When I lived in New Orleans

… I was witness to a particularly obnoxious bartender hitting on a woman of our mutual acquaintance and getting turned down with the standard “let’s just be friends,” whereupon he spat out “I’ve got enough friends,” and stalked off. I have always thought it was just face-saving behavior, and/or some ninnified attempt to deploy the oft-discussed “game,” and/or pure being a dickhead. But perhaps he meant it.

In other, totally unrelated, news, this explanation of why Jerry Fodor’s argument against natural selection is like Chomsky’s argument against Skinner, rather than making me skeptical about natural selection, has made me skeptical about the supposed refutation of behaviorism. One man’s modus ponens, yadda yadda yadda &c. It’s too long to reproduce here, so if you care, go read it now. I’ll wait.

So, take the explanation of Chomsky’s argument at the beginning. Really? (Ignoring the clearly nonsense extensional interpretation.) It seems to come down to the claim that the fact that stimulus X predicts response Y, but we can’t specify with complete precision what precisely it is that constitutes stimulus X, means that the theory is either vacuous or false? Kids, can we say “proves too much?” Like, oh, I don’t know, proves that all empirical science is either vacuous or false? Because all empirical science makes predictions of the form “if we do X to the treatment subject, Y will come out,” and there is always the problem that X will be specified by multiple properties, s.t. that by applying some X` that has some subset of the properties of X to the treatment subject you will either get Y, confirming the theory because it’s described in terms of that subset, or get ~Y, confirming the theory described in terms of the absent properties of X. The solution isn’t to throw out the theory, but to be more careful about which properties of X you describe the treatment in terms of.*

The behaviorist example given in that post is not vacuous, because, at a minimum, it predicts that identical red triangles will be followed by identical behavior: the theory will be falsified if an identical red triangle is flashed, by throwing the same switch you threw the last fifty times, and the behavior is not repeated. Generalizing to other red things, or other triangles, is a difficult problem, but not one that implicates the soundness of the theory.

Likewise, the adaptation example is not vacuous, because, at a minimum, it predicts that the repetition of all relevant facts will produce the same results — that is, if we moved the bears and the snow to an otherwise identical Twin-Earth, introduced the same mutations and predator/prey profile &c, we would again get white bears going to fixation.* If we didn’t, the theory would be falsified. Extending that theory to cases where the snow is a different color would require additional theoretical claims, but that doesn’t make the basic theory vacuous.

Also see Alex Morgan’s comment.

* That is, it’s always possible to take an implausible extensional interpretation of X, decompose it into the various properties of X, and watch the theory blow up as Y doesn’t happen. And it’s likewise always possible to go the other way and talk about the impossibility of completely defining X apart from, in the kind of extreme holism that Fodor seems to be pushing, defining the whole world along with it.
** One way of reading the objection, that might get raised at this point, is “but we don’t have a well-understood notion of what all relevant facts are — why can’t we say that location on Earth vs. Twin-Earth is relevant, s.t. the failure of white bears to come to fixation there would be no falsification of the theory?” To which I simply say: “Give me a break. Let’s be a little bit pragmatist here for a second. The scientific enterprise does not require some kind of deductive rule to pick out relevant experimental conditions from irrelevant ones in obvious cases like this.”


8 Responses to “When I lived in New Orleans”

  1. Matt Says:

    I think you’re more or less drawing the right conclusions re Skinner and Chomsky. Much of Skinnerian behaviorism is certainly false, but the arguments Chomsky used to “show” this were not so good and his own theory is really quite bad in explaining language learning, predicting all sorts of false things and having many parts that look, to the skeptical, more than a little like epicycles. Interestingly enough, his wife was an expert on child language learning, and nearly all of her work was in serious conflict with Chomsky’s on language acquisition. Since her work was empirical and his largely a priori I tend to give hers more credence.

  2. Michael Johnson Says:


    Thanks for joining the discussion. I’m going to have to respectuflly disagree with you: I don’t think either Fodor or Chomsky’s argument impugns empirical science wholesale.

    I didn’t have time to get into this in my post (it was long as it is), but I was going to explain what Chomsky proposes in place of Skinner: namely, cognitive psychology. Suppose that we hypothesize that the rat has a mental representation “pressing bar given red stimulus gets food” and that this representation means that pressing the bar given a red stimulus results in food. Then we predict that the rat will press the bar given a red circle (given the true theory of propositional attitude psychology). Our prediction may turn out to be false, and thus it is not trivially compatible with any data. This is in contrast to Behaviorism generally.

    You bring up Twin Earth. Now Fodor is a materialist, so he thinks that all properties are physical or supervene on the physical. So of course he thinks that in two physically identical circumstances, (provided the laws are deterministic) we can expect the same outcomes. But this follows from the laws of physics, not from psychology or biology. Furthermore, Fodor thinks that there are true scientific theories that are not reducible to physics. For instance, if we fix all the economic facts F at some time t, then the true economic theory will provide us with an operator E which determines the future economic facts (ceteris paribus) at some future time t’ by evolving the state from t: E(F, t, t’). This is true regardless of the physical facts, provided the economic facts are F.

    Fodor’s objection to natural selection is that it does not provide us with a way of taking the distribution of phenotypes P at t and evolving the state to get a new distribution of phenotypes at some later time t’. That is, it makes no predictions. This is consistent with physicalism– for instance, it is consistent with the claim that the future state of the wave function is determinable by the current state evolved in accord with Schrodinger’s equation.


  3. Paul Gowder Says:

    Thanks, Michael. Your elaboration makes me suspect that thereal problem wih Fodor’s argument is in that kind of non-reductive materialism. It’s hard to make sense of the predictions of any kind of theory other than physics on twin earth if you take that kind of position. On twin earth, all of the physical facts are the same, and it has to follow (from materialism) that all of the biological facts are the same. Do we, then, say that the white bears again come to fixation in virtue of our theories about the physical facts or about our theories of the biological facts? Does it even make sense to ask the question? The only way I can imagine to answer it would be to start talking about counterfactuals in which the physical facts are the same but the biological facts are different, and how is that possible?

    We can think a lot more clearly about this if we accept that the facts of biology are reducible to the facts of physics. Then natural selection is just a statement about how certain physical processes happen, and it is natural selection that gives us the claim that they will happen the same way, even apart from the higher-level deterministic claim that everything has to happen the same way given the identical physical facts. An analogy may make this clearer. I can drop a book from my hand, and it will fall. I can predict precisely the same thing on twin earth, but not as a result of general claims about the spatiotemporal arrangement of particles yielding a unique, deterministic outcome, but because of the theory of gravity. If I drop the book on twin earth and it doesn’t fall, not only determinism but gravity too is falsified. Same with natural selection.

    On the Chomsky-Skinner thing, I know much less about psychology than I know about evolution, so forgive any stupidity, but it seems like any objection to the behaviorist theory must also apply to the cognitive psych theory, because they depend on precisely the same observables — the cognitive psych experimenter too is predicting red triangles or redness or triangleness or shapeness or whatever causes the bar-pressing behavior, and too will have to deal with questions about whether his claims should be interpreted extensionally or not –he just adds in some extra unobservables about mental representation. Wasn’t this Skinner’s point?

  4. Steve M. Says:

    Maybe my philosophy muscles have atrophied from excessive lawyering, but Fodor’s objection (as construed by Johnson) seems very much like a workaday objection to causation. One of the chief problems for any theory of causation is that there are indefinitely many facts or events (or what have you–the substance of the ontology isn’t important here) that are necessary conditions for the occurrence of *any*fact or event. The primary work of any theory of causation is going to be sifting the wheat from the chaff.

    (I’m inclined to think that the objections hold water. The upshot, of course, is not that there is no such thing as causation (clearly there is), or that speciation is generally explained by something other than natural selection (it’s not–genetic-drift outliers excepted), or than garden-variety causal stories aren’t properly thought of as true or false, but rather that human minds are not well-equipped to conceptualize causation in the abstract. Billiard balls, yes. Causation as such, no. Of course, given that the billiard-ball model of causation seems to be a hard-wired feature of human cognition, this shouldn’t be surprising. Eyes turning on themselves, &c.)

  5. Paul Gowder Says:

    I think that’s quite right, Steve, which is why it does impugn all of empirical science… I suppose I’m enough of a pragmatist that that’s enough to count as an objection to the argument.

  6. Daniel Goldberg Says:

    I think Steve has it exactly right. But epistemically, I don’t think it helps that most Western human enterprises insist on working with models of causation which empiricism itself indicates are poor representations of the natural world (acknowleding that the natural world is not separate from human enterprises, of course).

    This — wait for it, Paul — reductionism has serious pragmatic consequences in context of illness and health, IMO. It’s not that we need to throw away our theories of causation — on this, Kant was right in that it is a category of our understanding. But there are better and worse ways of trying to think about causation, and it frustrates me no end that we tend to opt for the models we have every reason to believe won’t help all that much.

    Trying to open doors painted onto walls (W) . . .

  7. Michael Johnson Says:


    You ask how a nonreductive materialist can make predictions and explain things in any language other than that of fundamental physics. But the worry seems misguided to me.

    Yes, on twin earth, the physical facts are the same; economics, for instance, supervenes on physics; so one can give an explanation in fundamental physics terms of why the economy changed as it did (provided one has the supervenience relations to hand). So, you ask, what sense is there in saying that the laws of economics explain economic development.

    The sense is this: change the physical facts, but keep the economic facts the same. Replace dollar bills with sea-shells on twin earth, but don’t change supply and demand. Now we don’t even need to know the physical facts to predict the future economic facts. They’ll be the same on twin earth as here. That’s why economics is an autonomous science.

    Jerry’s view about biology [as I understand it] is that you can’t hold fixed the biological facts, while varying their realizers at will, and still expect to get counterfactual-supporting, ceteris paribus laws.

    As for your question about cognitive psychology, the basic idea is that cognitive psychology is all about making a principled distinction between those properties of the response that are outputs and those properties that are artifacts. For instance, the amount of time from stimulus to response is considered an artifact of the processing, not an output. That is, if response time is n seconds, the theory says that computation time was roughly n seconds and thus we can infer facts about the data structures and algorithms the animal is using. Behaviorist theory doesn’t allow such a distinction, or such representational posits, as a matter of principle.

  8. Michael Johnson Says:

    @Steve M.

    I’m not sure I understand. Consider a world with two point particles. They move toward one another at a rate proportional to the product of their masses and inversely proportional to the square of the distance between them. What are the infinite number of facts that are necessary for this situation to obtain?

Leave a Comment