Tuesday, 17 May 2011

Chemero (2009) Chapter 8: Neurophilosophy Meets RECS

Chemero's book finishes with two chapters on some philosophical consequences of taking a radical, embodied approach to cognitive science. Chapter 8 is about the mind-body problem, and how various attempts to reduce cognitive science to, say, neuroscience, can be vigorously resisted via the RECS approach, without being dualist about the mind. There are many people who think cognitive science can be reduced to neuroscience (intertheoretic reduction), but one plank of any embodied approach is that this won't work. RECS is particularly committed to a more extended notion of cognition and so a strategy for resisting reduction is critical. Chemero's plan won't rely on the usual philosophical manoeuvres such as Martian pain mechanisms or zombies. Like me, Chemero is concerned that these create the impression that philosophers aren't tackling real problems; he wants the philosophical conclusions of RECS to be grounded in data, and I thoroughly endorse this approach.

The main issue that Chemero wants to tackle is the status of the brain. Ecological psychology has a bad rap for ignoring the brain, and it's actually an obstacle to wider acceptance of ecological ideas in this age of neuroscience and imaging. I'm not a fan of neuro-imaging, in general, and reductionism has always bored me and struck me as destined for tedious failure. But I do accept that it might be time for us to get back in the business of including the brain in our discussions of behaviour - the question is just, what is the proper place of the brain in the overall system we study? (I've just bought Olaf Sporns' book Networks of the Brain: if anyone can help me to think about the brain properly, it will be Sporns. I knew him a little at IU and he does 'systems' thinking like few people can manage. He's the only neuroscientist I've ever heard speak who's made me go 'huh, that sounds about right'. High praise, I assure you!)

Clearly the brain is up to something - it's an expensive bit of biological equipment and damaging it has all sorts of consequences. The ecological approach is deeply committed to the notion that 'meaning ain't in the head', though, and so has so far resisted the trend in the field to try to explain behaviour in terms of brain activity. This trend, the idea that cognitive science can be reduced to neuroscience, has been around for a while (most vigourously from the Churchlands in philosophy). 

This kind of reduction works if and only if you can account for the phenomena of the higher level (cognitive science) using only the tools of the lower level theory (neuroscience) plus some bridging principles. The neurophilosophy of the Churchlands (e.g. Churchland, 2002) relies on the metaphor of computational representation; if cognition simply is computation, then the theory of how those computations are implemented (neuroscience) can readily account for the entities of cognitive science. A philosopher called John Bickle (2003) takes this all one step further, and argues that cognition will eventually be accounted for by a base theory grounded in molecular neuroscience. Bickle's ruthless reductionism is based in the idea that reductionist attempts should be driven by neuroscientists and the way they do their science - look at what they claim and rely on the data to decide between options. 

Chemero then constructs a rebuttal to these based in RECS which highlights how taking a reductionist approach can actually lead to bad science and therefore confounded results (Chemero & Heyser, 2005, 2009). I liked this section a lot.

The example is object exploration, a common task used with rats to study mechanisms of learning and memory (alongside various genetic and pharmacological manipulations). Experiments Chemero ran revealed that the results of a given exploration study depended heavily on the affordances of the objects in the task for the animal - rats only explore objects by climbing on them if the object affords climbing for the rat, for example. This wouldn't be a huge problem, except that the molecular neuroscientists studying the behaviour pay no attention to the properties of the object. Only 32 papers in a review of 116 used objects with equivalent affordances; the rest either had varying affordances or (even worse) insufficient information in the methods section to establish the affordances of the objects used. In other words, the vast majority of these neuroscience papers were either actually or potentially confounded, and many were not replicable. Bad science indeed, and all from taking the kind of reductionist stance advocated by philosophers of mind (and, less explicitly, by most cognitive scientists). This is also some good evidence for my typical complaint about neuroscience research: no theory of information, specifically no idea what they are pumping into the system before measuring it's response. Missing this step means you can't trust the interpretation of the data at the end.

Chemero's conclusion is straight-forward - RECS provides numerous theoretical and empirical reasons to resist reduction to neuroscience. Theoretically, the anti-representational stance takes away one of the prime motivating factors, and by replacing representations with an animal-environment system RECS is committed to the brain being only a part of the story anyway. Empirically, reductionism leads to poor science, and you can demonstrate this by doing the same experiments with a broader perspective which takes the relation between object properties (e.g. affordances) and animal capabilities seriously, and showing that this matters. 

These reasons are more convincing reasons for resisting reductionism than the typical philosophical arguments, which usually hinge on some thought experiment about Martians or some other spherical cow. I'm certainly much happier having arguments like this to rest on, as a scientist. I've also got an idea to pursue this analysis on a second literature (the effects of an internal vs. external locus of attention on motor control tasks such as a vertical jump); I think object affordances are likely driving the effects and I'm willing to bet I will find a similar lack of care and information in the external locus literature. So this chapter lays out a useful and productive line of reasoning that will allow a radical scientist actually make some empirical inroads on tricky questions, as well as allow the radical philosopher to have somewhere to stand while defending against an obvious critique. It's therefore a critical contribution to the broader RECS programme.

Bickle, J. (2003). Philosophy and the Neurosciences: A Ruthlessly Reductionist Account. Dordrecht: Kluwer Academic.

Churchland, P. S. (2002). Brain Wise. Cambridge, Mass.: MIT Press.

Chemero, A., and C. Heyser (2005). Object exploration and a problem with reductionism. Synthese, 147, 403 423. DOI Download

Chemero, A., and C. Heyser (2009). Methodology and ontology in the behavioral neurosciences: Object exploration as a case study. In Oxford Handbook of Philosophy and Neuroscience, ed. J. Bickle. New York: Oxford University Press.  Download

1 comment:

  1. churchland is modern day aristotle or plato. authoritative