The Mind Doesn鈥檛 Work That Way by Jerry Fodor, MIT Press, 拢15.95, ISBN 0262062127
鈥淪O FAR, what our cognitive science has found out about the mind is mostly that we don鈥檛 know how it works.鈥� Thus intones philosopher Jerry Fodor in the closing lines of his short new book The Mind Doesn鈥檛 Work That Way. As the title hints, this work was written largely as an antidote to psychologist Steven Pinker鈥檚 recent bestseller How the Mind Works, and in particular, it is a riposte to what Fodor labels the 鈥渞elentlessly cheerful鈥� tone adopted by Pinker.
Fodor鈥檚 key thesis is that much of human thought is 鈥済lobal鈥�, whereas computational models of thought are 鈥渓ocal鈥� or 鈥渟yntactic鈥�. Never, supposedly, shall the twain meet. So, Fodor says, there is something seriously wrong with the project of understanding the mind from the bottom up, where 鈥渂ottom鈥� is a set of relatively simple and independent 鈥渕odules鈥�.
Advertisement
There is ample evidence, from people with brain damage for example, that such local modules explain a great deal of what goes on when, spotting a lumpy inert beige object on a kitchen counter, we say 鈥減otato鈥�. Specific damage knocks out specific parts of the process: the ability to find the word for something you recognise clearly as a vegetable, for example.
Can science work upward from such observations 鈥� and from the thesis that such modules can be modelled computationally 鈥� to a full description of how a human mind works? This question goes to the heart of the endeavour known as 鈥渃ognitive science鈥�. If Fodor, or those whom philosopher Owen Flanagan has called the 鈥渘ew mysterians鈥�, is right, you can鈥檛 make this transition, and cognitive science is merely a hopeless pipe dream. How does Fodor make his case against cognitive science?
Unfortunately, in keeping with a long-standing personal tradition, Fodor has written a text as relentlessly obfuscatory as Pinker鈥檚 is relentlessly cheerful. Conducted at a breathtakingly high level of abstraction, it comes very close to being example-free. These sad facts place Fodor鈥檚 book at the opposite end of the spectrum from Pinker鈥檚, not only in terms of content but also in terms of style.
The primary type of 鈥済lobality鈥� that Fodor mentions is what he calls 鈥渁bduction鈥�. Astonishingly, the sole example that he provides for this concept is the dilemma of whether or not to invest in potato futures. In mock modesty, Fodor says: 鈥淣ot that I鈥檓 exactly sure what a potato future is, or of which potatoes have them. But I don鈥檛 imagine the details of the examples matter much.鈥� He declares that the problem with such a decision is its unlimited context-dependence. That is to say that it potentially depends on the relevance of facts that would seem to be arbitrarily far removed from the domains of potatoes and of finances. And here I, descending far closer to terra firma than Fodor ever does, would suggest that concrete factors such as recent weather trends, moves toward unionisation of farm workers and so on matter. Yet they might seem, a priori, quite distant from the matter at hand, in other words, irrelevant at first glance.
In discussion of his own theory of 鈥渕ental modularity鈥�, Fodor attempts to show why no one today has a plausible theory of relevance. Such a theory would supply an answer to the far from trivial question of how minds could possibly be so subtly organised as to allow a given situation to unconsciously coax from dormant memory not a gushing flood of utter irrelevancies, but rather a silver platter bearing a minuscule set of distant-seeming yet in fact perfectly on-target associations: an infinitesimal subset of the jam-packed larder of possibilities. In the absence of a theory of relevance, no one would have a clue how the human mind performs such 鈥渁bductive inferences鈥� as deciding about potato futures, asserts Fodor.
His point is well taken. Retrieving appropriate memories while not retrieving a flood of inappropriate ones is a deep challenge to those who model memory organisation, but there are many plausible approaches at this time.
As for Fodor鈥檚 style, a sample footnote will suffice: 鈥淕ood Quineian that I am, I suspect that (outside logic and mathematics) all rational estimates of relevance are empirically revisable (rather than stipulative, definitional, or otherwise semantic). I thus set my face against verificationists, operationalists, and creteriologists [sic]; and Granny sets her face against them too. You don鈥檛, however, have to accept this intransigent view to agree with us that estimates of relevance are sometimes empirically revisable, which is all that the point in the text requires.鈥�
This paragraph, crammed with allusions to nit-picky arguments among a certain inbred group of epistemologists, characterises Fodor鈥檚 prose style.
There are those in search of an authority who would scornfully attack computational models and cognitive scientists, claiming that the mind鈥檚 workings are murky and impenetrable. Such readers will be pleased with this book, finding in it ammunition to defend Fodor鈥檚 fortress of the mystery of mind.
Some of us, by contrast, not only share Fodor鈥檚 interest in the philosophical issues, but feel it is useful to formulate and test psychological theories, or perhaps to implement computational models of global-seeming features of thought, such as abstract reminding and creative analogical leaps. Indeed, it seems to me that the most interesting way to see whether local, computational models work is to build some and see what they do.
If you are of this view, then I would suggest that you might more profitably spend your time in trying to understand what potato futures are and whether they鈥檙e worth investing in, rather than in trying to read Fodor鈥檚 latest book. After all, ars longa, vita brevis. If you don鈥檛 believe me, ask Granny. She鈥檇 set her face against it, I swear.