Evans’s (2014) modularity myths

Evans’s (2014) modularity myths: A mental architecture digest Iris Berent Evans’s book has certainly gotten the attention of linguists, and it has done so for two good reasons. First, E reminds us that linguistics is first and foremost a cognitive science, so one’s account of language must be firmly grounded in a theory of mental architecture. Second, E insists that linguistic theory ought to be subject to careful empirical scrutiny from a multidisciplinary perspective. These reminders are certainly useful, and E should be commended for reigniting the discussion of these core theoretical and methodological issues. It is unfortunate, however, that The language myth does not abide by the same standards. At the center of this book is a fierce critique of the modularity of mind—the hypothesis that the mind is innately equipped with specialized domain-specific faculties. Being one of those putative faculties, language offers a test case for the modularity thesis as a whole. E believes modularity is dead wrong—it rests on an implausible account of neurobiology, is fraught with logical fallacies, and is supported by no empirical evidence. Regrettably, E’s own grasp of the theory is seriously flawed. Considering first the argument from neurobiology, E contends that innately specified modules are precluded by the brain’s plasticity, on the one hand, and the shortage of genes necessary to fully hardwire its synaptic connectivity, on the other. Indeed, an inspection of the brain language system reveals no brain region that is ‘dedicated to language to the exclusion of all else’ (142). Given that those presumed biological prerequisites for a language module are untenable, the thesis of innate modules is a priori doomed. E is of course right about brain plasticity and the absence of ‘language genes’ or ‘language-only hubs’, but these observations are hardly new, and their relevance to the modularity thesis is tenuous. Reviewing similar concerns raised by Martin Gardner in 1952, Jerry Fodor (1983:98) urges us to [r]emember that faculties are, in the first instance, functionally rather than physiologically individuated. And perhaps localization isn’t precisely the notion that Gardner wants, since, after all, there might be neural specificity of some functions that aren’t localized in the sense of being associated with large, morphologically characterizable brain regions. E does not acknowledge (let alone heed) Fodor’s warning and the many subsequent discussions in the literature. Brain plasticity—E’s other strike against modularity—is likewise a questionable challenge, as plasticity is hardly evidence for an unconstrained equipotentiality of the brain. Finally, E’s ‘gene shortage’ worries are based on unfounded assumptions about the genetic encoding of grammatical rules (Marcus 2004). These observations do not imply that evidence from neurobiology is irrelevant, and they certainly do not demonstrate that the modularity thesis is improbable or unfalsifiable. Rather, they remind us that the architecture of the mind cannot be reduced to the neural and genetic levels (Fodor & Pylyshyn 1988). Accordingly, modules can only be defined at the functional cognitive level. E, to be fair, does consider in passing the functional definition of modules, but he rejects it outright on logical grounds. Once again, however, his objections are rooted in a misconception. At stake is the question of how a module selects its input. A module (in the Fodorian sense; Fodor 1983) is both ‘eccentric’ with respect to the inputs it admits and encapsulated from external sources. Computations within a putative ‘face module’, for instance, are immutable by one’s beliefs and goals with respect to a given face, as those sources of information form part of central ‘horizontal’ systems to which the module has no access. E wrongly believes that these two properties—the selectivity of modules with respect to their inputs and their encapsulation from horizontal systems—are mutually incompatible. He worries that a ‘face module’, for instance, cannot sort out faces from, say, vases, so one must invoke central intelligence (‘big brother’, in E’s [End Page 195] words) to direct the ‘perceptual traffic’ into the appropriate module, and this, of course, violates the encapsulation tenet. But this entire concern is misplaced. Fodorian modules are entrusted with their own ‘traffic policing’, and this is possible because they

Argument (complex analysis) · Cognition · Cognitive science · Epistemology · Linguistics · Mental representation · Modularity (biology) · Political science · Scrutiny · Architecture · Categorization, perception, and language · Computer Science · History · Language and cultural evolution · Law · Neuroscience · Philosophy · Psychology · Syntax, Semantics, Linguistic Variation

• The many errors of Vyvyan Evans’ The Language Myth

  Nicholas Allott, Georges Rey•The Linguistic Review•2017

  Add to list

• Rules and representations

  Noam Chomsky•Rules and representations•1980

  Add to list

• Connectionism and cognitive architecture

  Open Access•Jerry A Fodor, Zenon W Pylyshyn•Cognition•1988

  Add to list

• On The Language Instinct

  Steven Pinker, American Psychological Association•PsycEXTRA Dataset•1994

  Add to list

• The Modularity of Mind

  Jerry A Fodor•Modularity of Mind•1983

  Add to list

• Science and Core Knowledge

  Open Access•Susan Carey, Elizabeth S Spelke et al.•Philosophy of Science•1996

  Add to list