that it is false but that it is not coherent. It makes no sense, because the idea of an immaterial, spiritual substance is incoherent. It lacks both synchronic and diachronic criteria of identity. The idea that an immaterial substance might be an active agent in the material world, with the powers to causally affect the pineal gland by acting on it thereby releasing animal spirits into the nerves, makes no sense. It is not a false theory but an incoherent conjecture.
Conceptual analysis is to neuroscience what the differential calculus is to physics: a neuroscientist cannot do without it, whether he likes it or not. He can take for granted a received conception, often rooted in past defective analyses (e.g. by William James or by von Helmholtz ), or he can confront the received assumptions and reflect on them. It is for those striving to think for themselves that we offer our work as a set of guidelines. For on most of the major themes of cognitive neuroscientific research we try to provide the major conceptual characterizations, depicting the logical geography of the domain.
Conceptual mistakes have real consequences, for our concepts unavoidably guide both practice and interpretation. These consequences in neuroscience are of at least three kinds. First, if a neuroscientific experiment misconstrues its actual subject matter, then no matter how internally consistent it is, its results will have no practical, real-world implications, since its results are hermetic to the misguided framework within which the experiment is constructed. An example here is Libet’ s conception of voluntary action, which has not in practice stopped anyone from changing their behaviour in the several hundred millisecond interval preceding the movement with regard to which we are allegedly mere automata. Second, an incoherent scientific theory built on perfectly valid data will fail in inductive, predictive reasoning, just as an equation in physics that contains a covert division by zero will fail in its predictive use. Once the incoherence has been realized, one need not wait for the zero values to be entered into the equation to know that a theory based on such an equation is going to be worthless. Hence investigations of the neuroscience of memory that rest on the assumption that memory is of or about the past, or that memory is exhibited by any change of behaviour consequent upon prior experience, can be rejected in advance of any experiment. For memory is acquired in the past, but it need not be of the past – it may be of or about the present, the future, omnitemporal or atemporal; and wricking one’ s ankle and thereafter limping is not remembering anything. Third, whether predictive or not, a scientific theory that is conceptually flawed cannot be explanatory. For explanation, even more than prediction, depends on the coherence and integrity of the conceptual framework within which it is constructed, no less than upon the empirical data it brings to bear on the explanandum. Of course, the empirical consequences of conceptual errors in neuroscience are not typically immediately obvious, especially when widely accepted by groups of working neuroscientists. However, such errors are not stochastic: if something makes no sense, there is no chance that it will later spontaneously acquire sense as the experiments unfold. It is true that in the past important discoveries have been made despite incoherent theories. But that is not a recommendation for incoherence. Incoherence is not an infantile disease that neuroscientific theories catch before they mature into received neuroscientific wisdom.
Does all this mean that neuroscientists have to study philosophy or to become philosophers? No – there are great swathes of philosophy, such as metaphysics, logic, philosophy of mathematics, or moral and political philosophy that are of no concern to the neuroscientist. Nevertheless, many of the conceptual structures presupposed in neuroscientific research and the specific conceptual forms invoked are, by their nature, highly problematic. For they unavoidably link the conceptually heterogeneous domains of the behavioural, the psychological and the neural. Furthermore, neuroscientists are sometimes required to extend existing concepts or to introduce new concepts for specialized purposes. It is all too easy to do so in inconsistent and incoherent ways. It is desirable that neuroscientists be familiar with connective analytic methods in order to sharpen their sensitivity to conceptual unclarities, errors and confusions. They must be able to realize that, for example, memory need not be of the past, that one cannot order someone to be conscious of something, that if one is conscious of something’ s being so, then it follows that it is so, that something may be voluntary without being intentional or intentional without being voluntary, and so on. They cannot stand aloof from conceptual questions or refrain from committing themselves to conceptual forms.
It will come as no surprise to those who have read the first edition of this book that a spirit hovers over its arguments: the spirit of Ludwig Wittgenstein. Why Wittgenstein rather than Popper (who inspired Eccles ) or even Locke (who inspired von Helmholtz )? Because Wittgenstein’ s reflections on the philosophy of psychology surpass anything previously achieved in this domain of philosophical thought. Our conceptual analyses in this book echo and elaborate Wittgenstein’ s not because we are fond of Wittgenstein but because his intellectual approach in this domain is precisely what is needed by cognitive neuroscience. Philosophy of psychology has to elucidate the conceptual relations between the behavioural and the ‘inner’, ‘mental’ or ‘psychological’. It has to account for the conceptual structures that inform the asymmetries between the first-person utterance and the third-person description of ‘experience’ or ‘states of consciousness’. Wittgenstein broke with a long tradition of conceptual confusion characteristic of both rationalist and empiricist thought, indeed a tradition dating back to antiquity. He ploughed up the field of philosophical thought afresh. Cognitive neuroscience was firmly planted in that field, with roots reaching down to the late fifteenth and sixteenth centuries, as we describe in chapter 1. It has to cope with a subject matter that straddles the behavioural and the ‘inner’, although here the ‘inner’ is neural and cortical. The basic observation statements of the science are what people do and say, and it is these that need to be related to an ‘inner’. Here too there is an illusion that just as introspection seems to bypass behaviour and give one direct access to an ‘inner’, so too positron emission tomography ( PET ) and functional magnetic resonance imaging (f MRI ) give neuroscientists the illusion that they can bypass behaviour and study the ‘real thing’ directly. At last, as one well-known neuroscientist exclaimed, we can ‘actually see thought’. But that too is an illusion. For neuroscientists, the ‘hard facts’ are recurrent and reproducible human behaviour. We can see thought when we watch Le Penseur and hear thought when we listen to a great thinker lecturing, but not when we watch the computer-generated image of an fMRI scan.
Finally, to mention a theme that will recurrently preoccupy us throughout the methodological sections of this book, critics seem puzzled at the idea that examination of language can have anything to tell empirical scientists other than linguists. The puzzlement seems to have two sources: first, the thought that language is trivial, whereas neuroscience is deep; second, that an interest in current psychological or mental vocabulary precludes conceptual and linguistic change. Both are mistaken. Since it is the capacity to speak and the mastery of a language that is a condition of all that is distinctively human, and hence too a condition for the sciences and the arts of humanity, it is hardly appropriate to trivialize what differentiates us from monkeys and rats. Without a mature language there is no formulation of scientific hypotheses and so too no discovery of general truths of neuroscience or of science in general. The scrutiny of the language of psychology and neuroscience will not discover new neuroscientific truths, but it will guard against the inadvertent adoption of incoherent hypotheses and the illusory confirmation of unintelligible neuroscientific conjectures. Language, and the concepts expressed in our languages, are the medium of scientific thought. If one ties knots in the web of words, one can get nothing right, save per accidens – and even should one be so fortunate, one will not understand what one has discovered. Second, that we need to examine our means of representation carefully in order to avoid the manifold confusions to which we are prone does not imply that we are confined to our existing vocabulary and the concepts it expresses. Nothing prevents linguistic stipulation and conceptual innovation. But we must be forewarned to ensure coherence, and to assure that the innovation dovetails smoothly into antecedent usage. Analogical extension needs meticulous monitoring, lest it be stretched beyond the bounds of intelligibility. The step from ‘as it were’ or ‘it is as if’ to ‘it is so’ is far too
