Carl Craver
Assistant Professor of Philosophy & PNP Program, Washington University

"Of Layercakes and Learning Mechanisms: Some Constraints on a Metaphysics of Levels"

Abstract: In science and its philosophy, we find repeated reference to levels of analysis, of behavior, of control, of description, of explanation, of organization, of sciences, and of theories, all without specifying what a given levels hypothesis asserts or how it differs from the others. This is unproblematic so long as reference to levels carries no metaphysical burden— merely pointing vaguely at the domain of a particular field, or grouping objects of roughly the same size. But the levels hypothesis does carry such a burden in many recent discussions of causation, emergence, explanation, realization, reduction, physicalism, and supervenience. In such discussions, we should take more care to specify what items are beings sorted into levels, what distinction sorts them into different levels, and what common feature binds them into a single level. I explore different answers to these questions in an effort to develop a taxonomy of levels hypotheses. By considering an exemplary multilevel theory from the contemporary neuroscience of learning and memory, I introduce a central and primary sense of levels— levels of mechanisms— and I argue that it is central and primary by contrasting it with the other items in the taxonomy. I close by arguing that the much-vaunted problems of causal/explanatory exclusion and interlevel causation simply do not arise for levels of mechanisms.