¹ Ehring, Douglas, “Accidental Functions”, Dialogue 25/2 (1986), 291–302.CrossRefGoogle Scholar

² Enc, Berent, “Function Attribution and Functional Explanations”, Philosophy of Science 46 (1979), 343–365.CrossRefGoogle Scholar

³ Adams, Frederick, “A Goal-State Theory of Function Attributions”, Canadian Journal of Philosophy 9 (1979), 498–518.CrossRefGoogle Scholar

⁴ One of the more recent reviews of the literature on functions occurs in Peter Achinstein's The Nature of Explanation (Oxford: Oxford University Press, 1983).Google Scholar

⁵ See Hempel, Carl, “The Logic of Functional Analysis”, Aspects of Scientific Explanation (New York: Free Press, 1965).Google Scholar

⁶ See both Enc, “Function Attribution”, and Adams, “A Goal-State Theory” for elaboration and examples of this intensionality.

⁷ Attempts at developing a naturalistic theory of intentionality share a core of identical problems with functional analysis. Hence, it is no coincidence that many philosophers have availed themselves of a naturalistically conceived account of functions in their dealings with intentionality. See, for example, Millikan, Ruth, Language, Thought and Other Biological Categories (Cambridge, MA: Bradford/MIT, 1984)Google Scholar, Stampe, Dennis, “Toward a Causal Theory of Linguistic Representation”, Midwest Studies 2 (1977), 42–63CrossRefGoogle Scholar, Enc, Berent, “Intentional States of Mechanical Devices”, Mind 91 (1982)Google Scholar, Adams, Fred, Goal-Directed Systems, Ph.D. Dissertation (Madison, WI: University of Wisconsin, 1982)Google Scholar, and “Comparison Shopping In The Philosophy of Mind”, Critica 27 (1985), 45–70.Google Scholar

⁸ Adams, , “A Goal-State Theory”.Google Scholar

⁹ Ene, , “Function Attribution”.Google Scholar

¹⁰ Since Enc's original paper explains the conditions of his model and our defense of it does not now depend upon its intricacies, we shall not elaborate upon the model here. We list the conditions only for fullness of reference. They are as follows: The function of X in W is to do Y only if given the kind ofthing X is and the kind ofthing W is, (1). If X were under conditions C1, then normally (i.e., unless a set of specifiable but not necessarily specified, conditions C' occurred) a series of events, S (movements and changes of state) would occur in X; (2) If a series of conditions C2 are satisfied, X's going through S will be causally sufficient for Y (or will be one of the several causally relevant factors which are jointly sufficient for Y); (3) Y is necessary for the maintenance or the realization of some property D or W; (4) At least in those cases of X where part of the identity conditions for being an X involves the function of X, X is the only kind of thing which under normal conditions (C2 and not-C') is capable of causing Y in all W's; (5) Part of the explanation of why X normally maintains S, or undergoes S, lies in the fact that S results in Y.

It is condition (5) which is designed to separate mere effects from functions. Those structures in which S would be maintained e ven if S did not result in Y violate (5), and in those structures S is unmasked as a mere effect. Ehring thinks his example shows (5) not to be necessary. In the example to be discussed, he says Jones would be pushing a certain button regardless of its outcome. Thus, it cannot be the outcome's serving some function which explains Jones's pushing it.

¹¹ Surprisingly, Ehring says of this example that it shows “that there exist functions which are ‘accidental’ in the Encian sense [sic], i.e., [Enc's] condition fails to be satisfied for this class of functions”. Ehring, “Accidental Functions”, 293.Google Scholar

¹² It is no accident that both of our accounts (and Wright's, the third target of Ehring's analysis) stress the importance of kinds of systems, and kinds of activities in function attributions. See, for example, Wright, Larry, Teleological Explanations (Berkeley, C A: University of California Press, 1976)Google Scholar as well as either of our accounts.

¹³ The problem of “one time shot” cases in the analysis of goal-directed systems is not new (not even to Ehring's criticism of Adams). See Adams, , “Feedback About Feedback: Reply to Ehring”, The Southern Journal of Philosophy 24 (1986), 123–131.CrossRefGoogle Scholar

¹⁴ See Peter Unger's sensitive discussion of different uses of “accident” in “An Analysis of Factual Knowledge”, Journal of Philosophy 65 (1968), 157–170Google Scholar. Also see Adams's discussion of the same in “The Function of Epistemic Justification”, Canadian Journal of Philosophy 16 (1986), 465–492.Google Scholar

¹⁵ One of the conditions of Adams's model is that there be a dependency of functional output upon an unbroken control loop involving a goal-directed system. The model reads: A structure x has a function y just in case: (1) x does y in system S; (2) y causally contributes towards S's outputting 0 (through the causal feedback mechanism); (3) 0 is (or itself contributes toward) a goal-state of S. If for no other reason, the rock has no function since it violates condition (2). As is made clear in the context of the original paper, for artifact functions, the causal feedback mechanism includes an information flow through the goal-directed system S. If this information flow is broken, the conditions of the model are violated.

¹⁶ So, for example, if by a miracle a heating system complete with feedback control circuitry were to have popped into existence yesterday on Mars, it would not as a whole, nor any of its components, have a function. However, it could acquire one if suitably conjoined with a goal-directed user. (See the discussion of the “functional supervenience” thesis in the next section.)

¹⁷ See Adams, , Goal-Directed Systems.Google Scholar

¹⁸ For a review of the action-theoretic literature on deviant causal chains see Brand, Myles, Intending And Acting (Cambridge, MA: Bradford/MIT, 1984)Google Scholar. Also see Adams, , Goal-Directed SystemsGoogle Scholar and Adams, Fred and Al Mele, , “The Role of Intentions in Intentional Action”Google Scholar (forthcoming in the Canadian Journal of Philosophy) for the control-theoretic solution to the problem and see Berent Enc, “Causal Theories and Unusual Causal Pathways” (forthcoming in Philosophical Studies) for a slightly different solution.

¹⁹ See Adams, , “Comparison Shopping”Google Scholar for a discussion of the co-relative problem of “narrow” and “wide” specification of mental states now known as the problem of “Methodological Solipsism”.