The main issues raised in the regrettably brief discussion session after I’d shut up last week concerned, I think, the levels of explanation involved in cognitivism.
As Haugeland makes clear, reduction in the sciences doesn’t supplant the level of explanation being reduced – there are plenty of cases from the more established sciences where explanations are still valid at a certain level, even if the regularities at this level can be explained in terms of (or mathematically derived from) further sets of regularities and laws.
Lycan [http://mugwump.pitzer.edu/~bkeeley/class/FoNS/Lycan.PDF sorry, bit of a rubbish copy…] provides some nice further clarification of this idea. One thing he is very clear on is the fact that distinctions between function and structure (eg. between a cognitive process and the (sub)components realising it) are relative to the level of explanation – nature is continuous and whether or not we see something as an IBB composed of smaller functional subcomponents, or as one of the functional components comprising a larger IBB, depends on the level of explanation.
But this isn’t just an epistemological convenience. For Lycan, I think, nature is genuinely organised hierarchically, with the smaller simple bits making up ever larger, more complex bits – all the way up. And this bottom-up ‘aggregative ontology’ is best approached with the kind of top-down explanatory strategy that I was trying to clarify last week (in terms of Haugeland’s IPSs being decomposed into component IBBs and so on).
Thus, biological organisms are composed of atoms collected into molecules collected into macromolecules, organelles, cells, organs…and so on. And we understand these organisms at any given level by appeal to the components which realise the function we are interested in. So, an explanation of the heart proceeds by describing its blood-pumping function in terms of the muscular and neural cells composing the heart, which cells can in turn be decomposed further into their functional components, and so on… And at the same time, we can always go the other way and move up the hierarchy, explaining the heart’s place in the wider circulatory system and eventually the organism as a whole.
What was novel and exciting about cognitivism and homuncular functionalism (if I understand correctly) is that it promises a way of bringing meaning and intentionality within this kind of aggregative ontology. And thus the mind, like everything else in nature, could be understood by top-down decomposition of highly intentional and meaningful cognitive states and processes, into gradually simpler less intentional components, all the way down through neurobiology and eventually to the level of fundamental physics.
Anyway, this is a gross simplification on my part and it’s definitely worth reading the Lycan chapter (or various other papers in which he defends this kind of functional explanation) to get a better understanding of these ‘levels of nature’. Also if anyone has or knows of some other related reading, it’d be interesting to get some further clarification, and perhaps compare with Haugeland’s characterisation of cognitivism.
With regard to causation and how it fits into this kind of picture, I’m still pretty clueless. I know that amongst those interested in complex systems and dynamical systems theory as a way of explaining the mind, the idea of ‘circular’ or ‘reciprocal’ causality is pretty popular. The idea being that as well as upwards causation, from smaller subcomponents to higher-level global behaviour, there is also such a thing as downward causation, whereby behaviour which only emerges at this higher global level causally influences the behaviour of lower level components. This might be a way into understanding how cognitive states can be explained reductively but still retain their causal role (as well as just their explanatory role).
But as I say, I can barely understand, let alone explain this stuff. So if anyone can give a better/different/more convincing explanation of emergence, or of how and why cognitive states can preserve causal efficacy at their particular level of organisation, whilst simultaneously being reducible to lower-level components, that’d be good!