The Adaptable Mind: What Neuroplasticity and Neural Reuse Tell Us about Language and Cognition - John Zerilli 2021
Are Modules Innate?
Asking whether modules are innate is problematic for at least three reasons. First, the conclusion of the previous chapter suggests that the subject of the question is not appropriate: we would do better to ask whether brain regions are innate. Second, there is the rather thorny issue of what one actually means by “innate.” Third, the question assumes that a general answer can be given, when it is unlikely that all (or even most) brain structures will have the same developmental story to tell: “there will be cases and cases” (Mameli & Papineau 2006, p. 564). This last concern can be alleviated by concentrating the weight of one’s empirical attention on a particular module, if not by having regard to as much brain-wide evidence as possible. Chapter 2 was my attempt to incorporate a wide survey of the evidence of neuroplasticity—with as many caveats and limiting clauses as its interpretation reasonably warrants—while Chapter 7 is my attempt to home in on one particular system (namely, language). In the present chapter, I aim to build on the interpretation of neuroplasticity that I began in Chapter 2.
This still of course leaves us with the problem of having to define what we mean by “innateness,” a far from trivial matter (Griffiths 2002; Mameli & Bateson 2006; Bateson & Mameli 2007; Mameli & Bateson 2011). The trouble is that the term is as ambiguous as it is entrenched, and some have wondered whether it can perform a useful function in the sciences at all. With so large a variety of distinct notions lying beneath the surface, it becomes very easy to commit fallacies of ambiguity (Griffiths et al. 2009). One might, for example, infer that a trait is species-typical by virtue of its being the product of natural selection, or developmentally fixed by virtue of its being species-typical (O’Neill 2015). In the result Griffiths (2002) recommends having done with the term altogether, and suggests that scientists should specify explicitly what they mean on any given occasion. I do not take an “eliminativist” stance myself, but, along with Griffiths, do think it absolutely essential to make explicit the sense in which the term is being used. Following O’Neill (2015), what I have in mind is insensitivity relative to some specific set of environmental variations. This is the idea of developmental robustness or environmental canalization, broadly speaking (Ariew 1996, 1999, 2007; Mameli & Bateson 2006; O’Neill 2015), except that it is explicitly relativized to specific environmental factors (Bateson & Mameli 2007, p. 823).1 After all, no trait is developmentally robust in an absolute sense, yet the claim is frequently made without specifying the environmental factors with respect to which the trait is supposed to be robust. For the most part this is not a problem, since it is usually clear in a given context which environmental factors are relevant (O’Neill 2015, p. 212). Still, it is important to bear in mind that a trait’s invariance (or otherwise) is always relative. In the present context, we are concerned with the innateness of modules—iterated cortical structures with distinctive columnar and laminar patterns of organization. It should by now be clear that modules are not insensitive with respect to such experiences as learning, injury, and sensory deprivation, regardless of how young or mature the organism happens to be. The extent of both intramodal and crossmodal plasticity, as well as evidence for the extensive rewiring of latent supramodal connection channels, does much to discredit the traditional nativist assumption of “hardwired” cognitive capacities with rigid developmental schedules (Marcus 2004).
Yet this cannot be the full story. For one thing, sensitivity with respect to a particular set of environmental factors does not entail sensitivity with respect to others; and in the absence of factors to which a trait is sensitive, its development might well be considered robust. For instance, when developmental biologists speak of “activity-independent” cell differentiation, which results in cortical areas’ acquiring fixed structural characteristics in utero, they can be taken to imply that at least some aspects of modular development are insensitive relative to certain factors, although obviously not with respect to the factors that can be expected to become influential at a later stage of development; i.e., during postnatal “activity-dependent” cell differentiation (Saitoe & Tully 2001, p. 193; Kolb et al. 2001, p. 225; Sanes & Jessell 2013, p. 1259).2 Furthermore, sensitivity admits of degrees (Collins 2005). Granted that cortical development is robust in certain respects, how robust is an important question in each case. Thus there are really two senses in which we can speak of invariance as being a matter of degree: along one dimension, we can say that the more factors with respect to which a trait is robust, the more invariant it will be; while along a second dimension, the more a trait is robust to variation in any single factor, the more invariant it will be (Griffiths & Machery 2008, p. 399). One may therefore legitimately inquire as to whether cell differentiation results in a stereotyped but essentially crude pattern of synaptic connections and brain regions before birth, or whether it results in more robust operations that limit and constrain the functions these regions can later take on. There is scope for genuine disagreement here between those who think there is a lot of prewiring, combined with some inevitable rewiring during development (Marcus 2004), and those who think there is comparatively little prewiring, with a lot of rewiring during development and later life (see the discussion by Mameli & Papineau 2006, pp. 563—564).
In this chapter, I shall argue that the evidence of neuroplasticity supports neither a traditional nativist nor yet a strictly antinativist interpretation of development. Rather, we seem to be confronting a phenomenon that falls somewhere midway between the two extremes of developmental hardwiring and original equipotentiality. While the extent of the neuroplastic responses we considered in Chapter 2 is undoubtedly impressive, and sometimes vast, a closer look at these cases suggests that the pattern of responses is constrained. For all their plasticity, brain modules and regions appear to be significantly robust in the presence of such environmental variables as learning, injury, and sensory deprivation. More precisely, the changes that do occur are exactly what one would expect to find under the assumption that cortical regions have robust processing capabilities and clear input preferences (what I earlier described as a “bias”). This is not a traditional nativist picture, to be sure, but neither is it antinativist. (Warning: the evidence to follow is circumstantial, the argumentation non-demonstrative and abductive. In the realm of cognition, however, we frequently find ourselves with little else to go on.)