The Necessity of Embodiment: The Dreyfus-Collins Debate

Abstract

In order to contest Hubert Dreyfus's claim that having human body is necessary condition for exhibiting intelligent behavior in our world, artificial intelligence researcher (henceforth, AI) Douglas Lenat presents what he takes to be the counter-example of Madeleine, a wheel-chair bound woman described by Oliver Sacks who was blind since birth, could not use hands to read Braille, and yet acquired common sense knowledge from books that were read to her (Dreyfus 1992, xx).1 Vexed by Lenat's association of Madeleine with AI, Dreyfus claims that the very existence of Madeleine's skills reinforces, rather than refutes, his phenomenological concerning expertise being rooted in the human body. Describing Madeleine, Dreyfus writes: has feelings, both physical and emotional, and body that has an inside and an outside and can be moved around in the world. Thus she can empathize with others and to some extent share the skillful way they encounter their (1992, xx). Although not fully able-bodied, Dreyfus characterizes Madeleine as embodied like other humans; this seems to mean that she has physical body (although not the typical 'body schema'), an affective relation to learning and improving skilled behavior, and the ability to store and organize facts, which is aided by empathy and imagination. Even though Harry Collins, like Dreyfus, is critical of Lenat's optimism regarding the progress of AI, he still finds it instructive to return to the Madeleine example. Coming from sociological perspective, Collins interprets Dreyfus's comments about as unwittingly proving that his phenomenological commitments are based on an inconsistent understanding of embodiment. Specifically, Collins takes Dreyfus's remarks to suggest that phenomenologists vacillate between conceiving of embodiment as something physical and conceiving of it as something conceptual: But under this argument [e.g., the Madeleine example] body is not so much physical thing as conceptual structure. If you can have body as unlike the norm and as unable to use tools, chairs, blind persons' canes and so forth as Madeleine's, yet you can still have common sense knowledge, then something like today's computers-fixed metal boxes-might also acquire common sense given the right programming. It is no longer necessary for machines to move around in the world like robots in order to be aware of their situation and exhibit 'Intelligence.' (1996, 104) Despite this last comment about robots, Collins does not try to reveal problems with Dreyfus's notion of embodiment in order to argue that computers are more intelligent than Dreyfus thinks; rather, he intends to prove that attributions of intelligence are based on linguistic socialization into form of life. To this end, Collins proposes that being could exhibit intelligent behavior in our world without sharing the common form of human embodiment. To clarify this proposal, Collins claims that we need to distinguish between what he calls the embodiment thesis and the minimal embodiment thesis, noting how each relates to the possibility for intelligence being attributed by human community (Collins 2000, 185-88). According to the embodiment thesis, the shape of the body is important at the collective level for establishing social norms, but an individual who lacks the type of embodiment found in given society can nevertheless still be socialized into that society. The minimal embodiment thesis is Collins's concession that intelligences need enough of body to have some inputs in order to be socializable (2000, 188). Collins's main point, then, is that so long as an individual can respond to sensory inputs, he or she can come to be viewed as exhibiting intelligent behavior by society whose members share completely different form of embodiment. To win the reader's assent to this view, he proposes an analogy between Madeleine and lion cub. …