As published in The Oxford Magazine:

Readers of the recent exchange between Ray Guillery and Peter Hacker may be left with the impression of a straightforward conflict between two disciplines: philosophy and neuroscience. But there is a third discipline here―medicine―that gives more substance to scientific enterprise than mere curiosity, and so ought not to be ignored.

The perspective of a good physician mirrors that of a patient. He cares less about the bare facts than about the grounds for future action they provide. He is concerned with explanations of previous observations less than with prediction of future events yet to be observed. For him, the general matters only to the extent to which it informs the particular: the singular, real, living, patient he faces. Reputations matter little; that Tom, Dick or Harry is FRS, or even en-Nobeled, is immaterial. Indeed, he may recall that a recipient of one of the few Nobel prizes within the field of systems neuroscience was Egas Moniz, the originator of radical psychosurgery.

In short, his focus is on practical action, in the context of the unknown, the yet-to-be-observed. This may seem even further removed from conceptual issues than the usual concern of the scientist. In fact, his concern is far greater. The conceptual here is not a mere playground for rival, arbitrarily favoured, ways of talking about things—as many imagine it to be—but a direct determinant of the predictive power of any theory in which the empirical is couched. For a theory implies a conceptual framework where the empirical is only fuel, and may fail not only because it runs out of gas but because the conceptual structure is fundamentally broken. These two elements are sharply distinct, and neither is remedied by more of the other. Just as adding more fuel to your car is not going to get you anywhere if your axle is fractured, so doing more experiments is not going to help you if your conceptions are mistaken. Either defect may destroy your theory, and rob it of the predictive power that justifies our interest in it.

Physics is at ease with this dichotomy, here corresponding to mathematics and physical experimentation respectively. A physicist may complain to a mathematician that the tools he gives him are not good enough to describe the reality he observes, but he is unlikely to say “I’ll divide by zero if I want to!” (Or at least if he does he will listen to what the mathematician says about the consequences of the move for the predictive power of his theory). Equally, if a physicist uses equations to describe his theory he will not pride himself on remaining in the dark about the underlying mathematics, or ignore mathematical analysis because “it has never discovered anything”. Nor will he casually dismiss valid objections from a mathematician because he has “changed his mind once before”.

Now cognitive neuroscience is not typically framed in mathematical terms. But that the terms are instead logico-grammatical does not mean one can arbitrarily ignore them, anymore than one can ignore an operator in a mathematical equation just because one does not like the look of it. The equivalent of mathematical analysis is here conceptual analysis, a distinctive conception of philosophy that does not seek to usurp empirical endeavour but to give it predictive power. Far from arrogating authority it sharply limits what philosophy can licitly do*, it is a philosophical vision that is modest, cautious, and at its core collaborative with biology, for it acknowledges that the nature of a human being―a biological entity―can only be comprehensively disclosed by the joint enterprise of conceptual analysis and science, just as the nature of the universe can only be disclosed by mathematics and physics, acting hand in hand.

So far from being “anachronistic” Peter Hacker’s contribution to the field is long overdue, and desperately needed. If it is destructive―and there is no doubt about that―it is so only because it seeks to clear the ground of conceptual structures that are crumbling anyway. However well a conceptually-flawed theory may fit the extant data now, it is bound to fail to fit them in the future: that is what a conceptual error ultimately amounts to. Moreover, Hacker gives the scientist applied tools to test his hypotheses as he builds, so that he does not find himself, at the end of an illustrious career, hanging in mid-air as his marvellous structure crumbles to dust beneath his feet.

If on reading Bennett & Hacker all you see is the mereological fallacy then you need new conceptual glasses. The mereological fallacy is merely a synoptic description of a varied family of errors― discussed in detail in the book―borne of a natural tendency to extend the use of expressions beyond their boundaries by attributing more to parts of the brain than is intelligible. It is discernible even in some of the greatest successes of the neuroscientific literature. Take, for example, Ralph Adolph’s beautiful study of the dependence of “fear recognition” on the amygdala. It took him―a brilliant scientist―a decade to discover that the defect of the index patient with bilateral amygdala damage he and his colleagues originally described was not “fear recognition” but the direction of gaze to the eyes in which fear happens, incidentally, to be manifest. Were it not so tempting to try to attribute the entirety of a behaviour to the area on which it depends this kind of advance would surely have happened faster. Similarly, Milner & Goodale’s celebrated dissociation of “vision for action” and “vision for perception” was widely seen as surprising, yet the exact opposite would be astonishing, for action and perception are different things, and so must be dissociable in the brain at some level. Here again, the natural impulse is to try to localise the concept of vision―as applicable to the whole organism―somewhere within the brain, and it continues in misguided attempts to localise regions that explicitly “bind” various dissociable streams.

I have only mentioned two landmark studies based on lesions, inferentially amongst the strongest techniques we have at our disposal in cognitive neuroscience. If the conceptual terrain here is perilous, what of purely correlative techniques such as functional imaging? A scientist is free to ignore the helping hand being offered here, but if his aim is the kind of understanding of human nature that achieves more than just the transient admiration of his colleagues, he would do well to think again.

 

——————————————–
*This is why so many contemporary philosophers cannot forgive Wittgenstein, for he robs them of the possibility of greatness, and it hurts their egos.