The Idea

To be in pain is not to have some particular brain state. It is to be in whatever state is typically caused by tissue damage, produces beliefs like “I’m hurt” and desires like “make it stop,” and issues in withdrawal and complaint. The identity of a mental state is its causal role between inputs, other mental states, and outputs. What matters is the role, not what fills it.

The main argument is multiple realizability. Octopuses plausibly feel pain with a radically different neural organization; if pain were identical to a human brain state, they could not. So mental kinds are functional kinds, and mind is to brain roughly as software is to hardware. This is the license under which computational cognitive science operates: study the algorithm without waiting for the implementation story.

Why I Keep It

It justifies working at the algorithmic level. I can study parsing or bilingual lexical access as functional organization, ask whether two speakers realize the same switching mechanism differently, and treat constructs like working memory as functionally defined and behaviorally measured before anyone locates them in tissue.

Tensions

• Qualia: inverted spectra and zombie thought experiments suggest functional role might not capture what experience is like.
• Liberalism: the right causal structure in water pipes should not feel pain, but functionalism struggles to say why not (Searle’s Chinese Room pushes here).
• Embodied cognition presses from the other side: maybe the body is part of the role, not an interchangeable realizer.

Key Sources

• Putnam, H. (1967). “Psychological Predicates”
• Lewis, D. (1972). “Psychophysical and Theoretical Identifications”
• Block, N. (1978). “Troubles with Functionalism”