There is much of interest in this article on misbeliefs and their “adaptive” value in evolutionary terms, but I have a serious problem in knowing how to assess the arguments: I cannot see what all this has to do with evolution, understood as natural selection of traits inherited though the genome.
McKay & Dennett (M&D) are perfectly well aware of this overall requirement: “To be sure, it might plausibly be argued that delusions are psychologically adaptive in certain scenarios (as the above reverse Othello case suggests). But this does not establish a case for biological adaptation” (sect. 10, para. 5, emphasis in original). The issue of whether beliefs or other mental representations could affect gametes has surfaced in the past in the Artificial Intelligence (AI) context: Longuet-Higgins (in Kenny et al. Reference Kenny, Wadding, Longuet-Higgins and Lucas1972) discussed how such brain modifications might conceivably affect the gametes, yielding a sort of post-Lamarckianism (cf. Lamarck Reference Lamarck and Hugh Eliot1809/1914/1984) with a possible underlying mechanism stated in genetic terms. I am not sure how seriously that discussion was intended to be taken but, unless some such discussion is set out, what can we make of the arguments of M&D except seeing them as simple Lamarckianism without any mechanism at all – that is, mere adaptivity of traits transferred to a successor generation, such as a tendency to create life-prolonging misbeliefs about, say, health (sect. 10) but with no suggestion of how this could be done?
The only alternative account – and M&D do not give this either – is to fall back on a long and honorable tradition of “cultural transmission” from Lovejoy's Great Chain of Being (Lovejoy Reference Lovejoy1936) to Dawkins' memes (Dawkins Reference Dawkins1989) and perhaps even Sheldrake's morphic resonance (Sheldrake Reference Sheldrake1987). The problem with all this is that it has no known empirical, biological, substrate and hence nothing to do with evolution as the latter is normally understood.
Our authors' dilemma is very like Chomsky's in his lifelong campaign to establish a notion of “universal grammar” that is the biological and inherited substrate that allows humans, but not other species, to develop languages of such and such a type. Chomsky (Reference Chomsky1965) has never established clearly what features constrain the class of languages we call human and which, if found, would presumably correspond to the coding in gametes. Chomsky had additional difficulties because of his distaste for the whole idea of evolution as applied to languages, but the problem is not far from Herb Simon's closely related speculations (Simon Reference Simon1996) about the need for an inherited capacity to manipulate hierarchically structured representations to explain the mechanisms underlying vision, language, and reasoning. There have also been comic diversions like the discussion following Fodor's Language of Thought (Fodor Reference Fodor1981) about the need for concepts as complex as telephone to be innate, because they could not be decomposed into simpler concepts without loss of meaning. Again, it is not clear how serious the discussion of the innateness, and inheritability, of the concept telephone was intended to be.
But is the discussion of the target article any more firmly based than Fodor's? Can we imagine that special classes of helpful misbelief, and their associated concepts (e.g., “supreme being” in the article), are any better candidates for heritability through genetic representations than Fodor's telephone? And if they are not heritable, why is this discussion taking place in the context of natural selection and Darwinian evolution?
It may be said that the above criticism is unfair and beside the point: If we accept that there is a heritable feature corresponding to human languages, broadly conceived as a class, and that we have the feature and other higher mammals do not, then we can speculate about its content, as Chomsky and his school have, without any knowledge of coding within organic matter of the structures that heritable features require. Similarly, it may be said, we can speculate about the features of a general belief and reasoning system, in just the way that AI has for 50 years, without brain codings, but under general assumptions about which aspects of such a system will aid survivability or reproduction.
But Chomsky did realize that none of this could concern itself with particular languages, since a new brain can learn any language, but only with underlying representational features common to all languages. It seems clear to me that the target article does not observe that constraint for beliefs and therefore has even less plausibility than Chomsky's long-running campaign. The article is largely about particular (mis)beliefs and their function: “I am sexually attractive,” “I will get better from this disease,” “A supreme being watches my actions.” Is their coding any more plausible than that of the concept telephone, or some structural feature of Indo-European languages? Is it not much more plausible that all this is learned within some very general structure we can barely specify, though Simon (Reference Simon1996) probably made the best attempt to guess it.
Suppose we stand a little further back, and accept the general Quinean view (see Quine Reference Quine1953) that we test not individual beliefs against the world but whole belief systems: anthropological, Newtonian, Einsteinian, religious, naive physics (Hayes Reference Hayes and Michie1978), folk psychology, and so forth. These are accepted or not, give satisfaction to individuals or not, but none can be completely right or true as a whole; and no scientist thinks any such theory complex is final or correct, though some are plainly better than others, and may have served our species well for millennia. It can be very hard to separate out individual beliefs as misbeliefs, without regard to the whole complex they come from; this is a cliché of the philosophy of science as much as it is of anthropology. It is also very hard to get any clear evidence on how much, if at all, of these complexes could be hard-coded in genetic material: Is jumping out of the way of rapidly approaching large objects innate in humans, as it is in some insects?
I believe one needs to hear more about these issues, and above all about the relationship to a biological substrate, before discussions of the adaptivity of particular beliefs can be more than re-warmed Lamarck – no matter how much fun the discussion is.
