dogs and monkeys the default meaning is threat, stress, or defensive
submission.
On the
other hand some form of communication probably is fundamental to intelligence
as such, even if only because, as I propose elsewhere, it probably is not
practical to implement intelligence in any form that we understand the term,
unless the components of a system communicate with each other. In communities this aspect is visibly
essential; a crowd hardly constitutes a community in any useful sense when the
effectively meaningful communication between the elements is insufficient.
Even
conceding this need, there is a tendency to think of communication as a
transmission of passive, coded items of information, but there is more to it
than that. In intelligent systems the
communication processes and media generally include part of the data processing
mechanism. In the interface between units,
information may be emphasised, attenuated, inverted, combined constructively or
destructively, even supplemented or newly generated by what, in human
conventions and techniques are called clandestine channels. An example would be when a bridge player conveys
information to his partner by tone of voice or deliberately timed hesitations
in his bids. We see all these and more
in the interaction between neuron and neuron, organ and organ, person and
person, and most definitely between members of colonies. Signals, whether visual, auditory, chemical,
or mechanical, go far beyond simple single-channel one-way transmission.
Just for
one example, the same signal emitted by an organism can carry information
concerning distance or time as well as identification or stimulus to
activity. Such extra information is not
intrinsic to the signal as emitted, but is the product of modification, of
processing or dilution or decay, that the emitted signal undergoes once it has
been released. Examples include the
state of decay of pheromones or excretions, and state of construction of
physical structures.
Signal
context too is crucial. A signal at low
level might be attractive, at a higher level or repeatedly increasing in
intensity, it could be a threat or even an attack. A signal received in one physiological state
could stimulate aggregation, while in a different state it could cause
dispersal.
There are
many elaborate examples of signal processing between transmission and
reception, whether within a body, between individuals, or within
communities. Without such information
processing en passant as it were, it is hard to imagine how insect
communities in particular, but any eusocial communities, could attain anything
like their elaborate and apparently intelligent structures and control. It also gives one pause to imagine the
complexity of the intra-community control.
It leaves
one too with an interesting speculation: the signalling structures within say,
the mammalian or indeed the insect body, are of stunning complexity and are
nowhere near being properly analysed as yet.
And yet within a community we have enormously complex communication
structures too. How confidently can we
say whether either of those two is greater than the other? How meaningfully can we say that the
complexity of the community is the sum of the internal and external signalling
of its members?
Not only do
individuals signal to external members of their own species or community, but
whole communities, social or not, signal or suppress signals, collectively or
individually, consciously or not. Consider shoals of some fishes, that form
shapes suggesting that they are a single large organism, or the shimmering
behaviour of Apis dorsata colonies, which might function as a flash
threat, or as disguising the colony as a large or nondescript object.
One way or
the other, communication in several senses at least, could be argued to be a
constraint that any claim of intelligence must meet, if only because so much of
the function of communication is as a component or amplifier of
intelligence.
. . . Say, from
whence
You owe this strange intelligence?
William Shakespeare -- Macbeth
Except for
trivial examples for purposes of illustration, components of an intelligent system
necessarily include subsystems of various levels of complexity.
You
never see animals going through the absurd and often horrible fooleries of
magic and religions.
Only man behaves with such gratuitous folly.
It is the price he has to pay for being intelligent but not, as yet,
intelligent enough.
Aldous Huxley
For such
systems to function, requires a high degree of intercommunication between
modules. The intercommunication may be
one-way, two-way, one-to-many, many-to-many or more than one of those modes
simultaneously.
The
intercommunication may be broadcast, as with pheromones or alarm signals, or
point-to point as in nerves or tactile communication. It may be discrete (in that respect being in
some sense digital) or analogue. It
might function by language, whether human and largely conscious, learnt and
formally sophisticated, or rudimentary, instinctive, and stereotyped as in apes
and some birds and insects.
Such
intercommunication need not be rigidly hierarchical in terms of the relative
size of the components. In fact, the
relationship is largely a network of modules rather than in the form of a
loop-free hierarchical tree. However,
hierarchical or not, the various modules generally are “information hiding” as
in “good practice” in computer programming: that is to say that components
inside one module typically have no insight into the processing tasks or data
of different modules.
Conversely,
components in general might not even be aware of their own function in
terms of the effect they have on other levels or parts of the “organism”. They might not be aware of any organism or
intelligent entity at all. Here
Hofstadter’s “Ant Hillary” in “Goedel, Escher, Bach” springs to mind.
This “good
practice” in biological modularity, whether in brains or in colonies, does not
so much reflect an explicit design decision, as the fact that the typical
module is neither equipped to inspect, understand, or even care about the
insides of other modules. In fact,
one-to-many and many-to-many communication tends to have wide ranges of
“unintended” effects, and many of the attributes of pheromonal or hormonal
systems are adapted to minimise the harm done by those unintended effects.
Such
“unintended effects” are denounced as abominable practice in current software
engineering. Nature however, has not
read the training manuals. Natural
selection works on all the effects of a genotype, a “design”, and anything
really inviable goes extinct fairly quickly, whether it is at the level of a
gene, an individual, a community, or an ecosystem. Whatever is not too drastically inviable
eventually gets sorted out in a stable, established ecosystem, but in a disturbed
ecosystem or where there are new, potential niches, those unintended effects
may be the basis for a new, drastic evolutionary development within one species
or more, what some people nowadays call a punctuation event. It usually takes a while for selection to
adapt all the attributes to establish a new ecotype, but evolutionarily the
process is rapid. A lot of selection can
happen in a few thousand years, sometimes even in a few decades.
When the
controls of a system that relies on internal communications are insufficient or
get interfered with, the results may be disastrous. In humans we see this in the complications
following on hormonal treatments or abuse.
Among eusocial insects we see an analogous example of disaster when a
hive of the much maligned “African Killer Bee” (Apis mellifera dorsata)
is invaded by fertile workers of the Cape
Bee (Apis mellifera
capensis).
The
principle is as follows. Unlike most
Hymenopteran worker castes, unmated Cape bee
workers can lay fertile female eggs.
Their own colonial control communications have adapted to the situation,
but other subspecies, including dorsata, have not. As a rule, the disturbances in the hive
pheromones wreck the dorsata colony till sooner or later only Cape bees are left.
Without a queen, their hive balance is unstable and the colony is likely
to perish entirely, though it may sometimes muddle on till a viable queen takes
over and establishes a viable colony.
As we have
seen, intercommunication between components is not just there for simplistic
control. The signals themselves are the
sites where much of some kinds of information processing takes place. Probably a great deal of functional
intelligence depends on that principle.
There are many examples. Synapses
may be excitatory or inhibitory.
Pheromones may vary
in their stability or mobility, so that the signal permits the appropriate
reaction to old or distant signals.
Receptors vary in their reactions to persistent signals, many actively
destroy the signalling molecules that they receive. They thereby permit habituation or clear the
medium for successive signals. In fact,
whole classes of deadly poisons work by inhibiting the breakdown of the
compounds that carry synaptic signals.
...it has yet to be proved that intelligence
has real survival value ...
We have, as yet, no
definite proof that too much brain, like too much armor,
is not one of those unfortunate evolutionary accidents
that lead to the annihilation of its possessors.
Arthur C. Clarke
If the
capacity for learning is essential to intelligence, it raises difficulties in
formulating the concept of entity. A
given organism or indeed, a colony, might or might not adapt its behaviour in
the light of experience, but a species might, through natural selection, adapt
to circumstances. Such a species might
do so in ways logically difficult to distinguish from learning. If the species is defined as a population
sharing a common fertilisation pool and having dimensions in time and location,
is it an intelligent entity? Is the
learning capacity of any component, either more important or less important
than that of the total system? Important
or not, it certainly is not generally
equivalent. “Learning” by a species is
not the same as learning by a particular colony, which in turn differs from
learning by an individual, and differs even more so from learning by a neuron
in an individual.
At the same
time, these entities are deeply different in their respective natures as
entities, and perhaps even more different in the respective natures of their
intelligence. What “learning” means in
one differs from what it means in all the others.
Temporal
factors also apply. Many activities of
long-lived communities or species are hardly perceptible to the faster, but not
necessarily more complex, minds of their members or observers. This would make a difference even if the
learning process were logically the same.
In human
history it frequently happens that quick-thinking individuals first exploit
community weaknesses or stock market patterns, then ossify their thinking. Typically they fail to recognise that their
views were far too simplistic to represent the complexity of the behaviour of
the community in general. Sooner or
later new developments bury them, either within their lifetimes, or within a
few generations. It is not clear how far
such contrasts indicate a fundamental difference between the natures of the
intelligence of individuals and communities respectively. One could argue that such examples suggest
differences in speeds as well as types of learning, and that the communities
had more simultaneous modes of learning than their individual members.
Conversely,
in many communities the individuals can learn far faster than the community
can. It is a long-standing source of
ridicule among Young Turks, that as Max Planck wrote: “An important scientific
innovation rarely makes its way by gradually winning over and converting its
opponents: it rarely happens that Saul becomes Paul. What does happen is that
its opponents gradually die out, and that the growing generation is
familiarised with the ideas from the beginning.” Insofar as the assertion has merit, this is
by no means true only of scientific disciplines, in fact as a rule far less
true of scientific disciplines than of prejudices in everyday life. Some scientists certainly are
unscientifically hidebound all their mature lives, but they should at least in
principle have a discipline according to which they may proceed to explore or
reject new ideas. Among laity who have
not developed such a discipline, novelties are likely either to become fads or
unjustly get rejected.
Inappropriate
rejection of novelty by the way, is not only true of old fogies; young fogies
often are much, much worse.
Anyway
neither age nor newness is any guarantee of truth. Facile new superstitions are no likelier to
be true than centuries-old superstitions.
However, traditionally the very stability of communal beliefs is what
preserves wisdom in hard times. Old
beliefs may have survived simply because they have been seen to work. The general problem of what to learn and what
to unlearn is no better conditioned than the general problem of what defines
evolutionary progress or retrogression, and for analogous reasons. In vertebrate individuals, including humans,
the programmed reduction of learning capacity with maturation is an important
and universal process.
In insect
societies in particular, learning is a cumbersome and vague process. There might be a minor reaction to experience
by individuals, a “change of mood” because of continual stimulation in the
community, or genetic adaptation by the species under the influence of
selection.
.
. .let not thy left hand know what thy right hand doeth. . .
Coupling of
components may be intimate to any degree and in several dimensions. It is fundamental to any form of group or
colonial intelligence or within a body.
Consider some degrees and aspects of the coupling of such components in
nature:
When people are free to do as they please,
they usually imitate each other.
Eric Hoffer
Coupling between components of a loosely
coupled system may be facultative, as when either conspecific or unrelated
organisms combine in hunting, guarding, nesting, grooming etc (such a community
exhibits intelligence in some form of apparently conscious interaction, not
necessarily consciously or consistently aimed at the benefit of each other or
the community as a whole.) Conversely,
though one does find adventitious, episodic examples, the behaviour commonly is
consistent and continuously adaptive. On
what grounds do we deny intelligence in such systems? What kinds of intelligence do we deny
them?
A Brother may not be a Friend, but a Friend will always be a Brother.
Examples
include hedgehog fleas, that appear to be necessary to the health of the
hedgehogs, scavenger interactions where scavengers remove matter that would be
unhealthy for the creatures that produced the litter, cleaner wrasse,
protective relationships between ants and plants or plant bugs or both.
Such a
community of multiple species exhibits apparent intelligence in some form of
obligate interaction. Often the
participants display no apparent teleological, mutually beneficial
intentions. However, their relationship
does so, even though the teleology is not apparently conscious. Any disturbance of the short-term balance
between the benefits of participants is likely to be fatal in the long term, in
the absence of “enlightened teleology”.
For
instance, oxpeckers take no interest in the well-being of their hosts; for them
a tick is a convenient means of rendering giraffe or buffalo edible. They will greedily cluster round a wound and
do their best to eat the host alive, bypassing the ticks. Excessive demands on one partner might kill
it, in some cases leading to the death of the greedy partner as well. One could regard cancers as examples. Conversely, modest, consistent demands are
likelier to lead to adaptation that might develop into mutualism.
Those friends thou hast,
and their adoption tried,
Grapple them to thy soul with hoops of steel;
But do not dull thy palm with entertainment
Of each new-hatch'd, unfledged comrade.
William Shakespeare
More
complex relationships include vector relationships in worms, protozoa,
Legionella, typhus. The vector
interaction comprises an intelligent system, in many examples harmful to some
parties, in some cases beneficial to both.
Metabolic
explanations are both proximate and ultimate,
in the same way genetic explanations are.
Endosymbioses, therefore, point evolutionary biology toward
an important dimension of evolutionary explanation.
Maureen A. O’Malley
Examples
include ruminant gut flora, and gut symbionts in cockroaches and termites (several
levels and dimensions of endosymbiosis in the same termite or cow!) Particularly stunning relationships include
some types of mycorrhiza. Another
extravagantly baroque example is “the
medusa & the snail”, as related by Lewis Thomas. Such communities of more than one unrelated,
but physiologically interdependent species exhibit intelligence in some
form.
Examples of
endosymbiotic relationships include eukaryotic cells. Various species contain
plastids, hydrogenosomes, and mitochondria.
Such a cell itself comprises a community of genetically diverse
organisms interacting as an intelligent structure. Such communities challenge our concept of
“entity” to the breaking point.
Hydrogenosomes in particular are almost the inverse of the medusa-snail
relationship, in that the DNA relocation to the nucleus has left the
endosymbiont sometimes without any DNA of its own.
When all night long a chap remains
On sentry-go, to chase monotony
He exercises of his brains,
That is, assuming that he’s got any.
Though never nurtured in the lap
Of luxury, yet I admonish you,
I am an intellectual chap,
And think of things that would astonish you.
W S Gilbert. Iolanthe
Non-eusocial
communities are sometimes termed “sociable”.
They differ from eusocial colonies mainly in the following particulars:
Members may
not be family members with overlapping generations, such as sibs;
There may
be no co-operation in care of young; and
There is no
question of castes.
Examples
include certain seabirds, colonial caterpillars, and wasps that nest in close
proximity, often in regular arrays. Sociability has many advantages. Arrays of nests may deter predators or
decrease individual exposure. Combined
labour can achieve things beyond what individuals could; for instance sociable
weaver birds construct huge nests that no one pair could build alone.
In such a
sociable community each individual selfishly seeks its individual comfort
level, but the community profits and adapts accordingly, and on average, so do
the individuals. There is a gradation of
sophistication of such nesting patterns, from mere mobs to spectacularly
regular arrays. In a still more radical
adaptation to predators and cold, emperor penguins have dropped their mutual
close range repulsion in favour of huddling.
In our
context, this last reversed adaptation is particularly suggestive. At various levels of adaptation, the nature
of the association changes, or in some cases even reverses. Such change in the species is reminiscent of
the process of flexible learning in individuals. It is just that in communities, particularly
a community comprising a species, the “learning process” is not trivially
visible and may take many generations of selective adaptation. However, a sudden change in the selection
applied to the population may achieve changes within a few generations. Examples include changes in the flight
behaviour of game birds under pressure of hunting or road traffic.
When in that House M.P.s divide
If they’ve a brain and cerebellum, too,
They’ve got to leave that brain outside,
And vote just as their leaders tell ‘em to.
But then the prospect of a lot
Of dull M.P.s in close proximity,
All thinking for themselves, is what
No man can face with equanimity.
W. S. Gilbert Iolanthe
Familial
communities involve parent-young interaction.
There is some degree of care for offspring, the simplest form of which
is selective oviposition, grading to elaborate parental care. At higher levels there may be co-operation
between siblings or between offspring and parents, such as when scrub jays,
foxes or carpenter bees share the burden
of raising genetically related young.
The
components of the intelligently altruistic system are derived from those innate
behavioural patterns that are necessary in non-social care for offspring. I discuss examples in the section on
inhibition of aggressive behaviour.
Familial altruistic behaviour grades into true social communities, for
example, we get solitary mole rats such as the Cape
mole rat, Georychus capensis, sibling assistance in family parties in
the Hottentot mole rat, Cryptomys hottentottus, and then eusocial
species of naked mole rats.
Such
familial communities appear to have been a stage in the evolutionary
development of all the eusocial species we know. It is not good to be simplistic about it
though. There are examples, such as some
of the paper wasps, where the original family party is not a single queen and
mate, but a small group of cooperating queens.
The place of the father in the modern
suburban family is a very small one –
particularly if he plays golf, which he usually does.
Bertrand Russell
Examples
include naked mole rats, sponge-living shrimps, termites, ants, wasps and bees,
where there are non-breeding castes that are pheromonally and behaviourally
determined. Quite minor interference
with such pheromonal controls can be fatal to the colony, such as the effect
when Cape bees (Apis mellifera capensis)
invade hives of African “killer” bees (Apis mellifera dorsata).
In
order to be an immaculate member of a flock of sheep,
one must above all be a sheep oneself.
Albert Einstein
There are
many examples in nature, of effectively sedentary organisms that live in close
proximity, sometimes physically connected and reproducing vegetatively; many
such creatures behave in orchestrated fashion in their feeding, defence or
mating, even though their physiology and organisation is neither that of a body
nor of a eusocial colony. Examples
include beds of mussels, bryozoans, barnacles, and corals. Typically they all snap shut if a single
member is alarmed, open together for feeding in synchrony, and release their
gametes or young in synchrony. Their co-ordinated movements resemble those of
shoals of fish or flights of birds.
Their growth patterns resemble the growth of organs in animals or
branches or roots in plants.
Relationships
between cells or organs within the body’s ontogeny, including cell death where
it is necessary, irresistibly recall colonial interrelationships. Consider tumours as counter-examples. Where cells refuse to die on cue, one gets
malformation and perhaps death of the colony.
Note how the nature of the genetic relationship affects the employment
of cell lines in sacrificial roles.
Since the somatic cells in most organisms are all genetic clones, it is
well worth a cell’s while to sacrifice itself if that improves the survival of
the organism. Consider for example cells
in slime moulds or in embryonic tissues of multicellular organisms, and compare
them with competing or cooperating team members in a colony.
All such
co-ordinated behaviour is closely related to the intelligence of communities,
even though the range of the intelligence in the individual members varies from
practically zero in the amoebae to birds with some of the highest intelligence
in the animal kingdom, and yet the apparently intelligent behaviour of the
various communities is strikingly similar.
Crowd traffic behaviour is in many ways similar in everything from ants
to humans. The reasons seem to have a
lot to do with mechanical constraints, but that does not reduce their
significance for the concept of community intelligence.
. . . An expedient was therefore offered, that since
words are only names
for things, it would be more convenient for all men to carry about
them
such things as were necessary to express the particular business they
are to discourse on. . . I have often
beheld two of those sages . . . who,
when they met in the streets, would lay down their loads, open their
sacks, and hold conversation for an hour together. . .
Jonathan Swift A Voyage to Laputa
Pending
anyone demonstrating the contrary in some way I cannot imagine, I regard
information processing as one of the essential components of intelligence. In information processing, communication
necessarily entails the encoding and transmission of information to be
processed. In turn, this entails
abstraction; some device must be put into a state that reflects at least the
relevant parameters of the object that is the subject of the information. It also must not present too much information
besides. In other words, the signal to
noise ratio must be manageable, or the abstraction is likely to fail.
Consider
for example, the conversations of Swift’s Laputan sages, in which they displayed things instead of using words.
This would not have worked very well because of a specific reason:
semantic ambiguity. The abstraction
would have been omitted and therefore the meaning would be ambiguous.
Depending
on how one looks at the act, to exhibit a knife in lieu of a word or a phrase could
amount to presenting anything from several dozen bits of information, to ten to
the several dozen bits. In “holding
conversation for an hour together”, the act of showing say, a knife, could mean
“a knife”, “this knife”, “knives in general”, “iron”, “a mass of 250 grams”,
“this colour”, or any of many other possibilities. Further information must be added to indicate
which of all the possible attributes of the thing
is being presented for consideration, in other words, to filter out some of the
critical noise.
No doubt in
practice sophisticated savants would add the necessary supplementary
information by employing high-flown conventions and techniques of the charade,
but in doing so, they would betray the ideal of displaying “such things as were
necessary to express the particular business they are to discourse on” instead
of using words; linguistically and
semiotically speaking, formalised gestures and grimaces are nothing less than
words. The medium also presents other
difficulties, such as how to express relevant non-material concepts such as
relationships or verbs, but those are minor compared to the abstraction
problem.
So it is
with say, the language of the bees.
Unlike the Laputan sages, bees specifically use abstractions, even if
their abstractions are not conscious.
Whatever consciousness a worker bee might have, if it has any at all, it
does not react to the food that the worker has recovered as “this nectar”, “this pollen”. It reacts
instead to the abstractions: “such
nectar, in that direction, so far away, at that time, to be identified by such
a smell”. Also, within the bee’s
information processing mechanisms, as opposed to its foraging mechanisms, it is
not pollen that gets passed on, but nerve impulses and synaptic signals that are
as abstract as any signal passed by a human or an artificial signalling
apparatus.
The very
fact that intelligent entities (apparently inescapably) are modular, implies
communication, at least between the modules.
Otherwise the modules cannot non-trivially constitute an entity.
Some kinds
of communication, as we have seen, are so simple that they barely qualify as
language in any non-trivial sense. On
the other hand, for non-trivial intelligence we need to convey concepts and
data in non-trivial notation. A few
whiffs of hormone molecules really cannot suffice as non-trivial examples. In the brains of the most spectacularly intelligent mammals and birds, modules in the
brain must convey complex, sometimes intrinsically abstract, concepts for internal
manipulation. “Think when we talk of
horses, that you see them printing
their proud hoofs i' the receiving earth. . .”
Thinking thus, it is not little horses that we pass about the affected
parts of our brains, but impulses and synaptic states. Merely explicitly thinking about anything
implies some level of ability for communication, abstract communication within
the brain. In the light of information
processing practicalities, this implies at least the first requirement for
abstractly communicating the same information outside the brain.
Now this
raises practical and philosophical difficulties in imagining recognisable
advanced intelligence of species or social communities. If all the extant specimens of a species are
subjected to the same selection, then the species will respond by adapting as a
unit. Is that intelligence? If it is, then is it intelligent for a
snowflake to grow symmetrically, even though its various points are not
communicating with each other? To claim
this is not persuasive, but the denial is not watertight. Unlike non-contiguous parts of the snowflake,
various parts of the species do in fact intercommunicate and adjust their
response to selection by the redistribution of DNA. Such responses could be seen either as very abstract
indeed, or barely abstract at all.
Obviously
the problem is tricky. There are several
aspects that may be regarded in various ways, and some of the parameters are
not well enough characterised for us to measure.
Within
large mammalian brains, such as in naked mole rat workers, some such processes
of communication between modules no doubt occur, and, at a lower level,
possibly also occur within the smaller and less elaborate brains of
termites. There always is communication
within colonies, even if it is only recognition of nest mates, alarm
pheromones, or recruitment to activity patterns. However, within the beehive is where we see
the most advanced and explicit approach to language. It is impressive in its own terms, being
specific, semantically abstract, and fairly parsimonious.
None the
less it is not comparable to communication among humans, nor possibly to
communication among some other mammals or birds. For one thing, it is not a general means of
communication, but highly constrained to particular subject matter. It would fare poorly in a Turing test.
If we are
to find a truly advanced community intelligence, it must take some radically
different form, possibly at some higher, more abstract level.
And, in
spite of our human intelligence, how much more intelligent are human
communities than insect communities?
Certainly in some senses our communities can converse with apparently
indefinite flexibility. If we judge by
the more sensible members of society, one would like to think our communities
are very intelligent, but it is sobering to remember that majorities of the
intelligent public concerned, supported Hitler, Stalin and some more recent
political bodies that have highly invalid arguments in support of their
policies. To the extent that we as
humans are influenced by the pathological, the influence of the intelligent in
anything but the persistent improvements in technology and infrastructure, is
debatable.
One part of
the problem with our community intelligence as opposed to our individual
intelligence, is that the complexity is intrinsically greater than in a unit
brain, and it lacks the equivalent of many of the controls in healthy unit
brains. Managing such complexity is
currently neither practical, nor even foreseeable. This is presumably why communities so often
behave apparently more stupidly than most of their members, even when leaders
are well-meaning and intelligent. They
are necessarily chaotic systems and we have not yet learnt how to damp their
oscillations and excursions.
How does
one measure the relative intelligence of a system that involves stereotyped
behaviour of its components, as opposed to systems that incorporate putatively
subjectively conscious components? It is
a messy business to compare the intelligence of a termite nest with that of a
human community. Bear in mind that it
does not follow that the intelligence of the community exceeds that of each, or
even any, of its members.
Possibly it
is as well for animal colonies, that they have not yet achieved the internal
complexity and communication characteristic of human communities. As a rule evolutionary selection work more
effectively on innate, stereotyped behaviour such as in insect colonies, than
on the disorganised innate impulses and weakly-stereotyped rationalisations of
human colonies. However, it does not
follow that human communities are exempt from such selection. Lorenz speculated on personality traits such
as degrees of aggression, being selected for in particular communities. Possibly his views were simplistic, but like
many of his speculations, they were intellectually fertile.
The
language of the honeybees is a useful model in several ways, though it is too
complex for detailed discussion. The
nature of its abstraction is interesting at several levels.
Warning:
here follows a Just-So Story informal enough to drive any positivist, let alone
any serious ethologist, to distraction, but it will have to serve as convenient
illustration.
The
successful foraging bee returns from the field reeking of the food source,
tired in proportion to the wind and the length of the trip. The direction of the sun is her most
important landmark and clue to her bearings over a distance of anything from a
few metres to perhaps a few kilometres. She
cannot speak directly, but what she can do is to convey her condition to her
hive-mates. They in turn must accept her
(necessarily) abstract message and encode it for comparison with their own
internal condition.
First she
must pass the guards at the hive entrance.
They are there for several reasons, but a major one is to protect the
hive from strange bees; in spite of their many virtues, bees have a practical
outlook on honey: wherever you find it, grab it! Accordingly, if they find it in any hive that
lets them in, they immediately load up all they can, and take it home. For the guards charity begins at home, so not
many strangers get in.
Mind you,
the guards are not really perfect. I
have already mentioned the Cape bees invading
the hives of “killer” bees. The reason
that they can do so is the slightly leaky protection of the guard bees.
Now, it
sometimes happens that a bee eats some fermented food, rich in ethanol, and the
effect on bees is much like that of humans.
Any crapulent bee that finds its way home tries to enter, but it is
likely smelling of the wrong mix of scents, so the guards will first shove her
out, or if she persists, may bite or even sting her. So we can say that her first signal on coming
home amounts to: “I smell right and behave right; that is my password.”
A frequent
communication on the way in is for the returning forager to make sure that
foragers on the way out get a good sniff of the flowers they have been
collecting from. She even might feed
them samples. If it is a kind of flower
that the others know where to find locally, that might be sufficient to send
them off to the right source.
Investigators have used sneaky tricks to fool bees into visiting the
wrong flowers to prove that this scent and feeding trick is a non-trivial
message. It also proves that the bees
knew where to find the wrong flowers; they must have memorised where to find
many of the flowers in the neighbourhood of the hive. This too is a non-trivial form of
intelligence. In fact, when bees are to
be used for pollination of crops, it sometimes pays to feed the bees with syrup
scented with the appropriate flower.
Then when the flowers open, the bees are ready and willing.
Often the
bees also mark a rich nectar source or a good source of water with their own
scent glands to signal nest mates that they have reached the right place. This is especially interesting because less
extremely adapted eusocial bees that do not have the dancing language, use scent
marked trails to guide nest mates to food sources.
Honeybees
also use their own scent marking to indicate that the food source is very rich,
rather than just telling where it is.
When they do that, they tend to attract a lot more followers than just
by doing their dance. It is all part of
the signal that adjusts the number of foragers to match the richness of the
source. If every bee that found enough
to dance about would recruit the whole hive, that would be worse than useless,
it would be disastrously wasteful of effort.
Instead she visits a food source several times before doing her scent
marking. That way she is sure that she
won’t be sending her sisters after a temporary attraction. It also is a good mechanism for assessing
when to release the scent, instead of relying on the equivalent of a conscious
estimate.
When the
forager gets into the dark of the hive, she cleans up, stores what she has
brought, or passes it onto young workers who are not yet foragers. If the source is a good one, plentiful and
rich, and she has visited it enough times to move her to communicate the fact,
the bee will start to dance. She does
this in the dark, on a vertical surface, usually a comb. It is a complicated business with several
variables. First there is the type of
dance. If the food is just outside, say
up to 25 metres away, she will do a round dance, roughly in a circle one way,
then turning round and doing a circle the other way. She may keep this up for a minute or so, or
just a few seconds, with or without repeated encores. By and large a specially good source moves
her to longer, livelier dancing. The
only bees that follow her around are other foragers. Young nurse bees and other non-foragers keep
away, as well as one can in a crowded hive.
The dance carries no message for them!
Between the
scent and the performance, and the number of other bees that go foraging and
repeating the message, each food source generally gets the most suitable number
of foragers. If there is a famine,
practically any reasonable source gets recruits, but if there is a honey flow
on, only the best sources set the bees dancing.
At slightly
greater distances to the food source, the circle that the bee dances becomes
sloppier and no longer closes. In fact
it begins to take on a sort of distorted figure eight shape. If the distance is much longer, say over 100
metres away, the dance is a proper figure eight with a good long middle stroke,
not a wasp waist.
At such
distances the foragers are beginning to take note of the direction instead of
just searching round the hive. Here is
one of the most remarkable aspects of the dance. So far we have had fairly modest levels of
abstraction and not much indication of anything but lot or little food
nearby. But for directing bees to
distant sources we really need some good indication of direction and
distance. Otherwise we have not told the
hive much more than that there really is food somewhere. . .
The
directional information is supplied by the direction of the cross-stroke of the
figure eight. The way it is indicated is
so unexpected and so hard to detect that von Frisch richly deserved his Nobel
prize for spotting and elucidating it.
What it does is to show what the bearing is relative to the bearing of
the sun away from the vertical. So at
local noon in the high latitudes of the southern hemisphere, the sun would be
due north.
The problem
remains that the bee is dancing in the dark of the hive. How is it to represent the direction of the
sun? This is one amazing thing; the
direction of the sun gets abstracted and represented by an alternative
stimulus. Usually the alternative
stimulus is gravity. It is as remarkable
as humans graphically indicating a horizontal direction on a map hanging
vertically on a wall.
Much as we
represent with a streak of ink on a vertical map, a direction to walk
horizontally or up hill and down dale on the ground, the bee represents the
journey with the middle stroke of the figure eight dance. A dance with the middle stroke of the eight
going straight up, means: “Fly towards the sun”. If the stroke is at 45 degrees to the right
of straight down, it means: “Fly away from the sun at an angle of 45 degree
toward the right. In other words, if the
sun is due north, fly at a bearing of North 135 degrees.”
There are
various exciting exceptions to some of the conventions, but the most important
is what happens when the sun is obscured by heavy cloud. Then if there is any sizeable patch of clear
sky at all, the bees deduce the position from the pattern of the polarisation
of the light from the visible sky. One
amazing thing is their accuracy. They
can estimate the azimuth of the sun when it is barely 3 degrees from the
vertical. Given a fair sighting, their
dances can direct their sisters to a mean accuracy of about one degree. Even given a compass, an inexperienced human
might do a lot worse.
The
accuracy of the direction is all the more amazing when one thinks how the
distance is conveyed. The fellow workers
crowd round and feel her movements with their antennae. The bee does not do the middle stroke as
smoothly as possible, but waggles her abdomen from side to side as she
runs. Now comes another amazing
thing. When the bee is on the middle stroke
of a serious dance, she begins to buzz with her wings, and the length of the
buzzing run tells her sisters how much flying to do to get to the food. Roughly speaking one second of buzzing run
means one kilometre of flight. Don’t
take the exact figures too seriously; they vary. The graph of distance as a function of the
length of the run is not perfectly straight, and the bee does not dance just
once. Usually, especially for rich
sources of food, she goes over the course repeatedly and her sisters track her
time after time. They base their
conclusions roughly on the mean of the performances. The message they get is remarkably accurate,
as I think you will agree, considering how poorly humans typically estimate and
convey such information without special instruments.
Another
very interesting fact is that novice foragers take a while, typically a day or
so, to learn to interpret the dance properly.
In fact, in the dark of the hive novices have difficulty following the
dancer with their antennae. This is
another dramatic example of how the innate mechanism of particularly complex
behaviours often needs practice before it works at its best. You might draw comparisons with sheepdogs,
whose handlers speak of “teaching” the dogs to use the innate skills of the
breeds, not “training” them. Another
analogy would be the learning of language by human children; large components
of their learning seem to be innate, a mapping of the mother tongue onto an
inborn universal human language.
There are
many amazing things about the communication system of the bees. One is that the bee has an internal clock
that allows for the movement of the sun.
She changes her direction of dancing even over a period of several
hours’ delay when indicating which way to go.
Also, her information concerning distance is not the distance as a
surveyor would measure it, but a measure of the work it takes to get
there. This includes allowance for up
hill, down dale, crosswinds and the like.
But there is more.
However
convoluted the course that the bee flew in finding the food, she flies back in
about as straight a course as possible and on her subsequent trips, she also
flies efficiently. So do her followers
in their turn when they respond to her dancing.
At first it
is hard to imagine how such an elaborate and abstract a language could have
evolved. Fortunately we get very
suggestive clues to the history if we compare the honeybee language with the
languages of some other species of bees.
Think again
about that buzzing run. The bee in
effect runs in the direction of the food. It is not the actual direction, but
one that is isomorphic to the direction.
After all, remember that in the convention of the honeybee dance, the
direction of gravity substitutes for the direction of the sun. Honeybees do occasionally dance on horizontal
surfaces in the sun, and some related species of bees always do so; and when
that happens, they do the buzzing run in the actual direction of the food.
Now, as it
happens, some other species actually recruit followers with smells and other
clues, but then fly off and lead their sisters in person. Some species leave scent markings at various
points along the way. In the light of
this, it is easy to imagine a honeybee ancestor doing just that. Then, when the navigation got good enough,
the scent marking could be reserved for just the target area. Then maybe the bee got good enough in passing
on the message, not to have to fly all the way.
Finally the pathfinder flight got stylised into just the length of that
buzzing run that forms the middle stroke of the figure eight.
Proof? Of course not. The foregoing discussion is an exercise in
the assessment of a reasonable conjecture for a course of adaptation in
information handling. It is a course
that involves no teleological leaps of ability or drastic genetic
discontinuities on the way. For the
purposes of this informal essay, that is sufficient.
Or is
it? Where did the bees in the first
place get the ability to convey to fellow bees that there was something they should be doing? There are two aspects to this. Firstly, sociable (nothing like eusocial)
organisms commonly are adapted to feed together. Except for creatures that live actually in
their source of food, I cannot think of a eusocial animal that does not
cooperate in finding food and bringing it back to the nest. Those exceptions are the likes of the shrimps
that feed on the sponges inside which they live, or thrips and aphids that live
in the hollow plant galls that they themselves caused to grow. The majority of those species that cooperate
in foraging, lead their nest mates to food, if only by the smells they release
in feeding or in storing the material they have collected. Food and protection are the most obvious
primary benefits of sociality.
Secondly,
decades ago studies of non-social species such as various kinds of flies,
revealed some suggestive patterns. Some
of their behaviour patterns in feeding are suspiciously reminiscent of some of
the bee movements. In particular, they
are delayed responses to food, say after a morsel of food has been removed. If all the ancestral bees did was to be
attracted to bees that had just been fed, that could well be the original basis
for the whole development.
There are
other bee dances too, such as a buzzing run to indicate the direction of a
desirable residence for a swam to move to, but many elements of such honeybee
language have not been decoded yet. Some
of the putatively communicative behaviour might even have no operative
significance; they may be vestigial movements prompted by assorted
stimuli. It is not a simple matter to
decide such questions before one manages to identify the semiotic significance
of a language.
Loose systems last longer and function
better.
John Gall Systemantics
Another
thing to get clear is that the behaviour patterns that support foraging and
communication in eusocial species are not in general automatic and
deterministic, even though they incorporate a good deal of stereotyped
behaviour. In many kinds of eusocial
colonies there typically will be several foraging recruitment efforts going on
at any one time in a good season. Which
invitation will succeed in recruiting a given new forager, will depend on that
forager’s own apparent whim and the intensity of the performance.
In turn, in
bees for example, the quality of the source, the intensity of the competition
at the site, the exertion of the flight, all affect the recruiter’s dance one
way or the other, so there is are continual changes in the numbers of bees that
visit each particular feeding site.
This sort
of chaotic democratic process is characteristic of practically all co-operative
eusocial endeavours. In fact it is
fundamental to eusocial control. Note
that such non-deterministic classes of algorithms need not at all necessarily
be inferior or unsophisticated. Even in
human programming random resolution of conflicting demands for resources has
proved to be quite efficient in most cases, and above all, has a very reliable worst-case
performance. In eusocial colonies the
resolution is not truly random, but is more like what we might call the P. T.
Barnum approach. Whatever keeps catching
the attention of colony members is likeliest to recruit them, and if they are
disappointed, they are likely to follow rival recommendations. Such classes of resolution are particularly
suited to the needs of eusocial colonies because they are simple, flexible, and
degrade gracefully except when a pathological agent exploits them.
This is a
prime example of why I speculated whether there are multiple types of
intelligence. Barnum influencing of the
community looks as convincingly like a component of intelligent information
handling as any synaptic or hormonal communication.
Some of the
most impressive examples of Barnum persuasion occur during swarming in honeybee
colonies. Swarming is a variable and
complex process and the following remarks are nothing like a full or coherent
discussion, touching only on a few illustrative points. Typically, early factors promoting swarming
include crowding, overfull stores of food, and the construction of queen
cells. Such stimuli lead to
discontinuities in colony behaviour that provoke workers to go scouting for
nesting sites and prepare the queen to stop laying and join the swarm.
On leaving
the hive, the swarm typically assembles at some intermediate point. It may stay there for some time, typically
several hours. Sometimes it might take
several days before the bees agree on where to go. In deciding whether such an assembled swarm should
move to one new home or another, the scout bees strut their stuff on the
surface of the swarm itself, and as one group or another of the followers moves
to prefer one option or the other, the swarm may vacillate more or less
persistently. If no clear choice
emerges, the swarm might even split, each going to a separate new home, one
swarm necessarily being queenless. Such
a splinter swarm lacks the queen pheromones, so they soon have to unite again
with the other.
Usually
however, the end result of such plebiscites is a working solution, one way or
another. When ants move their nests,
there will generally be whole trails carrying larvae, eggs and pupae hither and
yon and, depending on the species, possibly even queens. As one side or another prevails, individual
ants might change their allegiance several times, doing and undoing. This is more or less how driver ants, that
are almost always on the move, decide where to assemble in their protective
bundles or which trails to follow. When
nest building, eusocial insects might tear down each other’s structures to use
the wax, paper, mud, or whatever it might be, for their own efforts.
Once again
the thing to note is that generally the ultimate outcome is not just any
workable result, but a something that corresponds to that particular species’
style of doing things. The comb, termite
nest, anthill or trail activity is recognisable as typical of that type of
colony.
It all
looks not only like intelligence, but personality and style as well. When we as humans decide on a preference,
who are we to claim that our cerebral neurons behave any more efficiently or
sensibly than the individual colony members?
Bee dances
do not offer the only examples of interesting behaviour patterns and
information transmission in recruiting foragers. Ants also recruit fellow workers in various
ways, and in particular, a recent observation is interesting in that the
transmission of information is reminiscent of teaching, rather than simple
telling and showing. Foragers of the ant
Temnothorax albipennis will lead fellow workers to a source of food, but
the leading ant only proceeds while it is tapped on its abdomen by a
follower. Whenever the follower stops
tapping and inspects its environment, the leader stops too, and waits for the
follower to resume. The whole procedure
considerably slows down the leader ant, so it has an associated cost. It follows that the mechanism is an important
adaptation for efficient recruitment of foragers so that they can find their
way to the source of food and back again.
In some
ways this is not radically different from the way that some species of bees
(not honeybees of the genus Apis)
wait to recruit hive mates before going out again to the source of food.
In much the
same way as we have difficulty on defining intelligence when there is little
reason to believe that activity is consciously teleological, so such behaviour
leaves us in difficulties in trying to define unambiguously, the concept of
teaching.
There is
yet more to it than that. Many control
systems in nature work, not on a basis of logical propositions, but on a basis
of thresholds and majorities. This is a
very pervasive principle. For one thing,
it is fundamental to the control of cellular activity and growth; when things
go wrong in cellular control in a an embryo, it causes anything from minor
birthmarks to horrific abnormalities.
The slightest, subtlest changes affect the form and character of tissues
and organs, the fit of teeth and joints, the shape and texture of bones,
leaves, and flowers. Such logic is the
basis of hormonal control, physiological life histories and rhythms, circadian
and other timing rhythms, and the movements of cellular organelles and of gross
organs such as muscles.
Above all
in our context, such logic is the basis of the reaction, communication and
control of individual neurons and of neural systems. Unlike the deterministic threshold logic of
most electronic devices, brain neurons fire only when triggered by a not very
deterministic combination of trains of input pulses from various synapses or
sensors. Some inputs might be inhibitory
and others stimulatory. Sometimes
particular combinations of inputs are necessary, such as in the eye, where
retinal receptors detect features such as edges, horizontal or vertical lines,
blocks of colour, or particular textures.
And such
mechanisms are irresistibly suggestive of control in communities, all the way
from bacterial colonies, through the coordinated behaviour of slime moulds, the
control of foraging or bivouacking driver ants, the recruiting of foragers or
swarms in bees or ants, flock manoeuvres in shoals of fish or flying birds,
even unto the behaviour of human communities.
Sociology has a lot to learn from such complexity, control, and
information theory, before it can assume its place as a predictive and
constructive applied discipline for control in human communities.
The
question arises why such an unobvious and complicated mechanism should be so
pervasive in biology, while human control engineering predominantly relies on
either analogue control or finite state control, increasingly often nowadays,
in the form of explicitly Boolean logic.
The most
obvious suggestion is that threshold and majority controls could most simply evolve
in cells that already contained mechanisms that supported such functions. The next is that such mechanisms could work
in many ways and perform wide ranges of functions. They also are fail soft; they often continue
to work under unfavourable conditions.
They are versatile in that they sometimes continue to work validly when
presented with unfamiliar challenges.
Another
point is that such mechanisms also lend themselves well to genetic control and
natural selection. It is not easy to
imagine the functional equivalent of electronic circuits being specified
genetically in such a way that the resulting brain contains more information
than the specifying genome. Nor is it
easy to see how such a system could undergo random modification without gross
failure. And yet, that sort of
adjustment is just how natural selection routinely adjusts the logic of
thresholds and majorities.
A point of
particular interest is that in communities of largely independent units, like
ants in eusocial colonies, instead of somatic cells fixed in the body,
individual members seem to function in ways reminiscent of neurons. The resemblance still is tenuous, but if it
develops to a higher degree, it may be the basis for the formation of a
recognisable community intelligence, literally comparable with the kind of mind
supported by a unitary brain, and for the same reason.
MIND: A mysterious form of matter secreted
by the brain. Its chief activity
consists
in the endeavor to ascertain its own nature, the futility of the attempt
being
due to the fact that it has nothing but itself to know itself with.
Ambrose Bierce -- The Devil’s Dictionary
It is not
clear what the relationships might be between apparent intelligence and
subjective consciousness, but it is hard to imagine subjective consciousness in
a system that has no relevant and elaborate form of intelligence. Subjectively, subjective consciousness seems
to me to require not only intelligence, but meta-intelligence, by which I mean
intelligence aware of (at least its own) intelligence, but I am not at all
certain that my impression is correct.
Certainly there seem to be various degrees of self-awareness among
animals of different levels of objective intelligence, and yet among at least
the most patently intelligent mammals and birds, their behaviour suggests some
sort of subjective consciousness. And
the clearest suggestions are to be seen in the most socially communicative of
species.
Such
remarks are normally taken to imply anthropomorphism, but I reject that charge
in this connection. If anything, it is
the converse of the usual forms of anthropomorphism. In practice we have no more cogent evidence
for subjective consciousness in fellow humans than in fellow animals. This is a (rather weary) problem frequently
raised by existentialists. The fact that
behaviourist experiments and scripting software can produce impressive
demonstrations of apparently conscious behaviour, does not logically compel one
to accept that all apparently conscious behaviour is unconscious and
mechanical. Just because one cannot
demonstrate the fact cogently, sceptics have no need to refute the suggestion,
but this still does not establish that the idea is illogical or even that it is
not a useful basis for criticism of experimental models.
To pile
difficulty on difficulty, it by no means follows that because an empirically
intelligent entity passes the Turing test, including (truthfully) producing
evidence that it is empirically aware of its own mental processes, it has a
subjective consciousness such as humans do.
A mechanical or electronic computer could in principle be programmed to
access and discuss its own data as well as external sensory data.
In other
words, subjective consciousness might very temptingly suggest
meta-intelligence, but meta-intelligence need not imply subjective
consciousness.
The
significance of such meta-intelligence, and perhaps subjective consciousness in
general, would extend beyond intelligence purely and limitedly applied to
external problems. It is quite easy to
accept that meta-intelligence could be of functional mental importance, but it
is commonplace to assume that therefore subjective consciousness also is
functionally important, almost a basis, an essential aspect of intelligence in
any meaningful sense. That this is so,
is not obvious. It is not clear that
subjective consciousness is not an emergent, possibly non-functional, effect of
the way that our brains function as intelligent systems, while say, an
electronic device might pass the Turing test at an arbitrary level without any
subjective consciousness at all, as far as we could tell.
Various
writers have pointed out that subjective consciousness seems to be uncommonly
closely associated with short-term memory.
This might of course be no more than the consequence of the fact that
our speech processing and idea retrieval and co-ordination depend on that part
of the brain. There might be a lot of
other consciousnesses in the brain that never connect directly with the
subjective consciousness that has contact with the outside world. Speaking purely for myself, I often get a
vague impression that there is a consciousness behind my primary consciousness,
and that that is what feeds thoughts, words, intentions and so on to that
accessible consciousness. Uncompelling
though such subjective anecdote might be, it remains suggestive in a field
where suggestiveness often is as useful as anything one may reasonably expect.
For
instance, assuming that what we might call the short-term memory consciousness
is indeed the main, definitive, subjective consciousness, that would have some
important implications. They are not
definitive, conclusive, unique implications, but they open important
possibilities. Suppose that the
short-term memory consciousness is somehow intrinsic to the short term memory
as a region of the brain. That leaves
open the possibility that subjective consciousness is generated by the logical
control functioning of a brain region.
If this is indeed the case, then we have no basis in logic for denial of
subjective consciousness in the brains of animals, including insects.
It also
leaves us with serious questions concerning the possibility of subjective
consciousness in social communities. If
regions of the brain that access and process data for immediate use are
responsible for our subjective consciousness, then if particular
sub-communities are responsible for logical control of the community and its
data, how do we logically deny their having a literal, genuine consciousness of
some sort? Or even a number of partly or
wholly independent consciousnesses in a single community? Equally we cannot logically assert it, to be
sure, but the importance of the implications, or the importance of the
implications if we can show the speculation to be wrong, or even the importance
of our contemplating the problem and exploring means of investigating the
alternatives, forbids us to dismiss the concepts too facilely.
In general
then, discussion of subjective consciousness and its biological significance is
fraught with difficulty. We need a major
scientific breakthrough, analogous to the first advances in the study of
electricity. Such a breakthrough would
have to be on the scale of the first recognition, measurement and generation of
electric current, before we could study the matter meaningfully. At present we are no better equipped to
discuss the nature of subjective consciousness, than we were to discuss the
nature of disease before the invention of the microscope. (And don’t be impatient, the germ theory of
disease took some 200 years to develop after the first microscopic observations
of bacteria and cells.)
If an animal does something they call it instinct. If we do exactly
the same thing for the same reason they call it intelligence.
I guess what they mean is that we all make mistakes, but that
intelligence enables us to do it on purpose.
Will Cuppy
One major
value of the discussion of the intelligence of communities is that it throws
the fallacies of composition and division into stark contrast. Works such as “The Mind’s I” by Hofstadter and Dennet have elegantly
illustrated some of the problems, and Hofstadter’s “Ant Hillary” in Goedel,
Escher, Bach” showed some others. In
nature we have cases where an intelligent system, such as the brain, consists
of neither particularly nor obviously intelligent units, such as neurons.
Conversely
it is hard to tell how much intelligence a human community has, and it is not
clearly possible at all to compare it with the intelligence of any component
human, let alone each component human.
What about idiots and geniuses who happen to be members of the same
population? Can we meaningfully compare
the intelligence of the community with the mean, modal, or extreme intelligence
of its individual members? Does one
measure community intelligence in terms of the technology that a few ingenious
members have implemented and the rest hardly understand at all? Or in terms of the infrastructure that a
major sub-community have established? If
so, does it matter whether the infrastructure was imposed by a few in
authority, or by conscious co-operation, or spontaneously by not consciously
intelligent mass behaviour? In castes in
a community, it is very likely that different castes have different levels of
intelligence, and different kinds of intelligence to suit different roles. For
example, in the leafcutter ant genus Atta, the minims that ride on the
heads of foraging workers are very likely to have brains that differ from those
of the workers. And the repletes of honeypot ants that hang upside down for
months, might well have brains that differ from those of active workers.
In Apis
colonies it is certain that drone brains differ from worker brains, and from
queens’ brains, but it also seems likely that worker brains differ as they
mature after eclosion from the cocoon, and undertake successively different
roles in the colony.
If the
community is less intelligent than its sentient members, then how does the
intelligence of a honeypot replete or of a mobile worker ant or bee compare
with that of its community? How does the
intelligence of a worker or queen termite compare? How does the intelligence of a hive compare
with that of a worker?
The replete
really is just a worker ant that has developed a huge crop that fills its
abdomen with stored syrup. Once it has
achieved that stage it requires little more intelligence than is necessary for
accepting or dispensing honey, plus hanging from the roof in a suitably
comfortable chamber. The queen started
with enough intelligence for her mating flight, plus establishing a nest and
feeding the first generation of workers.
Once she has settled in though, she has little to do but lay eggs. We know that worker bees’ brains change in
shape and size as they change tasks. It
is not implausible that these other job-changing insects have equally flexible
brains.
Apart from
the question of the relative intelligence of different castes within a colony,
how does the intelligence of a worker bee compare with that of a fly? Houseflies are perhaps less inclined to get
trapped behind glass because they are less stereotyped in their positive
phototaxis, but not many of their behavioural patterns are nearly as elaborate
as those of bees. And yet the behaviour
of either insect shows little sign of insight or of anything but
stereotype.
But
both can apparently learn something, even if that thing is little more
than a state of arousal in response to an alarming or rewarding stimulus —
say, an encounter with a threat or with food.
This
problem expands to match other dimensions.
How does the intelligence of a species compare with its
members or with communities of its members? When two species form an association vital to
both, such as a specialist pollinator and its associated specialist plant, then
relatively, how intelligent is say, the yucca moth, the yucca, or the
association of the two? How intelligent
is the mitochondrion in the human, relative to the neuron, the brain, the
nation, or the species? It is not
obvious how far one should stretch such boundaries. Do we include domestic organisms as members
of our communities? Inquilines? Pests?
Crops?
It is
extremely difficult, even questionable, to compare the intelligence of
different species meaningfully, let alone the putative intelligence of
incommensurables such as individuals and communities, or devices and
organisms. We need a cogent discipline
of the systematics of types of intelligence, and measures of the various types
of intelligence before we can talk sensibly of such matters.
We also
need some measure of entity-hood. Formal
fuzzy logic barely attacks the problem.
Now
consider some levels of sociality and associated intelligence. Note that the levels neither are all
independent nor form a clearly ordered sequence. The problem is that apart from the poorly
defined nature of intelligence, it is a multidimensional phenomenon or range of
phenomena.
Eagles
commonly fly alone
John Webster
One might
expect that populations of socially independent, parthenogenic individuals
should be unintelligent both as entities or communities, but it is soberingly
difficult to find clear examples. Even
in many kinds of bacterial community there are interactions that look
suspiciously like altruism or organised control or competition. And yet, bacteria not only are brainless, but
lack the necessary complexity to form somatic structures the way the eukaryotic
cells of metazoa or metaphyta can.
The
principle extends still further. Viral
communities seem to work in unison to cause effects such as diseases in hosts,
or symptoms that aid the spread of the pathogens. For just one example, rabies viruses, lacking
any clear socially interactive behaviour, cause neurological damage in hosts,
such that it causes varied and elaborate behaviour that propagates the
virus. And they are by no means the only
pathogen to cause analogous behaviour that harms the host but propagates the
parasite. There are whole classes of
such behaviour among pathogens.
Conversely,
many other diseases, most notoriously myxomatosis, rapidly evolve into a less
malignant relationship with the host, not out of obvious altruism, but because
such aetiology maximises its infective success.
Selection for the success of pathogens works purely on effectiveness,
not specifically the harm or the health of the host.
How
intelligent are such pathogenic species as entities? Is the operative entity
the pathogen, or the host/pathogen/ecology structure? Where does the
information processing of such a structure reside? As individuals the microbes display nothing
resembling intelligent behaviour that is not stereotyped and mechanistic. And yet species that lack any such effective
aetiological strategy, tend to die out or at least become less successful as
pathogens except in exceptional circumstances.
From the point of view of the empirical observer, how do we logically
distinguish this sort of relationship from intelligent behaviour of the
community?
The popular type and exponent of obstinacy
is the mule,
a most intelligent animal.
Ambrose Bierce -- The Devil’s Dictionary
Most forms
of sexual reproduction entail apparently intelligent behaviour, usually in the
persons of its individual members, or they could not bring their gametes
together. In microbes such behaviour is
stereotyped, even mechanically controlled, and also in many metaphyta and
metazoa. A sexually reproducing species also must exhibit intelligent behaviour
as an entity, in the sense of reaction to relevant information, or its members
could not exhibit effective reproductive behaviour.
To
illustrate just how extreme the contrast may be between the functional
intelligence of a species and that of its members, consider the flowering
plants. The individuals are at most
arguably intelligent, but the majority employ sexual reproduction, including
examples of strategies that are intellectually baroque to put it mildly. Such selection does not come cheaply in
evolutionary terms, because within any species the reproductive apparatus tends
to be very conservative. This makes
sense because that apparatus is an example of where two sets of components must
not only work individually, but also must work in combination.
If they
fail to do so successfully, their line stops there.
This
principle: that systems comprising complementary components are generally
conservative, is of vital evolutionary significance. It is ubiquitous in many forms and
senses.
Firstly,
without complementary mechanisms we simply do not get complex organisms. Just to form bodies requires cells of
considerable complexity; no prokaryotes have managed it yet, let alone any
viruses. Even for eukaryotes the
development of cells that can be assembled into bodies is challenging. The mechanisms necessary for the formation of
bodies of metaphyta and metazoa are startlingly complex and specific. They rely on elaborate cell structures plus
elaborate molecular coding for intercellular matching, signalling, binding, and
release. Each of these is at once
complex and vital; let even one fail in just one type of tissue, and the
consequences can be anything from disease in the senile adult, to death of the
developing embryo.
In
assessing the non-triviality of the problem of the evolutionary development of
such complementarity, reflect that the best current estimates of the length of
the period it took for the first unambiguously multicellular organisms to
evolve from the first prokaryota, exceeds the length of the subsequent history
of life so far.
Even less
dramatic examples of the development of complementary biological mechanisms had
to overcome shocking challenges: consider the development of the Eukaryota from
prokaryotes. We do not know how long it
took, but it clearly was not abrupt — and their current biochemical and
biophysical mechanisms become the more imposing, the more we learn of them.
Equally
radical was the problem of the development of intracellular organelles such as
the ribosomes. Once having formed they
remained almost unchanged for perhaps some three billion years. And what is more, they look like staying that
way till the sun swallows this planet, short of some really aggressive genetic
engineering for as yet obscure purposes.
The
conservative nature of sexual morphology and physiology cannot compare with
those truly ancient examples, but it none the less is pervasive and stems from
the same logical problem. It is in fact
so pervasive that among fungi and flowering plants the comparative anatomy of
the sexual parts have presented far and away the most important taxonomic
features until comparative biochemistry and molecular biology began to advance. Among insects the principle is less extreme,
but still conspicuous; insect genitalia are of great importance in
characterising many groups. In terms of complexity theory there are good
reasons for this, especially in systems driven and constrained by biological
natural selection.
The
requirements of a sexual system vary with the nature of the organisms. Plants are generally sessile and depend on
mobile agents to move the pollen to the stigma.
Commonly the vectors are birds or insects, which means that not only
must the mechanics of pollination work properly, but the movement of the
pollinator must be guided as well. These
factors limit the rate at which flower architectures can be changed without
risk of effective sterility.
In insects
and in general among animals with small brains and highly stereotyped
repertoires of activity there are different reasons for highly conservative
mating anatomy, physiology and behaviour.
Although their behaviour is not strictly rigid, they never can allow too
much initiative in so critical an activity.
Often they get just one attempt and if a population misses that, it has
just become a selective dead end. The
challenges are too formidable for reliance on the intellectual originality of
individuals. Mobile species have have had to develop highly specific senses
instead, anatomical features and scripts for their mating, and the shorter
their adult lives are, the more specific their scripts are likely to be.
Commonly
the first attraction is chemical. It
serves for identification and attraction.
This sounds simple, but in fact recent work has greatly multiplied the
range of compounds that insects are known to use for signalling. In guiding mating activity, some wasps use
both visual clues, plus at least seven successive compounds for control of
seven successive stages of the mating act.
Amazingly, some orchids rely on those male wasps for pollinating them by
trying to mate with them, and even more amazingly their flowers not only
resemble the female of the relevant bee or wasp, but also produce all seven
chemicals! The very fact that the
orchids had time to evolve and specialise in the use of such an elaborate
system does imply that the wasps had been developing and using the same system
for a long time, probably many millions of years.
The reason
for the multiple attractants is that each one triggers a particular
activity. The first might be a
long-range attractant, the next short range, then a contact guide, another
might stimulate genital contact, and so on.
In experiments it has been shown that omission of any one signal can
interrupt the mating script at that point.
For all we know there may be other insects whose chemical mating signals
involve even more compounds. The point
is that where each signal is so specific, and their sequence so rigid, it
enables the species to rely on the script for automation of the process,
instead of on the discretion of the insect.
Konrad Lorenz has remarked that in fighting fish the courting procedure
is intense, not to say violent, but the actual fertilisation is apparently
dispassionate. This is true, and not
only in those fish, though of course we have no idea whether such behaviour
reflects actual states in the mind of the fish, if it has one.
There are
several variations of mating procedures among eusocial creatures. In naked mole rats it is not unlike a lot of
other mammals in which only alpha members of the community are allowed to
reproduce.
In many
eusocial insects there is a nuptial flight and the males and females seek each
other out and pair off. In many of ants
the flying reproductives form pillar-like clouds, much like the clouds of
courting, non-social Empidid flies, and for similar reasons. Being two-gender affairs, the visible clouds
are not really leks, but like leks they attract all the locally available
candidates. In fact, some such
mating clouds, multiple species of similar insects might combine — a
larger, denser, cloud may be more than sufficiently effective to compensate for
having to find a mate of the right species within it. This of course would
depend on the relative frequency of the distinct species; if one species made
up less than say, 1% of the cloud, it might never find a mate, but if 50% then
both species could benefit.
Most
species of termites do not form such clouds, or not for so long anyway. Possibly this has to do with the fact that
termites are more attractively edible.
Be that as it may, the participants tend to disperse as soon as possible
and the females generally attract males by scent. Once they have met and shed
their wings, termite pairs go off and the male joins the female in establishing
a nest. He typically remains with her
for years. Perhaps this is partly to
ensure an adequate supply of sperm for a single female that in some species
might need to reproduce for decades.
Honeybees
are notorious for the single mating flights of drones. It has long been known that the drone mates
at most once in his life, and some authors have overdramatised the event. They claim such things as that the female
also mates just once and that she tears the entrails out of he single
successful suitor. In fact a young queen
typically mates several times, once on each successful mating flight. She certainly does typically return with mail
parts trailing behind, but that is a consequence of the mating process, not any
particular impulse to female sadism.
Instead, what happens is that at the moment of coupling, the male’s
orgasm contracts every muscle in his abdomen with such violence as to expel his
genitalia, kill him instantly, and leave his corpse feeling peculiarly dense,
almost like a little pebble in one’s hand.
One effect
of this is to leave the female with a clogged genital passage, so that she has
to return to the hive and clean up before she can mate again. This ensures that no subsequent suitor can
displace the semen, so that no doubt the successful drone did not mate in vain,
even if he must share sperm space with his rivals.
In any
case, since the queen does not mate again after a few nuptial flights, it pays
for her to stock up on sperm, since she needs it to produce workers as well as
fertile daughters. Incidentally this
peculiarity of the honeybee mating process has led to a major debate concerning
genetic conflicts of interest among the workers and their siblings. On average they generally are less closely
related than full sisters. I do not
explore the implications here, but mention the point as yet another
illustration of the complexity of the factors that influence the strategies of
the species.
Sexual
strategies are generally species-determined; even humans only observed the
connection between sex and reproduction a few thousand years ago. There are examples of intelligent animals
practising sneaky sex, but there is no evidence that the proximal objective is
anything but the mating. In less
intelligent species such as fish and lizards one encounters some examples of
systematic sneaky sex, but those seem to be just as stereotyped as direct
sex. There also are many cases of where
one male will frustrate the sperm or kill the offspring of another, but the
elaborations of sexual strategies are too varied and too many to discuss in
detail.
The
important point is that nearly all the sexual strategies in nature are
strategies of the species, or at least of particular genetic lines, rather than
of individuals or communities.
A brother may not be a friend, but a friend will
always be a brother.
Benjamin Franklin
Not only do
communities of species that become mutually interdependent exhibit their
various individual apparently teleological behaviour, but each of the
individual species that form components of the communities has its own
teleological behaviour and so does each of the inclusive systems involved. However, it is not at all clear that there is
any question of individual teleology in the behaviour of particular
participating specimens, or that if there is, it corresponds to the teleology
of the species. Examples include:
· Symbionts such as cleaner wrasse
or oxpeckers
· Stranger-than-fiction symbioses
such as “The medusa and the snail” discussed by Lewis Thomas in his essay of
that name.
· Endosymbionts such as mycorrhizae
or organelles in eukaryotic cells. A
truly mind-stretching range of examples begins with the endosymbiotic bacteria
that developed into mitochondria. They
tend first to lose genetic material whose function is redundant because the
nucleus can supply it. Some go further
when some of the genetic material from the endosymbionts finds its way into the
nucleus and in turn becomes redundant in that species of mitochondrion in that
species of host cell, after which it might get lost from mitochondria as
well. After such process have gone far
enough, the mitochondria are left with very few genes indeed. The details of the next question remain open
to argument, but it seems that at least some classes of hydrogenosomes are
descended from mitochondria, and most hydrogenosomes have lost all their
genetic material. Truly a grin without a
cat! They do not however lose all their
information content; they retain the information embodied in their physical
construction; it seems unlikely that the host cell can create a new
hydrogenosome from scratch, even when it has all the necessary genes from the
parent hydrogenosome.
Now, this too is a very difficult concept to come to terms with. It might at first seem simple, but it leaves
us with difficulties in defining conflicts of interest, evolutionary strategy,
and identity.
How far you go in life depends on your being tender
with the young,
compassionate with the aged, sympathetic with the striving and
tolerant of the weak and strong. Because someday in
your life you will have been all of these.
George Washington Carver
Within any
species a crucial step toward sociality is mutual tolerance for purposes of
reproduction. This is necessary, first
for mating instead of prematurely eating the mate, second for sparing offspring
instead of eating too many of them.
Innate tolerance creates evolutionary opportunities for developing
altruistic strategies such as sib care in foxes, scrub jays, carpenter bees,
and paper wasps.
Mechanisms
for such tolerance are varied. Some
species manage it by smell: they recognise unacceptable company by the fact
that it does not smell like the right population. This is very common in social species. Many of them will not tolerate other
populations, even if the species is correct. Others recognise their own
neighbours as individuals. This is not
common in eusocial species, but many sociable species, especially birds and
some mammals, do that, including humans, to some extent.
Another
mechanism is that some species simply lose their appetite at critical
seasons. Some reptiles use such
tolerance for seasons when they are likely to hibernate together. Other species, particularly birds of prey in
the neighbourhood of their own nests, lose their aggression to other birds. Thus we might find a hawk ignoring the nest
of a dove in the same tree as its own nest.
Territorial
species such as some chameleons, may have young with different coloration from
the adults. Such adults, such as those
in the genus Bradypodion simply ignore their own tiny putty-coloured young. Predatory chameleons, such as Chamaeleo
namaquensis will not eat a juvenile as long as it is in profile, and as
such resembles its own species, but if the baby turns end-on, for example to
run away, the adult is likely to snap it up as once when it fails to recognise
it as non-prey. Conversely many
creatures that exhibit no special parental care, actually eat any of their own
young that they can, no matter what it looks like..
There are
other classes of examples, but the point is that eusocial species need just
such tolerance in one form or degree or another, or they could never have
developed any successful sociality. Many
social species are after all not just territorial, such as naked mole rats, but
also predatory, such as social spiders and paper wasps.
It is easier to love humanity as a whole than to love
one's neighbor.
Eric Hoffer
Similar
innate tolerance may arise advantageously between neighbours, such as Bembix
wasps or solitary bees nesting close together in earth banks, or seabirds
nesting in tight hexagonal array, or indeed, the flocking or shoaling of birds
or fishes.
Mammals and
birds often form mixed herds or flocks for purposes of security, feeding and
socialising. There may be squabbles
between neighbours, and problems with parasites, in fact, there are thousands
of species of parasites that specialise in their reliance on flocking
hosts. The hosts of such respective
parasites range from locusts and fish to starlings and cattle. In (human) modern warfare, they also included
convoys of ships trying to beat off submarine attacks.
Seeing the
dreadful effectiveness of attacks on flocks, one hardly can believe that there
are advantages to presenting such concentrated targets, but in most cases in
nature (and in shipping convoys in wartime) the advantage of improved security
exceeds such costs. It is mainly when
the predator is large enough and powerful enough to destroy the target
wholesale that flocking is a bad strategy.
The
behaviour of the flocking community may be largely selfish, for instance, when
attacked, each herring tries to hide behind its neighbours, and the results
might temptingly resemble neurosis or other dysfunctional behaviour. For instance, whales or large predatory fish
herd schools of small fish into effectively solid masses for easy eating. In such terms both the attackers and the
attacked communities constitute individual entities, or even resemble single
organisms.
The
effectiveness of the swarm tactics can be calculated on operational research
principles, or it can be seen directly in footage of predators attacking flocks
of bats or prey birds, or fish attacking shoals of pilchards or herring. Not many attackers will venture into such a
flock. The operative reasons for their
shyness include the risk of injury and the counter-intuitively low probability
of success. Each prey organism avoids
the attacker individually, while conversely attackers need to fix onto single
prey items and find it difficult to do so.
The true
effectiveness of such flocking as a defence becomes most obvious when it is
subverted and fails. A notorious example
is that if one marks a herd animal, such as a gnu, for study in an environment
of heavy predation, it is difficult to do so without dooming it. As soon as the predators can recognise the
marked individual, they can concentrate on it.
Normally hunters will harry a herd till they can identify an individual
by some peculiarity, and this is doubly effective because the most attractive
of such peculiarities are those that indicate that the candidate prey is not
well and is easier to catch. However, in
the light of the observation that even healthy animals fall to predators once
they are marked, it is clear that as a rule, identification is more important than
weakness.
Conversely,
there are potential advantages to the flock, that they might cooperate in their
defence. In its most sophisticated form
this is strangely rare as a strategy, presumably because it is difficult to
implement effectively, and it is costly to any individual if it relies on
support and gets let down. Such factors
militate against the development of the strategy. Buffalo
weakly apply herd defence against lions, but if they did so systematically and
aggressively, they would be effectively immune to predators such as lion or
hyaena. Musk oxen demonstrate that fact
in their famous defence against wolves, in which the adults, especially bulls,
form a protective ring around their young.
Except against humans with dogs, that defence is practically
impregnable. Presumably in the bleak,
ice-age north, wolves wiped out most herd species that failed to develop such a
defence.
Among
eusocial insects a different principle is added to the relevant variables. The workers or soldiers are not generally
reproductives. They have little to lose
by being killed in defence of their gonads and offspring, that is to say, the
gonads and offspring of their reproductive castes. Thus it is that we find ants, termites, wasps
and bees aggressively and almost uncompromisingly defending their colony or
attacking formidable prey, at the cost of their own lives. Most of them do so in an apparently
unorganised fashion, but even then it often is more organised than it seems. The majority often will not bite or sting
till they smell the attack pheromone released by the first hive-mate that is
injured or provoked. By that time many
warriors are likely to be in position, and they attack together from all
angles. This is far more effective than
just having the first bee that gets into in position attack at once before the
others can support it. Bees and many
species of ants are likely to attack in such a way.
Usually
when social species attack one member at a time, they do so because as individuals they have little to defend, or
because they are very formidable, so that a single sting or bite is likely to
rout the attacker. Some species of bull
ants (Myrmecia) and some paper wasps (e.g. Vespa mandarinia) are examples.
Apis cerana and some other Apis species offer an
example of coordinated attack in their balling behaviour, by which they kill
scouting giant hornets and similar predators, as I discuss elsewhere.
Notice that
all these examples might be seen as illustrating the intelligence of the
community; certainly there is very little evidence of any of them arising from
the intelligence of the individual, not even of any particular leader.
Notation is a tool of thought
K.E.Iverson
A very
important behavioural component in the origin and development of mutual adult
tolerance and care for offspring, let alone sociability and eusociality, is
context-sensitive inhibition of aggressive behaviour patterns. Lorenz was the major pioneering writer on the
subject, and his writings on innate, injury-limiting conventions for
intraspecific fighting were seminal.
Such
tolerance is innate, and is widely spread, both within species and even
interspecifically in non-colonial species such as hawks that will not attack
prey birds nesting close to their own nest.
Bembicine wasps hunt flies and carry their prey into the tunnels in
which they raise their young. Parasitic
flies wait around the nests for opportunities to oviposit on the wasps’ prey
when they carry it in. Observers have
wondered why the fly hunters do not simply catch those flies as well. It might be worth investigating whether the
wasps are not simply similarly inhibited from attacking flies so close to their
own homes. But it also may be that the flies are below the threshold size that
cues the wasps’ predatory attack
In any case
the importance of such innate inhibitions is that they can serve as a basis for
co-operation in both interspecific and intraspecific relationships.
Some therefore cried one thing, and some another:
for the assembly was confused: and the more part
knew not wherefore they were come together.
Acts of the apostles 19:32
Otherwise
independent non-social individuals of many species do assemble on occasion, and
in many ways and for many reasons. The
simplest examples meet just for mating, but in practice this involves visual,
chemical, or auditory signals that might assemble many more than two
participants, such that one sex or another has a choice of mate. Males might display at an assembly point,
such as a suitable tree for birds of paradise, or open space for grouse or
ruffs, or cicada males in particular trees.
Such assemblies are called leks.
The term
“lek” seems to be limited to male mating assemblies, but one does also get
two-sex assemblies and even a few female assemblies to attract males.
Whether to
call the female assemblies leks, and if not, then what, I don’t know, but never
mind.
An
interesting aspect of lek behaviour is that as a rule the assembly forms by positive
feedback. Some species of cicada males
that feed in trees and sing to attract females, prefer trees where there
already is a hubbub of rival males. This
might seem counter-effective, but it is not really; for one thing, the females
are prone to go where the boys are, and they choose a male where the communal
song is at its loudest. And secondly, the males in the chorus are subtle in
misleading predators: an approaching noise causes the nearer males to soften
their song or even go silent, so that a stalking predator’s direction finding
equipment systematically loses the nearest songster, and fixes temporarily on a
more distant target. It is a fine
example of how simple, innate activity on the part of individual members can in
effect combine in a sophisticated collective strategy.
Some
species of firefly males assemble where they see the most light flashing
synchronously in the patterns peculiar to their species. Their females accordingly are attracted and
can select mates. Interestingly, other species
of firefly females also assemble at such leks, but their objective is food;
they flash the feminine come-hither, and grab and eat any male that
responds. This is obviously an ancient
strategy, because research has demonstrated that there has been an evolutionary
arms race, with victim species evolving ever more complex signals, and
predatory species evolving correspondingly complex responses.
Another
class of assembly is common among insects.
Some species will fly to a prominent object, where they will find others
with similar behaviour. For example,
some beetles fly to the highest hill in the horizon. Empidid flies of many species assemble over
local items, such as a high nearby bush or rock. Sometimes they even assemble in a cloud above
a slowly moving human or large animal.
Some kinds of ants on nuptial flights assemble in a similar swirling
cloud above the nest from which they emerged, until they have attracted their
mates.
Some tiny
Diptera that form mating swarms near water bodies, are not selective about
whether more than one species, or even more than one genus, is represented in
the assembly, several species of the same size and general appearance may be
present. Obviously, this is beneficial for every species that has a high density
in the swarm, because a large crowd makes the swarm more visible and more
attractive and thereby recruits more potential mates. The fact that
participants in multi-species swarms need to be selective in avoiding trying to
mate with alien species, is outweighed by the advantages of the increased
swarming benefits.
A more
direct basis for assembly is where there is food or a nesting site. It is an impressive sight to see the huge
dung beetles that specialise in elephant dung, with their antennae spread, circling
in like so many vultures, not only to food, but to where they can find
mates.
By
assembling according to common or complementary stimuli, these organisms are
exploiting the principle of “Do you come here often? Small world!”. The seminal, but simplistic Milgram
experiments suggested that human society was “shallow” in that the
friend-who-knows-a-friend chain between any two people on Earth is likely to be
short, probably less than nine.
Importantly
however, further work shows that part of the reason for this turns out to be
because humans are not a random network of acquaintances, but inclined to
homophily; that is to say, birds of a feather flocking together (nothing to do
with homosexuality, for which the term “homophilia” seems to have been expropriated).
Also, even in an otherwise random network, just a low frequency of long-range
acquaintances shortens the maximum depth of the community drastically.
By thus
selecting or creating assembly points that attract participants from afar, the
species not only increases the chances of successful mating, but also of mating
with a low frequency of inbreeding. Some
of the assembly techniques more or less force outbreeding. For example, some ant nests release only one
sex of reproductives at a time and they accordingly are forced to find mates
from other nests. In such species,
nests in a given region tend to use similar triggers for releasing
reproductives synchronously, so that the new hopefuls at least stand a good
chance of finding mates.
Empirically
such assemblies are most intelligent strategies. The subject is a large one however, and there
are many counter-examples in nature.
. . .And how can man die better than facing fearful
odds,
For the ashes of his fathers and the temples of his gods,
And for the tender mother who dandled him to rest,
And for the wife who nurses her baby at her breast. . .
T. B. Macaulay Horatius
Going
beyond simple intraspecific tolerance, many of the behaviour patterns in
sociable and social species plainly are developments of behaviour patterns that
one might observe in solitary species that exhibit rearing care behaviour. For example, monkeys of many troop-forming
species will compete to handle the young of other members of their troop. Carpenter bees and paper wasps care similarly
for young in their nests, whether those young are their own or not. Some birds will accept the young of other
species in their nests, though others will kill or eject them. Mammals in their lairs often have been
observed to feed young of other species.
In
that it is less constrained, defensive behaviour for protection of the colony is mainly distinct
from defensive behaviour in solitary species.
For the genes of soldier castes, it is more important that the colony
survive, than that the individual survive.
In contrast, defensive behaviour in solitary iteroparous species usually
is limited to conflicts which the defender might be expected to survive,
because it commonly is more important for the parent’s genes, that the parent
survive to raise more offspring, than that any single threatened youngster
survives. For complementary reasons
semelparous species may be less compromising: for the genes of the parent that
will not propagate again, it is more important that the brood survive than that
the parent survive. Similarly, eusocial
castes might exhibit cleaning or foraging behaviour similar to that of solitary
species, but more specialised or intensive.
Another
interesting example of the use of castes for special defensive purposes is in
the minor workers of some leaf cutter ants.
The major workers sally forth to collect pieces of leaf. They are accompanied, or even ridden, by
minor workers who do no leaf cutting. On
the return journey the minor workers ride on the leaf sections, “parasols”
carried by the major workers. What they
are doing on the trips is not easy to tell, because their main duties are in
the nest, tending fungus gardens and the like.
In spite of the suggestions of early workers who studied leaf cutters,
it since has turned out that the minor workers are not passengers, but “riding
shotgun”.
The
function of these minor workers on foraging sorties is to repel parasitoid
phorid flies. The flies otherwise would
lay their eggs in the major workers, so that their larvae could eat them alive
from within. This employment of the
minor workers makes sense in simple economic terms. To the nest, the cost of raising a minor
worker is at most a few percent of the cost of a major worker. They are so cheap that a minor worker need
protect a major worker only slightly to justify its trip. I have more to say on this point later.
But
the point is that such a strategy, conscious or deliberate or not, is not a
simple item to distinguish from intelligence on the part of the colony.
Gnosin damazon, dedamakas panta
(Master knowledge, master all.)
Anonymous
There is a
serious evolutionary problem to the evolutionary development of specific
recognition signals. I first saw this
point raised by the late E. F. Whiteside, as I have described in another essay
on the evolution of pheromones. In
summary, the question is: in any genetically determined, functionally vital and
specific, signalling system, any deviation from the canonical signal must be
seriously disadvantageous, because such a deviant signal would be at best
relatively ineffective. For example, if
a female emperor moth’s signal failed to match the receptor of a male, she might
as well be sterile. If, on the contrary, there were several species of emperor
moths in a region, and their attractants were not distinguishable, that would
not be a great deal more effective.
How does it
come then, that we have more than one species of emperor moth, which in fact we
do? There are dozens of Saturniidae
known from Africa alone, and thousands of
relatives world wide.
As it
happens, I can propose a mechanism in which classes of highly specific signals
can evolve, as long as the receptor as well as the signal are determined by the
same genetic system.
This might
be seen as part of the intelligence of the species as a community. The principle is worth noting, but is not
highly germane to the current issue; I have discussed it in another article on
pheromonal evolution.
It is only to the individual that a soul is given.
Albert Einstein
I know of
no eusocial species in which a colony is not derived from a single nuclear
family, though in some species the queen (mother) may mate several times, and
there may be a succession of queens, sometimes even several reproductive queens
in the same nest simultaneously. The one
clear class of counter example is not eusociality in the sense intended in this
discussion, and that is in endosymbiotic or near-endosymbiotic cell colonies
such as lichens, algae, sponges and humans.
Such
endosymbiosis occurs in numbing variety; it ranges from unrelated species of
cells that interpenetrate to exchange materials or signals, to organelles in
cells that simply were not suspected of being separate species till well into
the mid twentieth century. The most
obvious examples are organelles such as chloroplasts and mitochondria. The very suggestion that organelles might be
endosymbionts at first was widely rejected, but once the evidence became
incontrovertible it caused a sort of witch hunt, or perhaps a gold rush.
Suddenly
zealots saw endosymbiosis everywhere and labelled practically every
identifiable cell structure as a relict endosymbiont. Flagella, cilia, ribosomes, nuclei, they all
were proposed by various enthusiasts.
Some proposals were reasonable, some simply silly, but none detracted
from the fact that cells and even multicellular organisms of various types,
related or unrelated, could connect and combine in amazing ways for amazing
functions.
It also
became obvious in retrospect that symbiosis and in particular endosymbiosis
offered the most powerful means by which natural selection could defy the
combinatorial improbabilities of simultaneous development of biochemical tools
that had developed independently in unrelated organisms. The need for simultaneous development of
mutually compatible adaptations, places an enormous selective burden on even
the largest populations.
We cannot
explore every development of this type.
It is the subject of whole sets of books. However, it is closely related to the subject
of this essay. There is good reason to
suspect that the principle of combining independently developed functions
involves as much of the mechanisms constituting community intelligence, as
co-operation among family members.
It would be
conceivable that eusociality could arise between conspecifics of different
parents, or even in multiple species, but obviously that sort of thing would
increase the range of genetic conflicts of interest, so that there are
probabilistic obstacles to the evolutionary development of a stable
relationship of such a type. In practice
it is very difficult to find clear examples where this actually has
happened. Inquilines usually turn out to
be either parasites, scavengers, or commensals.
In the case of ants, some parasitic species of ant have graduated to behaving
as actual predators.
Conversely,
all the most obviously social or eusocial species known, could easily be seen
as having arisen from ancestors with highly developed nuclear family care.
Societies need rules that make no sense for
individuals. For example,
it makes no difference whether a single car drives on the left or on the
right. But it makes all the difference when there are many cars!
Marvin Minsky
In the
sense I have in mind, reaction to the environment falls into two main
categories that are important in the current discussion. Both might easily be overlooked because in
solitary species we take such behaviour for granted, but there are intriguing
analogies to social species when one thinks of them in terms of apparent
teleology.
Brute force crushes many plants. Yet the plants rise
again.
The Pyramids will not last a moment compared with the daisy.
And before Buddha or Jesus spoke the nightingale sang, and
long after the words of Jesus and Buddha are gone into oblivion
the nightingale still will sing. Because it is neither preaching
nor commanding nor urging. It is just singing.
And in the beginning was not a Word, but a chirrup.
D. H. Lawrence
There are
many ranges of ways in which social insects establish colonies. The most primitive are presumably those that
most resemble the behaviour of solitary insects that patently exhibit parental
care. The most obvious examples of such
insects include solitary bees and wasps and some kinds of cockroaches, but one
could easily extend the category all the way to the even simpler behaviour of
insects that lay their eggs only on suitable food plants or in suitably secure
situations.
Other forms
of parental care include the raising of young to maturity, then dispersing them
to mate and establish new colonies of their own. Sometimes this requires the young to carry
quite elaborate resources, ranging from internal body fat, to cultures of the
necessary symbiotic microorganisms. Still others rarely send out reproductives
on mating flights, but instead raise many active reproductives in the colony,
and keep scouting for new nesting sites to take over on foot and without
nuptial flights. The Argentine ant for
example, rarely has fights between nests, but forms huge colonial complexes,
empires if you like, that include indefinite numbers of reproductives and
workers in nests scattered over large areas and connected by trails. When a nest becomes undesirable, there is
likely to be a lot of carrying of brood to neighbouring or new nests.
Not to
multiply examples however, modern solitary bees and wasps include examples of
behaviour that no doubt is functionally similar to that of the ancestors of
many modern eusocial species. Typically,
the young, mature female will seek out a spot suited to building the home or
digging the burrow where she will collect the food and lay the eggs. Obviously if she never gets this right, her
line ends there.
In
practice, there are thousands of extant species that routinely do get this
right. They include borers, miners,
builders, colonists, parasites, opportunists in bewildering variety, but they
all share one attribute: if they select the site unsuitably, or fail to prepare
or maintain suitably, that too is the end of the line. And such behaviour requires intelligence, in
the sense that the parent must react appropriately to environmental
stimuli.
Nest
selection is no simple task. Safety from
various enemies is one requirement.
Suitable moisture and temperature control are vital. A large percentage of nests fail on such
points. It takes some doing for a human
to judge the viability of a new nest, but as a rule the brood mother does
pretty well.
Here it
does not matter a bit whether the parent is subjectively intelligent in herself
or whether she is the purest automaton.
The same minimal amount of information collection and processing is
necessary in either case. In the context
of this essay that information collection and processing is what matters, and
in practice the behaviour of the parent can only be distinguished from
subjective intelligence by such experiments as Fabre would perform to
demonstrate the limits of their rational reactions to abnormal
circumstances.
This is a
difficult field in which to maintain good perspective. One is inclined to see the silly wasp
demonstrate its lack of mechanical understanding of the situation, and dismiss
its intelligence accordingly. However,
that is a treacherous line of reasoning.
Dogs for example, which we accept with good reason as being more
intelligent than wasps or bees, will also reveal a drastic lack of such insight
in analogous circumstances.
However,
none of that matters in our context. The
point is that the intelligence, if we meaningfully can call it that, in such a
case is intrinsic to the species, not the individual. Now, on that assumption we can move on from
the solitary bees to say, some carpenter bee species, in which instead of
keeping the larvae separate, they may be raised in communal cells and when they
are mature may even assist in raising their own siblings.
The wasp
equivalent is in paper wasps, such as Polistes or Belonogaster, in which the
queen establishes a nest in which the first young are likely to end up as
workers. In some species, when the first
workers reach maturity the queen stops hunting, lets the youngsters forage and
stays home to lay eggs.
In most
ants and termites the mated queen establishes a nest, with or without a
surviving male, depending on the species.
Usually this will be a burrow or a hollow under a stone, but many
species live under bark, in leaves and other situations.
As I said,
such behaviour requires decisions that would challenge human judgement, and
what is more, it is judgement that must be made in the context of external
circumstances. In other words the
behaviour must be flexible, adjusting to challenges of resources and
adversity.
Simultaneously
however, there is a great deal of stereotyped behaviour of great
sophistication. Sometimes queens will
carry necessary inquilines or microbes from the parental colony, from which
they can establish the necessary cultures of food or other functions. In at least the genus Carebara, the
flying queen was said to take along a few tiny workers that cling on to the
huge flying queen as best they may. No
doubt they are valuable, possibly absolutely necessary, in establishing the new
colony. Since this was first written, doubts have been raised as to the
validity of this observation, but the concept is interesting enough to bear in
mind.
Such
examples of behaviour have effectively nothing to do with the individual
intelligence of the queen founding the new colony.
Not to
labour the point too heavily, this all amounts to intelligence in the species,
if not always in the individual. In
their respective ways, the colony, the species as a whole, and individuals, may
each be seen in the light of dedicated, task specific, but flexible information
processing structures.
The work of construction seems to be a sort of balance
struck between many bees,
all instinctively standing at the same relative distance from each other,
all trying to sweep equal spheres, and then building up, or leaving ungnawed,
the planes of intersection between these spheres. It was really curious to note
in cases of difficulty, as when two pieces of comb met at an angle,
how often the bees would entirely pull down
and rebuild in different ways the same cell,
sometimes recurring to a shape which they had at first rejected
Charles Darwin. On the Origin
of Species
The next
phase, after establishing the nest, is maintenance. In many solitary species, that need does not
arise, because the adult abandons the nest after completing its structure,
provisioning, and populating it.
However, there also are many species that build just one nest or burrow,
in which they make separate cells that develop over an extended period. Such homes may need a fair amount of
maintenance during their period of occupation.
However, it
is mainly the social or eusocial species that really invest heavily in nest
maintenance. Some species of termites
are arguably the most spectacular examples.
Though it is hard to get precise data, the mounds of some species may
weigh tonnes and remain in use for more than one century. Probably most species’ mounds are in use for a
decade or more. One of the major factors
determining the longevity of a mound that survives the thousand natural shocks
that termite mounds are heir to, is the longevity of the queen, and if that is
too limiting, then of the secondary reproductives that finally replace the
female that founded the nest.
During its
occupation a mound is constantly remodelled, with most of its concrete-like
structures being temporary. The most
enduring parts are the outer walls, but even those get remodelled. In the building season, usually after rain,
worker termites tunnel to the surface and begin to construct new, soft,
thin-walled chambers on the outer wall.
As these new shells harden, they get reinforced till they become the
lasting new outer walls. Then the old
thick walls, progressively softened by the moisture of the internal
environment, get perforated with new chambers and tunnels. They soon become indistinguishable from the
rest of the internal spongy structure of the nest.
The mound
also gets restructured internally as needs change. Sometimes changes adjust the overall shape of
the nest, for instance, in compass nests.
Sometimes there is tunnel digging or tunnel building to look for food or
water supplies. In some regions termites
tunnel so deeply for water, tens of metres at least, that minerals in the walls
of their mounds have been used successfully for mineral prospecting on an
industrial scale.
Furthermore,
there may be substantial remodelling to establish or adjust food storage or
other structures, such as fungus chambers and living space for queens, eggs and
young. The queen in particular cannot be
shifted without a lot of serious restructuring of partitioning and
passages.
These last
examples are particularly significant in the discussion of nest
intelligence. They entail both
homeostatic adjustments to the internal environment, and modifications to
accommodate new or seasonal requirements.
In other words, to meet the needs and “comforts” of their various
subpopulations, they require the workers to respond both to the internal and
external environments, and to apply the appropriate negative feedback in their
construction.
Nor are
these static requirements. The
homeostatic requirements of the colony change all the time, according to
season, weather, food supply, threats, damage, or changes in the surroundings,
and anything of the kind requires building or remodelling activity. It is easy to imagine the workers migrating
from one part of the nest to another as things become too hot, or cold, too
soggy or too crowded, but they do more than that; they move eggs and food, and
thicken or replace walls to suit other things than just their own comfort. In those species that have air conditioning
requirements, the maintenance can be quite obscure, including convection
chimneys whose functions consciously intelligent human naturalists simply
failed to understand at first.
Special
reactions are necessary on occasion, for example when it is time for the mating
flights of their reproductives. These
are particularly demanding, because they should coincide with suitable weather
and the mating flights of neighbouring nests.
The more neighbours that fly at the same time, and the shorter the
period during which they emerge, the better the chance for outbreeding, and the
better the chance that the flying termites will survive by overwhelming the
predators that gather to feed on them.
As an
intelligent information processing system, such a colony compels respect,
whatever the intelligence of its individual members, or whether or not the
colony embodies some sort of mystical consciousness. However hard this might be
to believe, or provide for the existence or nonexistence of, bear in mind that
we still are unable to do the same for humans, whether individually or in
communities.
Ants are
not far behind termites in these respects, and various species probably must
accommodate even wider ranges of needs of tenants, eggs, larvae, pupae, and
inquilines such as aphids. Some of them
also must suit the needs of the plants they live in. In all cases they do not place items
arbitrarily. Unlike termites, they have
immobile pupae to care for, and practically immobile larvae as well. Some species also have repletes, gorged
workers that form effectively a separate caste.
The job of repletes is to store food, and they hang upside down with
bulbous abdomens filled with syrup. Once
replete, they too are practically immobile and must remain in a part of the
nest that keeps them healthy. And like
some species of termites, some species of ants rely on fungus gardens that
require special climatic conditions inside the nests. Ants apply less obvious air conditioning
engineering than termites, but some species probably do apply the venturi tube
principle to air exchange.
Eusocial
bees vary in their nesting behaviour.
Most hives are in hollows in rocks, earth, or wood, but some build combs
in the open. However, the hive does
require adaptive climatic control depending on the region. Such control can be quite sophisticated. In cold weather workers ball around brood and
the queen and vibrate their wing muscles to warm the core of the hive. In hot weather they not only have teams of
bees that stand in the entrance and fan fresh air into the hive, but water
carriers will collect water and evaporate cooling drops of water in the current
of air.
Bees are
less assiduous constructors than termites, but mainly from plants, they do
collect gummy, resinous materials. These
they knead into “propolis”, a sort of all-purpose sealant and adhesive in
adjusting the hive’s structure and layout.
Using the propolis, they are likely to narrow the entrance to the hive
to a comfortable size, and some, especially species of stingless bees, may form
it into a rainproof, wasp-proof, intruder-proof spout as a nest entrance.
The combs
themselves, marvellous as their construction is, seem to be rather featureless,
rather an exercise in industrial uniformity than anything else, but this is
misleading. At the edge of the combs one
finds uncompleted cells, and they show irregularities in depth and thickness of
their walls. To complete the
construction takes a great deal of remodelling, as described beautifully by Darwin a century and a
half ago. Larger cells also are built to
accommodate drones, but they are sufficiently close to the normal size of comb
cells, that it is easy for the inexperienced eye to overlook them. Queen cells on the other hand, are much larger
and of a totally different shape. As usual,
the variety and complexity within the hive’s repertoire is astonishing, even
when one examines something that looks simple or uniform.
Furthermore,
the combs of honeybees usually are not hung randomly in the hive, but parallel
to each other. When this fails, if combs
meet at an unsuitable angle, a great deal of reconstruction or reorganisation
results. It is another example of where
the logic of thresholds and majorities settles things democratically. Bees will take wax from the competing comb, and
use it in their own constructions. This
usually leads to a lot of wasted and repeated labour before calm returns.
In a
sufficiently prosperous hive, the bees also will construct combs in every gap
large enough to take a comb plus workers walking on it. This complicates the construction of
artificial hives. They may look like
simple wooden frames in a wooden box, but in fact they are very precisely
constructed with spacers and measurements that will suit the bees’
comb-building tastes, without interfering with the apiarists’ harvesting and
husbandry. If the spaces are too wide,
the bees build combs in positions that cause jamming and waste; if too small,
they refuse to build and may abscond in search of a more suitable hive.
When one
goes further and looks not only at the combs, but also their contents, more
complexity emerges. They are not only
used for brood cells, in which the larvae and pupae develop, but also for
storage of honey and for pollen.
Typically the brood is in the middle of the comb, the pollen is stored
around the brood region, and the honey around the edge. This means that if the bees have to warm up
the hive, they can do it round the brood in the middle. Around the brood the pollen will come to no
harm, while the honey round the outside is least prone to become too warm and
runny.
Space and
time prevent anything like a coherent comparison of the various kinds of combs,
ranging from the single paper combs of wasps, to the waxen combs of bees. A point of interest is that though the
honeybees have the most precise constructions, and though they suffice for all
their needs, except arguably for queen cells, the various kinds of stingless
bees make totally different separate honey pots in which they store their food.
Maintenance
and management of the colony comprises whole ranges of information processing
controls and functions. There is the
question of the homeostasis of the environment necessary for the colony members
and their functions. There is the
homeostasis of the nest materials and defences.
There is management of the costs of maintenance, of the efficiency of
foraging and reliability of defence.
The nest is
not the only aspect of colony homeostasis.
There also is the matter of population homeostasis. A colony with the wrong constitution of
castes will be inefficient, insecure, or ineffective, perhaps all three at
once. A colony that forages too little
in times of shortages will starve. If it
forages too much in times of plenty, the food will go to waste.
In honeybees
for instance, the hive workers begin to refuse to accept new nectar once the
storage space gets too full. For one
thing, if they cannot find space to store the content of their crops, they
simply do not have room to take on more; their crops become de facto storage
depots, much like the repletes of honeypot ants, though on a smaller
scale. This is one of the controls on
foraging. Such control may redirect
foragers to find water or propolis or return to other hive duties.
Meanwhile
the bees with the full crops are likely to metabolise the nectar they are
carrying and convert it to wax. The wax
goes into building new combs if there is room, which in turn makes more room
for what the foragers go out and collect.
After all, depending on the quality of the honey, it might take over ten
parts of honey to supply material for one part of wax.
This is a
neat feedback system, but there is another dimension. When the nest gets too full for new combs,
that is one of the stimuli that sends scouts out to look for new homes for
swarms.
The ranges
of community intelligence required for all these functions, and for invoking
the functions in response to the appropriate circumstances, is not only
imposing, but repeatedly astonishing.
They not only are astonishing in themselves, but in the apparently
mindless ways in which such complex tasks and programs can be implemented.
The implications for
cogent interpretation and classification of colony intelligence are troublesome
at the very least.
The sum of the intelligence on the planet is a
constant;
the population is growing.
Anonymous
A major,
even a limiting, factor in the development of sociality as an ESS
(evolutionarily stable strategy) is the conflict of interests that arises as
soon as there are distinguishable participants. Distinction in this matter might most
obviously refer to genetic or psychological differences. The latter are mainly obvious in large
brained creatures such as mammals, but subjectively it is not easy to distinguish
between conflicts of personal interest in queen bees, in fighter fish, and in
mammalian sibling rivalry.
Essentially
in the sense relevant to this essay, the concept of conflict of interests
arises only when the distinct attributes of conflicting parties are in some
sense transmissible, in particular heritable.
Throughout
evolutionary history genetic differences have been the most fundamentally
important, though genetic rivalries often manifest themselves as personal
conflict. Investigations in the last
century or so, and especially in the last few decades, have presented arguments
for pervasive competition within communities, within families, between
reproductives and their offspring, even within bodies.
Most
prominently, within Hymenopteran communities such as hives or colonies, there
is evidence for systematic competition between queens and workers, and between
workers and workers. The argument hinges
on the differences in genetic relatedness between hive members. These arise out of the fact that in the
relevant species Hymenopteran males are haploid and the females diploid. However, such fields are hazardous if one
takes them simplistically at face value.
For example, the view of worker bees as competing with their queens for
reproduction because of being less closely related to her than to each other or
their individual offspring, is open to question even in the face of numeric
predictions.
The most
serious problem is that no numerical argument has much cogency if it cannot be
shown to have appropriate isomorphism to the system it describes or
models. What is the difference between
the importance of genetic degrees of relationship within a colony, and the
degrees of importance of the relationship between the cells in one’s various
body parts? Unless the worker favours
the generation of reproductives more closely related to herself than to the
offspring of the queen or those of other workers, there is no evolutionary
advantage to frustrating the queen in any way.
There remains room for further analysis.
In say,
citrus and dandelion species, where, rather than zygotes arising from mating,
maternal tissue most often produces cells that give rise to the next
generation, one might see no conflict of interest between parent and young,
except that the parent might “prefer” to produce flexible generations that are
not parental clones, rather than only dead-end parthenogenic offspring. Some animals, such as many species of aphids,
do reproduce both parthenogenically and sexually in due season, and in fact do
very well. Citrus produces largely
parthenogenic seeds from maternal tissue, plus a modest proportion of sexually
produced seeds. It is semantically
hazardous to interpret such examples simplistically in terms of struggles
between genetic lines.
In humans
conflict of interest arguably is more elaborate than in any other species;
there is conflict of interest between cells and the organism that they
comprise, between individual and individual, between family members and
families, between factions within communities, between mind and genome, between
mind and body, between individual and nation, between individual and ideal, or
even between individual and species.
Arguably some of these conflicts entail major evolutionary penalties,
such as when a self-indulgent person of high intellect or other desirable
attributes chooses not to reproduce or becomes a parasitic despot, mating with
large numbers of sexual partners.
Much
argument and research has been expended on demonstrating the evolutionary
justification, or advantages, of particular behaviour patterns of
nonreproductives. Although such
arguments look very attractive in considering potential or prospective
reproductives, such as subordinate females in naked mole rat or Polistes wasp
colonies, they lose a great deal of persuasiveness in application to sterile
worker or soldier castes. Such
nonreproductives have little option for selection within the colony, and their
best strategy, insofar as the concept applies to their role, is to support the
reproductive success of the colony.
Conversely,
it is much easier to make sense of the situation from the point of view of the
evolutionary strategies of a queen or a dominant female. Even if all the offspring share all her genes
(in the ordinary sense, a most unusual situation) it might be more profitable
reproductively, for a queen to dedicate some of her offspring to assisting in
the production of larger numbers of more reproductively viable offspring,
rather than futilely maximising the number of zygotes produced in the short
term. There are whole ranges of
circumstances in which this is far more evolutionarily competitive. In this respect the strategy of a queen is
closely analogous to that of the zygote of a metazoan animal. Naively it might seem to make more sense for
the cell to feed and make more daughter cells to reproduce in their turn as
quickly as possible, but in reality this is neither practical nor effective for
competing with unicellular organisms.
Instead it pays to establish a smaller number of multicellular organisms
that can compete on a more favourable footing.
What
happens in practice is that some of the daughter cells become somatic cells and
never reproduce. Suppose that the need
is for one cell to split mitotically, such that one daughter cell becomes a
reproductive stem cell and the other a somatic cell. Which of those two daughter cells should
choose to be reproductive? The question
is hardly meaningful. The cells are not
genetically distinguishable and if both became either somatic or reproductive
then both would die without offspring.
Consider for example whether a muscle cell should try to favour its own
life or that of its daughter cells over the reproduction of other bodily
cells.
Now, that
muscle cell had had no say in becoming muscle instead of gonad. Having been born in that situation, it is
meaningless to speak of its options to favour its own type, and it would remain
meaningless if that cell were part of a genetic chimera, which is what many of
our bodies turn out to be in practice. A
surprising number of us are in fact mosaics of the cells of more than one
zygote. Would it make sense for those of
our cells whose genotypes are not represented in our gonads, to turn cancerous
so that they could produce more of their own type?
Similarly,
sterile nonreproductives in a colony may be seen as somatic cells rather than
reproductive. They can reproduce more of
their genes by assisting in the rearing of related offspring and feeding and
protecting related reproductives, than by favouring themselves. In fact, if it were possible (which it seldom
is) for a lost nonreproductive to join and assist a neighbouring colony of the
same, or even of a similar, species, that would do more for propagating its
genes than just dying. Related species
share by far the most of their genes.
Such lines
of thought lead to tricky problems when one considers behaviour such as slavery
(“dulosis”) where ant species capture pupae of closely related species. Workers that emerge from uneaten pupae then
join the colony as fully conforming members of the infrastructure. In extreme cases, as already mentioned, they
make up effectively the entire worker caste.
What evolutionary pressures apply to such slaves? Logically they should sabotage the nest of the
slavers, thereby favouring their own species, but that is not an option open to
them, given their phenotype (extended or not, depending on the point of view).
One
could see this in terms of rivalry between the respective intelligences of the
slaver and slave colonies or species; how the situation develops in terms of
long-term evolution of the slaver and slave species is another matter: any form
of specialisation in evolution is suicidal in the long run, however
opportunistically effective it is in the short-to-medium term, because the
specialist falls hostage not only to its own attributes, but to the survival of
the entities on which it becomes dependent.
What
one might in such concepts see as something like evolutionary “intelligence”
would thereby seem to be self-defeating in the long run.
Turn on, tune in and drop out.
Timothy Leary
This
subject is far too large for serious attention here but it is important enough
to demand cursory attention.
Both
communities and their
individual members as as entities, are subject to immense ranges of
pathological behaviour, some of it interestingly reminiscent of the behaviour
of creatures with integrated brains.
Drug addiction is one example, most notoriously in human communities and
in ant colonies that have been invaded by various parasitic inquilines. For example, larvae of some Lycaenid butterflies (“blues”) enter ant
nests and secrete narcotic substances that the ants eagerly consume. It paralyses them, and the butterfly larvae
then eat either ants or their larvae, depending on the species in
question. From the caterpillar’s point
of view the animal food is more concentrated nutrition than plant material, and
many Lycaenidae have largely become dependent on it at certain stages of their
life cycle at least. Some ants actually
feed the larvae to the caterpillar. This
suggests that the caterpillar gives a particular pheromonal signal that moves
the ants to do so. As a speculation I
conjecture that it might be a signal that normally stimulates the ants to move
the larvae about the nest.
Just as in
human communities, a heavy infestation of narcotic suppliers can have a harmful
effect on the community.
Actual
poisons also may cause behavioural pathology.
Apart from simple toxicity, in particular neurotoxicity, mercuric
chloride poisoning provokes at least some species of ants, such as the
Argentine ant, to begin fighting their nest mates. This is interesting because Argentine ants do
not normally fight other colonies of the same species, let alone nest
mates. Patently one pathological effect
of the neurotoxin affects the social signalling and feedback mechanisms of the
ants.
The most
pernicious and varied of such aberrations, with or without the mediation of
drugs, occur most obviously in human society.
Examples include faction fights, riots, and sport hooliganism.
There is
not much evidence of play behaviour in invertebrates, not even in bees, whose
workers go through various roles during their lives, and learn part of their
skills on the job. Offhand I can think
of no example of anything resembling sport in invertebrates, nor even play in
general, except in some Cephalopoda.
Bees and ants do have factional disagreements of a sort, where there are
stimuli to change nests or split into new colonies, but these are generally the
natural mechanism for leading to what amounts to decisions on such matters.
Such disagreements are more like eusocial democratic disagreements in the
colonies, than like social pathology.
The reason that group disagreements often are pernicious in human
society is that human society and communication are so much more complex and
less specific than in any other species.
Such complexity gives more scope for the development of social pathology
such as lynching or hooliganism.
Stress
certainly can affect the behaviour and the health of any colony. A colony that lives in a disturbance-free
environment is likely to be less easily aroused and less aggressive than after
regular disturbance. Part of the reason
is perhaps because of the persistence of traces of alarm pheromones after
disturbances. Part might be because of
accommodation of the respective nervous and hormonal systems of the colony
members to constant arousal. In this
respect colonies behave in ways that closely resemble the behaviour of most
animals, whether colonial or not.
Interestingly,
the reactions of many colonies, as opposed to their individual members, to some
kinds of disease or parasitism also resemble those of individual animals. In beehives for instance, some kinds of
infections cause the bees to clump and increase their temperature by vibrating
their flight muscles. The effect is much
like a fever. Fevers are surprisingly
ubiquitous as a reaction to disease in the animal kingdom.
For the same thing that might, perhaps with some
reason, seem
very imperfect if quite alone, may be very perfect in its nature
if it is looked upon as part of the whole universe.
Descartes
It is by no
means clear in what ways the intelligence of a community is commensurable with
that of its members, or even in what ways the two are the same type of
thing. Apart from communities of cells
that form complex organisms with the most complex of central nervous systems,
we have no clearly recognisable, let alone unambiguous, examples of
consciousness, let alone self awareness in colonies.
And yet the
nature of collective behaviour seems almost inseparable from collective character. Every species of tree has its characteristic
shape, often ineffable, but positively shouting its nature at those with eyes
and experience to see. A naturalist at
home in his territory often can tell the species of a tree where he can make
out no more specific detail than its motion in the wind. He might see a flying bird at distances where
he cannot make out anything at all except the motion and rhythm of flight, the
area and direction of its course, and yet recognise not only the species, but its
probable activity as well.
As I shall
point out again, these characteristics are not random. Their causes may not be traceable in
detail, but like an individual’s voice,
gait, and handwriting, they are real and they are the consequences of informational
states within the colonies of cells and within the individuals that constitute the colonies.
Increasingly advanced observations of the apparently functional intelligence of colony members belie traditional views of the behaviour of say, insects within colonies, as that of pure automata. For example, bumble bees have been trained to perform various tasks for reward, or to avoid noxious stimuli, and to cooperate with partners to perform tasks that they could not perform alone. The objects that they cooperated on were not such as to occur in nature, such as objects to push or doors to open. Significantly, untrained bees failed to cooperate with each other, whereas a trained bee was seen to retreat from such a challenge when it encountered it alone, but to turn around and approach the task again when it encountered its companion on the way in. One such study has been published at: https://doi.org/10.1098/rspb.2024.0055
The
intelligence of individual colony members obviously forms the basis of some
differences between the behaviour of communities of insects and of birds, jail
or otherwise, but it does not do to be too dismissive of colonial
intelligence. As far as we can tell,
after all, the intelligence of an individual bird stems from the behaviour of
the communities of neurons in their brains.
Those individual neurons in turn are less impressively intelligent than
individual bees or ants.
There are
many ways in which it is very difficult to say how to compare individuals with
the communities in which they live, or with which they deal. For instance, one normally expects the
individual to adjust a great deal more quickly to intellectual pressures, while
communities often have to wait for the powers that be to be replaced before
there is any effective change of mindset.
And yet, one must be cautious in condemning the relative hidebound
stupidity of the group. For one thing,
there is a certain robustness of a culture that can call on the greater
redundancy of multiple members to substitute for each other.
Then also,
temporal factors are not always a matter of “fast mind good, slow mind
bad.” Many activities of long-lived
communities or species are hardly perceptible to the faster, but not
necessarily more complex, minds of their members or observers.
It is easy
to think of examples. What good would
quick thinking do a tree or a limpet?
For a slow mover to profit from from thinking faster than it can react
to a situation, would require foresight or clairvoyance rather than
intelligence. It has been suggested that the outbreak of World War I was
largely the result of newly developed high-speed of communication technology
outstripped the ability of the great powers to adapt to emerging circumstances.
Modern
human politics is rife with examples where individuals lead their communities
to exploit the weaknesses of other communities, then fail to recognise when the
victim communities, typically on a time scale of decades, adjust so effectively
to the challenges as to turn the tables.
It is an open question how far these contrasts reflect different
complexities, different speeds, or even different types of intelligence.
The intelligence of any discussion diminishes with the
square of the number of participants.
Adam Walinsky
Since
intelligence is a multidimensional variable and we have no precise and unambiguous
measure for most of those dimensions, interspecific comparisons of intelligence
are suspect. Often they apply to such
narrow aspects of behaviour as to be practically useless for comparisons of
general intelligence. Also, there has
been little work on the intelligence of individual members of eusocial colonies
of any species, and less on comparisons between such individuals and the
individuals of related species that are not eusocial. It is in any case difficult to investigate
the intelligence of insects meaningfully, and the very concept of the relative
intelligence of eusocial individuals is extremely tricky.
It is open
to speculation whether there are at least two classes of intelligence in
colonies. One type might be where the colony has a level of intelligence of a
higher order than any of its members, such as possibly a termite colony, or the
intelligence of a human body relative to its barely intelligent component
cells. Another type could be where the
intelligence of at least some individuals within the colony is greater than
that of the colony, such as when some few of the citizens of a nation can tell
when the state is steering on breakers, but cannot do anything effective to
overrule the mob.
Whether a
third type would be where the intelligence of the most intelligent member
determines the intelligence of the colony, I cannot guess. Comparisons of different types of
intelligence is a very moot problem.
Though I do
not go into an analysis of the point (I do not know how to!) there are some
deep difficulties to distinction between communities with intelligence based
primarily on the structure of the community, and those with intelligence based
on the intelligence of subcommittees or of individual members. As humans we might well be biased in favour
of the latter, but then human communities should be far and away more
intelligent than those of other species.
In practice human communities do not put up a very impressive showing in
comparison to other communities. In
particular, the superiority in intelligence of human communities are nothing
like as radical as the comparison between the intelligence of individual
members would lead one to expect.
A jaundiced
view might suggest that the intelligence of human communities rather reflects
the nature of the most selfish, obstructive, cross-grained, dominating,
parasitic, vindictive and malicious of their members, rather than the most
intelligent.
This is not
necessarily self-indulgent, simplistic misanthropy. Humanity is adapted rather to life in small
groups where everyone knows everyone else, and everyone knows who is the most
muscular. Instead we live in faceless
masses in which social parasites can prey on socially supportive members.
Structures
of unselfish, mutually beneficial and supportive social interactions are
metastable in most circumstances because they are vulnerable to
parasitism. Among humans it seems to be
easier for betrayal to change them into distrustful, resentful, selfish relationships
than the other way round. In human
families and intimate communities this can lead to intense embitterment. In larger, anonymous, urban communities the
personal embitterment may be less intense, but alienation, distrust and
selfishness may become a way of life if the powers that be are not sufficiently
firm and far sighted.
And if
humanity wipes itself out, that is likely to be how it happens, when demagogues
steer nations as mobs.
These are
not problems that arise in most social animal communities. In eusocial communities the necessary
structures just do not exist. Not that
one gets no strife in such communities, even deadly strife, such as queens
destroying queen cells or younger queens, but such things are totally
impersonal competition. In a healthy
hive they do no lasting harm. There is
no indignation or obvious resentment.
In
contrast, in many species of monkeys, biting is the standard form of
chastisement, particularly of subordinate by dominant troop members. However, in many such species the canines are
large, dangerous weapons, and disciplinary nips involve only the incisors and
are not much resented. Biting with the
canines on the other hand, invites instant retaliation, much as biting or
kicking might do on a first world school playground where there are conventions
for the use of fists above the belt only.
Such
conventions for the control of social interactions in communities are necessary
in species in which personal recognition is essential. In eusocial species this is not a
consideration except for such things as recognition of the fact that
individuals share the same hive smell, or possibly that worker bees share the
same floral smell, or things of that type.
In human communities there are subcommunities such as families, gangs,
neighbourhoods, and the like. These may
overlap or even demand conflicting or contingent loyalties.
Loyalties and intergroup enmities are among the most important and
basic controls of mammalian or avian social interactions, and in humans they
are the most abstractly and variously developed. Without following this line of thought too
far, note the way that neighbouring gangs, schools, or sporting teams and their
supporters will treat each other with pointless, automatic resentment rather than
amiable rivalry, but when they are pooled for regional competition, the
supporters unite almost seamlessly and almost unconsciously.
Such
hierarchical resentment and loyalty may be traced on all scales up to the
international. It is widely, almost
automatically encouraged in education, but repeated experimental investigation
has demonstrated that it also is innate and arises in totally meaningless and
artificial associations even between strangers.
It makes patriotism one of the most powerful and resources of the
cynical demagogue, not, as Johnson said, the “last resort of a scoundrel”, but,
as Bierce observed, “the first”.
Simply put,
the population is a fertile medium for such exploitation. George Orwell reacted to such facts with the
remark: “Serious sport has nothing to do with fair play. It is bound up with
hatred, jealousy, boastfulness, disregard of all rules and sadistic pleasure in
witnessing violence. In other words, it is war minus the shooting”. It is
hard to deny his point when one contemplates the last two centuries’ international
sport events, but actually, such feelings and activity were remarked upon in
classical times.
I do not
intend this as an irrelevant Jeremiad or satire, but as an illustration of a
mechanism of control and characterisation of community interactive control. It depends on innate attributes of the
community members, is rarely recognised consciously as such by participants,
has little to do with any participant's conscious intelligence, and yet
controls the bulk of the population without appeal, often at the whim of ruling
subpopulations, cliques with their own internal loyalties. It has been like this for all human
history. “All were for the party. None were for the state.”
Furthermore,
the smaller the human community, the more directly the intelligence of the
leaders is likely to be reflected in the policy of the tribe. In large nations one is likely to encounter
the curse of the committee, encapsulated in many a quip about the standard of
the intelligence of any community varying with some inverse function of the
number of members.
Quips are
all very well, but there are material reasons for taking such relationships
seriously, and some of the reasons favour neural networks in modular brains on
the one hand, and on the other, eusocial insect societies. They may not be as intelligent in some
senses, but they also interfere less with each other’s intelligence. Each member of such a community simply reacts
to neighbours’ behaviour for local decisions, and to major signals such as
sounds or pheromonal or hormonal diffusion for long distance signalling. The signal may be unable to carry many bits
of information and accordingly it may be simplistic, but that need not be a
problem if the repertoire of reactions is small and suited to the community's means
of dealing with its environment. When
that is the case, then the effect is for the community to display startling
sophistication in its reactions as an entity.
When human
communities try to give everyone a voice and to consider each voice before taking
action, then everyone (in theory at any rate) should hear every opinion, weigh
it and combine it with every other opinion, and finally produce a digested,
hopefully optimal, conclusion as a basis for the policy to be implemented. In practice the problems of choosing the
right leader in each decision, the quadratic growth of two-way communication,
and exponential growth in the number of subcommittees, the increased path
length of lines of communication as the community size grows, the intellectual
limitations of certain members, the disagreements among others, the conflicts
of interest, lead to appalling results.
The more
passionate the sincerity of the reformer the worse the results one may
expect. As John Gall said in his
brilliant book “Systemantics”: “Reformers blame it all on "the system" and propose
new systems that would, they assert, guarantee a brave new world of justice,
peace and abundance. Everyone, it seems,
has his own idea of what the problem is and how it can be corrected. But all agree on one point: that their own
system would work very well if only it were universally adopted.”
In practice
no system of government follows such extreme logic. For one thing our species is not as well
adapted to such innate controls as the eusocial colonies are. Pretty soon, no matter what bliss it might
have been in such a dawn to be alive, it becomes necessary for the leaders to
reduce their discussions to cabinet level, and the inevitable slide into indirect,
“democratic”, party domination or even despotism, begins.
In short,
there is not yet much foreseeable hope for communities of intelligent
individuals being unambiguously more intelligent than the communities of
unintelligent, or at least less intelligent, individuals. The participation of too many independent
brains exacts crippling penalties. Their
mutual interference might not affect their respective effectiveness, but often
it certainly affects their communal effectiveness.
Still, the
intelligence of community members does have some effects. Social structures based on more stereotyped,
genetically determined behaviour are more subject to direct evolutionary
selection, because counteradaptive behaviour carries a more direct penalty for
the genes responsible.
In contrast
communities of organisms with more flexible behaviour might be dysfunctional
for contingent reasons that are not a direct function of a particular heritable
trait. Although they could not escape
the consequences of severe dysfunctionality, they would not be not subject to
the same precision of adaptive selection, but the consequences of the looser
selection should include a wider range of social behaviour patterns.
Such
variability would increase as a function of the intelligence of the individual
members of the communities. There is
some support for this among primates, but far and away the most spectacular
variability is in human communities.
Apart from
the variability of human communities, there also is greater complexity of
community structures and activities.
Some of those forms of interactions could be seen as components of
community intelligence that are as different from the community intelligence of
eusocial insects, or naked mole rats, as the brain of a primate differs from
that of a fish. The major part of the
primate brain consists of extensions to the brain stem that simply do not occur
in fish or amphibians. Analogously,
non-human eusocial communities have nothing that corresponds to business,
religion, art, scientific disciplines or many other activities that deeply
influence human societies. It is hard to
see whither such developments will take human societies, but they are
materially different from any components of community intelligence in extant
eusocial species.
There have
been some speculations to the effect that eusocial individuals would be
unusually intelligent in specialised activities appropriate to their castes,
and unusually unintelligent in other respects, as compared either to other
castes or to non-specialised individuals.
Sometimes this seems to be so, and certainly the brains of some eusocial
insects have local enlargements that differ both from the brains of nestmates
of other castes, and from the brains of related non-eusocial species.
This is
very suggestive of course, but it is not much more than suggestive. The ratio of brain size to body size is a
useful rule of thumb for estimating intelligence in mammals, and it surely is
not without significance in other animals, but it also is highly imprecise and
untrustworthy as a measure of intelligence.
And in any case it is not based on any clear functional
relationship.
It is not
at all clear that the size of the brains of say, crows and parrots, gives a
fair indication of their intelligence compared to that of mammals of similar
brain to body size ratios. It is quite
possible that different neural organisations might give greater compactness of
brains with competitive levels intelligence.
Pigeons, not normally regarded as mental giants, have demonstrated
behaviour that suggested startling intelligence by mammalian standards. (Of course, we must be cautious. It is not many decades since the high
intelligence of pigs was formally recognised, and anecdotal evidence suggests
that black rhinos are disconcertingly intelligent.)
At the same
time, absolute brain size remains inescapably important. Obviously no single neuron could have a very
high IQ, but less trivially, a little girl with a particularly vicious case of
persistent epilepsy had “half her brain” removed (presumably a defective
cerebral hemisphere). She was very young
and everyone was delighted with the speed with which her half brain adapted and
took over the running of both sides of her body, leaving her superficially normal
in most obvious ways. Subsequently however,
it seemed that there were in fact limitations to her learning ability, as
though her brain just were not large enough.
Anyway, it
cured the epilepsy.
This case
is of course very far from cogent, being abnormal, probably pathological,
without controls, and open to rival interpretations. However, it is sufficiently suggestive to
justify suspicion of clichés such as that humans use just X% of their brains,
where X an arbitrary small number, varying according to the source of the
assertion. Such clichés too, are very
far from cogent, being abnormal, probably pathological, without controls, and
open to rival interpretations.
Anecdotal
evidence of purposive behaviour suggests that some very small-brained animals
are disproportionately intelligent in particular ways related to their
adaptations to their ways of life. For
instance, I have personally observed that Cape
chameleons (Bradypodion sp.) can rapidly trace an unfamiliar, unobvious
and circuitous path to a perceived rival several metres away, and follow the route directly without
preliminary trial-and-error exploration.
This is far beyond what say, a dog, with a brain both absolutely and
proportionately larger, could do. It is
an innate skill that might reasonably be associated with short range
navigational problems commonplace in shrubbery.
Taking this
even further, I have been astonished to see a noctuid moth (Sphingomorpha sp.) fly through an open window into a lighted
room at night, and when it was alarmed while still in flight, turn without
alighting, and unerringly navigate its way out again by the same dark window a
few metres away. It was one window of
several. This was astonishing behaviour
in an alarmed insect at night; as a rule one would expect automatic flight
towards or around the lamp. Even diurnal
birds alarmed at night will usually fly blindly towards a lighted sheet. What passed for mental processes in the moth,
I cannot say, but the subjective effect was that it gave an impression of
surprising purposive intelligence.
Interestingly,
in the laboratory at least some species of eusocial insects, particularly bees
and ants, show intelligence thoroughly comparable with other insects. This recalls Lorenz’s opinion that social
animals are more intelligent than solitary species because they need the extra
facilities for social interaction. He
mentions foxes as being less intelligent than wolves for example. There is room for argument on either side of
this point. I mention it just for
consideration in context.
It also is worth
noting that the brains of many species of worker ant are exceptionally large
relative to the animals’ size. I have
already mentioned some reservations on the significance of this ratio, but it
is interesting that their maze running behaviour is only a few times less than
that of a rat. Significantly however,
unlike a rat, they do not obviously benefit from having previously learnt the
maze when they have to learn to run it in reverse.
Then again,
both bees and many ants are good at navigating their way back after a
convoluted outward foraging sortie. What
is more, species that do not follow trails routinely, tend to follow a direct
route back to the nest, often amazingly accurately. Even the trail-bound species tend to shorten
a long-used trail progressively by cutting corners.
Many
insects show elementary learning abilities when confronted with the standard
laboratory challenges, maze learning, Pavlovian responses, and the like. However, insects are usually not able to
generalise from learnt experiences, much less show any comprehension of the
tasks they perform. Still, occasionally
there is an innate ability to perform surprisingly sophisticated tasks when the
species is adapted to particular demands.
For
instance, bees, with their need to find flowers, and, having found them, locate
or evaluate the nectar and pollen within them, are surprisingly good at pattern
recognition and memory of patterns once they have found food associated with
them. They also can to some degree
recognise categories of patterns, such as the same pattern in different
colours. Another very impressive feat is
a worker bee’s ability to find the direct way back to the hive when having
found food after a convoluted search.
Especially
in more stereotyped, but apparently intelligent behaviour, one is at a loss
whether to ascribe intelligence to the species, the colony or the
individual. Consider the ability of some
species of ants and bees to navigate by the sun. Is this their own intelligence, or that of
the colony? In the latter case we should
regard the “intelligence” of the individual ant as being analogous to the
intelligence of an optic neuron in the human eye, specialised to identify a
vertical or horizontal edge. The ability
as such is trivial, but it gives great power to the intelligence of the
colony.
The local
enlargement of regions of the brains of various castes of insects is of course
highly suggestive. To the extent and in
the ways that intelligence is a function of the amount of brain tissue available
to support it, that phenomenon makes it extremely likely that such insects
differ in the degrees and types of intelligence they exercise.
There is
more to it than that. Not only are all
the specialisations within a given species the product of effectively the same
genome, but in some species, such as eusocial bees, the same worker undertakes
different duties at different stages of its life, in a more or less fixed
sequence. In the case of the honeybee,
which probably exhibits this effect in its most advanced form, a typical
sequence would be something like: nursing older larvae, nursing younger larvae,
receiving and processing nectar from foragers, housekeeping and cleaning
functions, comb maintenance, comb building, water carrying and foraging.
Such
functions vary with the colony’s requirements, and often workers differ both in
the duration of each duty in the sequence and in the degree to which they
deviate from it. Patient observation has
shown that many bees and other social insects seem to spend a lot of time
apparently loafing. This should not be
taken too seriously, on the principle that they also serve who only stand and
wait, and idle bees consume little food.
Furthermore, during the hive phase of the life of a mature bee, once it
is say, a week or two old, it “patrols” the hive, apparently looking for odd
jobs. It deals with each job as it
encounters the need for it.
Even some
of the jobs that sound very organised, are tackled in a very arbitrary
fashion. For example, comb building, that
most precise of the engineering tasks that bees undertake, begins by the
deposition of tiny pellets of wax where some of the bees happen to leave
them. Other bees might pick up the wax
or add to it according to individual reaction.
As work progresses, more and more bees join in, adding, removing, and
sculpting wax. It begins crude and
thick, and ends up exquisite and precise.
However, it
is not necessarily consciously well-planned; where a swarm rests for hours or a
day or two, in a solid ball in a bush before taking off for the permanent new
hive, one often will find that they leave behind scraps of wax where the
building of a new comb had begun prematurely. Possibly that might be seen as an
example of displacement activity.
This
illustrates principles that are very important in many ways. The fact that a large part of the colony’s
survival and efficiency depends on such ad hoc recognition by individual
workers, of when there is work to hand, shows how complex a society can be
maintained by the apparent whim of individuals.
A great deal of the activity of humans in a free enterprise system is
strongly reminiscent of that mechanism.
It would be
interesting to see whether it would be possible to have a lot of human workers
build a structure or work of art out of say, clay or wax, or computer graphics,
volunteers free to fiddle with the material as they please, perhaps for a
maximum of half an hour each, no talking allowed. It is commonplace that a singing crowd, as
opposed to a choir, hits the correct notes with remarkable precision.
Be all that
as it may, some of those bees’ duties, such as the feeding of the young and the
production of wax for the building of combs, depend on stages of physiological
developments. The main examples are the
development of the glands that produce the bee milk that forms part of the food
of the larvae, and of the wax glands
that produce the comb wax. These
strongly influence the degree to which the bees can alter the sequence of their
functions. Conversely, some individual
worker bees specialise in particular duties more than some of their nestmates.
None the
less, not only is the sequence of the employment of a worker bee fairly well
marked, but some of the transitions have a tempting logic. It is for instance after a stint of feeding
the older larvae that the bees begin to feed the newly hatched larvae. It is after comb maintenance experience that
they join clusters for comb building. It
is reasonable to suspect that the comb cleaning and smoothing duties act as
development initiators for innate comb building skills.
Such ranges
of behavioural development within individuals suggest intelligence and mental
development of various kinds. However,
the behaviour, though flexible in duration and in response to need and
circumstances, is largely stereotyped.
Consider pollen collection. This
is a vital activity because pollen is the main source of most body-building
nutrients other than the carbohydrates in nectar.
Collection
of pollen depends on availability and on need, and it is a complex activity
that varies with the plant that supplies it.
Different plants signal the availability of pollen and nectar in
different ways, some of them by changes of colour and scent. Some do the pollinator violence, for instance
the stamens of some plants swipe the bee to daub it with pollen. Bees visibly behave as though the experience
were aversive, and have been seen apparently trying to avoid the blow after
they have had sufficient experience to teach them what is coming. Furthermore, some kinds of pollen are not of
value to the bee and are not actively collected.
What is
more, flower species differ in the nature of the duty they require. Some, such as Eucalyptus macrocarpa, offer more nectar than any one bee can
carry, for pollinating just one flower.
Strictly speaking, Eucalypts with such large nectar supplies are not bee
plants, but adapted to pollination by vertebrates. This does not discourage the bees though;
they avidly pollinate many species of Eucalyptus and other Myrtaceae, including
those with the largest burdens of nectar.
It does not follow that this is bad for the plant; such a flower is
likely to be visited by several bees, and receive pollen from many other
plants.
It shows
that the simplistic view of plants offering the least reward possible to make
it worth pollinating them, is too naïve a view of the selection pressures on
the species in the relationship. If
there is some form of intelligence in such relationships, it is obviously not
trivial to characterise it in any predictive form, however facilely one might
spin Just-So stories after the fact.
Other major
honey plants, such as alfalfa, give minuscule amounts of nectar from each of
thousands of florets. For each form of
flower the bee has a particular strategy to exploit the flowers, for example
working its way up one column of flowers on a raceme, and flying off to another
raceme when reaching the top. This may
sound counterintuitive, but some research suggests that for some flowers it
actually may be more energy efficient.
However debatable this point, the fact remains that there are many more
possible inefficient than efficient algorithms for pollinating a complex
inflorescence, and the bees use at least reasonably efficient algorithms.
The
relationship between flowers and specialist pollinators gets even more complex
in some species, with for example, the flower changing in colour and in scent
once the nectar has been removed and the stigma presumably pollinated. It might then produce pollen, and get visited
mainly by pollen collectors. Later on it
is no longer in the interests of either party that bees should disturb the
flower, and the appearance changes again.
Foragers from eusocial bee colonies not only specialise in the type of
collection they perform during a given few days, but in the time of day. They visit the same spot and seek the same
scent, and collecting the same type of food until the supply dries up. During the rest of the day they are very
likely to remain in the hive while other bees undertake other duties. Interestingly, within the hive they often
form groups, cliques if you like, of similarly scented bees that have been
visiting the same patch or type of flower.
Solitary
bees often specialise in the species they visit, but cannot afford the luxury
of moping around the nest between chores.
Also, some
species of flowers, including rich sources of nectar such as Aloes, might keep
their nectar in tubes that the bees cannot reach into. They normally might be bird or moth
pollinated. Often the bees will discover
the fact and learn to chew through the side of the tube and rob the flower
without pollinating it. There is dispute
about whether they learn such tricks independently or from colony mates or from
other species of cheating pollinators.
Then again
there is the point that bees soon learn to visit particular flowering plants,
often just at particular times of the day, How much of this behaviour is
stereotyped, and to what degree, is
unclear, but it certainly is intelligent by some criteria.
A vivid
example of where behaviour adjusts homeostatically to match circumstances, is
the way in which bees will ventilate their hive when hot, and bring water which
they fan for evaporative cooling. This
is reminiscent of the way that a very hot cat will slather itself with saliva,
cats having hardly any ability to sweat.
Such behaviour seems to be equally innate and stereotyped in both
cases.
There are
other similarly illustrative analogies.
In cold weather bees will bundle for warmth much as emperor penguins
will, and they stock their combs in zones that reflect the need or tolerance of
the contents for warmth or coolth. As
for how much more intelligently vertebrates behave, any poultry farmer will
know how easily chickens may smother each other when large flocks huddle for
warmth at night.
Homeostatic
shivering in vertebrates has direct counterparts in moths warming up the wing
muscles in their furry thoraxes, and in bees that vibrate their wing muscles
either to warm a cold hive or cook an enemy smothered in a bundle of workers.
Then again,
some eusocial insects have very high degrees of task specialisation between
castes, and they typically have castes that, unlike honeybee workers, do not
much change function during their adult lives.
This occurs in many species of ants, and probably most termites. Among termites the soldiers usually are
physically well-marked castes, such as in the snouted harvester genus Nasutitermes,
other examples include both the soldiers that plug nest openings with their
heads, in Cryptotermes, and in some Camponotine ants. There also are
various kinds of large-jawed, heavy-headed soldiers in many species of both
ants and termites.
For
example, apart from the reproductives, some ants have two, some more than two
obviously different castes. Such
anatomical specialisation of castes goes with behavioural specialisation. In fact, many such specialised castes are simply
unable to undertake the duties of other castes; non-worker castes often cannot
even feed themselves.
Interestingly,
in many species of eusocial insects, particularly ants, such as the Argentine
ant, Linepithema humile (formerly Iridomyrmex humilis), there is
just one multipurpose non-reproductive caste, while in say, Pheidole
megacephala there are few large-headed soldiers and they seem not to be as
active in combat as the workers are. I
suspect that such soldiers are on the way out, and that, given time, Pheidole
megacephala will become an ant with just one non-reproductive caste. I furthermore suspect that some species of
non-specialist ants already have undergone such changes and have lost castes
that were present in their ancestors. In
other words, their condition of having just one non-reproductive caste would be
secondary.
Apart from
distinct castes, there are other examples of function specialisation within
eusocial colonies. One example is the
use of immature colony members for particular duties. For instance, larval termites (“nymphs”,
though some entomologists seldom use the term nowadays) of many species will
wall up small enemies or undesirable matter by regurgitating building material
over them. Ant larvae of the weaver ants
are the source of the silk used in binding the leaves from which the workers
construct their nests. Furthermore, it
is practically universal among eusocial species that young adult workers barely
out of the teneral stage, begin with duties inside the nests, before going out
for external duties.
Also, some
species of ants, such as Anoplolepis
custodiens, have no clearly defined multiple worker castes, but just a size
range of workers, typically a log-skewed normal distribution that yields a
startlingly straight line on being subjected to probit analysis. Such observation strongly argues that the
distribution represents a dosage response curve that reflects random feeding of
the larvae, such that those most frequently fed grow into the largest
workers. Such feeding could easily
result from stereotyped feeding behaviour in worker ants. This suggests a phylogenetic point of origin
for the evolutionary development of multiple castes. If that is correct one might expect little
difference between any form of intelligence, between the larger and smaller
workers, because they are no more than developments of slight variations of the
same processes in the same organisms.
However, the argument is not very strong; even in honeybees various
parts of the workers’ brains change in size during their adult life, presumably
in association with their experience or changes in their duties or hormonal
development or all three.
As we have
seen, even in eusocial insects with just one worker caste of more or less
uniform size, such as honeybees, it is commonplace that some workers
concentrate on particular duties to a greater degree than others. It is only to be expected that where there
are differences in size and shape as well, there would be much greater
differences of bias in the types of work undertaken. In species where there are distinct castes of
different sizes, the types of work undertaken are strongly dependent on
physical differences between the workers.
A
spectacular example that I mentioned in another connection, is in leaf cutter
ants, species of Atta. Here we
find small workers riding shotgun on leaf segments carried by the large
workers. They are not just hitchhiking,
but protecting their larger sisters from parasitoid phorid flies that otherwise
would lay their eggs on them. Note that
these phorids are not the same species as the ones that attack fire ants, but
there is an interesting thing about the strategy of the Atta
colonies. Unlike the apparently
frightened behaviour that paralyses the working production of a fire ant nest,
this allocation of a part of the minor worker force to protection of the convoy
of large workers, permits the colony to remain productive, full speed ahead and
damn the parasitoids. As I already have
remarked, the cost to the colony of a minor worker is just a few percent of
that of a major worker.
In eusocial
mammals, naked mole rats, there is noticeable physical difference between
soldiers and workers. There also is
considerable, though not necessarily absolute, difference between their
respective habitual behaviour. It is
plausible that there is a significant difference in intelligence, but it is
clear that their behaviour patterns are not quite as stereotyped as in the
eusocial insect castes. It also seems
that their specialisation is more reversible than in highly eusocial
insects. If the queen dies then in a
healthy colony the changes in hormonal balance cause other members of the
community to re-adapt to the role of the queen.
In some eusocial insects, such as many species of termites, existing
immature larvae may develop into secondary reproductives of various degrees of
effectiveness.
Now, there
is relatively little published material on the subject of the brains and
intelligence of eusocial insects, and what there is, is largely speculative or
limited in scope. Not surprisingly there
is a great deal of stereotyped behaviour, and wherever this is demonstrated it
is likely to be characterised as non-intelligent behaviour. From our point of view in this discussion,
such non-intelligence is debatable. What
is more, even stereotyped behaviour may be fairly flexible, whether in insects
or in humans.
One example
is in web-building spiders. On one hand,
most species will repeatedly run out to attack a wax-covered insect when lured
with a suitable tuning fork. Conversely,
though web building is a largely stereotyped behaviour, not only must the web
conform to local circumstances, such as the available points for tethering, and
the suitability of the site for catching prey, but in species that build orb
webs and repair them every night, the repair activity seems to respond to the
success of capture of prey. If one
places prey consistently in the same corner of the web, then during
reconstruction, the spider progressively builds that corner of the web larger
and larger.
Is this
intelligence? Learning? It certainly is flexibility, though it is not
clear whether that flexibility is a “mental process”, or the modification of
one component of stereotyped behaviour by another.
In either
case, why or in what way is that not “learning”?
There was a cage with several apes in it. Inside the cage
a banana hung on a string, and there were stairs below it.
Before long an ape went to the stairs to get the banana,
but as soon as it even touched the stairs, all the apes were sprayed with
water.
After a while the same ape or another
one repeated the attempt, with the
same result: all apes got sprayed.
Pretty soon they all got the idea
and whenever another ape tried to climb the stairs, the others would
try to prevent it.
Now the researchers took one ape from the cage and replaced it with
a new one. The new ape saw the banana, and tried to climb the stairs.
To his horror all other apes attacked him.
After another attempt he
knew: if he wanted to climb the stairs, he would get beaten up.
Then the researchers removed a second old ape and replaced it by another
new ape. The newcomer went to the stairs and got beaten up in its turn.
The previous new ape participated in the punishment with enthusiasm.
A third old ape was replaced by a third new one.
The new one made it to the stairs and got beaten up as well.
Two of the apes who beat him had no idea why they might not climb the stairs.
They replace the fourth old ape, and the
fifth,
until all apes that ever have been sprayed with water have been replaced.
Nevertheless, no ape but another novice
ever would try to climb the stairs thereafter.
One day a new young ape asks, "But Sir, why not?
"Because that's the way we do things around here, my boy."
Anonymous (traditional on the WWW)
Reflection: the way they did things around there might have seemed
arbitrary to the novice, and the uncomprehending dogmatism of the elders might
indeed reflect intellectual limitations, but unless the research workers had
turned off the waterworks in the mean time, the first novice to buck the system
would demonstrate inadvertently that a little learning is a dangerous thing.
If it is
difficult to compare or even define the intelligence of individual organisms,
it is far more so with communities. As I
have remarked, it is hard enough to define the intelligence of a community at
all. There are grounds for arguing that
if group intelligence means anything at all, then some kinds of intelligence
and temperament in the group are not the same thing as in the individual.
For
example, any teacher or drill sergeant knows that each group of pupils or
rookies has its own character. One gets
them good and bad, affable and sullen, biddable and unwilling, stupid and
talented. Sometimes it is easy to tell
what some of the sources of some of the attributes are; here there is a class
member with high skills, there we find a couple of rotten apples, again, we
find a class that has not met its prerequisite training goals, or one class
comes from an educationally backward area, while another comes from a
university town teeming with gifted children.
And yet,
that explanation is nothing like sufficient, let alone of predictive
value. Given the most consistent
possible source of members, each batch differs from the rest. In fact some teachers will swear blind that
the years of good classes alternate with years of poor classes. Similar differences appear with sporting
teams, teams of workers, company staff, public servants, teams of dogs, and
almost any type of community one might mention.
Sometimes
one can change the nature of such a community by replacing or removing a few
members. Sometimes, perhaps more often,
a particular attitude, atmosphere, character, remains associated with the team
indefinitely. In commerce and industry,
unthinking tradition often will determine the strengths or the ultimately fatal
weaknesses in an erstwhile dominant company until it goes under and its staff
disperse into other institutions.
Certain
approaches that have been of limited value in exploring the human mind can be
of more limited use in studying eusocial communities; The main approaches have been to examine the
behaviour of the colony and the contribution of certain members of the colony,
and to observe the effects of removing or interfering with the activity of
certain members of the colony.
So for
example, we can see that say, comb-building and foraging in bees are worker
activities. We find it unsurprising that
the worker brains and anatomy are the only ones in the hive that equip them for
such activities. In much the same way,
we do not expect mature human retinal neurons to be interchangeable with say,
neurosecretory cells in the hypothalamus, any more than worker castes are
interchangeable with queens. We also
find that typically, soldier ants and certainly mature soldier termites,
contribute nothing to the routine tasks of workers.
Although
objectors might point out that in fact neurons do have considerable functional
plasticity, that is nothing like enough to suggest that mature neurons are
anything like interchangeable. And even
if the occasional worker does lay an egg, or the queen does display a bit of
incidental worker-like behaviour, it does not imply that their roles are not
distinct.
Conversely,
by removing a source of particular pheromonal or behavioural control from the
colony, we can tell how rigidly the behaviour is limited to a given population
in the colony. In some kinds of
termites, losing the queen causes suitable immature colony members to become
fertile, though they seldom can rival the function of the original queen. Vigorous colonies of naked mole rats or of
some kinds of paper wasps can replace a missing queen in a fairly short time.
However,
the fact that some members of the colony have particular functions and are
specialised for those functions, does not mean that they understand their own
functions, any more than a neuron in the human visual cortex knows what a face
is, even though its function is vital to face recognition. Most often any particular member almost
certainly has no explicit understanding of its own role and actions, any more
than a neuron in the human cerebrum could understand its own role in art
appreciation. “All” it does is to
modulate its output according to the interaction of its initial state at any
given time, with the rates of pulses that it receives from it various input
synapses or external stimuli.
Like the
actions of insects in eusocial communities, the neurons’ behaviour is
stereotyped, though flexibly so. And yet
that very behaviour is by far the most apparent contributor to the intelligence
of brains. Rival contributors so far
identified are minor in comparison, such as hormonal distribution in the body,
or falling down when fainting, which maximises the body’s chances of
recovery. What is more, such mechanical
controls have even less pretension to intelligence of their own than neurons
do.
Repeated
observations have shown that most of the activities of individual social
insects are stereotyped, with very limited (though vitally important)
flexibility. In fact, even apparently
purposive architectural activity such as mound building in termites or comb
building in bees or wasps, patently are not understood by the individual
participants. The bees use very simple
actions in combination with the geometrical effects of crowding, and the
termites use feedback from initially random placement of pellets of nest
material.
Granted, it
is not all they do; when there is a small breach in the wall, they place their
pellets just in the breach, but within such limits, that activity too, is
stereotyped. The nature of the way in
which they cooperate depends in some ways on the pheromones released at the
site of the activity.
Still, it
is very important to note that an entomologist knowledgeable about termite
nests can recognise a considerable number of species from the morphology of the
nest alone, even when the insects themselves look very similar. The most famous example is the nest of the
Australian compass termite, but that is by no means the only one. For example fungus grower nests such as those
of Macrotermes may look less spectacular, but they are more complex
internally. Each species builds nests
adapted to its needs, often marvellously so.
Some species build their nests by tunnelling into wood, some altogether
below the ground, some in mounds on the surface. Most are distinguishable by texture, colour,
size, shape, orientation and position.
Their designs reflect the available materials, climate, threats,
substrates, and resources.
This is a
crucially important point; the nests vary as characteristically as birdsong, as
handwriting, and they vary for similar types of reasons. Though we cannot trace all the details, it is
clear that the character of one’s handwriting stems from one’s mental
structure, as reflecting certain control systems in one’s brain, partly innate,
partly developmental and partly modified by external stresses and stimuli. Similarly, the control of the building activity
is partly innate, and partly in the systematic informational interaction
between the nervous systems of the participants. It furthermore is modified by environmental
stimuli, such as the daily passage of the sun, the depth of the water table,
the presence of food, attacks from enemies, and so on.
One way or
another, all the most advanced colonial species rely on their own types of
infrastructure, the nest itself, the stores, the labour and security
communities, and in each case the form of the infrastructure is characteristic
of the particular species as influenced by the local environment.
It gives
one pause to compare the shape of the termite nests with the nests of solitary
mud wasps, or the mouth of the male mole cricket’s tunnel, shaped like an
exponential horn, the better to propagate his song. Some solitary species build their own
infrastructure, as characteristic as the communal constructions of termites,
and sometimes of startling sophistication.
This implies that the solitary organism interacts not only with the
environment as it was, but with the products of its own earlier activity and
constructions. Here there is something
vaguely reminiscent of the way that a single quantum can interfere with itself
in passing through a grating, but more realistically, the creature is
responding to the feedback from its own previous labours.
Some very
important work has shown how flexibility in response to feedback can be
superimposed on stereotyped repertoires of activity, to achieve building activity. For example, particular kinds of ants collect
scattered refuse such as grains of sand or corpses into patterns. Such patterns are characteristic of the
species. Similar behaviour determines
the placement of nest material in the building of termite mounds.
Some such
patterns are so simple that they can be simulated graphically by fairly
elementary computer algorithms, thereby revealing the simple nature of such
aspects of the colony intelligence.
However, to find one simple principle is one thing, to assume that it is
the whole story would be simplistic.
Note that it is far more difficult to simulate anything like the total
building of the nest. Simple as they
presumably are, some aspects of such communal behaviour still have not been
worked out.
Though it
is clear that the intelligence of the colony is neither the intelligence of the
individual colony members, nor yet the sum of their intelligence, it still
looks like a definite form of intelligence, even if it is rudimentary. What to compare it with is hard to say. How to characterise it may be harder. The beehive might be seen as being like an
octopus that locates and retrieves food with its tentacles. The driver ant colony might seem very like a
wandering predator, shedding cells (colony members) in capturing food to grow
more cells, and so on. Some colonies,
particularly beehives, learn where there is food and what the food looks and
smells like. Some, like foraging ant
columns, learn where food seems to be absent, by marking unproductive paths as
such.
Bee
colonies may become inhibited from working when predatory wasps, such as the
South African bee pirate, Palarus latifrons, are about. The wasp lurks outside the nest and captures
water carriers or foragers as they emerge to go about their duties. This suggestively resembles the fear that
say, a herbivore might display when predators are about.
This
panicky effect on bee colonies is not unique.
Phorid flies, of the genus Pseudacteon in particular, are being
used as biological control agents against the various South American species of
fire ants (Solenopsis) that have become pests in the southern United States. They lay their eggs in worker ants and the
larvae eat them from within. On first
thought this seems a little puzzling, because the flies are not particularly
fecund and it takes the death of a lot of workers to destroy a fire ant
nest. In fact, estimates are that only a
few percent of the ants get infected, far too few to harm the colony much.
It turns
out that the ants detect the presence of the flies and instantly go into
defensive mode, turning over to fend them off.
The whole workforce is likely to run underground as soon as they detect
the flies. Just a few flies are enough
to waste a whole workday for a colony, and that is more expensive than the loss
of a few dozen workers.
Interestingly,
although the reaction of the workers is costly, its very existence, violence,
and specificity suggest that the attacks by Pseudacteon have been a
feature of the environment of Solenopsis for millions of years. This also is consistent with the fact that
the flies are so very selective in the species of ants that they attack.
But then
why such an expensive strategy that it paralyses the productive activity of the
nest? Atta have a similar problem
and their strategy has reduced it to the level of a nuisance. The most persuasive argument is one of
evolutionary opportunism. Solenopsis
does not indulge in any activity in which it would be practical and obvious for
one worker to protect another, plus, they do not have multiple worker or
soldier castes to take over such duties.
Presumably those facts denied them the opportunity of developing the
“shotgun riding” strategy.
Conversely,
as a species they cannot afford to ignore the flies; the current rate of loss
is trivial, but if they did nothing to frustrate the flies, they would increase
indefinitely and might wipe out the ants.
If this were not so, there would not have been any selective pressure
for them to develop the behavioural patterns for avoiding the flies.
To return
to the honeybees, other species such as Apis cerana in Japan, will
ball around certain types of enemy wasps till they kill them with heat, most
famously the giant hornet, Vespa mandarinia. This suggests a tempting analogy to metazoan
fevers in reaction to certain diseases, particularly viruses such as influenza.
Studies on
beehives have shown that there are particular stimuli, smells, movements,
colours, textures and so on, that are likeliest to provoke any hive to
attack. None the less, some hives are
more aggressive than others and breeders successfully select for docile
colonies. Different strains have
different personalities. South African
beekeepers returning from exhibitions overseas often are dumbfounded to see the
casual behaviour of European or American colleagues. In South Africa foreign beekeepers
would change their modus operandi pretty smartly, especially in dealing
with Apis mellifera dorsata, which has a notoriously low threshold of
intolerance.
Another
observation is that beekeepers dealing with the more irascible breeds of bees
find that hives differ in their preferred target. Some go for the eyes, others for the
buttocks. I am not aware of formal work
on this matter, but certainly in South Africa this observation is a
commonplace among long-suffering practitioners.
Whatever the mechanisms behind these patterns of behaviour, they are
inescapably analogous to temperament in individuals in say, humans and dogs. Here too, temperament is affected by both
genetics and environment.
What is
more, it cuts no ice to claim that differences between the hives are a trivial
matter, being due to the relative sensitivities of the individual members. A perfectly valid counter observation is that
slight differences in the receptor and neuronal sensitivities of persons and
animals lead to drastic differences in temperament and behaviour. So does the history of the individuals and
the community; a colony that has lived undisturbed for longer than the life of
any of its workers is likely to be more tolerant than one that has frequently
been provoked.
And yet, we
are left with a serious question. In
analogy to an individual animal, does the colony learn to recognise particular
patterns? Actually they do. An example is the appearance of its own nest
opening. This is so specific that if one
moves a hive a few metres, the effect on the colony is disastrous. The workers returning from foraging for food
or water will go to where the nest had been the previous day. Conversely there is no problem to moving the
nest several kilometres to outside the previous foraging range of the workers,
because then the workers do not recognise the new territory and the first thing
they do is to familiarise themselves with the new position of the hive.
Some tricks
do help for moving the hive over short distances. One can persuade the colony by moving it a
few centimetres at a time, say less than the width of the hive per day. This is a homing task that the bees can adjust
to. Or one can decorate the hive with
large projections, or paint the front of the hive a vivid colour combination in
a bold pattern that dominates the territory for a few metres around. This trick also is particularly useful when
one keeps many hives in close proximity.
It reduces the tendency for workers to enter neighbouring hives by
mistake.
Another
trick is to put a sheet of clear glass just outside the hive exit each time the
hive has been moved the previous night.
The emerging bees bump their heads on their way out, and that
concentrates their attention marvellously.
They then fly around the hive to confirm its precise location. A couple of days later one can repeat the
trick, till the hive reaches its new site.
These
observations are in no way trivial. All
these cases present examples of learning by individuals, in such a way as to
amount to learning by the colony.
But can the
bees learn recognition in the same way as the most intelligent
vertebrates? Can they learn to recognise
a face? I am not aware of any such
effect in insects. Worker bees certainly
have been trained to collect nectar where there is a picture of a particular
face, and no doubt some of the workers that they have recruited learn to
recognise the same picture. Personally I
regard this exercise as just a demonstration of the complexity of patterns that
the bees can learn to recognise. As far
as I know there is no evidence that a swarm can recognise the face when a
person passing, nor that they recognise different people from their faces.
It also is
not clear say, that bees learn to know any of their nest mates as
individuals. I already have mentioned
workers’ cliques within hives, but I know of no evidence that such cliques are
based on much more than shared smells and shared preferences for particular
zones in the hive. It is in any case
questionable how important recognition of individuals might be for insects that
live only a few months.
There is
however an important question concerning bees’ recognition of relatedness. I mentioned the subject of conflict of
interest in hives, arising from genetic differences between hive members. One basis for cliques might be recognition of
shared genes. Olfactory and visual clues
that support clique formation, mate choice, or family attachment have been
documented in mammals, including humans.
Some such clues are genetically determined. I do not know of similar research in bees,
but I may be behind the times. If such
effects can in fact be confirmed, it would go far to support the idea that
workers’ and queens’ relatedness could underlie various patterns of bias in
their respective support of reproductive activity.
Such
subpopulations within a colony could have their own “personalities” or
“intelligence”. Like regions in a human
brain, they could contribute to the intelligence of the entire colony.
One way or
another, it is not safe to be too dismissive.
Bee colonies certainly do react to certain stimuli such as particular
threats in the region to which they are adapted. In Japan Apis cerana react to
the giant hornet. In Africa
they react to the wasps called pirate bees.
Presumably such recognition is innate, but it still amounts to a
component in the intelligence of the community.
Conversely,
colonies of sociable birds such as jackdaws certainly can learn to recognise
individuals. Lorenz describes how, once
one flock member has been sufficiently alarmed by a given human or large
animal, it can pass on its alarm to others by example, and soon that individual
is persona non grata to that flock in perpetuity, or even all the flocks in the
neighbourhood. If several individuals
achieve that status, then soon all members of that species are automatically
regarded as enemies. In such a
situation, the intelligence of the individuals is the basis of the intelligence
of the flock.
There are
similar effects in human societies.
Among criminal gangs in various countries there are conventions by which an individual can be
named as fair game to the first person who gets a favourable chance to kill
him.
In South
Africa at least, it is recognised that such a sentence cannot be cancelled; not
because of any fixity of purpose, but because, while it only takes a few words
to start such an alert, one could no more recall it with any confidence, than
recall a rumour. In the USA there is
the Hell’s Angels cliché that “The Angels is like an elephant. It never forgets.” It is no surprise that such memories and
alerts are in fact highly unreliable and frequently lead to mistakes and
failures, but at the same time they are dauntingly effective and justly
dreaded.
In their
inefficient, unreliable, but effective and flexible way, the behaviour of
individuals who are party to such sentences strikingly resemble the behaviour of
neurons in mental systems. They really
are poorly predictable in their individual behaviour, in the pulses they emit,
and the combinations of stimuli in response to which they do emit pulses. And yet the systems not only work, but are
amazingly effective overall.
Such
effects also worryingly recall inter-group prejudice in human society as well
as among other animals.
The “Mongol
Horde” principle of attacking an objective is pervasive among large
populations, from virus to human, from evolution to military assault. There are various aspects that vary in
relative importance according to the population in question. The first is that if there are enough
attacking agents, one is likely to be lucky and strike a vulnerable point,
however obscure that point might be.
This is the kernel of truth in the designer’s traditional lament: “It is
impossible to make anything foolproof, because fools are too ingenious.”
Another
strategy for employing hordes of fools relies on the fact that if there are
enough participants, some tasks become worth while, even if they are
inefficiently approached. A simple
example is the way that ants carry loads too large for just one. Instead of the conscious sophistication of
human stevedores in cooperation, the ants tug at the load from all sides, at
considerable cost in labour. It works
well enough to have preempted costly development of more efficient algorithms
that would have required more elaborate communication and planning.
One item of
sophistication in the algorithm is that it can redirect the ants to divide food
items that are too large. If a group of
them cannot lift the prize, they are stimulated to begun the cutting
process.
Note that
none of the evolution of the behaviour patterns entailed a high selective cost;
their adaptation required very little that was not already present in primitive
ants.
Another
strength of the mass attack is that if there is to be any organised resistance
to the assault, then the sheer numbers of attackers are likely to disorganise resistance. For one thing, the defender cannot always
know which attackers are ineffectual and may be ignored. Among honeybees for example, it is well known
that they lose the sting after inserting it.
This ignores two salient points to the principle. Firstly, the sting goes on pumping venom into
the victim, behaving like an independent attacker until it is removed. It is not just a passive little bag of venom,
but a complex device that is continuing with the attack, so the advice to avoid
squeezing it, but to get it out with the back of a blunt knife, is an old
wives’ tale. The thing to do is get it
out fast to reduce the amount of venom injected; squeezing makes no significant
contribution to the injection and may even destroy the mechanism.
While the
sting still is dedicatedly pumping, the now stingless bees do not immediately
drop dead or go away. In pathetic
futility they go on trying to sting.
Only it is not as futile as it looks.
The victim cannot easily tell which bees still have stings, so the bee
that at first had been just one assailant, became two: one envenomator and one
distracter and intimidator. This is not
a trivial point; it is an example of a sophisticated application of hive (or
horde!) tactics and it is innate to the species.
Apart from
foraging and fighting, there are other aspects to the ways in which colonies
employ their population resources. In
fact, the sheer variety of colony dynamics in nature almost leaves one at a
loss to think of plausible principles that have never been exploited. Some kinds of colonies are small, just a few
dozen members to a mature nest. Ants of
that type usually are predatory and a single good kill is likely to feed a
large part of the nest for quite a while.
One also gets moderate-sized colonies of several hundred or thousand
members. Their size limitations vary
according to seasonal constraints, available food, nest space, and other
considerations. For instance, a hollow
root or thorn might not accommodate a nest for a large population, not even of
ants.
Either way,
modest colony sizes offer certain options, but large colonies offer different
options. Large colonies can use
strategies of defence, foraging, food culture, reproduction, and building, that
simply are not viable for modest-sized colonies. Driver ant colonies just have to be large, or
they could not accept the attrition of open-air bivouacking and attacking of a
wide range of prey sizes. Termite nests
of certain kinds simply have to be large, which means large populations. It takes thousands of foraging trips for
honeybees to make the merest spoonful of honey.
In turn, it takes a great deal of even the best honey to make up a
little wax. Only a large community can
afford such costly materials, especially in large quantities.
In short
the use of population numbers is another option that may be used by either the
colony or the species in meeting environmental challenges. The strategies not only vary, but vary in
several dimensions. For example, one way
is in the growth of the colony size.
Most ants, termites, social bees, and wasps propagate by sending out
young reproductive caste individuals to establish their new nests alone. This includes some of the species that build
the very largest colonies. In turn it
implies that they must start out with the smallest colonies possible and pass
through all the stages up to the largest.
At each
stage a successful colony must employ a strategy that suits its size and other
circumstances, or it will not survive to reach the next stage, or if it does
succeed, then at the least, its reproductive success will be lower than that of
a colony that adapted more effectively.
The
question of size seems at first to be a matter of a few trivial details, but
several of the principles are fundamental.
Size is a major factor in evolution, in interspecific and in
intraspecific competition, in the relationship between the creature and its
physical environment, its ability to hide or to survive exposure, in its
dependence on its food supply, in general its place and strategy in its
ecology.
There
definitely are differences between the relevance of size to individual animals
and to colonies, but actually the analogies are much more striking than the
differences. All the variables that
affect large animals reappear in related forms in colonies. The individual colony members may be able to
fit into smaller dens than lions, but a colony of a million members will not
fit into just any cleft in the rocks, nor can it feed as cheaply as a colony of
a few hundred members.
From the
point of view of this essay the details are less important than the fact that
the species must have a strategy for exploiting the powers and dealing with the
penalties that its particular size entails.
As I have explained, a strategy, evolutionary, communal, or conscious,
implies information processing in some sense that we still have not learned to
distinguish definitively from what we call intelligence.
Then again
consider interspecific inter-colonial associations of types that are analogous
to interactions between individuals. One
gets several kinds. Simple coexistence
occurs between some species of ants, where they share trails and nesting cavities
without aggression. This is startling
when one considers how viciously most ants fight even foreign colonies of their
own species, never mind alien species.
This mutual tolerance is rather like herds of antelope and zebra keeping
company, except that it is not clear in either case, how much of the
coexistence is simply for convenience rather than active mutual benefit.
Such
interactions shade into various degrees of mild to severe parasitism. Some just make mild inroads into the food
supply of the host, such as Megalomyrmex inhabiting nests of Sericomyrmex
and eating their fungal food without obvious opposition, and apparently
without causing much harm, rather like an infestation of rats or pets in a
human home. In contrast, Solenopsis
fugax, a small relative of the fire ant, builds its tiny tunnels to
interconnect with those of larger species of ants, and robs their larvae and
eggs. There are also many species that
undertake “slave raids”, robbing related species of their larvae, eating some,
and in some cases even relying on the survivors to act as the worker caste in
the nest.
Predation
is another type of colonial interaction.
I have already mentioned giant hornet colonies. If they are not stopped by beekeepers or by
the strategy of Apis cerana, they wipe out the bee colony by killing the
adults and carrying off the larvae. Such
conflicts are very suggestive of colonial intelligence, much like the
intelligence of bee colonies in scouting for food sources or new sites for
hives. The hornet scouts scent-mark the
site when they discover a hive.
Conversely Apis cerana workers react to the scent by instantly
attacking the scout and balling round it to kill it by heat. The behaviour is stereotyped and innate, but
the inter-colonial behaviour might be interpreted as quite complex, though
still stereotyped.
There is no
doubt in my mind that such effects vary quantitatively and qualitatively in
their underlying mechanisms, in the type, speed and persistence of the
communications between members of the colony, in the contributions and
intelligence of particular members of the colony, and in the genetics
responsible for particular types of behaviour.
Be that as
it may certain fundamental mechanisms in colonial behaviour seem to be directly
analogous to those within the metazoan body.
They include:
- Communication
by pheromones that are analogous to hormonal control within the body
- Communication
by contact and sensory signalling that are analogous to neuronal and
synaptic signalling in the bodies of creatures with elaborate central
nervous systems.
- Elements
of information processing within the communication process are temptingly
analogous to both synaptic amplification and inhibition in neuronal
signalling.
- There
are similarities between some aspects of memory in social and synaptic
systems.
Conversely,
we see repeatedly that there are certain differences. True, our insects (and shrimps and mole rats,
and even humans) have certain stereotyped reactions that they are poorly aware
of, if at all, but we also see repeatedly that they have some flexibility of
action that is vital to the effective activity of the colony. They have certain capacities for information
cultures of various types, that do not closely resemble cultures in the
cellular behaviour within our tissues.
Culture and intelligence:
Elephants, apes, orchestras and Selenites
They had no vision amazing
Of the goodly house they are raising.
They had no divine foreshowing
Of the land to which they are going;
But on one man’s soul it hath broken,
A light that doth not depart,
And his look, or a word he hath spoken,
Wrought flame in another man’s heart.
Arthur O’Shaughnessy. Ode. The music makers
We are left
with the question of the role of intelligence in directing communities, as
opposed to the nature of intelligence as such of communities. There are no impressive eusocial examples of
vertebrates on this planet; probably the closest we come to it is in the naked
mole rats. This is not very
satisfactory. Cerebrally a mole rat is
more impressively endowed than a bee for instance, but it is not clear that
this plays much more role in their eusociality than equipping them for
socialisation and domination of subordinates.
Among insects the nearest equivalent might be the power struggles among
the reproductive females of some paper wasp species. There is little sign of the development of a
body of learnt experience being preserved and passed on among such
colonies.
Among the
more highly intelligent sociable mammals things may be different. Experience in culling animals such as
elephants has shown that the social structure of herds and the experience of
the elders are vital to the well-being of the herd for generations to
come. Senior members not only are the
repositories of skills and wisdom, but of security. They are the community’s repository of
experience as well. They know where to
get water in dry seasons, what to eat, where to trek to when seasons change,
where there is danger, and what their social relationship is to neighbouring
communities.
They also
socialise the young, punishing harmful behaviour. Orphans without such guidance tend to grow
into delinquents. If communities that
orphans establish survive at all, they are likely to take some generations to
recover and develop advantageous cultures from scratch.
Interestingly,
though not surprisingly, experiments in pigtail macaque communities have shown
that removing the top-ranked males from a troop has unfortunate effects on
troop behaviour. It turns out that the
bosses suppress bullying and disruptive behaviour among hoi polloi. Furthermore the effect has material
significance. In the absence of the
dominant leaders, there is less beneficial social behaviour such as grooming,
playing, and sitting together; there is more violence, and less social stability,
with more clique formation and reformation.
The absence of authority radically reduces the quality of life in the
troop. The resemblance to elephant and
human sociology is striking. It is a
safe bet that many more such relationships will be discovered as studies
advance.
Communities
of other species of sociable primates are similarly dependent on the control
and leadership of the more experienced members.
The best-known examples are chimpanzees and gorillas, but there are
similar effects among all sizeable primate communities so far studied,
including various species of baboons and sociable monkeys. A species of Japanese macaque was the source
of the notorious “hundredth monkey” inanity, in which pseudoscientific Western
reports claimed that Japanese researchers had said that when enough monkeys had
learnt a new technique, say of washing food or floating grain in seawater, then
the technique mysteriously spread through the islands without interpersonal
contact and teaching.
On inquiry
the Japanese researchers scouted this claim as nonsense. They had reported nothing of the kind. What was far more important and interesting
was the fact that the techniques did in fact spread, and that they did so
neither mysteriously nor infallibly.
They spread as culture does, by contact, by example, and by fits and
starts, not every monkey understanding and adopting the new technology. As a rule the youngsters were the most
receptive learners and exponents.
Tool use
among chimpanzees, such as catching ants with straws, or cracking nuts with
branches or stones, spreads similarly, with variations of technique between
troops, and with the dimmer members of the troop failing ever to master some of
the techniques. This suggests that there
could be fairly strong selection for intelligence whenever it offers an
advantage in finding or utilising food sources.
Another point is that among at least the chimps, tool usage and hunting
techniques, neither of which is innate, both of which are components of the
cultural heritage of the troops, get passed on so stably that ethologists
familiar with the subject can identify a troop by the range of techniques they
use at a given period. Of course, such
technological accomplishments are not fixed, because occasionally some member
of a troop learns something new and may pass it on. Alternatively, if a
technique lapses into disuse for long enough to be forgotten, then it might
vanish or have to be redeveloped from scratch when needed.
Chimpanzees,
and to a lesser degree of sophistication, baboons also have their political
structures within troops, and as a rule the larger the troops, the more complex
the structure. In fact, because
alliances and parent-child bonds are such an important factor in their
politics, one actually can see rudiments of hereditary rule.
Social
insects have nothing of that type. The
closest any of them get to it is a simple version of pecking order that depends
on the ability to recognise individual community mates and remember who is
superior and who subordinate. Perhaps
some kinds of paper wasp communities have something like this. Mole rat social structures are far more
sophisticated in this respect.
Many
sociable birds and mammals that form flocks, herds and other communities become
far more sophisticated yet. Within
communities they form social substructures based on kinship and alliance. Each form of community control is effective
only within constraints of size and environment. Depending on the nature of their foraging,
very small groups may be able to survive better in times of famine, so that is
when large communities split up, perhaps only until times improve, perhaps
permanently.
Conversely,
large groups, especially of carnivores, can evict or dominate smaller groups in
times of plenty. However, if a community
grows beyond the size that the political structure can maintain, then the
community splits or becomes depleted sooner or later, often by really serious
fighting.
Interestingly,
it is mainly among chimpanzees that politics can become so dominant an activity
and so sophisticated, that pernicious violence can break out even in manageably
small troops. Whether this suggests
anything about human politics, I leave to the ideological tastes or personal
interpretation of the reader.
Culture and
politics of various degrees of complexity occur in wide ranges of animals, and
many more examples will emerge as more kinds of community are studied. Cetacean ethology has shown considerable
social complexity, and it also is probable that sociable species of pigs, such
as the peccaries, have social relationships far more complex than just a
dominance order.
It is very
important to notice that in those species that develop any form of cultural
infrastructure, whether innate or by experience, conservation of the
information resource and its processing and application become vitally
important. In sociable mammals and birds
the most flexible and rapidly developing of the information resources are those
in the minds and the educational procedures of the communities. In insects the innate, stereotyped cultural
infrastructures are more important.
However, the situation is not pure; insect control behaviour is not
strictly rigid, and often communication behaviour patterns and signalling
equipment in troops of mammals and birds, involve genetically determined
gestures and physical features. There
might be landing or take-off patterns in territorial birds, ear colouration
displays to highlight a threat in cats, or,
in monkeys, conspicuous fangs, or colourful gums or eyelids.
There are
thousands of such examples, and for the most part the animals themselves
patently are not consciously aware of their own signals. Humans are in no position to sneer; we have a
fairly elaborate repertoire of body and voice language. Part of it is culturally determined, and part
innate, but most of us are poor at reading it consciously. However, within our respective cultures most
of us signal and react to signals far more reliably than we realise. Most humans also are poor at consciously
simulating misleading messages.
Interestingly,
a fair number of mammals are fairly good at dissimulation. So are mating cuttlefish.
Such
interpersonal signalling within a community might be seen as analogous to
inter-neuronal signalling within a brain.
Within a species its development obviously is vital. For one thing it is
easy to see that anything that reduces waste of resources or risk of injury,
and increases one’s dominance in a community, is likely to increase one’s
fitness. For another, in many species
the sheer ubiquity and specificity of the selection for signalling and
reception of signals, conspicuously or subtly, is astonishing. Nothing of the type could have stood the
costs of such specific selection pressure if it were less than evolutionarily
functional. Even within most solitary
species it is important, but within social species, the variety and elaboration
of the signals in their contexts increase accordingly.
It does not
follow that the signalling becomes more spectacular; in social insects one does
not find much equivalent to anything in the line of antlers or the startlingly
extended eyes of non-social stalk-eyed flies (Diopsidae) or peacock
plumage. Social signals tend to be for
conveying specific information on a number of subjects, rather than for
competitive exaggeration of mainly one point.
Where eusocial insects do signal for spectacular effect, it tends to be
behavioural rather than anatomical. One
obvious example is the way bees, even of inoffensive strains, will buzz
threateningly around the face and eyes of an intruder. If intimidation succeeds well enough to
render physical attack unnecessary, that is a lot cheaper than stinging and
either dying if a bee, or risking death if a wasp.
Political
structures are of particular importance in the present context. They are an example of where individual
intelligence and whatever passes for the intelligence of the community affect
each other directly. Techniques such as
washing food or cracking nuts might be important; fortuitously they might even
be components of the culture that turn out to be vital to the survival of the
community, but political control is where the mind of one member directs the
action of the community.
Direction
of the action of the community by leaders sounds impressive, but not everyone
is convinced of its importance. Some
historians claim that nations such as the Roman Empire
had successions of rulers ranging from something like secular saints to the
most appalling despots and parasites, without detectably affecting daily life
and commerce of the bulk of the citizens.
Personally I regard this view with suspicion, though I grant that the
effect of the ruler gets filtered and moderated through the administration, as
brilliantly satirised by say, the “Yes Minister” television series. All the
same, the view of the historian is necessarily in many ways the view of the
outsider, at once simplistic, selective, and superimposing his own
interpretations with limited scope for falsification. One is reminded of the closing paragraph of
“The Star”, a short story by H. G. Wells, in which an event that nearly
destroyed humanity looked trivial to Martian astronomers.
For
counterexamples even in pre-modern times, there were the Chinese periods of
imperialism, followed by the inward-turning of their policies. Both affected every aspect of their people’s
lives. Genghis Khan, Alexander, and a
number of major autocratic leaders were of crucial importance in the lives of
their nations and of their people. Had
it not been for Wilhelm II and Hitler, the world wars might never have
happened, and if either or both did happen, they would not have been the same
in course, timing or possibly even in outcome. In modern times with the more
advanced technology permitting more elaborately controlled infrastructure, we
have seen a whole series of despots, from the last of the Tsars to the first of
the so-called communists and nationalists.
There has not been a time in the last few centuries when there have not
been peoples who lived and died under the control of pernicious despots, often
genocidal despots. Their control might
not have covered every detail, but it was enough to affect daily life in every
respect, whatever things seemed like to outsiders at a safe distance.
Be that as
it may, the point is that here too, we see the interaction between the
character of the community and of the individual brain.
Mode of
rule is only one aspect of such interaction.
Culture is another. So is
infrastructure. Both are affected by the
inventions and procedures instituted by individuals. This is not an effect that we can see in the
eusocial insects; palaeontology is not good at preserving the deeds that gave
rise to selection. This is a field where
the Just-So Story and the argument from analogy come into their own, not to
prove the detail of history, but rather to show that a particular type of
course of events is a reasonable conjecture, consistent with the realities of
biology as we observe them today, and a basis for research on principle and mechanism.
At least
one experiment has been tried at having an amateur orchestra play without a
conductor. I understand that the
orchestra acquitted itself very creditably, but it was a tour de force rather
than an influential development. In
fact, historically the original function of the conductor was far less formal
than we take for granted today, and the development of the role was suggestive
from the point of view of this essay.
One could say that the conductorless orchestra is rather reminiscent of
real life eusocial communities. There is
a score (analogous to the innate behaviour of the individuals), and a beat and
pitch (analogous to the environmental releasers of behaviour).
However,
once the role of the brain and creativity of the conductor became established
in musical practice, the interpretive enrichment and flexibility of the
orchestral behavioural repertoire attained new dimensions of complexity and
size. Conductorless performance was
effectively relegated to small groups of performers such as in chamber
music. Again, the analogy of the ruler
of the state becomes suggestive.
Certainly the conductor of a large orchestra cannot control the
character of his performers in detail, nor can he control the character of the
group in detail, but he certainly can affect them profoundly in many ways,
including in the speed with which his central authority can bring about changes
in performance.
It is
interesting to note that there has been no eusocial species whose individuals
have been of a high order of intelligence relative to say apes and pigs. Probably naked mole rats come the closest,
but to argue that they represent highly intelligent eusocial individuals,
involves slippery lines of reasoning that threaten to end in simplistic
arguments.
It is
arguable that such a thing, if possible in principle, is implausible in
practice.
However, H.
G. Wells, the greatest science fiction genius in history, head and shoulders
above any rival so far, supplied us with a thought experiment that was characteristically
ahead of his time. His Selenites with
their ruler the, “Grand Lunar”, in “The First Men in the Moon” were exactly
that, highly intelligent eusocial individuals, and his short story “Empire of
the ants” refers chillingly to related concepts.
There are no whole truths: all truths are half-truths.
It is trying to treat them as whole truths that plays the devil.
Alfred North Whitehead
A few
generalisations on this aspect of the subject may deserve consideration. Viewed simplistically, evolution of
intelligence is fairly simple in that it looks like any other adaptive
evolution. There is a genetically
determined set of variables in a given species in a given environment, and
where some particular values of those variables directly or indirectly affect
reproductive success in a population of significant size, adaptive selection
follows as a matter of course. Many such
genetically determined variables affect behaviour patterns, including flexible
behaviour in response to external stimuli.
Experiments,
both deliberate and incidental, in artificial selection for behaviour patterns,
have been very revealing. By artificial
selection breeds of dogs have had their natures changed within decades. Herding behaviour has been selected for in
sheepdogs, coach escorting behaviour in Dalmatians, and large breeds have
successively been bred for viciousness and gentleness. Within about a dozen generations of
concentrated selection of guinea pigs, hamsters and other small pets for either
docility or aggression, breeders succeeded either in producing almost paralytic
docility or practically pathological viciousness. And in Europe
selection of the honeybee for docility has produced colonies with behaviour
almost unbelievable to the African beekeeper who must deal with more recently
domesticated lines or subspecies.
Selection
for intelligence in the weak sense of behaviour adapted to their environment
has affected all fairly complex animals and colonies and has produced results
in various degrees of complexity and sophistication. The complexity of colonial behaviour (as
opposed to the behaviour of the members of the colonies) could match the
complexity of the behaviour of individual animals practically step for step,
and it proves to be roughly as modifiable through selection.
However,
one thing that I cannot find with any degree of confidence, is an example of
colonies that have generalised intelligence that permits flexible response to
almost arbitrary new challenges, such as individual specimens of primates,
pigs, some carnivores, and some birds exhibit.
Instead, colonies invade adventitious niches where their resilience,
small size, and inexhaustible opportunism are advantageous. Examples include the Argentine ant
(Iridomyrmex humilis), Pheidole megacephala, and the pandemic Formosan
termite.
However, a
few species of non-social cockroaches spread just as successfully. Clearly there is no general way to predict
when social intelligence will be responsible for the success of a species, much
diagnose that it amounts to intelligent response to complex challenges.
The
evolution of colonial behaviour however, is impressively supported by
observation of extant species.
Unfortunately that sort of evidence necessarily reduces to the classic
type of “Just-So Story”, but as long as one is not arguing with radical
fundamentalists or naïve positivists, such abductive reasoning is enough to
illustrate how innate intelligent behaviour may be adapted to the challenges
and opportunities of the environment. We
may not be able to say just how the species involved in fact, but often we can
point to other species that demonstrate that particular hypothesised
intermediate developments are indeed viable.
In
particular, in principle we also can see how the primitive foraging and
parental behavioural components of solitary insects could have become
elaborated progressively to amazing levels of sophistication. As things stand extant species illustrate
behaviour corresponding to putative stages in the evolution even of the dances
of the honeybee.
One remark
about evolution of intelligence in insects (in particular in social insects) is
that they have been extremely conservative in their modes of communication, and
that they use few chemicals for many purposes, and for example, the honeybee
uses similar dances for indicating sources of water, nectar, and new nesting
sites.
I have
severe reservations about this remark.
The
observation is true enough as far as it goes, except that recent work has greatly
multiplied the range of compounds that insects are known to use for
signalling. In guiding mating activity,
some wasps use at least seven successive compounds for control of the seven
successive stages of the mating act. I
already have remarked on the startling degree to which some orchids have
adapted to relying on those wasps for pollination. The key point however, is that the very fact
that the orchids had time to evolve such an elaborate system does imply that
the wasps had been using the same system for a long time, probably many
millions of years.
Anyway,
biological systems in general, not just insect signalling systems, are very
conservative when a change to a working system is difficult to bring about
without teleology. This is particularly
so when multiple complementary components of a complex system must evolve in
synchrony if the system is to survive at all.
Again, I have discussed the topic in an essay on pheromonal evolution.
Superficially
and qualitatively, there are few truly incomprehensible mysteries specific to
the evolution of intelligence as such, whether colonial or individual. When genetic information supports the basis
for nervous control, adaptation in response to selective pressures seems to be
quite as effective as adaptation of genetically determined somatic
features.
Just as the
dinosaurs and whales produced the largest bodies, complete with the necessary
adaptations of support and circulatory systems, so Homo has produced the
largest increase in abstract, flexible intelligence so far. It remains to be seen whether we can thereby
match the duration of the dominance of dinosaurs, which (as an enormously
assorted lot with a vast range of internal succession), lasted for well over
100000000 years.
As for the
idea of a subjective colonial intelligence, I find it hard to imagine one in
the colonial organisms we know, if one omits the view of metazoans as colonies
of cells. However it remains difficult
to justify my failure of imagination as a valid argument, seeing that we have
not yet learned even how to measure subjective intelligence in ourselves, let
alone in alien communities.
There has
however been a major evolutionary leap in our intelligence. It is not at all clear that any other species
ever has developed non-trivial levels of inductive or deductive thought, or
imagination, either individually or in the form of community thinking.
Still,
anecdotal evidence suggests that something like imagination or deduction might
exist in non-primates. I have seen a zoo
hyena lying immobile till it saw me trip over a raw chicken leg camouflaged
with dust. The morsel no doubt had
dropped from the feeding trolley some hours before. Even as I stooped to pick it up without so
much as a betraying glance at the hyena, it charged the fence, correctly
anticipating my throwing it into the enclosure.
The leg barely had time to reach the ground. To me this suggested awareness that the food
had fallen there, and that a human might pick it up and oblige. It also suggested a keen personal interest in
the outcome.
Again, a
hand-reared black rhino has been known to open its enclosure, and release the
dogs to find its human foster mother, who had gone out onto the reserve,
leaving the animals behind. Pigeons have
been experimentally shown to be able to learn specific techniques to open food
containers by watching other pigeons experienced in manipulating the
apparatus.
I am not
aware of anything of such a type in insects, much less in eusocial
communities. In eusocial species? Do symbolic language, complex nests,
abandonment of nests threatened by ants or fire, mean anything? There is a long road before we can formulate
such questions usefully. The day that we
succeed, we will have made giant strides in the comprehension of intelligence
in any form, and that would make the study of communal intelligence worth while
whether it were interesting in its own right or not.
Where we go
from here, in what form, and how fast, is unclear. The fact that we are monkeys rather than
termites precludes human society in its present form from undertaking really
worthwhile, large, long-term projects.
We operate on conflicting interests that bedevil our individual tribes,
let alone our planet. Until, as a
species, we master the virtues of the insect colony, our long term future as a
species is discouraging.
References:
- Gall,
John Systemantics: "How
Systems Work and Especially How They Fail" ISBN: 978-0812906745
- Hofstadter,
Douglas R & Dennett, Daniel
C.: "The Mind's I" ISBN: 978-0465030910
- Hofstadter,
Douglas R: " Goedel, Escher, Bach" ISBN: 978-0465026562
- John R.
Searle: "The rediscovery of
the mind" 1992 ISBN
0-262-19321-3
- Konrad Lorenz: “On
Agression” ISBN: 978-0-415-28320-5
- Sacks,
Oliver:
"Musicophilia"
2008 ISBN: 978-1-4000-335 3-9
- Smith, John Maynard: “On
Evolution” 1972 ISBN: 0 85224 229 8
- Thomas, Lewis: “The medusa and the snail” 1979 ISBN” 0-670-46568-2
- Wittgenstein, Ludwig:
Philosophical Investigations
German & English: ISBN:
978-0-631-23127-1 and : 978-0-631-23159-2
and: ISBN 0-631-23159-5
