Reductive, Non-Mechanistic Explanations in Neurobiology

Jacob Reimer
Conceptual and Historical Studies of Science - University of Chicago

     Full text: Not available
     Last modified: June 15, 2005
     Presentation date: 07/14/2005 4:00 PM in MACK 236
     (View Schedule)

Abstract
I identify a reductive but non-mechanistic mode of explanation that is commonly used by neurobiologists studying higher brain functions and ask how it is related to the kinds of mechanistic explanations at the cellular level that are often considered the ultimate explanatory aim of systems neuroscience.
Recent discussions of explanation in the neurosciences have emphasized the importance of mechanisms (Machamer et al., 2000). I take it as evident that one of the primary goals of systems neuroscience is to provide descriptions of causal chains that “bottom out” (Craver, in submission) in entities at the cellular or molecular level. In practice, descriptions of this sort are rarely possible. The causal chains (of neurons) involved in higher brain functions are long and complex, and in the intact organism investigators are usually limited to investigating the properties of one or a few cells at a time.
In lieu of mechanisms, neuroscientists often give explanations in terms of correlations between the activity of a particular cell and some property of the animal’s behavior or environment. The unknown causal chain connecting the periphery to the cell of interest is modeled implicitly as a noisy channel or black box, and the activity of the neuron is said to “encode” the relevant aspects of the stimulus or behavior. Descriptions of this sort represent a non-mechanistic mode of explanation that allows neuroscientists to study features of higher-level perception, cognition, and behavior while still cashing out their accounts at the cellular level.
I address several questions that I believe are raised by explanations of this kind. What is the relationship between these sorts of explanations and the ultimate explanatory goals of neuroscience? (i.e. are explanations of this sort merely a stop-gap measure?) What is the status of “partial” explanations in a science of complex systems? What explanatory value is gained by referring to the cellular level without the ability to identify the functional role of a cell’s activity in a particular mechanism? To what degree is the idea of "encoding" unique to neuroscience (i.e. because of the information-processing character of the brain), and to what degree does it apply to the study of complex biological systems more generally?