Recent discussions on the purported 'reversal' of the relation between 'machine psychology' and physics seem to be running, as ever, into the sand over disagreements on the meaning and significance of rather complex arguments like the MGA. I'd like to try another tack.
The computational theory of mind (CTM) asserts, in effect, that all experience is a simulation - i.e. is the net effect of some form of computational activity. Bruno's starting assumption, at the beginning of the UDA, is that a 'computation' be understood, conventionally, as any sequence of physical actions whose net effect adequately approximates that computation. This is essentially what I understand to be the standard physical notion of computation. One of its consequences, noted in step 7 of the UDA, is that a physical computer capable of instantiating the trace of a universal dovetailer (UD) would thereby simulate all possible experiences. If a computer running such a program were indeed to exist, it would be impossible to distinguish whether any given experience was a consequence of its activity or that of some other 'primitive' (i.e. non-simulated) physical system. Indeed, the quasi-fractal, super-redundancy of the trace of the UD would render it overwhelmingly improbable that the origin of any given experience lay outside of its domain. Of course, such a notion can be attacked by denying that any actual physical universe in which we are situated is sufficiently robust (i.e. extensive in space and time) to support the running of such a computer, or even if it were so robust, that any such device must necessarily be found in it. However, even at this point in the argument it may be a little disturbing to realise that we might escape the 'reversal' only by appealing to what might appear to be contingent, rather than essential, considerations. In order to torpedo these final objections, Bruno deploys the MGA, which is intended to show that any brute equivalence between net physical activity and computation, accepted previously, is in fact unsound. However, the issue of what the MGA does or does not demonstrate seems to open up a never-ending conversational can of worms. Perhaps there are simpler arguments that can be accepted, or at least that might lead to a clearer form of disagreement. My suggestion would be to re-examine the notion of computation itself as a foundation for a theory of mind. ISTM that as long as we restrict discussion to third-person (3p) notions, there is no unusual difficulty, in principle, in justifying an equivalence between some psychological state and the action of some physical system, understood as approximating a computation. This is the sort of thing we mean (or at least is implied) when we say that human psychology supervenes on the activity of the brain. According to the tried and tested principles of physical reduction (which essentially boil down to 'no strongly emergent phenomena') a psychological state supervening on the physical activity of the brain (at whatever level) should be understood as being nothing over and above the combined effects of more fundamental physical events and relations that underlie it. In other words, both 'psychology' and 'computation' should here be understood as composite terms that subsume a great mass of reducible sub-concepts, 'all the way down' to whatever level of physics we consider, for present purposes, as 'given'. None of this, as said before, occasions any special difficulty in explaining correlations between such concepts as psychology and computation, as long as it is realised that any new effects 'emerging' from the underlying physical sub-strata are ultimately to be understood as merely composites of more fundamental events and relations. If none of the foregoing presents any special theoretical difficulty so long as we restrict our arguments to the familiar 3p mode of discussion, the same can't be said of its application to first personal (1p) concepts. This is the point, I feel, where sheep and goats begin to shuffle apart (sheepishly or goatishly) in the matter of theories of mind. What too often gets lost in our discussions, ISTM, is the essential distinction between any third-person account of the first-person (e.g. as I am now doing in these paragraphs) and the 1p phenomenon itself. Whereas the former can be understood without special theoretical difficulties as a weakly emergent (i.e. composite) effect, the latter cannot, at least not without implicitly dismissing its status as an independently real phenomenon, in the manner of the Graziano theory recently discussed. It's perhaps not so surprising that this distinction is elusive, as there is no other circumstance, AFAIK, in which this consideration arises. Putatively parallel examples of emergence, such as the 'liquidity' of water, aren't directly comparable, because no other phenomenon demands that we 'stand in its place', as distinct from being characterised at second or third hand. Because our stance on emergence is almost always of the latter kinds, it's all too easy to miss that any purely 3p account of consciousness as a weakly emergent phenomenon cannot ultimately escape this radical ontological reduction. Whereas such a reduction may be theoretically or explanatorily inconsequential to a 3p account, any properly 1p aspect is thereby effectively eliminated, or else can only hang on in the guise of what would be a genuinely novel, 'strongly emergent' or quasi-dualistic phenomenon. If the first of these alternatives is frankly incoherent on its face, the latter would be a state of affairs hitherto unsuspected and unjustified in nature. In effect, what I am saying is that if computation is to be understood as a basis for a theory of mind that properly encompasses 1p phenomena, it cannot be accepted as such merely in the form of its physical approximations. Perhaps this is just a long-winded way of pointing out that computation (at least the kind that would be of any use to a theory of mind) is a mathematical, not a physical, concept. The consequence of accepting it as a fundamentally physical concept is its radical ontological reduction, and this inevitably results either in the elimination of any putative 1p phenomena or their elevation to an unevidenced and frankly implausible dualistic status. On the other hand, if computation is understood in a fundamentally mathematical sense, the way is left open for the 1p phenomena to be justified in a manner that is integral to a suitable mathematical theory (e.g. as an intrinsic aspect of arithmetical truth). The problem for any such theory is indeed a 'reversal', in that instead of the difficulty of extracting a (1p) account of psychology from physics we are faced with hardly less thorny one of extracting physics, as observed, from the basis of a machine epistemology emulated in a fundamentally arithmetical ontology. One saving grace of the latter, however, may be that the explanatory appeal to an embedded and explicitly justifiable epistemology may act as the 'vaccine' that immunises 1p phenomena against an otherwise equally radical ontological reduction to 'mere numbers'. David -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/d/optout.
