Representational parts

In this reply we claim that, contra Dreyfus, the kinds of skillful performances Dreyfus discusses are representational. We explain this proposal, and then defend it against an objection to the effect that the representational notion we invoke is a weak one countenancing only some global state of an organism as a representation. According to this objection, such a representation is not a robust, projectible property of an organism, and hence will gain no explanatory leverage in cognitive scientific explanations. We argue on conceptual and empirical grounds that the representations we have identified are not weak unprojectible global states of organisms, but instead genuinely explanatory representational parts of persons. Growing dissatisfaction with traditional artificial intelligence over the past decade or so has resulted in an increasing number of research projects that take up the cause of anti-representationalism. We think this is wrong-headed: not because we are enamored with traditional AI, but rather because the problem with traditional AI was not its representationalism, but its vision of representations as logical symbol structures, and its conviction that intelligence is best understood as problem solving via manipulation of such symbol structures. Accordingly, we think Dreyfus' anti-representational challenge is misguided. There are a number of points at which one could quarrel with his article. For example, even if one grants phenomenology the explanatory authority Dreyfus grants it, it is clear that his description is incomplete. Often when playing chess, one not only visually recognizes certain configurations and opportunities on the board, and perhaps feels the right thing to do. But there are often times that one imagines making non-actual moves: What would happen if I moved the black bishop there? Then my opponent would likely do that, and then I could do that. Such bouts of reasoning are undoubtedly common not only in playing chess, but in many cases of problem solving, and are curiously absent from Dreyfus' phenomenology. What one is exactly not doing in such a case is letting the world be its own representation; rather, one is letting an internal representation (image, whatever) stand for the real chess board.