Robin Dunbar's ideas on the origin of language

I recently finished reading Robin Dunbar's "Grooming, Gossip and the Evolution of Language", a fairly short and quite readable book which advances the hypothesis that the human language faculty evolved initially for the purpose of "social grooming", i.e. strengthening social bonds between individuals in larger social groups, and then later was co-opted for the purpose of "gossiping", i.e. talking to each other about each other. More generally the book is concerned with the connection between cognition and social behaviour (and contains discussion of the well known "Dunbar's number").

The ideas in the book are well argued and I find myself without any good reason to reject most of them, although the possibility exists that there are better argued alternatives I'm not aware of. My biggest complaint is that Dunbar's rhetoric seems to blur the distinction between what language fundamentally does and how we preferentially use language. I don't have exact quotes, but there are several places in the book where Dunbar says things to the effect of "language evolved not for the purpose of exchanging information about the world around us, as we so often assume, but for exchanging information about each other for social purposes". This phrasing baffles me somewhat, because other human beings are part of the world around us. Language is, by Dunbar's hypothesis, for exchanging information about the world around us, it just happens that kind of information about the world around us which we preferentially exchange - and possibly the only kind that our ancestors exchanged - is each other.

This might seem like an odd or insignificant complaint to make, but my motivation for reading Dunbar's book (and for reading more in general about the evolutionary origins of language) was considering the applicability of Andersonian rational analysis to language. My recent research has centred around the Uniform Information Density (UID) hypothesis, which is theoretically motivated by the idea that human language is a roughly optimal solution to the problem of high-speed, high-reliability exchange of information (rational analysis in general holds that human cognitive faculties represent roughly optimal solutions to specific problems). I have heard people question the applicability of this view to language, and in particular suggest that language is not essentially "for" information exchange but rather for various social interactions. It seems implicit in this line of argument, and in much of Dunbar's rhetoric, that these are orthogonal or mutually exclusive goals, but after reading this book I think that in fact one is simply a special case of the other, which is good news for rational theories of language.

This reassurance aside, I think the other thing I got most of reading Dunbar's book was an appreciation for the continuity between the behavioural (and so presumably cognitive) traits of primates, including humans. Cognitive science seems to be a fairly anthropocentric field - for obvious reasons - but it seems to me like it would often make a lot of sense to take a more inclusive view. Perhaps attempts at modelling individual differences should also try as best they can to model inter-species differences at cognitive tasks which non-human primates are unanimously considered capable of?