Book Review: Mostly Harmless

Radical as this claim may seem to a casual reader, the actual alternative that emerges from integrating Noë’s arguments with the state of the art in vision research is in fact rather tame, if not entirely mainstream (hence the title of my review). The main reason for this happy, if somewhat anticlimactic, conclusion is that the first part of Noë’s thesis—that perception cannot be fully internal— is perfectly in tune with quite a few existing theories of cognition [3, 13, 6, 5, 7]. The unifying framework behind these theories is this: they all take heed of the injunction that the external world should bear the main burden of being its own representation [17, 15, 16]. Some of the many existing arguments in favor of this approach are computational: reconstructing the world in an internal memory space is infeasible, ultimately futile, and actually unnecessary [17, 6]. Others are psychological: people’s performance in perceptual tasks suggests that no reconstruction is attempted [2, 6]. Excellent philosophical works compatible with the idea that the world is its own best representation are also available [13, 5]. Much of the present book is a useful contribution to this latter literature, made all the more valuable by the fine balance it strikes between the roles of internal and external components of representations (rather than rejecting the former entirely; cf. [6]). What about the second part of Noë’s thesis—that ‘‘to perceive you must be in possession of sensorimotor bodily skill ’’ (p. 11)? In my understanding, this statement (and many others similar to it in the book) unpacks into the following postulate: to perceive the world as a human does, you must reside in a human body, and, moreover, be in constant, full sensorimotor control of that body. We thus discover that there are actually two independent claims here: (i) the body must be human, because it is human perception we are interested in, and (ii) active control of that body is required to support perception. Let us examine these in turn. Embodiment in a human form is clearly only an issue if it is specifically humanlike perception that is at stake. This view hardly needs special justification: denying that animal brains, alien brains, or computers are capable, respectively, of animal-, alien-, or machinelike perception would amount to an untenable anthropocentrism [19], or worse (cf. ‘‘What do you mean . . . human?’’ [4]). Indeed, Noë acknowledges that ‘‘to perceive like us [ . . . ] you must have a body like ours’’ (p. 25; my emphasis).