Function and Functionalism: a synthetic perspective. Parte 3

6. From Function to Functionalism. With the help of the preceding elucidations, we can now take a look at functionalism. Having distinguished six concepts of function, we may in principle form six corresponding notions of functionalism. Whether there actually are six such notions will be examined in the following sections.

Functionalism is usually understood as the (ontological) thesis that function is all-important and stuff (or composition) nothing; more carefully stated, stuff is relevant at most as the material carrier of functions, but, inasmuch as two things perform the same function1,2, their material or compositional differences do not matter. Witness Putnam’s classic dictum: “We could be made out of Swiss cheese and it wouldn’t matter.” (1975,291) If functionalism claims that the internal activities (or functions1) of two compositionally different things can be essentially the same, we call it, for reasons to be given in the next section,formalist functionalism (or functionalism1). If functionalism claims that two compositionally different things can perform the same external activities or roles (or functions2), we call it black box functionalism (or functionalism2), because it disregards the inner workings of the things in question. The same name will be given to its epistemological counterpart, namely the claim that all we need to know about things is what they do or what roles they perform (functions2), regardless of substance and mechanism.

The term ‘functionalism’ also designates the thesis that every item of certain kinds (biological, social, or technical) serves some function (“purpose”). In this case ‘function’ may refer to an aptation, an adaptation, or even an intention or teleofunction. We call the former two adaptationist functionalism (functionalism4-5), and the latter teleological functionalism (or teleofunctionalism). Teleological functionalism proper would hold that all socio- and technofunctions, whether or not they are aptations or adaptations, are in any case teleofunctions, in being useful to somebody or at least in being intended to be useful to somebody.

In the following we shall examine which functionalisms discussed in the philosophical literature do match the above distinctions. We shall also assess these functionalisms with respect to their explanatory power.

7. Functionalism in Biology and Psychology. The best-known functionalism is the one discussed in the philosophy of mind and in Artificial Intelligence (AI) (see, e.g., Putnam 1975; Dennett 1978; Block 1980; Fodor 1981). The latter’s counterpart in biology is the Artificial Life (AL) project (see Langton 1989; Langton et al. 1991). In both cases, two versions of functionalism are commonly distinguished: weak (or moderate) and strong (or radical). Weak functionalism claims only that some biological (or psychological) functions can be successfully imitated by artificial systems or machines. By contrast, strong functionalism claims that mind as a whole-or life as a whole, as the case may be–can be replicated (not just mimicked) by machines, and perhaps by other things as well.

Obviously, certain artifacts can imitate the net effect or outcome of certain biological functions1. That is, they can perform certain roles (functions2 even though they differ from the corresponding living thing in both composition and structure; in other words, they are functionally equivalent as defined in Section 5. For example, when adequately programmed, a computer can perform certain computations, such as calculating the value of an atomic energy level. But it does so through processes (functions1) widely different from the corresponding processes in the living brain; moreover, the computer does not know what an energy level is. Therefore, any such imitation is partial. Thus, weak functionalism is a functionalism2: it claims only that certain functions2 are substrate-neutral in the sense that different mechanisms can perform the same role. In disregarding the mechanisms underlying the roles in question, functionalism2 is a black box functionalism: it may provide functional accounts, but no functional explanations.

The strong functionalism that rules in both AI and AL is quite another matter. The former is the thesis that computers (or computer programs) are equivalent to thinking brains (or minds), in that they can replicate in toto whatever the original things do. This equivalence is said to be functional (or structural), in being substrate-neutral, i.e., independent of the stuff that “embodies” the ideas concerned. What matters are computations or algorithms, not material mechanisms. Similarly, AL holds that life is a property of the organization of matter, not so much a property of the matter itself that is so organized. Thus, AL is concerned with the “formal basis” of life, which would allow us to replicate or synthesize genuine life processes in computers and perhaps other artificial things. (Langton 1989)

In both cases, the functions in question are not certain roles (functions2) that might be performed with different underlying mechanisms (functions1): The mechanisms or processes themselves (functions1) are said to be replicable in different “substrates”. Thus, in contradistinction to weak functionalism, strong functionalism is an instance of functionalism1. However, to a materialist, this position should be unacceptable. The point of materialism is that only matter matters and, in particular, that the functions1 of any thing depend critically upon the kind of stuff the thing consists of. Change it, and the organization and mechanism of the thing will change accordingly. Functionalism,, in contrast, assumes that processes (mechanisms, functions1) are not changes in concrete things but instead stuff-free algorithms, which are purely formal objects. For this reason, we call it formalist functionalism.

Although strong functionalism is a functionalism1 for referring to mechanisms, these mechanisms are, as we have just seen, formal, not material, mechanisms. Unless we share the idealist belief that formal objects can steer material processes, or be embodied in material objects, we cannot admit that formalist functionalism has explanatory power. (Further criticisms are made in Boden 1999.)

8. Adaptationist Functionalism. Traditional teleological functionalism, viewing everything as being purposive or goal-seeking, is no longer tenable. Its scientifically respectable, but not necessarily truer, descendant is adaptationist functionalism. In biology, this is the thesis that all traits are ultimately adaptive (or functional5, i.e., the product of natural selection. In teleological parlance, it asserts that “selection acts for the benefit” of something, be it the individual (organism or gene) or–nowadays less fashionably–the species. Thus, if observation shows that A does B, the adaptationist will say that A has been designed to do B, or that B is the proper function or purpose of A, or that B has evolved for being beneficial to its bearer, or at least to the latter’s genes (see, e.g., Williams 1966). Thus, it maintains that the concept of function is that of function5 (or proper function or adaptation), also known as the etiological concept of function.

Some adaptationists regard even disease as a means for survival and reproduction–though admittedly one that can occasionally turn on its bearer. (Nesse and Williams 1994) These authors refrain from asking whether every disease has an evolutionary origin: instead, they assume this to explain all diseases in such terms. Indeed, they assert that “we must discover the evolutionary causes for each disease.” (6) We submit that this way of approaching biological features is precisely the main methodological flaw of adaptationism in general.

Nevertheless, if an aptation can be shown to be also an adaptation, this may provide a genuine explanation, for it will refer to evolutionary mechanisms such as natural selection.

9. Pan-Functionalism in Evolutionary Biology and Psychology. Daniel Dennett, who has long espoused functionalism in the matter of mind (1978), has recently extended it to evolution (1995). In so doing, he combines formalist with adaptationist functionalism. Dennett’s formalist thesis is that evolution is a substrate-neutral algorithmic process. However, if bioevolution were indeed substrate-neutral, it would be the only process of this kind. The thesis that evolution is an algorithmic process is equally dubious, for it assumes that evolution reduces to selection. But selection is a sorting process, not a source of qualitative novelty. (Sober 1984; Mahner and Bunge 1997; Buller 1998; Walsh 1998) And to account for the emergence of qualitative novelty (not just its subsequent distribution in organismal populations) is what we expect from a satisfactory theory of evolution. In any event, Dennett’s thesis is just hand-waving, for he exhibits no evolutionary algorithm that we could feed into a computer to check whether, indeed, “it” correctly predicts speciation and extinction.

The assumption that some kind of algorithmic-selectionist formalism is the driving force of everything evolutionary leads Dennett straight to adaptationist functionalism: Since everything biological is the result of selection, all features are adaptive–if not prima facie, then at least ultimately. Thus, panselectionism is paired off to panfunctionalism. (Ahouse 1998) No wonder, then, that Dennett devotes a considerable portion of his book to attacking Gould’s and Lewontin’s (1979) classic critique of adaptationism (see also Gould 1997a, 1997b; Mahner and Bunge 1997; Ahouse 1998; Godfrey-Smith 1998).

Combining the adaptationist and formalist approaches to biological evolution with the view that the mind is a sort of computer program, or information processor, yields the latest version of evolutionary psychology and its foray into social science. According to its parents (Cosmides and Tooby 1987), the main theses of evolutionary psychology are the following. first, psychology deals with function or purpose, not organ, and every function can be “instantiated” in a variety of systems. “Just as different kinds of hardware can run the same computer program, different physiological mechanisms can accomplish the same adaptive function.” (283) Hence, psychology needs no neuroscience. Second, “natural selection theory is a theory of function.” (284) Third, there are innate psychological mechanisms, all of which are information-processing programs that “extract information from the environment” and process it: they “map informational input onto behavioral output.” Fourth, the psychological mechanisms in charge of learning special tasks are “Darwinian algorithms” that “organize experience into adaptively meaningful schemas or frames.” (286) Chomsky’s mysterious “language acquisition device” would be one of them.

As will be obvious from the preceding sections, these views are open to serious objections. To begin with, functions3-5 are not substrate-independent. If in doubt, try to build a computer chip with air or iron rather than silicon. For this reason, no two organs will be able to perform the same functions1, although some may indeed perform the same functions2 (or roles). Hence, again, the retreat to formal mechanisms, i.e., algorithms. Moreover, function presupposes organ, but the converse is false. Hence, natural selection does not only select “for” external activities of organs–which is what the selection-sees-only-function thesis asserts–but indirectly also “for” internal activities.

Furthermore, in our view, a mechanism is a process in a concrete thing; hence there are brain mechanisms but not psychological ones. Besides, contrary to what the defunct “ecological psychology” held, the environment contains no cognitive information. But even if it did, there is no evidence for the algorithmic nature of all cognitive processes, from face recognition to concept formation. Rather, our tentative and often unsuccessful performance at such tasks is in sharp contrast with the certainty inherent in any algorithm-steered process. finally, the claim that “learning is accomplished through psychological mechanisms (whose nature is not yet understood)” (Cosmides and Tooby 1987, 292) reminds one of Moliere’s doctor, who solemnly declared that opium makes one sleepy because it possesses the “dormitive virtue.”

To conclude, the currently fashionable version of evolutionary psychology is purely speculative: it thrives on concocting adaptationist just-so stories, which may be plausible but are as yet untested–and it is not obvious how they could be tested unambiguously. We submit that it may become a field of scientific research, but only once the adaptationist, formalist, and computationalist dogmas have been jettisoned in favor of the psychobiological approach to mind. (For further and different criticisms see Lloyd 1999.)

Mahner, Martin, Bunge, Mario, Philosophy of Science 


Introduce tus datos o haz clic en un icono para iniciar sesión:

Logo de

Estás comentando usando tu cuenta de Cerrar sesión / Cambiar )

Imagen de Twitter

Estás comentando usando tu cuenta de Twitter. Cerrar sesión / Cambiar )

Foto de Facebook

Estás comentando usando tu cuenta de Facebook. Cerrar sesión / Cambiar )

Google+ photo

Estás comentando usando tu cuenta de Google+. Cerrar sesión / Cambiar )

Conectando a %s

A %d blogueros les gusta esto: