Token Identity and Physical Realizability

Core Idea

Token identity theory holds that individual mental events are identical to individual physical events, even if mental types are not identical to physical types. This allows multiple realizability: the same mental type could be realized by different physical types in different creatures or systems.

Explainer

To understand token identity theory, you first need the type/token distinction from your prerequisite. A type is a general kind or pattern; a token is a specific instance of that kind. The word "cat" appears three times in this sentence: that's three *tokens* of one *type*. Applied to mental states: pain as a *type* is the general category; this particular pain I'm feeling right now is a *token* — a specific, dated mental event.

Type identity theory — your prerequisite — made the bold claim that mental types are identical to physical types: pain (as a kind) = C-fiber firing (as a kind). Every pain, anywhere, in any creature, would have to be realized by C-fiber firing. You know from multiple realizability why this fails: an octopus feels pain with completely different neural architecture; a silicon robot might experience pain with no neurons at all. The same mental type appears in wildly different physical substrates, so type-type identity is too rigid.

Token identity theory retreats to a more defensible position: each individual mental event is identical to some individual physical event, but the physical realizer can vary. *My* pain at 3pm on Tuesday is identical to *some specific neural event* in my brain — perhaps this particular C-fiber activation, or this pattern of distributed cortical activity. Your pain at a different time is identical to a different neural event. An octopus's pain is identical to yet another physical event, using entirely different biological hardware. There is no single physical type that all pains share; but every pain *token* is a physical event.

This move preserves physicalism — nothing mental happens without something physical happening — while respecting multiple realizability. It's a form of non-reductive physicalism: mental types don't reduce to physical types (no psychophysical type-type laws), but every mental particular is a physical particular. The mental and physical descriptions pick out the same events under different concepts. Token identity naturally underpins functionalism: what makes something a pain isn't its physical constitution but its functional role — what causes it, what it causes — and that role can be physically realized in multiple ways. This sets up the broader question of substrate independence: if token identity holds, could a sufficiently organized computer token the same mental events as a brain?