Dear Jim, Re the 23rd April post (?Australian time): That makes an awful lot of sense. There is the caveat that for Chalmers's 1995 solution, at least, function entails phenomenal content (as I understand it) but I agree that phenomenal quality totally resists input-output functional  analysis. If it is allowed I would argue that phenomenal quality could be subject to input functional analysis. That is to say that phenomenal quality would be entailed by the nature of a local immediate interaction between influencing (or informing) entity and influenced (informed) entity. In a third person analysis you end up with the same causal chain with links as pairs (input-output or influencer-influenced) but carved at different joints and currently in an incommensurable descriptive language. One might say that instead of a passing-the-buck analysis we need a 'buck-stops-here' analysis. As you say - experience is the place. That handles inverted spectra and suchlike comfortably, I think (and there are ways to deal with fading qualia as well).

The interesting corollary is that zombies really do get squeezed out, as Derek would like. If the Whiteheadian occasion is the same in terms of giving and receiving then the qualia are the same. A robot with the wrong local occasions, contrary to the multiple realisability doctrine, will get no qualia like ours even it passes every Turing test in town - which it well might.