Design Principles and Mechanistic Explanation

Metascience, 2010

The SAGE Handbook of Theoretical Psychology. (Eds.) Hank Stam and Huib Looren de Jong.

Philosophers of psychology debate, among other things, which psychological models, if any, are (or provide) mechanistic explanations. This should seem a little strange given that there is rough2 consensus on the following two claims: 1) a mechanism is an organized collection of entities and activities that produces, underlies, or maintains a phenomenon, and 2) a mechanistic explanation describes, represents, or provides information about the mechanism producing, underlying, or maintaining the phenomenon to be explained (i.e. the explanandum phenomenon) (Bechtel and Abrahamsen 2005; Craver 2007). If there is a rough consensus on what mechanisms are and that mechanistic explanations describe, represent, or provide information about them, then how is there no consensus on which psychological models are (or provide) mechanistic explanations? Surely the psychological models that are mechanistic explanations are the models that describe, represent, or provide information about mechanisms. That is true, of course; the trouble arises when determining what exactly that involves. Philosophical disagreement over which psychological models are mechanistic explanations is often disagreement about what it means to describe, represent, or provide information about a mechanism, among other things (Hochstein 2016; Levy 2013). In addition, one's position in this debate depends on a host of other seemingly arcane metaphysical issues, such as the nature of mechanisms, computational and functional properties (Piccinini 2016), and realization (Piccinini and Maley 2014), as well as the relation between models, methodologies, and explanations (Craver 2014; Levy 2013; Zednik 2015). Although I inevitably advocate a position, my primary aim in this chapter is to spell out all these relationships and canvas the positions that have been taken (or could be taken) with respect to mechanistic explanation in psychology, using dynamical systems models and cognitive models (or functional analyses3 ) as examples.

The Routledge Handbook of Mechanisms and Mechanical Philosophy

Systems biology is a new and highly interdisciplinary field that combines elements from molecular biology and physiology with quantitative modelling approaches from disciplines such as engineering, physics, computer science, and mathematics. The term 'systems biology' was used originally in 1968 by Mesarović to urge the use of systems theory to understand biological systems (Mesarović 1968); some commentators would trace the historical roots even further back (Green and Wolkenhauer 2013). But when the term is used in the context of contemporary bioscience it typically refers to a much more recent approach, initiated in the late 1990s as a response to the new experimental techniques and fast computers that allowed the rapid sequencing of DNA and automated measurements of molecular interactions (Ideker et al. 2001; Kitano 2001). These innovations afforded major new initiatives in the life sciences but also produced unforeseen challenges. Systems biology addresses one of these, namely the interpretation of extensive quantitative data via mathematical and computational modelling (Alberghina and Westerhoff 2005; Boogerd et al. 2007). Research in systems biology is driven by complex problems that require multidisciplinary integration (Carusi 2014; MacLeod and Nersessian 2014; O'Malley and Soyer 2012). Consequently, it is a diverse field. Some proponents pursue strategies that extend molecular biology with sequence-based tools (see Chapter 24), while others explore the relevance of abstract mathematical systems theory to molecular interactions (O'Malley and Dupré 2005). Common to all branches of systems biology is the willingness to borrow reasoning and representation tools from engineering and the physical sciences, including network diagrams and graph-theory, other types of mathematical modelling (primarily ordinary differential equations) and computational simulations. We focus on just one of the many possible questions about systems biology: To what extent can the modelling strategies and explanations in systems biology be characterized as mechanistic?

Model-based reasoning in science is often carried out in an attempt to understand the kinds of mechanical interactions that might give rise to particular occurrences. One hypothesis regarding in-the-head reasoning about mechanisms is that scientist rely upon mental models that are like scale models in crucial respects. Behavioral evidence points to the existence of these mental models, but questions remain about the neural plausibility of this hypothesis. This chapter will provide an overview of the psychological literature on mental models of mechanisms with a specific focus on the question of how representations that share the distinctive features of scale models might be realized by neural machinations. It is shown how lessons gleaned from the computational simulation of mechanisms and from neurological research on mental maps in rats can be applied to make sense of how neurophysiological processes might realize mental models. The goal of this chapter is to provide readers with a general introduction to the central challenge facing those would maintain that in-the-head model based reasoning about mechanisms in science is achieved through the use of scale-model-like mental representations. 1.1 Overview A central form of model-based reasoning in science, particularly in the special sciences, is model-based reasoning about mechanisms. This form of reasoning can be effected with the aid of external representational aids (e.g., formalisms, diagrams, and computer simulations) and through the in-the-head manipulation of representations. Philosophers of science have devoted most of their attention to the former, but the latter is arguably at the heart of most of what passes for explanatory understanding in science (Sec. 1.2). Psychologists have long theorized that the humans and other creatures (e.g., rats) reason about spatial, kinematic, and dynamic relationships through the use of mental representations, often termed mental models, that are structurally similar to scale models, though clearly the brain does not instantiate the very properties of a modeled system in the way that scale models do (1.3). A key challenge facing this view is thus to show that brains are capable of realizing representations that are like scale models in crucial respects. There have been several failed attempts to show precisely this, but a look at how computers are utilized to model mechanical interactions offers a useful way of understanding how brains might realize mental representations of the relevant sort (Sec. 1.4). This approach meshes well with current research on mental maps in rats. In addition, it has useful ramifications for research in A.I. and logic (Sec. 1.5), and it offers a promising account of the generative knowledge that scientists bring to bear when testing mechanistic theories while also shedding light on the role that external representations of mechanisms play in scientific reasoning (1.6). 1.2 Mechanistic Reasoning in Science A common reason that scientists engage in model-based reasoning is to derive information that will enable them to explain or predict the behavior of some target system. Model-based explanations provide scientists with a way of understanding how or why one or more explanandum occurrences came about. A good model-based explanation will typically provide the means for determining what else one ought to expect if that explanation is accurate 1that is, it will enable one to formulate predictions so that the explanation may (within widely known limits) be tested. 1 One must bear in mind, however, that models are often accurate only in certain respects and to certain degrees [2]. Model-based reasoning can, corresponding to the diversity of representational structures that count as modelsincluding external scale models, biological models, mathematical formalisms, and computer simulationstake many forms in science. As for what models represent, it is now widely accepted that mechanisms are one of the principal targets of modelbased reasoning. This is most obviously true in the non-basic sciences (e.g., biology, medicine, cognitive science, economics, and geology). In philosophy of science, much of the focus on mechanisms has thus far been on the role they play in scientific explanation. The idea that all genuine scientific explanation is mechanistic began to gain traction in contemporary philosophy of science with the work of Peter Railton, who claimed that "if the world is a machinea vast arrangement of nomic connectionsthen our theory ought to give us some insight into the structure and workings of the mechanism, above and beyond the capability of predicting and controlling its outcomes…" [1]. Inspired by Railton, Wesley Salmon abandoned his statistical-relevance model of explanation in favor of the view that "the underlying causal mechanisms hold the key to our understanding of the world" [3]. On this view, an "explanation of an event involves exhibiting that event as it is embedded in its causal network and/or displaying its internal causal structure"[4]. Salmon was working in the shadow of Carl Hempel's covering law model of explanation, according to which explanations involve inferences from statements describing laws and, in some cases, particular conditions. Salmon tended, in contrast, to favor an ontic account, according to which explanations are out in the world. He thought that progress in understanding those explanations requires 'exhibiting' the relevant mechanisms. However, even though he rejected representational and inferential accounts of explanation, he naturally recognized that reasoning about mechanisms, which requires representations (models), plays a big part in the process of exhibiting those mechanisms. A more recent formulation of the mechanistic account of explanation is supplied by Machamer, Darden, & Craver, who claim that "Mechanisms are entities and activities organized such that they are productive of regular changes from start or setup to finish or termination conditions" [5]. A central goal of science, on their view, is to formulate models, which take the form of descriptions of mechanisms that render target occurrences intelligible: Mechanism descriptions show how possibly, how plausibly, or how actually things work. Intelligibility arises...from an elucidative relation between the explanans (the setup conditions and intermediate entities and activities) and the explanandum (the termination condition or the phenomenon to be explained).... As with 'exhibiting' for Salmon, the process of 'elucidating' how setup conditions lead to termination conditions requires a significant contribution from model-based reasoning. Bechtel offers a related account of mechanisms. He claims: "A mechanisms is a structure performing a function in virtue of its component parts. The orchestrated functioning of the mechanism is responsible for one or more phenomena" [6]. As compared with other mechanists, Bechtel is much more explicit about the role that model-based reasoning plays in science and about the diverse forms of representation that may be involved (e.g., descriptions, diagrams, scale models, and animal models). He is, moreover, among the few to acknowledge the importance of 'in-the-head' model-based reasoning. He suggests that its central form may

Journal of Cognitive Science, 2013

Recently, Piccinini and Craver have stated three theses concerning the relations between functional analysis and mechanistic explanation in cognitive sciences: No Distinctness: functional analysis and mechanistic explanation are explanations of the same kind; Integration: functional analysis is a kind of mechanistic explanation; and Subordination: functional analyses are unsatisfactory sketches of mechanisms. In this paper, I argue, first, that functional analysis and mechanistic explanations are sub-kinds of explanation by scientific (idealized) models. From that point of view, we must take into account the tradeoff between the representational/explanatory goals of generality and precision that govern the practice of model-building. In some modeling scenarios, it is rational to maximize explanatory generality at the expense of mechanistic precision. This tradeoff allows me to put forward a problem for the mechanist position. If mechanistic modeling endorses generality as a valuable goal, then Subordination should be rejected. If mechanists reject generality as a goal, then Integration is false. I suggest that mechanists should accept that functional analysis can offer acceptable explanations of cognitive phenomena.