Abstract
In addition to his famous Chinese Room argument, John Searle has posed a more radical problem for views on which minds can be understood as programs. Even his wall, he claims, implements the WordStar program according to the standard definition of implementation because there is some “pattern of molecule movements” that is isomorphic to the formal structure of WordStar. Program implementation, Searle charges, is merely observer-relative and thus not an intrinsic feature of the world. I argue, first, that analogous charges involving other concepts (motion and meaning) lead to consequences no one accepts. Second, I show that Searle’s treatment of computation is incoherent, yielding the consequence that nothing computes anything: even our standard personal computers fail to run any programs on this account. I propose an alternative account, one that accords with the way engineers, programmers, and cognitive scientists use the concept of computation in their empirical work. This alternative interpretation provides the basis of a philosophical analysis of program implementation, one that may yet be suitable for a computational theory of the mind.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Searle’s charge here is thus much more radical than that made by his Chinese Room thought experiment or by Ned Block’s Chinese Nation thought experiment. (Searle 1980; Block 1978) In these cases, program implementation is seen to require severe physical regimentation either of an inner rule-following human or by an entire population of humans. Searle’s wall, however, can be left alone. All that is required is that there be some way of interpreting whatever it happens to be doing that is isomorphic to the formal structure of some desired program.
Many philosophers (Chalmers 1996; Copeland 1996; Block 2002; Haugeland 2002; Rey 2002) have objected that a thing such as Searle’s wall cannot have the behavioral dispositions that would meet counterfactual constraints imposed by some programs. But, intuitive as such appeals may be, they face two problems. First, some authors claim that it doesn’t always make sense to require counterfactual constraints. (Maudlin 1989; Scheutz 1998; Bishop 2002) Second, it has not been shown that the right counterfactuals cannot be secured; in fact, the means (however dubious) by which one can argue that a sufficiently complex wall displays the requisite actual behavior can be generalized to argue that the wall is disposed to display the requisite counterfactual behavior. The idea here is that whatever descriptive liberties apply to the actual physical state changes of the wall can be applied to counterfactual state changes as well, putting these enhanced counterfactual-supporting descriptions on equal footing with the original inspirations we find in Putnam (see footnote 1) and Searle. David Chalmers (1996) exploits this opportunity in order to enhance Putnam’s argument, adding to the physical system a dial that has, for each other way the program could have been run, a position the dial could have been in. Teller, in an unpublished draft (1999), exploits the unbounded potential to use refined physical state descriptions (also exploited by Putnam) to secure not only the requisite counterfactuals, but input–output relations and a kind of internal complexity. The foregoing gives some indication why I think Searle underestimates the power of his objection. For the purpose of this paper, which is to identify the real problem with Searle’s Wall, I grant that it can be shown that Searle’s wall is so disposed.
Isomorphism is widely acknowledged to be cheap. For some arguments establishing isomorphism (albeit in these cases between a variety of significantly different kinds of physical and computational entities), see Putnam (1988); Chalmers (1994, 1996), Copeland (1996), and Scheutz (1998, 1999). These arguments may require using seemingly ad hoc definitions in order to acknowledge odd, disjunctive, or gerrymandered physical states, but as I think a survey of this literature shows, justifying their rejection has proven difficult and controversial.
Although this may not have bothered Galileo, speeds greater than the speed of light are included.
Searle frequently insists that computational features are not intrinsic features of the world but are instead observer-relative ones. This may leave the impression that he and I agree at least on the three-place nature of the implementation relation, disagreeing only on whether the third relatum is an observer, as he insists, or a description, as I propose. But this is to overlook the important distinction just made, that distinction between a two-place relation that is merely conditional on a third thing and a three-place relation that includes that third thing as a relatum. In light of this distinction, it is more accurate to say that Searle has been treating program implementation as being merely observer contingent, as if it holds whenever an observer can see it as holding. But this cannot be right, as I think my second argument, presented in the next section, will show.
By the same token, the fact that we rarely cite frames of reference or linguistic interpretations is not taken to imply that motion or meaning does not hold in ways that are relative to these things.
Mappings have been specified in a variety of ways. Copeland (1996), for example, takes a model theoretic approach, while Putnam (1988) stipulates “definitions” of computational states in terms of physical states. But however a mapping is specified, it must at least be equivalent to a function from physical states of an object to computational states of a program such that the causal relations among physical states preserve the transition relations among computational states. What exactly counts as a physical state or a computational state I leave open. (Putnam 1988; Chalmers 1996; and Copeland 1996 are willing to consider grossly disjunctive physical states.) Moreover, what counts as an appropriate causal relation is left open. The strength of my point that implementation is a three-place relation over causal structures, programs, and mappings does not rely on the specific nature of mappings.
Turing’s original exposition (1936) of computing machines makes it clear that states are mutually exclusive and that state change is thus nontrivial, and they are routinely treated as such by practitioners.
Similar remarks hold for meaning.
References
Bishop, M. (2002). Dancing with Pixies. In Preston & Bishop (Eds.).
Block, N. (1978). Troubles with functionalism. In C. W. Savage (Ed.), Perception and cognition: Issues in the foundations of psychology. Minneapolis: University of Minnesota Press.
Block, N. (2002). Searle’s arguments against cognitive science. In Preston & Bishop (Eds.).
Chalmers, D. J. (1994). On implementing a computation. Minds and Machines, 4, 391–402.
Chalmers, D. J. (1996). Does a rock implement every finite-state automaton? Synthese, 108, 309–333.
Chrisley, R. (1994). Why everything doesn’t realize every computation. Minds and Machines, 4, 403–420.
Copeland, B. J. (1996). What is computation? Synthese, 108, 355–359.
Haugeland, J. (2002). Syntax, Semantics, Physics. In Preston & Bishop (Eds.), pp. 379–392.
Maudlin, T. (1989). Computation and consciousness. The Journal of Philosophy, LXXXVI(8), 407–432.
Preston, J., & Bishop, M. (Eds.). (2002). Views into the Chinese room: New essays on Searle and artificial intelligence. New York: Oxford University Press.
Putnam, H. (1988). Representation and reality. Cambridge: MIT Press.
Rey, G. (2002). Searle’s misunderstandings of functionalism and strong AI. In Preston & Bishop (Eds.), pp. 201–225.
Scheutz, M. (1998). Implementation: Computationalism’s weak spot. Conceptus JG, 31(79), 229–239.
Scheutz, M. (1999). When physical systems realize functions. Minds and Machines, 9, 161–196.
Searle, J. (1980). Minds, brains and programs. Behavioral and Brain Sciences, 3, 417–457.
Searle, J. R. (1992). The rediscovery of the mind. Cambridge: MIT Press.
Teller, P. (1999). Rocks can think (unpublished draft).
Turing, A. (1936). On computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society, 42(1936), 230–265.
Acknowledgments
Special thanks to two anonymous referees for this journal who made helpful comments and to Andrew Melnyk for helpful comments on a previous draft presented at the 2011 APA Conference, Central Division. I am also very grateful to Paul Teller, Robert Cummins, Jonathan Dorsey, John Keller, and Michael Jubien for valuable discussions and commentary regarding earlier drafts of this paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Blackmon, J. Searle’s Wall. Erkenn 78, 109–117 (2013). https://doi.org/10.1007/s10670-012-9405-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-012-9405-4