| Home | About Edge| Features | Edge Editions | Press | Reality Club | Third Culture | Digerati | Edge:Feed | Edge Search |

THE REALITY CLUB
Organs of Computation
A Talk With Steven Pinker


Steven Mithen, Steven Quartz, Nicholas Humphrey, Patricia S. Churchland, Sandra Blakeslee, David Lykken, Steven Pinker, Steven Pinker (2), Nicholas Humphrey (2), Rick Potts, Steven Pinker (3), Arnold Trehub, Steven Quartz (2), Steven Pinker (4), Arnold Trehub (2), Steven Quartz (3), Arnold Trehub (3) on Organs of Computation by Steven Pinker.


From: Steven Mithen
Date: 1-13-97

A First Impression of Organs of Computation

Asking how the quirks of the mind might have made sense as solutions to the problems that our ancestors faced in negotiating the world is a powerful strategy for working out how the mind works. But there are quirks and mega-quirks. The latter include those things that Pinker is not over impressed with, such as the accomplishments of Mozart, Shakespeare and Einstein. Well those may be rare, extra-ordinary feats but other mega-quirks are more widespread. Apparently more than 50% of the population of the USA believe in divine creation as opposed to human evolution by natural selection. The last two decades has seen the growth of fundamentalism throughout the world, even though it is precisely within this period that human evolution has become so well documented. How can the belief in super-natural beings, life after death, miraculous transformations of matter beliefs that are so resilient when faced with evidence to the contrary be explained as solutions to the problems that our ancestors faced? The archaeological evidence is quite clear that such beliefs did not arise until very recently in human evolution, a 100,000 years ago at most, and yet they dominate the minds of the majority of people alive today. Racist thought is another mega-quirk, manifest recently in the attempted genocides in Bosnia and Central Africa. Such genocide is justified on the basis that some people are less than human, a racist idea that has plagued the twentieth century and which cannot be acquired simply by processing information about those peoples behaviour. How can such racist thought be explained as solutions to the problems that our ancestors faced in their evolutionary environments? To suggest that such thoughts are the 20th century manifestation of ways of thought and behaviour that were once rationale in an ancestral environment, as one might when considering the ex-lover problem that Pinker describes, is simply incompatible with the nature of racist thought that has pervaded so much of human history (but only the very latter stages of prehistory). And tell those oppressed people that the minds of their oppressors are organs of extreme perfection (to either ancestral or modern conditions) and I suspect that they will hasten to disagree.

To make any impact on the academic or popular understanding of the mind, evolutionary psychologists need to tackle these mega-quirks rather than staying on the safe ground of explaining how the human ability to recognise faces, manipulate objects and engage in complex social relations are products of an evolutionary history. That these arise as adaptations to ancestral conditions is not seriously challenged, especially as so many of these mental attributes are shared by our living relatives. Pinkers description of lifting and squeezing the milk carton, for instance, has an uncanny resemblance to Dick Byrnes descriptions of the careful manipulation of thorny plant material by gorillas. If we are to explain the human mind as a product of biological evolution, then we must deal explicitly with those ways of thought that are unique to humans the peculiar feats of bizarre, creative and destructive thought for which all humans, not just a few individuals, have the potential. We can find such explanations, but they require us going beyond the idea that our minds are simply computational, information processing devices for solving the problems that were faced in ancestral Miocene/Pliocene/Pleistocene environments. There is a far more interesting and important evolutionary story to be told.

Steven Mithen


From: Steven Quartz
Date: 1-13-97

Does Steven Pinker really think the Internet isn't catching on because the "mind evolved to deal with physical objects?" If the mind is so wedded to physical objects how did the phone line that makes faxes possible ever replace the written letter as a primary medium of communication? It would be hard to think of a technology that has been more eagerly (and in many cases uncritically) adopted than the Internet-just ask the U.S. Postal Service, who is losing billions every year to e-mail, or the original stockholders of NetScape.

To me this sort of "the mind was designed to (fill in your favorite behavior)..." argument typifies the vague half-truths of evolutionary psychology. It is an old point, but still timely: Looking just to the constraints imposed by natural selection wildly underdetermines the diversity of human behaviors seen across the spectrum of human cultures (to say nothing of the possible range of human abilities). The mind didn't evolve to write books about its own evolution, ponder the paradoxes of set theory, mow the lawn every week, or waste time surfing the net, but there it is. The Internet, remote access technologies, and new forms of computer-mediated knowledge representation and manipulation will be deeply transforming because the mind isn't limited to life as our Pleistocene ancestors knew it. The mind is constrained by natural selection, not shackled by it.

By looking simply at the constraints natural selection imposes on the mind's capacities, evolutionary psychology invites us to bite into a false (and vampire-like) dichotomy: Is it learned, or is it innate? When we look inside the brain what we find is something far more interesting, a self-organizing system in which structure is generated from the interaction between a structured world (which we ourselves structure with culture) and a brain with its own intrinsic constraints. The really interesting research program isn't to suppose everything has to be built into the brain via organs of computation (incidentally, contemporary brain science provides no evidence for innate organs of computations). The interesting problem is to decipher the principles of this interaction, the nature of the biological constraints, and how that interaction can produce the extraordinarily diverse behaviors and cognitive capacities we see all around us. Terry Sejnowski and I have suggested (http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.quartz.html is an online version of a forthcoming article in Behavioral and Brain Sciences) that the principles of this interaction are rich enough to minimize the amount of domain-specific knowledge that must be built-in--that the computational machinery is built by this interaction.

Understanding the interaction between brain and world that builds the mind will take the perspective of all the human sciences. With a maturing brain science, a new generation of computer modeling, and cognitive neuroscience's naturalistic approach, it would seem the final pieces are falling into place for such a collaborative project. But first we need to move beyond the old dichotomies.

One more impression of the Pinker interview. Another one of evolutionary psychology's gambits is the zero-sum game metaphor, one Pinker uses in discussing happiness: If I'm happy it's because someone else isn't. The fact is, most people are happy-80% of Americans describe themselves as happy. Objective life circumstances play little role in determining who is happy and who isn't-the vast majority are happy despite being poor, downtrodden, or even severely handicapped physically (a good review of the literature is Myers and Diener, "Who is Happy?" Psychological Science, Jan 1995). Not everything need be a scarce resource in a Hobbes-like struggle of all against all.


From: Nicholas Humphrey
Date: 1-13-97

Steven Pinker is right about so much and argues the case so well, that if he ever has a bad idea it stands out like a sore thumb. I think there is one really bad idea in his discussion with JB. This is the suggestion that when a man threatens to kill a potentially unfaithful mate, he had better be prepared to carry out the threat even if it would hurt him to do it, or else the threat won't be credible.

It's true of course that he wants his threat to be believed, and it probably won't be believed unless other men have in fact carried out such threats in the past - so that his mate has reason to believe that such semi-suicidal acts of revenge are in general "typical" of male behavior. But this does not mean that he himself would stand to gain anything if , when his threat failed to deter her, he did in fact behave as a typical male. Clearly, in such circumstances he would in fact do better to "cheat" on the sexual stereotype - and walk away.

It's the old issue of a group policy that potentially requires self-sacrifice being liable to invasion by an individual free-rider. (There was a similar paradox with the cold war policy of Mutually Assured Destruction - but that's another story).


From: Patricia S. Churchland
Date: 1-14-97

I have a small challenge for Steven Pinker. Could he actually name a real living person (i.e. nonstrawman) who thinks that we can understand cognition by looking only at the brain itself?

Cheers,

Pat


From: Sandra Blakeslee
Date: 1-17-97

Message: I would like Steve to tell us more about what he means by a "mental organ." It seems to be a metaphor more than a biological reality. Is there any evidence that mental "organs" exist? How are they organized? What is known about their wiring, physiology, dynamics, etc. Or this is a made-up term?



From: David Lykken

Date 1-21-97

There is much to agree with among Steven Pinker's interesting comments. Evolutionary psychology does provide a provocative and heuristic perspective, helping us to "reverse engineer" and understand the species- specific characteristics of the human brain. John Locke and J.B. Watson both were wrong in supposing that the mind begins as a tabula rasa and the many modern day cognitive psychologists who still treat the brain as only a general-purpose computer also are wrong. In addition to its general-purpose computer function, the human brain contain numerous ROMs or special-purpose modules for learning language, recognizing faces, and the like.

BUT, like Tooby & Cosmides, Gould, and other evolutionary biologists, Pinker makes the odd mistake of arbitrarily assuming that all of the genetic diversity that permitted natural selection to evolve the human brain has now been exhausted and that individual psychological differences that we observe today are all environmentally produced (as in his example of one- and no-legged men). In other words, all human babies today, unlike Paleolithic times, have brains like so many new Mac computers, waiting to be programmed. This is wildly improbable on evolutionary grounds (e.g., if there is abundant genetic diversity in the psychology of domestic animals, as Darwin himself observed, as well as in the anatomy and physiology of humans, as any child can observe, why is the human brain the sole exception?

And it is a notion that can only be entertained by resolutely ignoring several decades of behavior-genetic research. At my University, we have shown that 69 pairs of monozygotic twins, separated in infancy and reared apart, correlate about .75 in IQ and .53 in a measure of happiness or subjective well-being. (This includes two sets of MZA triplets, each treated as three pairs of MZA twins.) The corresponding correlations on large samples of MZ twins reared together and also tested as adults are virtually identical. What this shows (except to eyes that will not see) is that from about 25% to 75% of the population variance in psychological traits is associated with genetic variance. These variance or heritability estimates apply of course only to the broad middle-class of the western countries from which these samples were drawn and would be smaller in countries or sub-cultures where the environmental variance is greater. But the fact remains that human brains not only have an innate structure that was "engineered" by evolution, but they also vary innately one from another in as many ways as we have been able to measure.

The only explanation I can see for the refusal of evolutionary biologists to acknowledge these facts is political correctness and/or fear of controversy. Those are both bad reasons.

David Lykken, University of Minnesota



From: Steven Pinker

Date: 1-23-97

Some Comments on the First Impressions of the Interview "Organs of Computation"

I thank Steven Mithen, Steven Quartz, Nicholas Humphrey, And Patricia Churchland for their thought-provoking comments. The oral interview format forced me to simplify and abbreviate arguments, and I apologize for those cases in which my points didn't come through clearly.

==========================================

1. Steven Mithen asks how an evolutionary-psychology or reverse-engineering approach would shed light on "mega-quirks" like religion and racism, which have such profound effects on human affairs but do not seem particularly well engineered. Excellent questions, which I deal with at length in How the Mind Works.

First, a few points on the general theory. I *don't* believe that human behavior is adaptive or well engineered in an everyday sense of the words, for three reasons:
(i) The criterion for "good engineering" in the evolutionary sense is replication of genes. That often diverges from other criteria that we think of as "adaptive" in nonbiological contexts: happiness, moral values, harmony, efficiency of the group.
(ii) The brain is engineered for successful outcomes in the environment in which we spent most of our evolutionary history, and may lead to wildly non-adaptive behavior in the recent, topsy-turvy world of civilizations.
(iii) Like all biological structures, some features of the brain are adaptations, others are the inevitable nonadpative by-products and spandrels of the adaptations. The challenge for psychology is to figure out which is which.
(iv) Behavior didn't evolve; the mind (i.e., brain) did. Overt behavior is an outcome of intricate interactions among mental faculties and among other behaving human beings. The mind's structure is bound to be more orderly than actual behavior.

About religion itself: In HTMW, I rely on the anthropological literature to try to figure out why in all cultures, including ours, the so-called rational animal believes in things that are patently false, such as ghosts and spirits. It's a complicated question, one that I do not claim to answer, but here are some helpful ideas:

(a) The question "Is X adaptive?" must always be completed with "adaptive for whom?" One has to distinguish among parties with competing interests. It may not be adaptive for a typical person to believe in hocus-pocus, but it may be adaptive for a typical person to defer occasionally to experts, and it may be adaptive for "experts" to inflate their status by alluding to a world of great power and wonder reachable only through their services. Tribal shamans are flim-flam artists who supplement their practical knowledge (herbal medicine and such) with cheap stage magic, and enjoy great prestige and all its perquisites. Readers may make their own comparisons to Western religions.

(b) There are hints that religious beliefs are spandrels of cognitive faculties that in ordinary circumstances accomplish more mundane feats, such as figuring out how objects, plants, animals, and other people work. First, religious beliefs are recognized by their believers to be out-of-the-ordinary, accompanied by a sense of awe and wonder. Second, they are minor modifications of ordinary cognitive categories (spirits, for example, have the standard inventory of human beliefs and desires that people impute to other people; the spirits are just stipulated to lack bodies or some physical property of bodies). Third, religious beliefs are typically invoked when a person is in desperate straits and the usual remedies for success in the physical world have failed. "Religion is everywhere a recipe for success," said Ruth Benedict; people pray (or in our society, call the Psychic Hotline) for success in love, competition, finance, and overcoming illness.

(c) The evidence that makes religious beliefs so obviously wrong to us scientists is recent, and, lamentably arcane even in our society. And the phenomena that most directly inspire spiritual beliefs worldwide -- dreams, death, altered states -- involve consciousness, which, pace Dan Dennett, even some contemporary scientists feel has not yet been explained.

2. Racism (more generally, ethnocentrism, for most hatred worldwide is within-race). Mithen writes, "Tell ... oppressed people that the minds of their oppressors are organs of extreme perfection (to either ancestral or modern conditions) and I suspect that they will hasten to disagree." See my general point (i) above. Mithen substitutes our ordinary sense of perfection (moral rightness, personal happiness, societal harmony and efficiency) for the one that is relevant to a scientific understanding of the brain, namely gene replication. Ethnocentrism may or may not be biologically well engineered, but the misery and tragedy it causes is irrelevant to that particular question (compare: "Tell the family of a murder victim that a gun is a well-engineered machine." or: "Tell the zebra that a lion is an organism of extreme perfection." It all depends on which criteria for "well-engineered" and "perfection" you have in mind.)

Now, could collective aggression against conspecifics be adaptive in the narrow, biological sense? It has evolved several times, so it's at least possible. In human warfare before civilization, villages raid each other to abduct women or deter abductions, and Napoleon Chagnon and Laura Betzig have gathered extensive data from many sources showing that until recent conditions obtained aggressive men were reproductively more successful. (As Betzig points out, the entry in the Guinness Book of World Records for the human with the most children -- close to a thousand -- belongs to a man named Moulay Ismail the Bloodthirsty.) Such evidence doesn't settle the question (and it certainly doesn't render ethnic warfare "natural" in the sense of "good" or "inevitable"), but it shows that morally repugnant mega-quirks are not necessarily design flaws in the narrow biological sense.

========================================

2. Steven Quartz asks whether I really think the Internet isn't catching on because the "mind evolved to deal with physical objects," pointing to the telephone. The telephone cunningly simulates a different kind of human interaction with the world, oral conversation, so it doesn't need to tap object cognition. And I would, indeed, cite the recent success of the Web as an example of the overall point. Web-browsers don't literally deliver much more information than the old text-based Internet interfaces (telnet, ftp, gopher, etc.), but they are responsible for the explosion of Internet use outside the university. Why? Because they tap in to visual-spatial cognition, and better simulate physical interactions with the world like pressing buttons, turning pages, and traveling along paths through space. One of the biggest current efforts in the development of web interfaces is how to use graphic representations to give the user a better sense of currently confusing information such as document contents and web topology.

Quartz also points out, "Looking just to the constraints imposed by natural selection wildly underdetermines the diversity of human behaviors seen across the spectrum of human cultures (to say nothing of the possible range of human abilities). The mind didn't evolve to write books about its own evolution, ponder the paradoxes of set theory, mow the lawn every week, or waste time surfing the net, but there it is." I agree with the latter -- see (i) to (iv) above, which explain why evolutionary thinking is not about constraints on behavior at all.

However, I do disagree with the following: "By looking simply at the constraints natural selection imposes on the mind's capacities, evolutionary psychology invites us to bite into a false .. dichotomy: Is it learned, or is it innate?" That dichotomy is indeed all too common (see, for example, the recent paper on language acquisition by Safran, et al., in Science, which frames its hypotheses in almost exactly those words). But it is evolutionary psychology that argues most strenuously that the dichotomy is incoherent (see, e.g., Tooby & Cosmides' The Psychological Foundations of Culture; Marler & Gould's paper Instincts to Learn; Gallistel's contribution to the Gazzaniga volume; Chapter 13 of my Language Instinct, and soon, HTMW).

More generally, I don't think that "constraints" are a particularly useful way of characterizing the mind, either in the human-potential sense of "Don't try to change blah-blah-blah, it's constrained by the genes and hence inevitable," or even in Quartz's framing of development as an interaction between a structured world and a constrained brain. I don't disagree outright that there is such an interaction, but think we can do better. The brain is not a list of "constraints" but a highly structured and adapted organ, which is designed to process information in clever ways both as it assembles itself in utero and when it guides the whole organism. As for whether "contemporary brain science provides no evidence for innate organs of computations" -- well, so Quartz and Sejnowski argue, but other brain and cognitive scientists interpret the same evidence differently. That will have to be a discussion for another day.

On happiness: I'm confused by the criticism that "objective life circumstances play little role in determining who is happy and who isn't." That was precisely my point! Happiness, according to the research Quartz and I both cite, is determined in large part by how much better or worse off you are now compared with how you recently were, and how much better or worse off you are relative to members of a group you compare yourself to. (Note, by the way, that that's very different from Quartz's summary, "I'm happy because someone else isn't".) A huge literature in social psychology documents these statements, and they are better explained by thinking of happiness as part of a striving-calibrator than as our everyday concept of just desserts.

=========================================

3. Nicholas Humphrey disagrees with the theory (which I borrowed from several economists and game theorists) that irrational passion may have evolved to make our threats, promises, and bargaining positions more credible. He notes that a vengeful estranged husband, for example, would not "stand to gain anything if, when his threat failed to deter her, he did in fact behave as a typical male. Clearly, in such circumstances he would in fact do better to `cheat' on the sexual stereotype - and walk away." First, the theory is not about people's reactions to stereotypes or group precedents; it is about their reaction to other individuals. Not, "Does a typical male kill his estranged wife?" but "Is this guy crazy enough to come after me?" Indeed, it is Humphrey himself who deserves the credit for alerting us all to the highly developed Machiavellian intuitive psychology in the human mind. And, there lies the problem in Humphreys' cricitism here -- of course, the man would do better to walk away, but that's exactly what the wife would predict if she had reason to believe he was a passionless cost-benefit appraiser -- in turn allowing her to call his bluff. Only if some part of him was *irrational*, in this specific sense of being willing to carry through a threat regardless of its costs, would the threats be effective.

A corollory, by the way, is that "What does X stand to gain by carrying out his threat?" is the wrong question. It enters the story too late -- after the target has defied the threat -- and presupposes that at that moment X is calculating the costs and benefits of alternative acts, which is precisely what is at issue. The right question is, "What do X's genes have to gain, over the long term, by wiring X up so that X is compelled to carry out his or her threats?" And the answer is, deterring adversaries. Deterrence is a proactive notion; it always looks pointless post hoc. In cases of brinkmanship and bluff-calling, it may have terrible costs, but its average benefit is in all the examples of brinkmanship and bluff-calling that it prevents from ever taking place. (See also my general point (i) at the beginning.) Humphreys' analogy of the paradox of Mutually Assured Destruction is apt, as is the moral paradox of criminal deterrence ("Hanging the murderer won't bring the victim back to life"). I think human passion is paradoxical in precisely those ways.

=======================================

4. Patricia Churchland challenges me to name a nonstraw person who thinks that we can understand cognition by looking only at the brain. I would begin with 90% of the membership of the Society of Neuroscience. The ignorance of psychology and cognitive science among neuroscientists was considered a serious enough problem by a recent executive of the society that she convened a strategy session at the annual meeting to brainstorm about what to do about it. And it has real consequences for how research is conducted. Two examples are neuroimaging studies of language, which until recently were carried out as if the topic had never before been studied, and the search for the neural basis of learning, which has been equated with the search for a neural version of the associative bond, a notion considered antique within ethology, animal learning, and most of cognitive science.

As for prominent individual spokespeople: admittedly, my remark was not carefully worded, and no one would let himself be characterized in exactly that way. But there are real people with whom I disagree on whether our understanding of the brain must be couched exclusively in neurophysiological terms (and perhaps that is a better statement of what is at issue). For example:

"I claim that the entire structure on which the cognitivist enterprise is based is incoherent and not borne out by the facts. ... [It is an egregious category mistake to think that] the whole enterprise can proceed by studying behavior, mental performance and competence, and language under the assumptions of functionalism without FIRST understanding the underlying biology." [emphasis added] -- Gerald Edelman: Bright Air, Brilliant Fire.

"There are brute, blind neurophysiological processes and there is consciousness, but there is nothing else. .... [There is no] rule following, no mental information processing, no unconscious inferences, no mental models, no primal sketches, no 2 1/2-D images, no three-dimensional descriptions, no language of thought and no universal grammar." -- John Searle: Consciousness, explanatory inversion, and cognitive science, in Behavioral and Brain Sciences.

The Emperor's New Mind -- title of a book by Roger Penrose attacking the idea that thinking can be understood as a form of computation.

Incidentally, my remark about people who want to look only at the brain was not meant to include Churchland herself (though I am sure we do disagree about many things). I loved the title of her excellent book with Sejnowski, The Computational Brain -- and point out that Searle, Penrose, and, if I read him correctly, Edelman, deny that the brain does computation.
============================================

Thanks again to the commentators for allowing me to clarify my remarks in the interview.



From: Steven Pinker
To: David Lykken

Date: 1-29-97

I agree with most of David Lykken's comments. Human brains do vary innately, and the research of Lykken and his collaborators has enormous implications for every part of psychology. Not only by measuring how much of individual variation is innate, but by providing rigorous tools that, for the first time, allow us to explore the causes of the variance that is not innate.

My caveats about species-wide versus individually varying innate psychology are just meant to separate some issues that are often run together. Variation in gross indexes of cognitive functioning, like IQ, don't speak to the issue of whether the mind has a complex innate organization, because it is compatible with the most radical empiricist models of the mind -- innately high IQ could come from faster neurons, or a bigger short-term memory, or the values of the dozen-odd variables that connectionist modelers fiddle with in setting up their neural networks prior to training. Conversely, you could have the identical-loaded-Macs scenario, with complex but universal factory-installed structure. I don't believe that either scenario is correct, but it's important to visualize them to understand that the two kinds of innateness are logically independent issues. They are often confused. A recent example is the anti-nativist, pro-connectionist book called Rethinking Innateness, in which the authors argue that their position is morally superior because it makes Bell-Curvish positions harder to maintain. Leaving aside the conflation of empirical and ethical issues, the argument is just logically wrong. The innate structure that connectionists typically build into their models, though less than in alternative models, offers plenty of scope for innate differences in IQ and other psychological traits.

As for the actual findings: I do find it interesting that Lykken and his colleagues have documented so many degrees of freedom in innate variation, not just IQ or the big five personality traits -- the identical twins reared apart who both wear seven rings, both like to sneeze in elevators, and so on. The implications for the mind's structure are not totally clear -- we certainly aren't equipped with ring-wearing modules -- but the number of dimensions of innate variation hints at the complexity of structure all parts of the mind must have.

What one makes of innate variation depends largely on the question one is interested in. I am mainly interested in "how" questions -- how does Homo sapiens recognize objects, learn language, fall in love, and so on. I suspect that the answers to those questions are not going to differ qualitatively for different normal people. That's because the neural structures responsible for those feats are surely complex and under the control of many genes. With sexual recombination, genes get scrambled and pasted together every generation. If Mom had one kind of neural device for recognizing objects, and Dad has a very different kind, how would Junior function with a random pastiche of the two? That does not say that there can't be quantitative genetic variation, or qualitative variation in simple traits, just that there is unlikely to be qualitative variation in complex abilities. Complex abilities were built out of genetic variance, of course, but that was by using up genetic variance over the past 50,000 to 500,000,000 years. The variance we see at any one time slice (e.g., today) is tiny in comparison, and is the raw material for the selection of today and tomorrow, not for the selection of yesterday which created today's human mind.

Still, none of this contradicts Lykken's main points or is incompatible with the fascinating results of his research group.



From: Nicholas Humphrey
To: Steven Pinker

Date: 1-31-97

Steven Pinker says "the theory [of male revenge killing of wives] is not about people's reactions to stereotypes or group precedents; it is about their reaction to other individuals. Not, 'Does a typical male kill his estranged wife?' but 'Is this guy crazy enough to come after me?'".

However I don't think the two issues are so easily separable. The question has to be: how could a wife possibly get to know that her husband is so crazy? Presumably it isn't, as it were, "transparent": she cannot actually see the wiring in his brain that might be responsible for launching automatic unstoppable revenge. So the best she can do is to infer it on the basis of indirect clues. But what could these clues possibly be, except either (i) evidence of this individual male having gone crazy in the past and killed a previous wife, or (ii) evidence of other males having killed their wives in the past. In neither case, however, could such evidence tell her for certain how her particular male will act on the next occasion. And so the possibility must still be there that, if and when a new occasion for revenge does arise, the male will decide not to follow the precedent that has been set (either by himself or his sex in general). The wife has therefore to rely on what she knows about typical male behavior. But in this case the male will indeed do better if he cheats on the stereotype and, instead of taking revenge, walks away.

The crucial difference between this example and other possible examples of effective deterrence lies precisely in the non-transparency of the system that brings about revenge. Deterrence in general can only work when there actually are public guarantees that revenge will follow automatically, come what may. This could be the case, for example, if the revenge mechanism were in fact open to inspection, so that the potential victim can be in no doubt about what will happen. During the Cold War, certain deterrence strategists suggested that each side should allow the other to inspect its arrangements for nuclear retaliation for just this reason. But the problem is that this kind of guarantee can never be provided when the mechanism is hidden inside the brain and cannot in principle be open to inspection from outside.

But Steven Pinker acknowledges that this isn't really his field (he borrowed the example from games theorists) - and nor, I hasten to say, is it mine. Would anyone else like to join in? How about opening up this and the other issues Pinker raised to a general discussion.

-Nicholas Humphrey



From: Rick Potts

Date: 1-31-97

Every structure, every behavior, every characteristic of an organism has a history. The human mind is no exception - as Steven Pinker notes, an evolutionary history shaped by (1) a long sequence of surrounding conditions and (2) novel changes in the brain's genetic underpinnings, which happened to give certain hominids a reproductive and survival advantage over others. This leads to Pinker's emphasis on "reverse-engineering" - an analytical metaphor on which his study of mind and the field of evolutionary psychology, in general, are based.

If you reverse-engineered the human shoulder, you would find (in comparison with most other primates) that it is designed well for brachiation. Swinging in the trees is what the human shoulder is about (seen by the upward angle of the glenoid fossa, the mechanical advantage of the "rotator cuff" muscles, the broad thorax to which it is attached, etc.). If you reverse-engineered the human foot, you would find a rigid paddle designed to bear the body's weight on the ground during striding. No tree-living in that creature's history, surely!

My point: Reverse engineering, as it relies only on present observations, inadequately distinguishes very old functions (so-called primitive e features shared with other organisms) from the unique operations of the organism we're interested in (so-called derived features that are, in a discussion of the human mind, particular to Homo sapiens).

As my example illustrates, our species is a complex mixture of structures and activities (including mental activities) that pertain to unique strategies of survival and reproductive success, and others that echo a more distant past. The latter may or may not still be engaged in human adaptation. To continue my analogy, a controlled laboratory test (a method so effectively carried out by evolutionary psychologists) would show that young humans can hang and swing by their arms (just as shoulder design indicates), and that they even enjoy these activities; still, arm-hanging has little to do with human foraging, basic moving about, or survival - I.e., how the human arm now works.

The analysis of "how the mind works" that Pinker suggests has strong possibilities for confounding different segments of our evolutionary past. Human-nonhuman comparisons are a way around this problem, and I look forward to a healthy dose of the comparative method in Pinker's new book. At the same time, evolutionary psychologists (and Pinker, too, given his own phrasing) adopt a monolithic and mistaken view of "the environment" of human evolution. (Brevity is risky here; nevertheless...) Their goal is to reconstruct that "environment" (sometimes called the "environment of evolutionary adaptedness," or EEA) by looking at present behavior and mental function. They then consider this to be an evolutionary "understanding" of human functioning. Well, what environment are they talking about?

We all agree that certain human decisions and brain-mediated actions are shared with many other mammals. In that case, we're talking about "an environment" beginning perhaps 200 million years ago, much of which represented ecological, social, and genomal conditions that were quite different from those experienced specifically by hominids over the last 5 million years. If we seek to understand uniquely human mental functions, then the environments of the past 5 myr are far more important to consider. And if we want to guess how modern human minds differed from those of earlier hominids, then only the past 1 million, or even the last few hundred thousand years, are relevant.

Analysts who employ reverse engineering keep referring to "the environment" but never really examine past environments or much about past hominids. According to evolutionary psychologists, the EEA is somehow encoded in the modern brain and mental functioning. Which environment? What time are they talking about? Failure to answer these questions leads, potentially, to misunderstanding the evolution and current functions of the human mind.

As I've argued (Humanity's Descent: The Consequences of Ecological Instability, Wm. Morrow, N.Y., 1996; and Science, vol.273, pp.922-923, 1996), the setting that shaped human adaptations over the past 5 million years primarily involved episodic fluctuation between habitat extremes. Resource distributions and abundances were repeatedly disrupted. As this happened, the factors that influenced natural selection were also greatly recast - affecting foraging, social life, mating, and competitive settings. Pleistocene environments, which involved threshold-type oscillations, differed in many ways from, say, Eocene or Oligocene environments.

So, to the extent that social/individual functioning (e.g., mental problem solving) is tied to habitat and resources, the answer to "which environment?" is that it depends! Brain-mediated functions shared by all higher primates originated in one set of adaptive settings. Functions unique to hominids evolved in quite different surroundings. The latter involved a dramatic sequence of diverse environments, which posed episodic and unpredictable adaptive challenges to hominids - not the same challenge over and over. Adaptive evolution in the genus Homo, I've proposed, responded mainly to disparities in selective conditions - in contrast with the uniformities or regularities of natural selection emphasized by evolutionary psychologists. If adaptive evolution in some organisms, notably later hominids, was a response to many diverse selective regimes, it rather dramatically alters the way we think about evolution and present function. At the very least, it means that, for some key aspects of human mentality, the monolithic notion of "the environment" or EEA, which is so important to the lingo and understandings of evolutionary psychologists, should be relegated to the dust bin.

In sum, I question exactly what the testing of present mental structure shows us about the evolution of the human mind. As the shoulder-foot comparison implies, design analysis tells us about present function but may confound early history with functions unique to human adaptation. I also question whether "presentist" analyses of mind indicate "the environment" in which the mind evolved. Reverse engineering is usually an analysis without past context. Natural selection ends up lacking time and place. It does not necessarily discern the crucial distinction between shared/primitive and unique/derived regimes of adaptation for the human species. To the extent that language is involved in organizing human mental function, social and ecological environments earlier than 1 or 2 million years ago may be irrelevant to understanding how the unique aspects of living human minds actually work. Natural selection, we've all learned, occurs within local settings. A "presentist" analysis of mental structure, ironically, lacks the time-place context so essential to the past evolutionary effects of natural selection.

Paleoanthropologists are beginning to build a better bridge to environmental sciences. Past habitats surely affected past and present adaptive strategies. In thinking about hominids, however, evolutionary psychologists have relied too much on arm-chair assumptions about how natural selection must work, without studying the past. It is exciting to think about building a better bridge to evolutionary psychologists, and how this would unify the study of human adaptation and evolutionary history.

-Rick Potts


From: Steven Pinker
To: Nicholas Humphrey and Rick Potts

Date: 2-10-97

While I see the value in John and Nick's invitation to open up the discussion to a wider circle, that could lead to an exponential series of answers to replies to remarks on comments, which I could not possibly keep up with. Regrettably, I have to make the following contribution to this discussion my last.

-Steven Pinker

PINKER ON HUMPHREY: I should make it clear that the theory in question here is not specifically about men killing their wives, but about acts of passion in general, by both sexes, of which revenge killing is just an extreme example.

Nick Humphrey asks, "How could a wife possibly get to know that her husband is so crazy? ... The best she can do is to infer it on the basis of indirect clues. But what could these clues possibly be, except either (i) evidence of this individual male having gone crazy in the past and killed a previous wife, or (ii) evidence of other males having killed their wives in the past."

There are many more cues to a person's likely future behavior than their having committed an identical act in the past and their belonging to a huge reference group, such as an entire sex, that statistically performs that act. Much of our everyday thought goes into interpreting a person's actions, words, and emotional reactions to predict what he or she is likely to do. In this case, a man who flies into rages, intimidates or assaults the woman, and shows a willingness to engage in reckless threats and displays, such as making noise and destroying property even if it attracts the attention of neighbors or the police, is giving ample evidence that he is crazy enough to punish desertion even at high costs to himself. Battered women frequently say "I'm afraid he's going to kill me," and they are often correct -- though judges and the police all too often criminally ignore their predictions. Nicole Brown Simpson, to take one example, predicted, probably correctly, that O.J. Simpson would murder her, based on his rages, assaults, and conspicuous vandalism.

Humphreys raises an excellent point when he says "Deterrence in general can only work when there actually are public guarantees that revenge will follow automatically, come what may. ...The problem is that this kind of guarantee can never be provided when the mechanism is hidden inside the brain and cannot in principle be open to inspection from outside." My own proposal is that this very consideration explains the age-old problem of why emotions are involuntarily expressed on the face and the body: in facial expressions, sweating, blanching, flushing, trembling, quavering, and so on. One signals that one's current course of action is under the control of the involuntary subdivisions of the nervous system, the parts tied in to regulation of the physical plant of the body, rather than under the control of the more conscious, rational, deliberative, voluntary subdivisions. These signs are a way of externalizing the fact that the reaction is automatic and uncontrollable, and serves as kind of guarantor of one's threats, promises, and bargains against suspicions that they are bluffs, double-crosses, or lowballs.

PINKER ON POTTS: Rick Potts seems to assume that reverse-engineering the mind requires that one ignore evidence from other species and the fossil and archeological record. I certainly do not subscribe to this mad view and neither does any evolutionary psychologist I know of. Paleoanthropological data clearly have limits -- reconstructing ancient minds is like trying to resolve current controversies in cognitive psychology by looking in graveyards and landfills -- but I certainly agree with Potts that we should squeeze every last drop of information that we can out of that kind of evidence (and of course comparative data as well). How the Mind Works discusses both comparative and paleoanthropological evidence in reasonable detail.

Don Symons' "Environment of Evolutionary Adaptedness" is not a specific time and place, but a composite: a weighted average of all the environments in which our ancestors were shaped by selection. Of course it is difficult to characterize the EEA in any kind of detail, but even characterizing it coarsely has radical implications for psychology that are seldom taken into account. It is not controversial that the vast majority of the span of human evolution, regardless of how far back you want to go, took place in an environment that lacked written language, institutionalized science and mathematics, police, a court system, contraception, substitutes for mother's milk, packaged food, money, and dozens of other features of modern life we take for granted. Contemporary psychology pays no attention to that fact, and proposes theories of human nature that are appropriate for survival in modern Western civilization. Even the crudest facts about the EEA go a long way in constraining psychological theories.


From: Arnold Trehub
Date: 5-13-97

Steven Quartz says "When we look inside the brain what we find is something far more interesting, a self-organizing system in which structure is generated from the interaction between a structured world (which we ourselves structure with culture) and a brain with its own intrinsic constraints." He then says parenthetically "(incidentally, contemporary brain science provides no evidence for innate organs of computations)." I wonder if we substitute the word "mechanisms" for "organs" if he believes that brain science provides no evidence for innate mechanisms of computation.

In THE COGNITIVE BRAIN (MIT Press, 1991), I described and simulated a number of biologically plausible brain mechanisms which could account for many of our basic cognitive capacities. At the conclusion of the book I wrote:

"Reasonable models of the human brain can provide a physical account of the mechanisms and processes of cognition. Interactions among such models in simulated social and physical contexts can provide an account of the evolution of the various contents of cognition. It is the total specific content of cognition, the current physical state of specialized mechanisms in an individual brain shaped by encounters in a world both real and imagined, that constitutes a mind."

It should be clear from the above statement that I agree that the brain is a self-organizing system which constructs (generates structure) in interaction with the world. But if Quartz does not allow innate computational neuronal mechanisms, then what constraints does the brain use to construct a mind?



From: Steven Quartz

Date: 5-21-97

Arnold Trehub asks, "I wonder if we substitute the word "mechanisms" for "organs" if he [Quartz] believes that brain science provides no evidence for innate mechanisms of computation."

The notion that the brain is a collection of "mental organs," or in more computational language, a set of functionally-dedicated computers, has its modern origin in Chomsky's view of language acquisition. He argued that the structure of the environment is so impoverished that the only way a child could acquire a language was if its principles were built into the brain. This claim took a very specific form: a universal grammar, capable of representing any possible human language, was somehow encoded in the structure of the brain.

When Chomsky first made this claim little was known about the brain's development. So little was known that Chomsky could freely ask us to think of the brain on a par with the liver or heart. Thirty years or so later, though, a rapidly maturing brain science reveals a very different picture.. A liver will never become a heart, no matter how sophisticated the transplant. On the other hand, cortical areas are capable of assuming each other's role. Indeed, children who lose the language dominant left hemisphere nonetheless typically become normal language users, provided the insult was early enough. So, where are these organs? Does the brain somehow redundantly represent universal grammar everywhere in the brain? The brain's plasticity coupled with the basic science of developmental neurobiology suggest that the organ metaphor no longer plays a useful role as an item of scientific explanation.

The challenge left for the "organ hypothesis" is to answer Sandra Blakeslee's query in this thread-what does it mean in neural terms to say there are organs of computation? Without an answer, organs of computation are like pulling a rabbit out of a hat-they don't tell you where the rabbit comes from, but only push explanation back another step. What we'd like to know is, how does the rabbit get into the hat?

To answer Trehub's question and to suggest where to look for the rabbit, abandoning organs doesn't lead to an unconstrained brain. I think the brain is richly constrained. These include mechanisms, such as certain receptor types, and also initial cortical circuitry, conduction velocities, subcortical organization, learning rates, and hierarchical development. But nowhere do any of these entail that domain-specific knowledge of cognitive skills is somehow embedded in the cortex. Even at the level of mechanism, there is no hard line in the sand separation between intrinsic and extrinsic. Activity-dependent gene expression, which acts on regulatory genes, for example, suggests that these interactions go all the way down.

I think these constraints lead to a much more interesting research program than did the nativist proposal. The question confronting us is how these constraints interact with structured environmental information (which recent work demonstrates is much richer than Chomsky's poverty of the stimulus arguments allow) to construct the brain through a process of self-organization. Trehub's interesting quotation from The Cognitive Brain resonates with such a program, which is fast becoming the standard model in developmental cognitive neuroscience (e.g., the excellent book, Rethinking Innateness). If this program turns out to be correct, then uncovering where the rabbit comes from will trace back to Piaget's constructivism, not Chomsky's nativism.



From: Steven Pinker
To: Steven Quartz

Date: 5-29-97

Steven Quartz, commenting on my interview "Organs of Computation," argues that recent research on brain development rehabilitates Piaget and casts doubt on innate specializations of the brain. That's the San Diego spin these days, but it's far from "the standard model in developmental cognitive neuroscience." Many cognitive neuroscientists interpret brain development differently, and the book Rethinking Innateness is an opinionated polemic, not an emerging consensus.

Quartz writes, "A liver will never become a heart, no matter how sophisticated the transplant. On the other hand, cortical areas are capable of assuming each other's role. Indeed, children who lose the language dominant left hemisphere nonetheless typically become normal language users, provided the insult was early enough. So, where are these organs? Does the brain somehow redundantly represent universal grammar everywhere in the brain? The brain's plasticity coupled with the basic science of developmental neurobiology suggest that the organ metaphor no longer plays a useful role as an item of scientific explanation."

Plasticity after hemispherectomy is a red herring -- we're bilaterally symmetrical organisms, and the asymmetry of language is imposed on top of that and highly variable even in the normal range (e.g., in left-handers and many women). But even if we look at plasticity within a hemisphere, the organ analogy is not far off. As with the heart and liver, a mature brain area does *not* take over an arbitrary function of any other brain areas (despite the common friend-of-a-friend exaggerations of the kinds plasticity that have been demonstrated). Immature brain areas are more plastic, but then of course if you go back far enough in development there is embryonic tissue that can differentiate either into heart or liver cells, too.

The point is that the brain and all the other organs have to differentiate in development from a ball of identical cells. Body parts take on their particular form as cells respond to cues in their neighborhood (chemical gradients, mechanical forces, molecular triggers, etc.) that unlock different parts of the genetic program. Similarly, the families of neurons that will form the different mental organs, all descendants of a homogeneous stretch of embryonic tissue, are surely opportunistic as the brain assembles itself, seizing any available information to differentiate from one another. The coordinates in the skull may be one trigger for differentiation, but the pattern of input firings from connected neurons may be another. (Quartz himself notes that there is "activity-dependent gene expression, which acts on regulatory genes.") Indeed, it precisely because the brain is is destined to be a set of organs of computation that one would expect it to exploit the capacity of neural tissue to process information as it assembles itself. The plasticity has been described as some kind of "learning," but much of it goes on in the pitch-black womb, before sensory receptors have even formed, so it's better interpreted as part of differentiation and development.

Quartz acknowledges that "the brain is richly constrained. These include mechanisms, such as certain receptor types, and also initial cortical circuitry, conduction velocities, subcortical organization, learning rates, and hierarchical development. But nowhere do any of these entail that domain-specific knowledge of cognitive skills is somehow embedded in the cortex." Actually, according to the computational theory of mind, it is precisly in microscopic details of circuitry, organization, conduction, and so on, that knowledge and skills must reside. What else could they be? If there can be innate circuitry and organization, there can be innate knowledge and skills -- the two are simply descriptions of the same thing at different levels of analysis. As for the question of where organs of computation ultimately come from, the answer is that they were shaped by natural selection and other evolutionary processes.


From: Arnold Trehub
Date: 5-30-97

If we put aside the question of whether it is fitting to say that the brain contains innate "organs of computation", it seems that Pinker and Quartz might agree that there are *at least some* minimal innate computational mechanisms which enable the neonate brain to begin the process of developing/constructing a mind. The crucial question is what are these minimal innate mechanisms and how do they work. I can think of no more important question, because it probes the foundation (and perhaps the limitations) of all human knowledge.

Steven Quartz says "I think the brain is richly constrained. These include mechanisms, such as certain receptor types, and also initial cortical circuitry, conduction velocities, subcortical organization, learning rates, and hierarchical development. *But nowhere do any of these entail that domain-specific knowledge of cognitive skills is somehow embedded in the cortex*." [emphasis added]. While I cannot disagree with Quartz's interest in the developmental plasticity of the brain, I must disagree with his last statement above.

It appears that Quartz is denying that there are *any* innate representations of domain-specific knowledge in the human brain. It seems to me that studies of infant perception tend to refute such a claim. For example, infants as young as 4-5 months evidence knowledge of (1) object permanence, (2) the essential properties of 3-D space, and (3) size constancy (a given object viewed at varying distance is perceived as having the same intrinsic size despite substantial changes in its projected retinal size). If it is claimed that the brain mechanisms which account for these very early manifestations of world knowledge are not genetically determined but, instead, depend upon post-natal experience of the detailed structure of the visual world, then it must be explained how the sensory experience of a 4-month-old infant is able to generate such competent mechanisms. While normal sensory stimulation may be essential for the general post-natal development and maintenance of innate visual-cognitive mechanisms, this does not imply that the basic structure of such mechanisms is established through interaction with the particulars of the visual environment. Like all biological structures, innate cognitive mechanisms depend on the sustenance of an appropriate proximal environment for their proper development. Manipulations such as temporarily blocking light input to one eye in a developing neonate, or selectively blocking neuronal input to particular functional regions of the brain may result in some reorganization of cortical structure and activity, but such findings do not entail that the *design* of *basic* cognitive brain mechanisms is determined by post-natal experience. Of course, this argument does not deny that some innate brain mechanisms/systems undergo progressive restructuring and elaboration in interaction with a complex environment. But it is precisely because we possess *particular* kinds of innate neuronal mechanisms that the adaptive restructuring and elaboration needed for effective human cognition is possible.

To return to the question posed at the beginning of this post, my own work suggests that there are two basic kinds of neuronal mechanisms with which we are genetically endowed and which work synergistically in interaction with the world to compose the mind. I call these minimal bootstrap mechanisms the *synaptic matrix* and the *retinoid*. (For a detailed model of their structure and dynamics, see THE COGNITIVE BRAIN (TCB), MIT Press, 1991). In brief, my claim is that retinoids and synaptic matrices are innately organized into brain systems which can do such things as (1) represent objects in 3-D space, (2) represent oneself with respect to other objects in 3-D space, (3) selectively represent salient sensory patterns by neuronal tokens in long-term memory (category learning), (4) detect stimulus novelty, (5) evoke previously learned patterns in the absence of sensory stimulation (imagination), (6) decompose and recombine previously learned patterns to create novel neuronal representations (creative imagination), (7) perform simple logical inference. These are just a few of the cognitive tasks modeled and simulated in TCB. There is no question that we routinely perform similar and much more demanding cognitive tasks. The challenge for those who deny that there are innate domain-specific cognitive mechanisms is to show how competent cognitive mechanisms can be generated without them.


From: Steven Quartz
Date: 6-3-97

"If there can be innate circuitry and organization, there can be innate knowledge and skills - the two are simply descriptions of the same thing at different levels of analysis."
- Steven Pinker

Pinker suggests that my list of constraints on the developing brain amount to innate knowledge, thereby reducing whatever genuine differences there might be between the positions. No differences-no problems. Well, not quite. I posited "initial cortical circuitry," not "innate circuitry." Although it might seem like splitting hairs, it is a key difference, one that when fleshed out reveals why Pinker's position is far removed from mainstream developmental neurobiology.

First, initial cortical circuitry (for sake of argument let's say the state of brain circuits at birth) can, and does, constrain without encoding domain-specific knowledge (as I consider below). Second, Pinker's theory of language as an instinct isn't just about whether the initial state provides constraints on development. It is a much more robust claim about how the mature state emerges. Pinker tries to minimize this difference by suggesting that much of development takes place prenatally, leaving little work for the postnatal period; he states:

"The plasticity has been described as some kind of "learning," but much of it goes on in the pitch-black womb, before sensory receptors have even formed, so it's better interpreted as part of differentiation and development."
This isn't the case for human brain development. Indeed, what makes human brain development so interesting is how slow and extensive its development is as it moves away from the sensory periphery. The real difference between these theories, then, concerns the processes underlying the generation of mature cortical organization, much of which operates postnatally.

Twenty or so years ago it was commonplace to encounter quotes like the following:

Linguistic information is innate in humans if and only if it is represented in the gene code
(Fodor, Bever, & Garrett, 1974, p.450).
These kinds of claims were a long way from specifying any kind of neural mechanisms. One of my frustrations with much of the evolutionary psychology literature is that it does not go beyond this level of generality. Pinker's writing, however, goes beyond these generalities to be very explicit about what he's proposing; for example, he states:
"grammar genes would be stretches of DNA that code for proteins, or trigger the transcription of proteins, in certain times and places in the brain, that guide, attract, or glue neurons into networks that, in combination with the synaptic tuning that takes place during learning, are necessary to compute the solution to some grammatical problem
(Pinker, The Language Instinct, 1994, p.322)."
This is innate circuitry-specified by genes that determine the pattern of cortical circuitry. We could imagine different sets of genes, each the product of natural selection, to encode the circuitry of each innate mental organ. In neurobiology, the closest advocate of such a position was Roger Sperry, whose chemoaffinity hypothesis once dominated developmental neurobiology. Actually, it is surprising that Chomsky never made reference to Sperry's theory. Sperry himself believed he had reduced developmental psychology to developmental biology, much as Chomsky argued would happen (i.e., that linguistics is ultimately a branch of biology).

I'm not sure how much work Pinker wants synaptic tuning to do. The traditional answer is not much. In fact, the hallmark of the tradition his position belongs to is that mature cortical structure unfolds through intrinsic maturational programs (grammar genes for language), in which the environment's role is exhausted by a simple "triggering" event. The environment is reduced to this role in part because it is too impoverished informationally to do any more.

The trouble is, Sperry's theory hasn't been viable for 25 years, and there are no contemporary neurobiological theories that could implement the sorts of developmental claims he and other evolutionary psychologists make. It is possible that Pinker will think I am creating a straw man, but I don't see any other candidate neural theories (though I'd be more than interested in hearing some candidates). There was a time when some hand-waving could be performed, citing how primitive our knowledge of developmental neurobiology is, but the rapid advancements of the field preclude this recourse.

What is so striking about contemporary developmental neurobiology (which was always thought to reduce to intrinsic programs) is how interactionist it is. This is the real and fundamental difference between nativist positions such as Pinker's and constructivist ones. The environment is not reduced to a minor role, but instead fundamentally shapes the developing cortex well into the second decade of life and indeed remains as lifespan plasticity. Among its defenders: Dale Purves, Larry Katz, Carla Shatz, Dennis O'Leary, Ed Callaway, and also selectionists: Edelman and Changeux. For this reason, I said that interactionist views are becoming the standard model in the field.

In his earlier work (a Cognition review of learning theory and some comments on Borer) Pinker speaks of the possibility that development is nonstationary, meaning that the mature brain organization (what used to be called the functional architecture) unfolds through interaction with the environment. Such a possibility was taken to be a major impediment to theory construction, and so was ruled out by methodological prescription, a principle Pinker called the continuity hypothesis. While this made developmental theory more tractable, it omitted what is most interesting about development-that the mature brain organization develops as a function of learning, becoming a qualitatively more powerful learning structure over time through its interaction with the world.

Such a possibility left an explanatory void-there was no vocabulary to describe this sort of developmental change. Learnability was stuck in inductive paradigms that were wholly inadequate to describe it. With the rise of computational neurobiology and theories of self-organization, however, a new vocabulary emerged, one that could begin to capture the richness of this interaction. This interaction is a form of learning, what Terry Sejnowski and I call constructive learning because it is a learning by building the brain. This is what makes the learning properties of developing cortex so intriguing, a type of learning that is much stronger than the learning straight-jacket traditional accounts shackled on the developing cortex.

In some ways this new perspective is even more radically different than what I intimated before. From its perspective, it is no longer even very interesting to be concerned about what must be innate. The old either/or dichotomies of rationalism/empiricism no longer generate the right questions. Arnold Trehub's interesting post was right concerning the need to understand how the initial state constrains development. And I think we are more or less in general agreement. But I would suggest that the generation of initial cortical circuitry does not reflect innate encodings of domain-specific knowledge. Instead, it constrains through more general properties, including a limited connectivity, that even in the human visual system is only rudimentary at birth. Because these constraints are not domain-specific, the developing cortex has a high degree of equipotentiality, a representational flexibility that is much better documented than Pinker's suggestion that they are friend-of-a-friend exaggerations.

From this perspective, the role of natural selection is also very different. But that's another story.


From: Arnold Trehub
Date: 6-11-97

Steven Quartz writes:

"Arnold Trehub's interesting post was right concerning how the initial state constrains development. And I think we are more or less in general agreement. [1] *But I would suggest that the generation of initial cortical circuitry does not reflect innate encodings of domain-specific knowledge*. [emphasis added] Instead, it constrains through more general properties, including a limited connectivity, that even in the human visual system is only rudimentary at birth. [2] *Because these constraints [initial cortical circuitry] are not domain-specific, the developing cortex has a high degree of equipotentiality, a representational flexibility* ......." [emphasis added]
If Quartz's claim [1] that there are *no* innate domain-specific cortical circuits were true then I can't see how normal postnatal development of the mind could occur. Nor have other investigators shown how it could occur. In my opinion, there are *some* minimal innate domain-specific mechanisms which are essential for later adaptive restructuring of parts of the cognitive brain. I am led to this conclusion by the weight of empirical evidence, and by many years of work on formulating a biologically plausible neuronal model for human cognitive competence. In debating the issue, it is easy for each side to talk past the other. Perhaps a concrete example will help. There is one particularly striking phenomenon which demonstrates the likelyhood of an innate domain-specific mechanism serving the perception of objects in 3-D visual space ___ the *moon illusion*.

We see the horizon moon as much larger than the zenith moon. This phenomenon has puzzled observers for at least 2500 years. It is now known to be an illusion because the true projected visual angle on the retina of either the horizon moon or the zenith moon is approximately 0.5 degree. Philosophers and scientists from Aristotle to contemporary psychologists have attempted and failed to provide a satisfactory explanation of the moon illusion. Some (e.g., Rock and Kaufman) have claimed that we judge the horizon moon as larger than the zenith moon because we compare the size of the horizon moon with known objects in our distant field of view, but are unable to do so with the zenith moon. It has been found that this explanation, which depends on experience-based learning of the relative size of objects, is wrong. In fact, there seems to be nothing in our developmental experience which can account for the systematic shrinking of the perceived size of the rising moon. A recent book (THE MOON ILLUSION, 1989) reviewed the entire corpus of proposed explanations. In a summary chapter, the editor of the book, Maurice Hershenson, concluded that an understanding of the moon illusion was unlikely without an adequate theory of visual space perception.

In the course of my own work, I found that I was unable to account for *basic* visual-cognitive competence by any kind of post-natal learning process and I then accepted the necessity of an innate neuronal mechanism for representing objects in 3-D egocentric space, i.e., an innate domain-specific brain mechanism. The model that I proposed was motivated by the need to explain the neuronal basis of normal visual cognition, but it also happened to provide a neat explanation of the moon illusion as well as other visual illusions (see THE COGNITIVE BRAIN, 1991, Chapter 14, "Illusions and Ambiguous Shapes: Epiphenomena of Brain Mechanisms").

A basic function of our visual system is to automatically maintain the perceived intrinsic size of an object despite changes in its projected retinal size as the distance between the object and the observer changes (size constancy). In THE COGNITIVE BRAIN, I detail an innate neuronal system that maintains size constancy by automatically expanding or contracting the brain's representation of the changing retinal image in compensatory fashion as an object's egocentric distance increases or decreases. Briefly stated, the model predicts the moon illusion as a necessary consequence of a natural (evolution based) anisotropy in the size-constancy mechanism of our terrestrial-bound ancestors. The relevant anisotropy is determined by distance and vertical angle of egocentric representation. I think that the success of the model in accounting for the moon illusion is convincing evidence that the human brain is innately endowed with a mechanism for representing objects in 3-D egocentric space.

Quartz's assertion [2] about representational flexibility implies that initial domain-specific cortical mechanisms would preclude a high degree of neuronal equipotentiality. There is no empirical or theoretical basis for this view. It is important to recognize that the existence of innate domain-specific mechanisms in the brain is easily compatible with an abundance of remaining neuronal tissue having a "high degree of equipotentiality" and "representational flexibility". If we assume that, on average, a *new* neuronal representation is learned every two minutes during each 16-hour waking day over an 80-year period, I have estimated that this would require (allowing for a ten-fold redundancy) that only approximately 2% of roughly 13 billion cortical neurons be uncommitted and available for adaptive tuning. From this we can conclude that even if there were many innate domain-specialized mechanisms in the brain, it is unlikely that the remaining capacity for constructive neuronal development would be strained.

Steven Pinker and Steven Quartz (together with Terrence Sejnowski) have staked their opposing claims near the extremes of the nature-nurture continuum. In their forthcoming books, each will undoubtedly have some persuasive criticism of his opponent's position. But I wonder how soundly each of them can argue against the more credible view that *innate domain-specific brain mechanisms* and *learning-dependent brain structures* are *together indespensible* for the creation of the mature human mind. In this perspective, it cannot be claimed as a general proposition that one process is more important than the other. The respective roles of genetically-determined mechanisms (innate) and experience-determined structures (learned) can be teased apart and evaluated only with reference to *particular* cognitive processes and tasks.


Back to Organs of Computation by Steven Pinker



| Home | About Edge| Features | Edge Editions | Press | Reality Club | Third Culture | Digerati | Edge:Feed | Edge Search |

| Top |