Re: Dunbar Grooming

From: Stevan Harnad (harnad@coglit.ecs.soton.ac.uk)
Date: Tue Dec 07 1999 - 10:55:48 GMT


On Mon, 6 Dec 1999, Pocs Kata Rita wrote:

pkr> *Is there any proof that the evolution of neocortex size is driven
pkr> purely by ecological factors that select for group size? Could not the
pkr> increase of neocortex size be attributed in part to language use rather
pkr> than just the need for language?

You are of course absolutely right, but let me put this in context,
methodologically nd empirically speaking:

(1) There is no question of 'proof' here at all. We only have proof
(which means that something is shown to be necessarily true, on pain of
contradiction) in mathematics. In science we have only evidence
supporting a hypothesis.

(2) The correlation is in itself 'evidence' in support of a causal
hypothesis, but is is evidence of the weakest sort. For A to cause B it
is a NECESSARY condition that they should be correlated (if they are
not, if their co-occurrence is random, then it is not possible that
their relationship is causal -- or rather, it is extremely unlikely),
but that is by far not a SUFFICIENT condition. The cause could be in the
other direction, or there could be a thrid factor, C, that causes both A
and B. Most likely: A, B and C will prove to be too simple and small a
subset of the variables involved, and sorting out cuase/effect will be
more like the blind men and the elephant...

Regarding group size, I agree with you that it sounds more plausible
that language caused bigger groups rather than vice versa.

Note that this sort of historical, backwards extrapolative
hypothesis-testing suffers from certain intrinsic disadvantages (which
is why it has often been criticized as circular, arbitrary, and
"just-so" stories).

Can you think of better ways to formulate and test hyoptheses on a topic
like this, which is essentially a one-off event in the distant past,
rather like the Big Bang?

  rd> Grooming as a bonding mechanism is highly inflexible:
  rd> -it is all but impossible to do anything else while grooming or
  rd> being groomed.
  rd> -it is an essentially dyadic activity: only one other individual can
  rd> be groomed at a time.

pkr> *Exept in the case of chain-grooming -but I guess that does not
pkr> make a significant difference.*

Correct. The Dunbar theory has much more serious problems than that!

  rd> Speech (can) be combined with almost every other activity,it can
  rd> also be used to address several different individuals
  rd> simultaneously. Thus, language introduces major savings by
  rd> allowing an individual to do two different things at once.
  rd> My suggestion, then, is that language evolved as a "cheap" form
  rd> of social grooming, so enabling the ancestral humans to maintain
  rd> the cohesion of the unusually large groups demanded by the
  rd> particular conditions they faced at the time.
  rd> That language (and hence speech) might have evolved as a
  rd> consequence of the need to increase group size raises the
  rd> question of just how it functions as a bonding mechanism.

pkr> *Is the enlargement of groups a purely ecological necessity, or could it
pkr> be attributed to the use of a possibility given by by previously changed
pkr> factors (language development, neocortex size),or simultanneously
pkr> changing factors.*

You are again, correctly, questioning the causal picture. Note, though,
that what you need is a rival hypothesis, rather than merely scepticism
about Dunbar's. Dunbar was trying to account for the origin of language;
to reply that language was the cause and not the effect is to beg the
question, is it not?

  rd> Conventionally, language has always been interpreted in terms of
  rd> the exchange of information, and this has usually been
  rd> understood as being the exchange of information about the environment.
  rd> However, the social intelligence hypothesis for the evolution of
  rd> large brain size in primates implies that the acquisition and
  rd> manipulation of social knowledge is the primary consideration.

Are you satisfied with Dunbar's partitioning of "environmental
information" and "social information"?

  rd> The fact that language can be interpreted as fulfilling the same
  rd> role as social grooming suggests that, rather than being the
  rd> selective factor driving brain evolution, ecologically-related
  rd> information-exchange might be a subsequent development that
  rd> capitalisedon a window of opportunity created by the availability
  rd> of a computer with a substantial information-processing capacity.

pkr> *Meaning that the ecologically induced increase in group size lead
pkr> to the need to develop language to facilitate social needs, which could
pkr> only be achived by and resulted in increase in brain size, which could
pkr> then be used for a number of other things.
pkr> Is (or Can) the change in the three 'variables' be treated as subsequent
pkr> or seperate events which did not keep on affecting each-other?

We are at the core of Dunbar's hypothesis here. You are right to say
that there may have been more variables involved and the causal picture
may be more complicated and interactive, with positive and negative
feedback effects -- but might there not be even deeper problems with
this hypothesis?

Great outcomes can come from small causes, but are we ready to conclude
that aa cognitive capacity as powerful and general as language came from
such modest origins -- and these modest origins in particular?

Does "social intelligence" -- understood in this sense -- really
contain, in microcosm, the germ of our general, all-purpose intellectual
capacity, as embodied in our linguistic capacity?

Child and animal "Theory of Mind" (ToM) capacities are currently a
fashionable research topic: Is linguistic capacity just a ramified ToM
capacity?

  rd> How might language function as a mechanism for social
  rd> bonding?
  rd> -at least two possibilities.
  rd> 1) enabling (participants) to acquire information about each
  rd> other's behaviour by direct observation. This appears to be one
  rd> way in which social grooming itself might work
  rd> 2)permit(-ing)s the acquisition of information about third party
  rd> social relationships, thereby enabling an individual to acquire
  rd> knowledge of the behavioural characteristics of other group
  rd> members without actually having to observe them in action -
  rd> considerably widening an individual's sphere of social knowledge
  rd> relative to what would be possible from direct personal
  rd> observation.

Is this 3rd party social information more important than information
about food and other resources, territory, predation, danger -- and the
all-important making "explicit" of learning, knowledge and memories
that are otherwise private and "implicit".

Did all-purpose symbolization and propositionality arise in the service
of perception of social relationships?

  rd> The acquisition and exchange of information about social
  rd> relationships is clearly a fundamental part of human
  rd> conversation.I suggest that it implies that this was the function
  rd> for which it evolved.

pkr> *Language as is shown, is more effective than grooming in
pkr> the sense that a larger audience or number of participants can
pkr> effectively be involved.

That is an fine reason why one of the EFFECTS of language might be
"social grooming" -- but is it also a plausible original CAUSE?

pkr> But this does not yet prove its effectiveness in the sharing of
pkr> information about patterns of behaviour regarding individuals not
pkr> present. Each individual, despite knowledge of them
pkr> might have to reappropriate his/her second-hand knowledge and negotiate
pkr> their relationship in interaction, as primates would have to test groom
pkr> outside their sub-groups.*

It certainly is a mystery how vocalization could gain this remarkable
symbolic power, to convey information about past, distant, and future
matters -- but why those distal matters have to be especially social
particularly, and how that would drive the advent of symbolization,
seems to me to be the QUESTION, not the ANSWER here.

  rd> (3) this explanation clearly stands in direct contrast to the
  rd> conventional wisdom that language developed in the context of
  rd> hunting to enable early hominids to communicate about the
  rd> location of possible prey and to plan coordinated hunting
  rd> expeditions. Indeed, the explanation for the increase in brain size
  rd> within the hominid lineage on which my argument is based itself
  rd> stands in contradiction to the conventional wisdom that these
  rd> large brains evolved to enable humans to hunt and/or
  rd> manufacture tools.

I would agree with Dunbar that the hunting- and/or tool-making/use-
hypotheses for the origin of language are no more convincing or
substantive than the grooming theory: Can anyone else suggest something
better?

pkr> *Meaning that, contrary to the general supposition that language
pkr> evolved to enable humans to conquer their natural environment,
pkr> it evolved to allow humans to handle their social environment,
pkr> the latter causing greater difficulties. The significance of the
pkr> thought cannot be overemphasized in my opinion, for language
pkr> which may be analysed as a social phenomena here is related to a
pkr> social origin.

You seem to be more sympathetic to the social-origins view -- and
undeniably there is an essential social dimension to language. We speak
to one another, after all, not to ourselves. (We speak to ourselves too,
but surely this was an EFFECT of language, not a CAUSE of it -- or was
it?)

But apart from the fact that language is necessarily a social ACTIVITY,
does it follow that what it was ABOUT, originally, must have been
social?

Also, what if language did not begin with vocalization, as is assumed by
Dunbar, but it began with GESTURE (as others have suggested)? What does
that do to the grooming theory (for gesture seems to have the same
disadvantages as the old kind of grroming!)?

  rd> (2) How is it that, despite these apparent cognitive constraints
  rd> on group size, modern human societies are nonetheless able to
  rd> form super-large groups (e.g. nation states)?
  rd> Two observations are worth making here.
  rd> -the structure of these super-large groupings is not particularly
  rd> stable through time

Tell that to the Greeks, the Armenians, the Kurds, the Israelites...

  rd> -language has two unusual properties that make it possible to
  rd> form groups that are substantially larger than the 150-200
  rd> predicted by neocortex size: it allows us
  rd> (1) to categorise individuals into types and
  rd> (2) to instruct other individuals as to how they should behave
  rd> towards specific types of individuals within society.

And how we could extend that list, inifinitely, in both social and
nonsocial directions. What evidence that our all-purpose categorization
capacity originated in social categories?

http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad87.categorization.html

pkr> * Much of human conversation, and most of (if not all) primate
pkr> grooming, is about the negotiation of relationships (of relative
pkr> power etc.) between the participants (acc. to sociopragmatic
pkr> theory).

On what sort of data is this quantification based?

It is not that I doubt that most human conversation is trivial (and
perhaps social); but the real question is whether that is representative
of original causes or just an effect of having this cognitive
nuclear-weapon for everyday use?

pkr> Social knowledge (or information about social relations)
pkr> is not only gained but created, interpreted and changed through
pkr> conversation and grooming. If considered so, the value of
pkr> information gained about a third person not present, may be
pkr> lessened as relations with the third will have to be renegotiated,
pkr> which would need direct personal observation and/or
pkr> participation.*

That does make us seem rather mechanical. Surely it's useful to be able
to talk/plan behind someone's back -- but that includes prey and
predators, not just conspecifics (is that still social?), and, while
we're at it, it includes all other distal objects, events and states
too.

  rd> This ability to categorise individuals into types clearly makes it
  rd> possible to create very much larger groups than is possible by
  rd> direct interaction. It is only necessary to learn how to behave
  rd> towards a general type of individual, rather than having to learn
  rd> the nature of each individual relationship.

That speaks to the value of category learning -- but why particularly
social categories (predators, prey, foraging opportunities, natural
dangers -- all these are "types" too)? and why/how language?

There is indeed a relationship between categorization and language, but
is this it?

  rdy By structuring
  rd> relationships hierarchically in this way, social groups of very
  rd> substantial size can in principle be built up.

Look at Aristotle's taxonomies to discover how hierarchical ALL kinds of
things are; nothing unique, special, or even representative about social
hierarchies.

pkr> *Categorization is indeed a great advantage of language, but I
pkr> would think it works in ways more complex than considered
pkr> here. Categorisation may be more effective in language but that
pkr> does not necessarily mean it only works through language.

I agree. Moreover, what is the cause/effect relation between
categorization and language?

pkr> Takig
pkr> apes as an example their groups are probably divided into
pkr> categories of members according to age, sex, physical power etc.
pkr> which are marked with observable differences and of which all
pkr> members are aware. Having first hand experience of how to
pkr> behave with members of a certain category, on meeting
pkr> individuals from other groups would probably clearly be able to
pkr> identify their category and act accordingly.

True. And equally true of prey/predator and inanimate object/event/state
categories. And equally true of the categories that nonlinguistic
species surely learn too!

  rd> Language seems to be a far from
  rd> perfect medium for acquiring detailed social knowledge about
  rd> other individuals: secondhand knowledge, it seems, is a poor
  rd> substitute for the real thing. Indeed, it is conspicuous that when
  rd> we do want to establish very intense relationships, we tend to do
  rd> so through the much more primitive medium of physical contact
  rd> rather than through language. The kind of "mutual mauling" in
  rd> which we engage under these circumstances bears a striking
  rd> resemblance to social grooming in other primates and suffers

This is all true, but sounds like bad news for the grooming hypothesis,
rather than support. And it could be said about ANY category: There are
always two ways to get to know it: By direct sensorimotor "acquaintance"
and by indirect symbolic "description." So?

  rd> (3) To what extent is language a uniquely novel solution
  rd> confined to the hominid lineage?
  rd> many species of anthropoid primates to coordinate spacing
  rd> between individuals of the same group.
  rd> (T)he calls allow the audience to infer a great deal about the
  rd> event or situation on which the caller is commenting, even in the
  rd> complete absence of any visual information.

True. But this is nonlinguistic. And it is by no means certain that the
origin of language was in the vocal modality...

  rd> This would seem to suggest that many of the basic properties of
  rd> speech and language were already available in the more advanced
  rd> nonhuman primates. What was required was their close
  rd> integration and elaboration, and this may have been dependent on
  rd> a significant increase in neocortex size in order to provide the
  rd> necessary computing capacity.
  rd> (T)he evolution of this increased capacity arose out of the need
  rd> to coordinate the large number of inter-personal relationships
  rd> necessary to maintain the cohesion and stability of larger than
  rd> normal groups.

pkr> *This may hinder the modification of the hypothesis, to what I
pkr> would like to see, but perhaps would not make it impossible.*

I1m not sure what you mean here, but it is certainly worth asking how
one could further test the hypothesis, how one could show it was
incorrect, and what might be more plausible altenative hypotheses.

--------------------------------------------------------------------
Stevan Harnad harnad@cogsci.soton.ac.uk
Professor of Cognitive Science harnad@princeton.edu
Department of Electronics and phone: +44 23-80 592-582
Computer Science fax: +44 23-80 592-865
University of Southampton http://www.cogsci.soton.ac.uk/~harnad/
Highfield, Southampton http://www.princeton.edu/~harnad/
SO17 1BJ UNITED KINGDOM



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:06 GMT