Content Individuation

Content individuation is the problem of specifying the conditions under which two mental states — beliefs, perceptions, desires, or intentions — have the same representational content. The question is not merely psychological but metaphysical: what makes a thought about water the same thought in two different heads, or a different thought in the same head at two different times?

The problem splits along the internalism-externalism divide. Externalists hold that content is individuated partly by environmental facts: the Earthling's thought about water and the Twin Earthling's thought about XYZ have different contents because their causal histories differ. Internalists hold that content is individuated by neural or functional structure alone: identical neural states carry identical contents regardless of external context.

For Artificial Intelligence, content individuation is not an abstract puzzle but an engineering question. When two neural networks have identical weights but different training data, do they share contents? When a model is fine-tuned, does its content change or merely its expression? Without a theory of individuation, claims about AI 'understanding' remain untestable.