Skip to main content Accessibility help

Pollock on How to Build a Person*

  • Ausonio Marras (a1)


Pollock's How to Build a Person is a highly sophisticated defence of the “strong” Artificial Intelligence (AI) program within the framework of a physicalist/functionalist ontology. Its central thesis is as robust as any defender of strong AI could possibly want it to be: to build a person—“a thing that literally thinks, feels, and is conscious” (p. ix)—is to build a physical system that adequately models human rationality; and building such a system is, essentially, a computational task. The epistemological groundwork for the computational task (a task already underway at the University of Arizona under the name “the OSCAR project”) is sketched in the sixth and last chapter of the book—a 64-page-long chapter (“Cognitive Carpentry”) where Pollock presents the main lines of a general theory of rationality which he has been developing over the course of many years. The bulk of the book (chaps. 1 to 4) is devoted to an articulation and defence of what Pollock takes to be the metaphysical underpinnings of his conception of a person as an intelligent machine: token-physicalism, agent materialism, psychophysical supervenience and (a form of) analytical functionalism. This part of the book, which will be the focus of this review, provides an account of “the physical basis for mentality”—an account of what it is that makes a physical structure capable of having mental states relevantly like our own. And since on the proposed account mental capacity “involves a way of codifying information—a system of mental representations” (p. 93), Pollock devotes a chapter (chap. 5) to the semantics of “the language of thought”; though basically a digression, since “semantics plays no role” in the computations involved in rational thought, the chapter is intrinsically very interesting, bearing as it does on current debates on thought ascription and content individuation.



Hide All


1 Schiffer, S., Remnants of Meaning (Cambridge, MA: MIT Press, 1987), p. 146.

2 Davidson, Donald, “Mental Events,” in his Essays on Actions and Events (Oxford: Clarendon Press, 1980), pp. 223–24.

3 Gibbard, A. (“Contingent Identity,” Journal of Philosophical Logic, 4 [1975]: 191) also suggests that the statue should be identified with a “piece of clay in that shape.” Pollock rejects the suggestion, but his reply seems to me beside the point.

4 Surprisingly, Pollock argues that it is not; but the argument relies on a thought-experiment which, like many thought-experiments in philosophy, is altogether inconclusive. Pollock himself candidly admits: “The example I have just construted would seem to be the best possible kind of example for showing that Oscar is distinct from his body. What is remarkable and deserves to be emphasized is how weak an argument this example provides” (p. 40).

5 Though Pollock does not say so, there are actually six distinct ways to interpret the question, depending on whether the conditions being sought are necessary, sufficient or necessary and sufficient, and on whether they are logically or nomically so.

6 Besides assuming a rationality principle (“a cognizer must tend to be rational”), Pollock also assumes that people are “overwhelmingly rational” (p. 70). But this panglossian assumption is actually not needed by the kind of theory Pollock envisages. For such a theory purports to be about mental structure, not about the actual cognitive behaviour of people: i.e., it purports to be a theory of competence, not a theory of performance.

7 Twin-Earth cases would not present a problem for Pollock even if he settled for local supervenience, for the only states that turn out to be relevant to cognitive processing on his account are those individuated by (introspectively accessible) syntactic properties. Individuation of states by their content is irrelevant to computation; hence there is no need for states so individuated to supervene on physical properties.

8 Pollock seems aware of this objection. For in “How to Build a Person: The Physical Basis for Mentality” (Philosophical Perspectives, 1 [1987]: 135) he rejects the proposal that “qualia might have some purely functional description” on the ground that “the only functional characteristics they have essentially are those involved in their rationality conditions and … these are isomorphic for all qualia.” Apparently, Pollock must believe that this objection no longer holds; but he does not tell us why.

9 In “Social Content and Psychological Content,” in Contents of Thought, edited by Merril, D. and Grimm, R. (Tucson: University of Arizona Press, 1988).

* Pollock, John, How to Build a Person: A Prolegomenon (Cambridge, MA: MIT Press, 1989), ix + 189 pp. Page references are to this work.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

Dialogue: Canadian Philosophical Review / Revue canadienne de philosophie
  • ISSN: 0012-2173
  • EISSN: 1759-0949
  • URL: /core/journals/dialogue-canadian-philosophical-review-revue-canadienne-de-philosophie
Please enter your name
Please enter a valid email address
Who would you like to send this to? *


Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed