Saturday, October 31, 2015

Tinker Bell and Fairy Dust: Joseph Carroll Doesn’t Know what He Doesn’t Know

Several years ago the journal Style devoted a double issue to evolutionary criticism. It had a target article by Joseph Carroll, responses from over thirty other scholars, and then a final reply by Carroll. One of the respondents was Ellen Spolsky [1].

In her first paragraph she notes that Carroll’s article lays out an evolutionary model of human nature in only 1600 words. She wonders by what authority Carroll presents this summary – does he think of himself as an evolutionary psychologist or an outsider to the field? – and then goes on to note (p. 285):
Beyond my general ability to follow rational argument, how can I evaluate the claims made therein? The most surprising is the claim that he reports finding a “consensus” model. Am I being asked to believe that although my own academic area of literary theory is cross-hatched by contested claims, the field of evolution and evolutionary psychology is blissfully free of conflict and competition? Especially when it comes to a model (of all things!) of human nature? I am being asked to believe in Tinkerbell [sic].
Those of you who know Carroll’s work will not be surprised if I tell you can he manages a superficially plausible response. He’s a skilled rhetorician. For one thing, no, he isn’t claiming that those 1600 words are all there is. There’s more to it, but, alas, you’re going to have to do some reading in the field [2].

Yet for all the apparent plausibility of Carroll’s response, I’ve read enough of his work, and know enough psychology (from reading it as well as using it in my work over the past four decades), to believe that Spolsky’s metaphorical dismissal is apt. Carroll believes in a Tinker Bell version of the newer psychologies and his arguments use generous sprinklings of fairy dust to make them glitter and sparkle.

Cognitive Science in Its Early Phases

In reading through several of Carroll's publications I found an astonishing statement near the end of Graphing Jane Austen [3] – authored with three others. By this point the main argument has concluded and Carroll is simply differentiating his approach from other psychological approaches to literary study. After running through Freud and Jung he comes to the recent interest in cognitive science. It’s in that context that he asserts (p. 167): “In its early phases, research in cognitive science typically operated in the discursive mode of formalistic speculative philosophy, and it was much preoccupied with models of the mind derived from analogies with computers.” If that statement were accurate if would score heavily for literary Darwinism. But it isn’t accurate. Rather, it is extravagantly – dare I say, defensively? – inaccurate.

It’s a bit like saying: “Little is known about Shakespeare. Consequently identifying the man behind the name has been a major project for Shakespeare scholars.” The first statement is true. And, while it is also true that people have worried about whether or not someone other than the man from Avon wrote plays under Shakespeare’s name, this is almost entirely the work of amateurs; it is not a focus of Shakespeare studies.

Another comparison: “For several centuries prior to Darwin biologists took notes, drew pictures, and collected specimens.” That’s true, though they weren’t called biologists then. And the statement completely misses classification and thus gives no sense of the systematic nature of the work. Nor does it indicate that description and taxonomy are major aspects of the foundation on which Darwin built his theory of evolution. Without it he’d have had no way to do the necessary comparative work.

Getting back to Carroll, yes in “its early phases” cognitive science was interested in computational models of mind. But characterizing that work as “formalistic speculative philosophy” is ludicrous, so ludicrous I can’t imagine how that statement got into print. Any time within the last decade a college sophomore could have gotten a more accurate picture in only five or ten minutes online. Carroll is not a sophomore. He’s a senior scholar with two-decades invested in the study of a psychological movement – evolutionary psychology – with substantial roots in cognitive science, including the early phases. How could he be so wrong about that?

By any reasonable account cognitive science in its early phases extends back into the 1950s and 60s. These thinkers, among others, became active and prominent in the 1950s [4]:
Noam Chomsky: The publication of Syntactic Structures in 1957 brought new interest to linguistics and stimulated psychologists to test his syntactic theories by experimental research.

Herbert Simon eventually won a Nobel Prize in economics but he is best known for his wide-ranging work in artificial intelligence, much of it in conjunction with his colleague

Allen Newell. In 1975 the Association for Computing Machinery awarded the Turing Award to Newel and Simon for that work.

Marvin Minsky has spent his career at MIT and is one of the best-known researchers in artificial intelligence and neural networks.

John McCarthy was a Stanford mathematician who organized the first international conference on artificial intelligence in 1956 and later invented the LISP programming language, which became the language of choice for work in artificial intelligence.

George A. Miller first became known for the use of information theory in studying language. His 1956 paper, “The magical number seven, plus or minus two”, is one of the best known in modern psychology. In 1960 he published Plans and the Structure of Behavior, in conjunction with another psychologist, Eugene Galanter, and a neuroscientist, Karl Pribram. I’ll return to this book in a minute.
These men were not doing speculative philosophy. There’s no quick and easy way to characterize what they were doing, but “cognitive science” is a good phrase, coined in 1973 by Christopher Longuet-Higgins [5], who was originally trained as a chemist. Just when did cognitive science pass out of its “early phases”? Who knows? And for the purpose of this post it doesn’t make any difference. That list alone makes it clear that Carroll’s characterization of those early phases is wrong.

Cognitive Science in Evolutionary Psychology

Plans and the Structure of Behavior [6] was widely influential, mostly for a problem solving strategy known as TOTE: test – operate – test – exit. We need not discuss it – this Wikipedia article will give you a feel for it [7] – but I mention it because one of the people who adopted it was John Bowlby, a British psychiatrist who revolutionized our understanding of infant-mother attachment and who was also influenced by ethology and thus helped introduce it into the psychological mainstream. As such Bowlby [8] played a significant role in the emergence of what became known as evolutionary psychology. Carroll devotes a section to him in Evolution and Literary Theory (pp. 435-452). That chapter has a footnote where Carroll notes (p. 439): “Within the field of cognitive psychology, Tooby and Cosmides attribute to Chomsky the kind of pivotal significance Bowlby attributes to Lorenz in the field of behavioral psychology.” So, at one time Carroll knew that Chomsky was important to “cognitive psychology”, though perhaps he didn’t and doesn’t know that there is a considerable overlap between cognitive science and cognitive psychology.

But it’s the Tooby and Cosmides connection I’m after. They’re the ones who catalyzed the emergence of evolutionary psychology as an interdisciplinary field of inquiry in the late 1980s and early 1990s. The various computational models and simulations developed in three decades of cognitive science are essential to their conception of domain-specific cognitive “modules”. The idea may have been sold to the general public with the Swiss Army knife metaphor, but that metaphor got its content, in part, from cognitive science. Those decades of computational work revealed that domain-specific programs were more effective than general-purpose computational routines.

In their introduction to an online primer in evolutionary psychology Tooby and Cosmides say [9]: “In this view, the mind is a set of information-processing machines that were designed by natural selection to solve adaptive problems faced by our hunter-gatherer ancestors. This way of thinking about the brain, mind, and behavior is changing how scientists approach old topics, and opening up new ones.” Where did this talk of mental information-processing machines come from? It came from the “early phases” of work in cognitive science. 

How could Carroll be so wrong about cognitive science? I do not know. I do know, from having read a good deal of his work, that he knows little or nothing about computational models. I also know, from my own experience, that you can’t learn much about such models simply by reading about them. You have to work with them.

But that’s not what we’re talking about here, actually working with computational models. We’re talking about Carroll’s understanding of the intellectual history of a discipline that’s central to his own thinking [10]. You don’t have to work with computational models in order to understand that they have had a formative influence on evolutionary psychology. Yet Carroll either lacks such knowledge or ignores it.

On what other matters is his understanding deficient?

My own suspicion is that, because the role that computational understanding plays in evolutionary psychology is so obvious, Carroll’s blindness is not mere inadvertence. He doesn’t want to go where the investigation leads.

Why not?

Carroll fancies himself, if not quite a theorist in the grand style, then as prophet announcing the coming of a Grand Theory – the title of the second section of his response to his critics in the Style symposium (p. 310 [2]). What do I mean by that? Carroll has been claiming we are on the verge of a solid evolutionary model of human nature, and it will look like this [insert appropriate summary here]. It is on that rock that Carroll and his disciples will found the One True Theory of Literature.

Alas, computation stands in his way. If computational models are important in evolutionary psychology, and to the human sciences generally, then Carroll is in effect announcing the coming of something that, when it arrives, will prove deeply mysterious to him – an uncomfortable situation for a prophet. Moreover Carroll appeals to a largely literary audience that is no more interested in such models than he is and so is not likely to critique his thinking on that score.

Human Nature Slips from Carroll’s Grasp

In a major article he published in the 2005 anthology, The Literary Animal [11], Carroll introduces and explains “A Diagram of Human Nature” (pp. 87-90):

JC on human nature

The shaded portion seems to represent the mind of a typical individual. Here is what Steven Pinker had to say about it in his review of the book (pp. 165-166 [12]):
Carroll, in contrast, does offer a theory of how the mind works in the form of a set of “cognitive behavioral systems” (what others might call modules). But his divisions strike me as arbitrary and unmotivated. For instance, “survival,” “technology” and “cognitive activity” are each put in a separate box, despite being heterogeneous categories with enormous overlap. And they are inexplicably lined up with the emotions “fear,” “joy,” and “surprise,” respectively.
He’s right, and one could easily go on in that vein for a paragraph or three. This diagram is certainly not a top-level view of a model of a mind, human or otherwise. Anyone with a modicum of experience working with contemporary computational behavioral models would recognize that at a glace. And someone having a great deal of experience with such models would hesitate to produce such a diagram at all, much less explain it in a mere three pages.

Pinker goes on to note (p. 165):
The chapter also thunders against the pioneers of evolutionary psychology John Tooby and Leda Cosmides for downplaying the notion of general intelligence, and of individual and racial differences. This, too, struck me as gratuitous. General intelligence is a dimension of variation among individuals (like “strength” or “health”), not a mechanism, so it is unhelpful as an explanation of how people think and feel.
That is, Carroll has made what philosophers call a category mistake, like mixing apples, rocks, and passion, or submarines, whales, and gas nebulae. These are different kinds of things, with different kinds of attributes and possibilities for use and action. Someone who tries to explain the behavior of apples in terms of the passions of rocks is not simply making a mistake. They're talking nonsense. Such is Carroll's diagram of human nature and his associated discussion: not even wrong.

Carroll hasn't presented a model. It’s a list of more and less general behaviors. That’s all. It won’t bear the weight Carroll gives it. This is not the work of someone with a reasonable grasp of the technical literature in evolutionary or any other psychology. Carroll has read widely in evolutionary psychology – more widely than I have – but his grasp of the subject seems second-hand. He may know well enough what a Swiss Army knife is, and see how its components can be analogized to aspects of behavior (like the columns and boxes in his diagram), but he knows little or nothing of the computational considerations that motivated the metaphor in the first place. In his hands the metaphor is a dead metaphor, not an instrument of thought.

Carroll presented that diagram a decade ago. At that time his account of human nature consisted of specialized behavioral modules of various sorts plus that flexible general intelligence. In his more recent thinking he’s added another level to this account. Here’s a paragraph from an essay from 2012 (p. 132 [12]):
Adding general intelligence gives a more satisfactory account of science and other rational and technical features of civilization, but the model still gives no good explanation for art and other products of the imagination. The early EP explanation for art as by-product remains active in the more recent, broader model that includes general intelligence. In contrast to both narrow-school EP and broad-school EP, several evolutionists in the humanities and sciences have argued that the imagination is functionally integral to the specifically human way of coping with the world. Humans live in the imagination; they create imaginative virtual worlds that contain past and future and that contain also their sense of relations with people and forces outside their immediate ken. Humans are the only species that can die for an idea. That is because they are the only species that lives by ideas, or more precisely, by emotionally charged imaginative constructs like religions and ideologies.
Does Carroll really believe that while art is among the “products of the imagination”, that the “rational and technical features of civilization” are not? Does he think that you can build a new tool without first imagining it? What about damming a river? Is that something people do without imagining how they would benefit from the water behind the dam, or how downstream locations would benefit from the dam’s protection? Did they start construction without imagining where the dam would go and how to construct it? How could anyone ever have proved that the earth moved around the sun without exercising a great deal of imagination? Are the gravitational forces of the sun, the moon and the planets not outside our “immediate ken”? Can one understand Hobbes’s conception of the Leviathan without exercising one’s imagination?

The idea that science, technology, and philosophy do not require imagination is so ludicrous that Carroll cannot possibly believe it. But how else are we to interpret his words? Perhaps he believes that there are different sorts of imagination. Science, technology, and philosophy call on Type 1 imagination, while the arts call on Type 2. If so, then he should have said that and justified it. Maybe Type 1 imagination is not “emotionally charged” while Type 2 is? Is that why Galileo ended his life under house arrest, because he wouldn’t recant the products of his emotionally neutral Type 1 imagination?

There is no good justification for this kind of intellectual chaos. But I can imagine at least one explanation: Carroll is not prepared to engage with the complexity and subtlety of contemporary psychology, especially at the level required for understanding the most complex and sophisticated human actions. I rather suspect that he is, at heart, an ideologue and he subordinates evolutionary social science to the demands of that ideology.

Just what that ideology is, that’s not clear. I’d be inclined to say that it’s political only in the broad sense that everything is political, though others are likely have a different view about the matter. What he seems to be after is a way of dressing common sense accounts of behavior in technical garb. That will allow him to do the practical literary criticism he wants to do while at the same time claiming intellectual progress.

And, of course, he wants to vanquish all varieties of post-modern literary criticism. While I have some sympathy for that desire, I can’t see that anything is gained by making an incoherent hash out of evolutionary social science. On the contrary, Carroll’s slipshod theorizing makes things more difficult for conscientious thinkers because it validates the fears of skeptical humanists. By exemplifying how biologically based thinking about human behavior can be opposed to open-ended inquiry Carroll has become his own worst enemy. His conceptual world is as closed and insular as the worlds of the  postmodern literary critics he has so vigorously criticized.


[1] Ellen Spolsky, The Centrality of the Exceptional in Literary Study, Style Vol. 42, Nos. 2&3, Summer/Fall 2008, pp. 285-289.

[2] Joseph Carroll, Rejoinder to the Responses, Style Vol. 42, Nos. 2&3, Summer/Fall 2008, pp. 309-371. He responds to Spolsky (and others with similar objections) on pp. 314-318.

[3] Joseph Carroll, Jonathan Gottschall, John Johnson, Daniel Kruger. Graphing Jane Austen: The Evolutionary Basis of Literary Meaning. New York: Palgrave Macmillan, 2012. Download URL:

[4] Thagard, Paul, "Cognitive Science", The Stanford Encyclopedia of Philosophy (Fall 2014 Edition), Edward N. Zalta (ed.), URL:

[6] Miller, GA, Galanter, E., & Pribram, KA (1960). Plans and the structure of behavior. New York: Holt, Rhinehart, & Winston. You can download a PDF of the book here:

[7] The Wikipedia entry:

[8] Bowlby is important enough to have a significant web footprint. You might want to try a Google search on his name. Here's his Wikipedia entry:

The Bowlby Centre:

FWIW I was introduced to Bowlby’s work as an undergraduate at The Johns Hopkins University, where I studied with Mary Ainsworth. She had been a student of Bowlby’s and had developed an important and original line of research on attachment theory that included fieldwork in Africa. She particularly emphasized the importance of the TOTE model – and she also had me reading early anthologies about primate ethology.

[9] Leda Cosmides and John Tooby, Evolutionary Psychology: A Primer, Center for Evolutionary Psychology, University of California at Santa Barbara, 1997, URL:

[10] Carroll presents some of that history in the context of a slide presentation: The Historical Position of Literary Darwinism. URL:

There’s no hint of computation in that presentation.

[11] Joseph Carroll, Human Nature and Literary Meaning: A Theoretical Model Illustrated with a Critique of Pride and Prejudice, in The Literary Animal: Evolution and the Nature of Narrative, Jonathan Gottschall and David Sloan Wilson, eds., Northwestern University Press, 2005, pp. 76-106.

[12] [12] Steven Pinker, Toward a Consilient Study of Literature, Philosophy and Literature, 31, 2007, pp. 161-177.

[13] The Truth about Fiction: Biological Reality and Imaginary Lives, Style 46.2 (2012): 129-60.

No comments:

Post a Comment