I am an Assistant Professor in the Department of Philosophy at the University of Western Ontario. Prior to moving here in 2012, I was an Assistant Professor at the University of Alabama at Birmingham.
My research is situated at the intersection of philosophy of neuroscience, philosophy of mind and philosophy of science and has its origin in a single basic question: What light does contemporary neuroscience shed on the relationship between mind and brain? My approach to this question is unique insofar as I contend that answering it requires directing analytical scrutiny at the investigative strategies neuroscientists use to probe this relationship. To this end, the project at the heart of my research program is to develop and refine a conceptual framework for analyzing experiments and experimental practice in the neurosciences of cognition (e.g., Sullivan 2009). By applying elements of this framework to cognitive neurobiological case studies, my work has illuminated an interesting set of epistemological problems that require solutions (e.g., Sullivan 2009, 2010a, 2010b, Forthcoming (as of May 2015 paper on natural kinds). Characterizing the nature and sources of these problems, identifying their implications for the explanatory goals of neuroscience, and developing viable strategies for overcoming them, are the primary aims of my current research.
One approach to answering the question of what light the neurosciences of cognition shed on the mind-brain relationship is to raise the question of what light they aim to shed—i.e., what the explanatory goals are. A substantial amount of work in philosophy of neuroscience has been directed at answering this question, and the received view is that the aim is to provide mechanistic explanations of cognitive functions (e.g., Bechtel 2008; Bickle 2003, 2006; Craver 2007 (See Sullivan 2009)). Taking for granted (for the moment) that such accounts are descriptively accurate, this prompts at least two questions. First, are the investigative strategies on offer appropriate for realizing these mechanistic goals? Second, is how the science is practiced on the whole (i.e., within and across laboratories) amenable to their attainment? During the past few years I have directed these questions at that area of neuroscience with which I am intimately familiar, namely, cognitive neurobiology. What follows is a synthesis of claims that I have defended in several recent publications as well as claims for which I am currently in the process of articulating good arguments.
First, appeals to the internal representational states of an organism may be regarded as playing a crucial role in shaping the development of experimental paradigms in the neurosciences of cognition. An experimental paradigm specifies how to produce a cognitive function of interest (e.g., spatial memory) and detect, by appeal to observable changes in an organism’s behavior, when it occurs. In designing an experimental paradigm an investigator must ask herself: “‘What’ do I want this organism to learn?” Answering this question informs stimulus selection, stimulus presentation and the arrangement of stimuli in the laboratory. Ideally, an investigator aims to determine what kind of information the experimental context “affords” the organism—i.e., ‘what’ it may learn and ‘how’. She uses such information to exert control over the learning context in a way that ensures the reliability of the paradigm for individuating a discrete cognitive function. However, this role that internal representational states play in the experimental context is only ephemeral; once an investigator believes she has hit upon an experimental paradigm that may be reliably used to produce a set of observable changes in behavior indicative of a discrete cognitive function, intervention experiments begin. At that point the investigator liberates herself from the framework of mental states altogether and replaces it with a framework of systems, networks, synapses, cells and molecules. The organism, in some sense, comes to be regarded as nothing over and above its working physical parts.
This strategy, however, is antithetical to the purported mechanistic goals of cognitive neurobiology for several reasons. First, cognitive functions are not identical to the observable changes in behavior used to detect them. Secondly, the relationship between those changes in behavior identified as relevant by an investigator for detecting a cognitive function and the actual changes in behavior produced during training in a learning paradigm is rarely one to one. In fact, many different changes in behavior and discrete learning events likely occur, and many different cognitive functions are likely involved. Thus, the phenomena an experimental paradigm circumscribes often outstrip the concept used to designate those phenomena. Unfortunately, however, these facts are missed precisely because cognitive neurobiologists fail to take an organism’s mental life seriously once they have located an experimental paradigm that seems to yield a reliable set of behavioral effects. Yet, the problems that arise from trying to sweep these facts aside rise to the surface sooner or later. For example, my recent analysis of an historical case study—spatial memory and the Morris water maze (Sullivan 2010)—reveals that one strategy for coping with an uncertainty about what cognitive function an experimental paradigm individuates is to increase the generality of the term used to designate the function. This approach, rather than moving the field closer to identifying the mechanisms of discrete cognitive functions, serves only to impede progress. For, as the term used to capture the phenomenon oscillates, it becomes increasingly unclear what cognitive neurobiologists are discovering the mechanisms of. This is an impediment to mechanistic explanation in so far as it requires that the phenomenon to be explained is distinct from what causes it.
These facts have been missed in the philosophical literature on neuroscientific explanation due to a failure to subject neuroscientific experiments to careful philosophical analysis. Furthermore, they raise the question of whether philosophical accounts of mechanistic explanation are descriptively accurate vis-à-vis the explanatory goals of neuroscience. Although on the surface the investigative strategies in the neurosciences of cognition appear suitable for achieving mechanistic goals, when we probe more deeply into the science, these strategies fall short. If we recognize why the science fails, then, as philosophers, we will be better poised to offer normative prescriptions appropriate for putting the science on the correct course. However, if the goals are not properly mechanistic (i.e., in the philosophical sense), then the task is to determine the nature of the goals and the kinds of approaches that are well-suited for achieving them. If I am correct about the source of the epistemological problems that I have identified, then one step in the direction towards a solution is readily apparent. The framework of internal intentional mental states should not be abandoned at the very point at which one has discovered that an experimental paradigm may be used to reliably and repeatedly produce a discrete set of behavioral effects. Rather, an investigator should, with respect to any given paradigm, continuously probe the possibility space for other kinds of information the organism may be experiencing or learning in the experimental context and then develop testing batteries aimed to tease these different cognitive functions apart. What I am advocating for, in other words, is a form of methodological pluralism in the context of experimentation and I am currently working out the details as to the form that such pluralism will take. One challenge I face is coping with a second type of epistemological problem that threatens the explanatory goals of the neurosciences of cognition, namely, the multiplicity of experimental protocols (Sullivan 2009).
What I will refer to simply as “the multiplicity problem” arises as follows. It is widely acknowledged by neuroscientists and philosophers alike that discovering the mechanisms of cognitive functions requires the collaborative efforts of investigators working across laboratories situated at the same and different levels of analysis. Ideally, then, explanatory claims emanating from different laboratories studying the “same” cognitive function should readily fit into a comprehensive explanatory model of that function. However, investigators alter certain mutable parameters of experimental paradigms (“subprotocols”) in ways that reflect differing intuitions about the experimental conditions that will be most reliable for identifying the mechanisms of a given cognitive function. Such differences translate into differences in the cognitive, cellular and molecular mechanisms implicated in the production of those functions (indicated by changes in behavior). The implication is that there is no sense in which results emanating from different laboratories will fit neatly into the same explanatory model of the same cognitive function.
My current intuition, contained in a paper I published in 2014 in Charles Wolfe’s Brain Theory volume is that the strategy of bringing back a consideration of an organism’s internal states into the context of experimentation as a means to improve the reliability of experimental paradigms for discriminating discrete cognitive functions will shed some light on how to cope with the multiplicity problem as well, in so far as investigators will be more sensitive to how variations in stimulus parameters may impact what and how an organism learns. Thus, it will at the very least prompt caution in trying to generalize results across laboratories and discourage the assumption that appears popular among both neuroscientists and philosophers of science that results emanating from different laboratories will neatly fit into unified coherent explanatory models of cognitive phenomena. I am also investigating several other areas of science (e.g., biology. anthropology, sociology) in which the multiplicity problem arises and trying to determine if the solutions that have been put forward in these contexts provide clues as to how to cope with it in those case studies I am considering. A term one commonly encounters in such literature for capturing how different areas of science come together is “integration”. Given that the concept lacks rigor, part of the task is to identify a more appropriate concept for capturing such relations (Sullivan in prep).
Thinking about these problems and potential solutions prompts a set of interrelated questions. Here are two: First, what are the implications of the problems I have identified for the cognitive neurobiological study of mental disorders? If the these kinds of problems arise there may they be subject to similar solutions? If methodological pluralism is the norm in the experimental context, should explanatory pluralism and a plurality of therapeutic approaches be the norms in the clinical context, too? I have begun providing answers to these questions in a paper (Sullivan 2014) in a recent volume in philosophy of psychiatry that I co-edited with Harold Kincaid that was published by MIT Press entitled Classifying Psychopathology: Mental Kinds and Natural Kinds.