New Look Faculty Psychology - Saving Faculty Psychology: Debunking the Argument from Multiple Realization

The Adaptable Mind: What Neuroplasticity and Neural Reuse Tell Us about Language and Cognition - John Zerilli 2021

New Look Faculty Psychology
Saving Faculty Psychology: Debunking the Argument from Multiple Realization

Exactly how the science of mental faculties will have to change to accommodate the reality of neural reuse is a matter of some uncertainty, but even those such as Michael Silberstein who assert that “the autonomy and irreducibility of folk psychology are assured” concede that reuse means “scientific psychology must be heavily revised” (2016, pp. 27—28). My position, and the position it was the aim of the previous section to make feasible, is that psychology and neuroscience are friends, not enemies. I start from the premise that the best way to understand something is to break it down, and that the best and most natural way to break something down is to carve it at its joints; i.e., in such a way as to respect its physical constitution and design. I take this to be near truism. Now the fact remains that, as I have tried to show, the endeavor to understand the mind has come a long way from the days when Herbert Simon and David Marr reasoned from evolutionary principles that the carve-up of relevance to the mind produces independently modifiable subcomponents that correspond to functionally specific units of high-level psychology (Sternberg 2011, p. 158). I am not saying that the mind is bereft of dissociable subcomponents (see Chapters 4 and 5), but our ideas about them have certainly changed. As long as psychology wishes to carve nature at its joints, then, it will have to update its ideas about what those joints are. At a minimum, neural reuse mandates an approach to decomposition that assigns domain-neutral functions to brain regions; and if such regions are the stuff of higher-level cognition, it is surely not reasonable to insulate the higher-level cognitive ontology from their effects. One does not have to be a ruthless reductionist or eliminativist to recommend that our sciences so develop as to facilitate mutual interaction and even potentially unification. If this is right, cognitive models of distinct domains should be placed upon such a footing as will best accommodate the possibility of interaction. We will learn more about the faculties, not less, if we can appreciate their deeper-level associations. It is interesting to note in this respect that the 2010 edition of David Marr’s Vision—a book that is (in)famous for having made the strict independence between levels of inquiry an article of faith in the cognitive sciences—contains an Afterword that chimes nicely with the message I am trying to convey here:

(1) insights gained on higher levels help us ask the right questions and do the right experiments at the lower levels, and (2) it is necessary to study nervous systems at all levels simultaneously. From this perspective, the importance of coupling experimental and theoretical work in the neurosciences follows directly; without close interaction with experiments, theory is very likely to be sterile. (2010 [1982], p. 364)

One example of this mutual endeavor can be seen in neurolinguistic work that tries to integrate formal results from the Minimalist Program in syntax. (In adverting to this work, I do not mean to endorse the Minimalist Program, only to illustrate how researchers in fields that have typically been seen as antagonistic to one another can come together in the interests of science.) David Poeppel (2015) remarks that the goals of systems neuroscience and research in syntax have aligned in the past two decades. I touched upon some of the relevant systems neuroscience in my discussion of canonical neural computations (§ 7.3). The discovery of primitive computations such as filtering, divisive normalization, and predictive coding bodes well for the basic assumptions behind the Minimalist Program. Both research programs seek to uncover fundamental and (as much as possible) domain-general operations underlying various cognitive phenomena. Merge can be seen as another such computation (§ 7.2). The part of Merge that combines elements is a close analogue of binding (or concatenation) operations within systems neuroscience (Poeppel 2015, p. 144). Given the expressions A and B, the binding operation produces a new expression (A, B →{A,B}). A separate procedure then labels the output. This work is encouraging in one important respect. Traditionally, a major obstacle to collaboration between neuroscientists and linguists was the abstruseness and intractability of transformational-generative rules. Simplification of these rules at least makes interdisciplinary collaboration possible and should be seen as a step in the right direction, inasmuch as many linguists now seem to have an eye to fundamental neural computations.

Returning to reuse, as I suggested, the most straightforward outcome on the table is for the higher-level ontology to incorporate the lower-level one—i.e., the level of fundamental computations performed in modules and discrete brain regions. But what is the nature of these primitives? Are we talking about discrete domain-general computations in specific cortical sites, with a one-to-one mapping between primitives and brain regions (in the manner contended by Russell Poldrack)? Or are we talking more about dispositions, so that an individual brain region represents a particular complex of primitives, with a many-to-many mapping between primitives and brain regions (in the manner contended by Michael Anderson)? Then there is the issue of how faculty psychology is to proceed in light of nondecomposition and network dynamics (see my discussion in § 5.1). Some structures (maybe many?), as we saw, are not classically decomposable—their properties are not additive in a bottom-up sort of fashion—even though functional decomposition is virtually an article of faith in cognitive science. These are questions to be clarified and hopefully resolved in coming years. What they provide is a sense of the terms that any negotiated settlement between psychology and neuroscience is likely to take, since it is certain that faculty psychology will have to reckon with these primitives one way or another.