The Mind Doesn't Work That Way: The Scope and Limits of Computational Psychology

Rating:
Summary: Wanted a fight, found a melt down
Review: Cognitive science is a failure. AI is a massive failure. Cognitive scientists know their computational theories of mind are wrong but they are in "deep denial". Fighting words from famous philosopher Jerry Fodor's response to famous psychologist Steven Pinker's "How The Mind Works". With fighting words like that, i was expecting a really good fight, with facts being offered, stats being analyzed and Pinker's analysis of topographical cortical maps and the visual processing map dissected

And instead i got... this

Let's start with writing. It's typical Fodor, which means it would put a coked-up Tigger to sleep and requires a Latin dictionary and the patience of Job to slog through. It's boring and difficult to read. But since it's a Fodor book, you probably already knew that. He's a fun guy and in previous books had some good points but man is he boring

But i didn't expect good writing, good structure or brevity. i expected good ideas. But unless they're hiding on the last page (sorry, i can't keep reading this thing), there are none in this book. You might wonder why a philosopher whose interest is in symbolic logic and grammar is writing a book about psychologists who use a very loose analogy about the mind being like computer software. After reading this book, so am i

So why is Pinker and everyone else in cognitive science wrong? Fodor's argument in a nutshell - because computer subroutines can't access information outside of the subroutine. To use Fodor's example (of which he has *very* few), you want to go to Chicago. It's not windy. Is that good? If you're sailing a boat, yes, otherwise no. A computer can't figure that out because it lacks context - it needs to know how you're traveling before judging the meaning of wind conditions. And computers, Fodor argues, can't do that. Since computers have subroutines and subroutines have no access to relevant data, computers can't solve simple problems and so the mind can't be like a computer. Take that Pinker! Makes sense? Of course not

Fodor approaches the topic of psychology and computers from a logician's standpoint, which is to say he makes some really bad, sweeping assumptions then uses really high level logic to prove that reality doesn't exist. For Fodor, it's just 5 steps - 1.A mental representation must have a syntax, 2.If you change the sytax the "Turing machine" can no longer function, 3.Therefore syntax can't vary by context, 4.Therefore mental processes can't be affected by context 5.But they are so the mind isn't a computer and cognitive science is wrong. (He later adds Principle M(CTM) which says sunroutines can access external data but an exhaustive search/tablescan of all memory would be needed and that would be stupid so cognitive science is stupid)

The book is a one trick pony. Computers supposedly can't solve the simple problems he describes (logical abduction), they are required to solve them based on the way he chose to define his terms and therefore the field and its theories are irretrievably broken. The book is laced with numerous, sweeping, unfounded assumptions that are glaringly wrong to anyone familiar with computers. The book lacks common sense. The book's main and only objection is an academic exercise completely divorced from anything resembling cognitive science

Rating:
Summary: Wanted a fight, found a melt down
Review: Cognitive science is a failure. AI is a massive failure. Cognitive scientists know their computational theories of mind are wrong but they are in "deep denial". Fighting words from famous philosopher Jerry Fodor's response to famous psychologist Steven Pinker's "How The Mind Works". With fighting words like that, i was expecting a really good fight, with facts being offered, stats being analyzed and Pinker's analysis of topographical cortical maps and the visual processing map dissected

And instead i got... this

Let's start with writing. It's typical Fodor, which means it would put a coked-up Tigger to sleep and requires a Latin dictionary and the patience of Job to slog through. It's boring and difficult to read. But since it's a Fodor book, you probably already knew that. He's a fun guy and in previous books had some good points but man is he boring

But i didn't expect good writing, good structure or brevity. i expected good ideas. But unless they're hiding on the last page (sorry, i can't keep reading this thing), there are none in this book. You might wonder why a philosopher whose interest is in symbolic logic and grammar is writing a book about psychologists who use a very loose analogy about the mind being like computer software. After reading this book, so am i

So why is Pinker and everyone else in cognitive science wrong? Fodor's argument in a nutshell - because computer subroutines can't access information outside of the subroutine. To use Fodor's example (of which he has *very* few), you want to go to Chicago. It's not windy. Is that good? If you're sailing a boat, yes, otherwise no. A computer can't figure that out because it lacks context - it needs to know how you're traveling before judging the meaning of wind conditions. And computers, Fodor argues, can't do that. Since computers have subroutines and subroutines have no access to relevant data, computers can't solve simple problems and so the mind can't be like a computer. Take that Pinker! Makes sense? Of course not

Fodor approaches the topic of psychology and computers from a logician's standpoint, which is to say he makes some really bad, sweeping assumptions then uses really high level logic to prove that reality doesn't exist. For Fodor, it's just 5 steps - 1.A mental representation must have a syntax, 2.If you change the sytax the "Turing machine" can no longer function, 3.Therefore syntax can't vary by context, 4.Therefore mental processes can't be affected by context 5.But they are so the mind isn't a computer and cognitive science is wrong. (He later adds Principle M(CTM) which says sunroutines can access external data but an exhaustive search/tablescan of all memory would be needed and that would be stupid so cognitive science is stupid)

The book is a one trick pony. Computers supposedly can't solve the simple problems he describes (logical abduction), they are required to solve them based on the way he chose to define his terms and therefore the field and its theories are irretrievably broken. The book is laced with numerous, sweeping, unfounded assumptions that are glaringly wrong to anyone familiar with computers. The book lacks common sense. The book's main and only objection is an academic exercise completely divorced from anything resembling cognitive science

Rating:
Summary: ANOTHER EXCELLENT FODOR BOOK
Review: Fodor is one of our greatest living philosophers. He has laid out groundbreaking theories on the philosophy of mind and language. This book is a bit of a departure, in that he is less interested in theory-building than in theory-destruction. You might think that the result would be rather dreary -- who wants to hear someone harping on other people's mistakes all of the time? -- but in fact the book is very engaging.

Fodor argues that while "computational" models of the mind (roughly, theories that the mind is just a computer) may be able to explain how the mind's modules work, they fail to explain how the mind's central processor works. (If the theory of modules and central processors is unfamiliar to you, then you MUST first read Fodor's excellent book "The Modularity of Mind" in order to understand "The Mind Doesn't Work That Way.") The primary problem, as Fodor sees it, is that central processors are general-problem solvers (or, more accurately, general-interest learners). They work with large databases of beliefs and are bombarded by immense amounts of information. If the mind were like a computer, it would experience the problem of "combinatorial explosion" as it tried to analyze all of this information, that problem being that with so many sentences in the language of thought, there would be far too many calculations to perform over these sentences in anything like a reasonable amount of time. Put another way, if you think your computer is slow loading Windows 2000, just wait until it had to load and analyze the entire database of a person's beliefs AND chew gum at the same time.

Although the problem of combinatorial explosion is the driving problem in the book, Fodor addresses many other interesting topics in his typically witty and insightful way. He gives glancing blows to connectionism, Darwinian approaches to psychology, the theory of heuristics, and more.

As always, Fodor presents many forceful and ingenious arguments, and even when I disagree with him (as I often do), I always walk away from his books understanding difficult issues more clearly and having a profound respect for his penetrating intellect.

This book is not a good introduction to Fodor's work or to philosophy of mind, but for those with some grasp of both, I highly recommend "The Mind Doesn't Work That Way." If only he had come up with a better title....

Rating:
Summary: Very Relevant
Review: Pinker and company needed some curbing of egos. This book does just that, and it explains some of the real problems. Fodor is an important author.

Rating:
Summary: We Control The Horizonal, We Control The Veridical
Review: Putting its concision (~100 pages) and extremely attractive presentation to one side, *The Mind Doesn't Work That Way* is actually Jerry Fodor's most ambitious effort to date and (grant protectors aside) the reigning champ of cogsci critiques. But the Giant of New Brunswick knows when to say when, and what is frequently presented as an "autocriticism" is really *nothing of the sort*: Fodor's task here is to align his version of the Computational Theory of Mind (capitals required) with Chomsky's somewhat "formalist" version, rather than the cooked-up "massive modularity" of Darwinian dreams -- Fodor is among the Coke-drinkers rather than head-splitters, and in this book (*not* soon to be a major motion picture) he begins to ask some well, Humean questions about *our* grasp of inferential processes in a way which derails nearly every major psychological research program of the present. A must-read from the veritable artiste of philosophers of mind, but how is that?

Rating:
Summary: Good-bye evolutionary psych?
Review: This book is an attack by an "insider" on the contemporary disciplines of cognitive science and evolutionary psychology.

Fodor does not deny that there are some valid aspects to these two disciplines. Rather, he rejects their extravagant claims to have successfully explained "The Way the Mind Works," to quote the title of a recent book by Steve Pinker, who is one of the leading evangelists for cognitive psychology and evolutionary psych.

Fodor's central complaint against evolutionary psychology is quite simple. Anyone who claims to offer an evolutionary explanation for the wings of birds can start with a great deal of solid knowledge about how bird's wings are in fact constructed, about how wings make flight possible given the laws of aerodynamics, etc.

But no one in fact yet possesses the equivalent information for the brain and the mind.

We do not yet know how the neurons are connected and in what manner they function so as to produce thought. More basically, we do not understand what "mind" really is from the viewpoint of the underlying physics of the brain (see, e.g., David Chalmers' "The Conscious Mind" or Colin McGinn's "The Mysterious Flame").

Fodor also has more specific objections. He is highly concerned with the issue of "abduction," the ability to make global judgments of simplicity, relevance, etc. over a broad intellectual domain. Fodor believes that humans are very good at this, but that the current "modular" approach pursued by cognitive scientists and evolutionary psychologists cannot explain how humans could be good at this.

I'm not sure human judgment is as powerful as Fodor believes, but he is correct that modular systems have difficulty making broad global judgments.

In his final chapter, Fodor directly addresses the issue of evolution, arguing that, for a feature to be the product of natural selection, it must be built up by a small number of steps. Using the example of the giraffe's neck, he argues, "Make the giraffe's neck just a little longer and you correspondingly increase, by just a little, the animal's ability to reach the fruit at the top of the tree; so it's plausible, to that extent, that selction stretched giraffe's necks bit by bit."

This example is somewhat misleading: there is no reason in principle why a single mutation could not have created huge giraffe necks in one fell swoop and natural selection then stepped in to preserve the mutation.

But Fodor is correct that such a "saltationist" explanation is not available to evolutionary psychology. The plethora of specialized mental modules favored by evolutionary psychologists (a language module, a "cheater detection module," a face-recognition module, a theory-of-mind module, to name only a few) are supposed to be carefully honed adaptations exquisitely polished by natural selection to serve human needs in the "ancestral environment" (the Paleolithic). Just as a complex organ such as the eye could not realistically be created in one single fortuitous mutation, so neither could these complex mental "organs" hypothesized by evolutionary psych.

But why does Fodor reject a gradual, multifaceted evolution of these hypothetical mental "organs"? He does not say, but there is a fairly powerful argument from the human genome project. We only have about 30,000 genes; most of these are shared with lower mammals and many with non-vertebrates and even non-animals. There just are not that many genes left which distinguish us from mice.

A change in a relatively small number of regulatory genes can bring dramatic changes in development -- our much larger brain, for example. But to actually create a number of new specialized "organs," not possessed by mice or cows places much greater demands on the genome. It's doubtful we have enough genes to handle it.

The evolutionary psych response, as made in Pinker's "The Blank Slate," to this argument is in essence that since these mental modules _do_ exist, our genes _must_ be able to produce the modules. That of course assumes what is to be proven, i.e., that the human mind is based on evolutionarily-derived specialized mental modules.

Fodor completely demolishes the claim that the unity of science demands that evolutionary psychology be true. The degree to which the science of evolution is relevant to the science of psychology is, he rightly argues, an empirical matter, just as (to use his example) it is an empirical matter whether "the theory of lunar geography constrains the theory of cellular mitosis." Not every science has to be relevant to every other science.

Fodor also shreds what he calls "neo-Darwinist anti-intellectualism," the view (he is quoting from Patricia Churchland) that "looked at from an evolutionary point of view, the principal function of nervous systems is to get the body parts where they should be in order that the organism may survive...Truth, whatever that is, definitely takes the hindmost."

Fodor counters that for humans "a cognitive system that is specialized for the fixation of true beliefs interacts with a conative system that is specialized to figure out how to get what one wants from the world that the beliefs are true of..." or, in simple English, humans engage in "rational actions predicated on true beliefs."

We are designed to pursue both truth and our own well-being -- there is no contradiction here. Not action instead of truth, but action based on truth.

Despite the brief length and Fodor's engaging style, this book is not easy reading. But it does raise questions which, if not adequately answered by Fodor's opponents, cast grave doubts on the grandiose claims of contemporary apostles of cognitive science and evolutionary psychology.

Rating:
Summary: Good-bye evolutionary psych?
Review: This book is an attack by an "insider" on the contemporary disciplines of cognitive science and evolutionary psychology.

Fodor does not deny that there are some valid aspects to these two disciplines. Rather, he rejects their extravagant claims to have successfully explained "The Way the Mind Works," to quote the title of a recent book by Steve Pinker, who is one of the leading evangelists for cognitive psychology and evolutionary psych.

Fodor's central complaint against evolutionary psychology is quite simple. Anyone who claims to offer an evolutionary explanation for the wings of birds can start with a great deal of solid knowledge about how bird's wings are in fact constructed, about how wings make flight possible given the laws of aerodynamics, etc.

But no one in fact yet possesses the equivalent information for the brain and the mind.

We do not yet know how the neurons are connected and in what manner they function so as to produce thought. More basically, we do not understand what "mind" really is from the viewpoint of the underlying physics of the brain (see, e.g., David Chalmers' "The Conscious Mind" or Colin McGinn's "The Mysterious Flame").

Fodor also has more specific objections. He is highly concerned with the issue of "abduction," the ability to make global judgments of simplicity, relevance, etc. over a broad intellectual domain. Fodor believes that humans are very good at this, but that the current "modular" approach pursued by cognitive scientists and evolutionary psychologists cannot explain how humans could be good at this.

I'm not sure human judgment is as powerful as Fodor believes, but he is correct that modular systems have difficulty making broad global judgments.

In his final chapter, Fodor directly addresses the issue of evolution, arguing that, for a feature to be the product of natural selection, it must be built up by a small number of steps. Using the example of the giraffe's neck, he argues, "Make the giraffe's neck just a little longer and you correspondingly increase, by just a little, the animal's ability to reach the fruit at the top of the tree; so it's plausible, to that extent, that selction stretched giraffe's necks bit by bit."

This example is somewhat misleading: there is no reason in principle why a single mutation could not have created huge giraffe necks in one fell swoop and natural selection then stepped in to preserve the mutation.

But Fodor is correct that such a "saltationist" explanation is not available to evolutionary psychology. The plethora of specialized mental modules favored by evolutionary psychologists (a language module, a "cheater detection module," a face-recognition module, a theory-of-mind module, to name only a few) are supposed to be carefully honed adaptations exquisitely polished by natural selection to serve human needs in the "ancestral environment" (the Paleolithic). Just as a complex organ such as the eye could not realistically be created in one single fortuitous mutation, so neither could these complex mental "organs" hypothesized by evolutionary psych.

But why does Fodor reject a gradual, multifaceted evolution of these hypothetical mental "organs"? He does not say, but there is a fairly powerful argument from the human genome project. We only have about 30,000 genes; most of these are shared with lower mammals and many with non-vertebrates and even non-animals. There just are not that many genes left which distinguish us from mice.

A change in a relatively small number of regulatory genes can bring dramatic changes in development -- our much larger brain, for example. But to actually create a number of new specialized "organs," not possessed by mice or cows places much greater demands on the genome. It's doubtful we have enough genes to handle it.

The evolutionary psych response, as made in Pinker's "The Blank Slate," to this argument is in essence that since these mental modules _do_ exist, our genes _must_ be able to produce the modules. That of course assumes what is to be proven, i.e., that the human mind is based on evolutionarily-derived specialized mental modules.

Fodor completely demolishes the claim that the unity of science demands that evolutionary psychology be true. The degree to which the science of evolution is relevant to the science of psychology is, he rightly argues, an empirical matter, just as (to use his example) it is an empirical matter whether "the theory of lunar geography constrains the theory of cellular mitosis." Not every science has to be relevant to every other science.

Fodor also shreds what he calls "neo-Darwinist anti-intellectualism," the view (he is quoting from Patricia Churchland) that "looked at from an evolutionary point of view, the principal function of nervous systems is to get the body parts where they should be in order that the organism may survive...Truth, whatever that is, definitely takes the hindmost."

Fodor counters that for humans "a cognitive system that is specialized for the fixation of true beliefs interacts with a conative system that is specialized to figure out how to get what one wants from the world that the beliefs are true of..." or, in simple English, humans engage in "rational actions predicated on true beliefs."

We are designed to pursue both truth and our own well-being -- there is no contradiction here. Not action instead of truth, but action based on truth.

Despite the brief length and Fodor's engaging style, this book is not easy reading. But it does raise questions which, if not adequately answered by Fodor's opponents, cast grave doubts on the grandiose claims of contemporary apostles of cognitive science and evolutionary psychology.

Rating:
Summary: Fodor has the guts to take on... Fodor
Review: This critique of the computational theory of mind and the pan-adaptionist tradition is clearly so honest that it goes after the ideas promoted by Fodor's own 1983 watershed book "The Modularity of Mind". In brief the essay is an attack on massive modularity by saying that there are things after all that escape the programming (encapsulation and opacity are key: how can we talk about something OPAQUE? We know nothing about a few critical things...).
Granted the book is horribly written (that is Fodor's charm after all) but his argumentation is so ferocious that he ends up loud & clear.
The man is critical of his own ideas, and of the current in thought that he he helped create --one may use Fodor-1 against Fodor-2. Perhaps persons I hold in highest respect are those who go after their own ideas!
Bravo Fodor. Even if I do not agree I can't help admiring the man.