Published online by Cambridge University Press: 01 January 2020
For those of us who are sympathetic to the research program of cognitive science, it is especially useful to face the deepest and sharpest critic of that program. Charles Taylor, who defines himself as a ‘hedgehog’ (1) whose ‘single rather tightly related agenda’ fits into a very ancient and rather elusive debate between naturalism and anti-naturalism, may well be that critic. My ambition in this paper is to distill Taylor’s central objection to the cognitive science approach to agency and the self as it is expressed throughout Human Agency and Language. After trying to set out the core of this objection, I want to remark on some rather curious aspects of the dispute of which it is a part, and then sketch, in relation to one or two examples, what I take to be the most promising line of resistance to Taylor’s attack. I conclude with a proposal as to how Taylor may – narrowly – escape one logical consequence of his position, according to which he should stop knocking the cognitive science program and instead go to work building a robot.
1 I shall be referring especially to parts I and II of Human Agency and Language: Philosophical Papers, Vol. 1 (Cambridge: Cambridge University Press 1985), using the following abbreviations: Agency for ‘What is human agency?’ (15-44); SIA for ‘Self-interpreting animals (45-76); HPhM for’ Hegel's philosophy of mind (77-96); Person for ‘The concept of a person’ (97-114); PCP for ‘Peaceful coexistence in psychology’ (117-38); Gen Psy for ‘What is Involved in a genetic psychology?’ (139-63); Mechanism for ‘How is mechanism conceivable?’ (164-86) and Cog Sci for ‘Cognitive Psychology’ (187-212). Also relevant, from Philosophy and the Human Sciences: Philosophical Papers Vol. 2 (Cambridge: Cambridge University Press 1985) are ‘Interpretation and the sciences of man’ (ISM II:15-57) and ‘Rationality’ (II:134-51).
2 Taylor's own formulation of the connection between the three features just listed is this: ‘The significance feature underlies the two features …. interpretation and the implicit/explicit distinction. We have these two … because we are agents with a linguistic capacity, a capacity to formulate the significance things have for us. But to formulate the significance of something, to make it explicit, is to alter it’ (Cog Sci, 198).
4 Mechanism, 1820 A few pages earlier the point is put like this: ‘What if, in order to generate the rich variety of distinctions which must be saved, future neurophysiological theory found it necessary to enrich its vocabulary and hence its conceptual armoury? And supposing that the new range of concepts incorporated some of the purposive and intentional force of our current ordinary vocabulary? … [W)e have no reason to look on our present categories as so fixed that a conceptual convergence of this form is ruled out’ (Mechanism, 179).
5 No doubt we should also posit a further intermediate level, maybe the neurophysiochemical, between the psychological and the physical. The existence of such a level may well be crucial to the difference between cognitive science and psychology.
6 Whether the case will prove otherwise for connectionist models remains to be seen.
7 Steven Davis's paper in this symposium provides an excellent illustration of this strategy. In what follows I give two other examples of how it might be implemented.