ABSTRACT
The joint advance of artificial intelligence and neuroscience has unseated the behaviorist and made it once again respectable to suppose that internal states cause behavior. The present task is to decide which species have representational states that approximate mental states. Historically, functionalism arose as a means of preserving the intuition, highlighted in the thought experiment above, that mental states can be enjoyed by any sufficiently sophisticated agent, regardless of its origin (natural or artificial), construction or, closer to home, its anatomy and physiology. The idea that mental states are things that must have not just many but also particular causal relations suggests another way to set up the criterion whereby the people could tell mental from the non-mental representational states. Classical conditioning might seem a rather optimistic place to go looking for tests of the presence of representational states, let alone mental states.
