Mental Mechanisms: What are the Operations?

Mental Mechanisms: What are the Operations? William Bechtel (bill@mechanism.ucsd.edu) Department of Philosophy and Science Studies Program, University of California, San Diego, La Jolla, CA 92093-0119 a level of organization and awaits the discovery of the na- ture of operations at the level required for successful expla- nation. Abstract An important goal of the cognitive sciences is to explain men- tal phenomena in terms of mechanisms. A mechanistic expla- nation requires characterizing the operations of the mecha- nism’s parts. The challenge in characterizing such operations is illustrated by an example from biology in which some in- vestigators tried to characterize the internal operations in the same terms as the overall physiological system while others appealed to elemental chemistry. Before biochemistry became successful, researchers had to identify operations at a new level of organization—operations over molecular groups. Ex- isting attempts at mechanistic explanation in cognitive science are in a situation comparable to the earlier stage in the bio- logical example, drawing their inspiration either from overall psychological activities or from low-level neural processes, neither of which is likely to provide a successful account of the operations in mental mechanisms. Mechanistic Explanations As philosophers of science increasingly focused on particu- lar sciences in the 1970s, philosophers focusing on biology noted the paucity of laws in biology. Some viewed this as a shortcoming of biology (Rosenberg, 1985, 1994), while others maintained that explanation in biology often takes a different form—articulation of mechanisms. Although there is much commonality in the accounts of mechanism that have been advanced (see, for example, Bechtel & Richard- son, 1993; Glennan, 2002, 1996; Machamer, Darden, & Craver, 2000), there are differences in terminology, scope, and emphasis. On my account A mechanism is a structure performing a function in virtue of its components parts, component operations, and their organization. The orchestrated functioning of the mechanism is responsible for one or more phenom- ena (Bechtel & Abrahamsen, 2005). A mechanism on this account is a system operating in nature whereas a mechanistic explanation is an epistemic product. To arrive at a mechanistic explanation, scientists must rep- resent (sometimes verbally, but often visually in diagrams) the component parts and their operations and the ways in which they are organized. A central feature of such mechanistic explanations is that they decompose a system that is responsible for a phenome- non into component parts and component operations. (Other features of mechanisms, such as the critical role played by organization and the fact that mechanisms are often con- strained by their environments, will not be discussed in this paper.) The parts and operations into which a mechanism is decomposed are closely related: the relevant parts are those that perform operations and hence are working parts, or are operated on by working parts. But it is important to distin- guish parts understood structurally from operations under- stood functionally. Although a full understanding of a mechanism requires both the structural and functional per- spective, different investigatory techniques are required to establish structural and functional properties of components. As a result, a given group of researchers may only be able to secure evidence about only one or the other. It is then often a challenge to link parts with operations (an activity I refer to elsewhere as localization). The cognitive sciences in general have been in the posi- tion of attempting to develop functional decompositions without the benefit of techniques to decompose the brain structurally in relevant ways. In neuroscience, researchers Keywords: Mechanism; explanation; operations; levels of or- ganization; connectionism; symbolic models Introduction Philosophical accounts of cognitive science still commonly treat explanation as a matter of subsuming descriptions of cognitive phenomena under laws in the manner character- ized by the deductive-nomological (D-N) model of explana- tion (Hempel, 1965; Suppe, 1977). Cognitive science expla- nations, however, typically do not appeal to laws. Cummins (2000), for example, argues that laws, or what are com- monly called effects in psychology, do not explain, but are what require explanation. Instead of appealing to laws, ex- planations in cognitive science, as in most of the life sci- ences, appeal to mechanisms. Recently several philosophers of biology have attempted to articulate the conception of mechanism and mechanistic explanation that figures in these sciences. In the first part of this paper I extend this account to cognitive science. In order to construct successful mechanistic explanations, a discipline requires an understanding of the types of opera- tions out of which explanatory accounts can be constructed. Historically developing the appropriate account of opera- tions for a given inquiry has proven challenging. In section two, I illustrate this with an example from biochemistry in which initial attempts to identify the operations that figure in physiological processes focused either on too low or too high a level of organization. Only once investigators learned to identify operations at the appropriate level of organiza- tion did biochemistry develop into the successful science that we know today. In the final section I will argue that in fact cognitive science is in the position biochemistry was prior to the articulation of an appropriate catalog of opera- tions—it too is looking for operations at too low or too high