Monday, August 24, 2015

On the Poverty of Literary Cognitivism 2: What I Learned When I Walked off the Cliff of Cognitivism

In the first post in this series I took a look at an essay-review Alan Richardson wrote about two recent books in literary cognitivism and asserted, in effect, that you can’t get there from here [1]. By “there” I mean a reciprocal relationship between literary study and “the mind and brain sciences” in which “methods, finds, or evidence from a literary field” (Richardson, p. 368) is important. And by “here” I meant cognitive literary criticism as it has developed over the last two decades or so. It’s not that I don’t think that such literary finds or evidence exist but that current literary methods enable scholars to find them and present them in a convincing way.

The problem, I asserted, is that the cognitive revolution was driven by the idea of computation and that neither the literary cognitivists, nor their twin, the literary Darwinists, have faced up to computation in a deep way. Until they do so, their efforts will be still born. No matter how widely they read in the newer psychologies, no matter how many findings, ideas, and models they incorporate into their theorizing, they won’t produce results that are compelling to thinkers in the mind and brain sciences.

The purpose of this post is to explain how I arrived at my belief computation must be dealt with. But the argument will be a strange one because it will tell the story of how my own journey computation semantics led to a satisfying failure and thereby forced me to reconceptualize what it means for a literary scholar to come to terms with computation. But I’ll save that reconceptualization for a later post.

From “Kubla Khan” to Computation

I was an undergraduate at Johns Hopkins when the French landed back in the Jurassic era of academic literary criticism. Though I didn’t attend the sessions of the famous 1966 structuralism conference, I fell into the orbit of one of its organizers, Dick Macksey. But I also studied with Earl Wasserman, Don Howard, and Don Cameron Allen, all of whom were more traditionally minded, each in his own way, and with J. Hillis Miller, who was a fellow traveler and who, as you know, went on to become one of our premier deconstructive critics. I was thus versed in New Criticism on the way to phenomenology, phenomenology giving way to structuralism and semiotics, and deconstruction bubbling up underneath it all. During my senior year I became interested in “Kubla Khan” and stuck around to do a master’s thesis on it. I read everything the Hopkins library had on the poem, threw all my critical tools into the fray and found myself walking in circles, getting nowhere.

And then, on a hunch, I decided to break the poem into pieces using the obvious markers, stanza divisions and line-end punctuation. Here’s one of my work sheets for the first part of the poem, 36 lines.

KK-triple-72.jpg

The tree structure to the left visualizes the result. Notice that we’ve got three sections at the topmost level; the middle of those is in turn divided into three; and the middle of that is again divided into three. All other divisions are binary.

It turns out that the poem’s second part, only 18 lines, has a similar structure. And now it gets really interesting. For the last line of the poem’s first part – “A sunny pleasure-done with caves of ice!” – is the middle, of the middle, of the middle of the poem’s second part – “That sunny dome! Those caves of ice” And that’s just a fragment of the elaborate patterning that exists in “Kubla Khan,” all there on the “surface” just waiting to be seen, analyzed, and described. But where did those patterns come from, what were they doing? [2]

And yet the literature I’d reviewed about “Kubla Khan,” one of the best-known and most-examined poems in English, knew nothing of this. And the body of literary and critical theory I’d absorbed, from New Criticism through structuralism and (a bit) beyond, had little to say about this kind of phenomenon. If I was going to figure out what that patterning was about, I was going to have to find the tools elsewhere or somehow create them myself.

I had been reading linguistics on my own and so I knew about tree structures as tools for describing linguistic objects. I’d also taken a course in computer programming – one of the first offered in a university – and had intimate knowledge of how the difference between a comma and a semicolon, for example, can be the difference between a successful program and a broken program. I also knew a little about nested control structures.

And that’s what I saw in those nested-triple branching structures. They ‘smelled’ of computation, not in the trivial – but then again, more abstractly considered, not so trivial – sense of arithmetic calculation, but in the deeper and more ‘metaphysical’ sense, shall we say, of the emerging cognitive sciences – the term was coined in 1973. I figured that’s where I had to do to understand what was happening in “Kubla Khan”.

Cognitive Networks and Literary Semantics

And so I headed off to get a Ph. D. in the Department of English at SUNY Buffalo. I had no reason to believe that there was anyone in that department that could be of direct help to me, and so I was anxious on that score. But it had the best experimental English program in the nation and the lines of communication between Hopkins and Buffalo were good.

I arrived in Buffalo in the fall of 1973 (while John Barth traveled back to Hopkins from Buffalo at the same time). While, as I’d expected, there was no one in the department ready to help me, I showed my work around and people were interested. They respected it. In the spring of 1974 a fellow graduate student, Ralph Henry Reese, introduced me to the late David Hays, a computational linguist and polymath in the Linguistics Department.

Hays was fascinated by my work on “Kubla Khan” and, while he had nothing of immediate use for me, he invited me into his research group. Computational linguistics certainly sounded like something I needed to know and, as one of the founders of the discipline [3], Hays was the one to teach me. And so I apprenticed myself to him, which was fine with the folks in the English Department.

That Fall Hays was teaching a graduate seminar on Language as a Focus for Intellectual Integration. Each student in the seminar got to put a book on the reading list. I put Northrup Frye’s Anatomy of Criticism on the list. Hays put several on the list, including a thick one by Talcott Parsons (who had been one of his teachers at Harvard), Gombrich’s Art and Illusion, and William Powers, Behavior: The Control of Perception. I think Bateson’s Steps to an Ecology of Mind also found its way to the list. At the same time I had weekly one-on-one tutoring sessions with Hays in which I learned his semantic model. At the end of the semester I submitted a joint paper for that course and a course in contemporary poetry was taking with Charlie Altieri. In that paper I had a bunch of semantic network diagrams, ending with a fairly complex series underpinning some passages from Patterson, Book V.

It must have been at the end of 1975 or very early in 1976 that I visited Baltimore and met with Dick Macksey and Sam Webber about the centennial issue of MLN. By that time I was well into my work cognitive work on Shakespeare’s “The expense of spirit” and offered that for the issue. I don’t recall exactly how that conversation went, but I remember the word “recusivity” and that was certainly an important point of the model. I think the general rationale was a continuation of the interdisciplinary spirit that was at heart of the 1966 structuralism symposium. Things were in flux and cognitive science seemed like it could be part of the flow.

While that didn’t happen back then, it was not, on the face of it, as strange an expectation as it seems in retrospect. The discipline was more fluid in the mid-1970s than it has been at any time since. The profession had not yet decided that deconstruction had dissolved structuralism and linguistics had not yet been sent packing; some scholars had actually read Chomsky’s linguistics. What came to be called Theory had not yet consolidated, though the term “literary theory” was rapidly coming to mean “instrument of criticism” rather than “theory of literature”.

Macksey and Weber accepted my offer of a cognitive piece on “The expense of spirit”. Cognitive Networks and Literary Semantics appeared in MLN in 1976 in the special centennial issue, The Responsibilities of the Critic, to celebrate the journal’s centennial [4]. Here’s a fragment of the model I developed; it’s Figure 9 (of eleven) in the article:

Shame s129 mln76

There are several things you should know about that diagram or, rather, about the model which the diagram depicts. First, it was in many ways like other models of that era, many of which used a network notation and so were called cognitive or semantic networks.

Second, though I did not do any computational work with it myself, two of Hays’s students, Teiji Furugori and Brian Phillips had done so [5]. That is, it is a real computational model and, like all such models, specifies a physical process. Third, during that 1974 seminar Hays and the rest of us students roughed out a scheme for grounding those networks in the sensorimotor system, making it a model of embodied cognition, though the term had not yet been coined. Nor for that matter was the idea of embodiment new in any deep way [6]. I point these things out to make the point that, while that (ancient) work was computationally inspired and realized, we did not conceive of computers as abstract alien beings and of computing as some disembodied process. We were very much committed to using computation as a way of understanding the actions of real minds in real bodies.

Text and Network Model Are Separate Entities

Fourth, though I employed the model in investigating Shakespeare’s sonnet, the diagrams I developed should not be understood as somehow representing the meaning of that sonnet. They don’t represent the meaning of anything. Rather, they depict mental structures (embodied in a nervous system that is) activated through living in the world and experiencing it as meaningful. Those mental structures are independent of any specific text and so can be called upon a wide range of life situations, including the reading any number of texts.

That diagram is one of eleven I developed for that article. In a sense, that suite of diagrams represents the major conceptual work in the article. I created the diagrams and then wrote the prose explain to explain. The diagrams thus are not illustrations of or aids to the understanding of ideas that exist perfectly well in prose. On the contrary, prose is a very poor medium for expressing and working with those ideas.

If those diagrams represent “pure” conceptual structures (rather like the “mentalese” of Chomskian discourse) what then of spoken or written text? It is conceptualized arising through tracing paths in the network. Here’s a simplified version of part of the model along with the final couplet of the sonnet:

had-having-quest

While the correspondence between the model and the words in the sonnet is not detailed – I don’t account for each word – it is direct. It is also clear that the text is one thing, the model is another. There is no confusing the two.

Let me repeat that: The text is one thing, the model is another. There is no confusing the two.

That separation is one of the most important lessons I got from that work. It is one thing to simply to assert that the mind is one thing, while a text, a string of signifieds, is another. It is quite something else to have a concrete representation of the mind that you can examine, manipulate, and change, and that is distinct from any given text. When you have that the mind is no longer some nebulous power operating behind the scenes in the cracks and everywhere, but always invisible and impalpable.

THAT was the force of the cognitive revolution. The computer gave us a way of conceptualizing the mind as a complex physical system. With that conceptualization we could on the one hand simulate mental processes using computers and, on the other hand, design precise experimental that would tease out the computational properties of the human mind.

Moreover, and in a peculiar way, the explicit separation of text and mental model seemed to dissolve the problematic of meaning. For that is what had been dogging me since my undergraduate studies; that is what I’d been pursuing: what do these texts mean? And that, of course, has been the central problematic once interpretation became the focus of academic literary criticism. Prior to World War II the discipline was based on philology and literary history, but interpretation moved into center place during the 1950s and 1960s. When I turned to linguistics I wasn’t looking for syntactic trees, fascinating though they were, I was looking for semantics. That’s what I needed to study meaning. That’s what I was looking for when I turned to David Hays and his semantic nets.

Now that I could, at least in principle, see the text over there, on the one hand, and the model right here, on the other, I could, if only in principle, see meaning as arising from the way the model moves through the text, or, if you well, the way the text activates the model. If I’ve hedged that statement ¬– “in principle” – that’s because I didn’t believe, then or now, that the model I was using was the Holy Grail of cognitive science, the key to all cognitivist mythologies. It needed lots of work, perhaps even to be scrapped and then reconstructed.

But I stand by the principle, the explicit separation of model and text, and that principle allowed me to dissolve the problematic of meaning. Meaning was no longer something I was seeking in the confusion between text and mind, in the inability to separate the signs on the page from the mind reading them. The mind was now, for better or worse, represented by an explicit model.

Beyond the Model

The fifth and last thing you need to know about that model is that, like all such models, of course it couldn’t do everything. Some things worked better than others. And one thing that model couldn’t do was tell me what’s going on in “Kubla Khan.” That’s what I went to graduate school to discover and that’s why I apprenticed myself to Hays. In our private tutoring sessions Hays would set specific tasks, things for me to diagram – I was working from a text he’d developed for one of his courses. Once I got through the basics I started working on “Kubla Khan.” That was interesting but it wasn’t quite coming together. I decided to work on “The expense of spirit” as a fallback. And that’s the culminating example I used in my dissertation, Cognitive Science and Literary Theory (1978).

Now, if you look at that dissertation you’ll see that it’s every bit as much an interdisciplinary extravaganza as anything produced by today’s literary cognitivists. Richardson tells us that David Herman “looks to neurophenomenology in his new book, as well as social psychology, cognitive linguistics, discourse analysis, ecological psychology, emotion theory, philosophy of mind, conceptual integration theory, cognitive ethology, and theories of extended or distributed cognition, among others” (p. 366). Among others! Brian Boyd, to mention one of the most interesting literary Darwinists, has a different but no less impressive inventory of conceptual tools in the kit he deploys in The Origin of Stories. And that’s surely a signal attraction of this enterprise, it provides a good professional excuse to read and think about a lot of cool stuff. Remember that seminar I took with David Hays in the fall of 1974 – Language as a Focus for Intellectual Integration? Hays was into a lot of cool stuff as well.

Cool as our model was, jammed with all that neat stuff, “The expense of spirit” is the last text I examined through that model. I continued working with Hays until his death in 1995. We talked and thought about literature, among other things, and published a series of papers, jointly and separately, on cultural evolution. But applying complex cognitive models to literary texts no longer held my attention.

Why not?

Superficially it was a matter of the sheer size and complexity of such models. The version I used in my dissertation was more complex than the one I’d used in the MLN article. I cannot begin to image what kind of model would be required for a Shakespeare play [7] much less a Victorian triple-decker novel.

What to do?

As real as those problems are though, as I say, they’re superficial. Recall that I’d done my first (student) work on cognition for both Hays and for Charlie Altieri. While Altieri was fascinated by the paper I wrote, he also said that it seemed to him that I was using Patterson V as an example within a theory that’s essentially about something else – the mind, cognition, whatever – but not poetry, not literature.

I bristled at the suggestion but he was right. And the same is true of the model I build for “The expense of spirit”. And I suppose that Richardson might say that I’d treated cognitive science as “a master discipline for literary scholars to study and incorporate into their own work” (p. 362). But I certainly did feel taken over. But it is true that I wouldn’t have gone so deeply into the cognitive sciences if the material hadn’t been interesting to me above and beyond whatever value it would have in understanding literature.

The question I faced then, as indeed I face it now, was: Where next? As I’ve already indicated, I worked on the cognitive model with Hays and have continued to work on it since he died. But I’ve also worked on literature (and film), largely independent of all those pesky, albeit very interesting, cognitive details. And a lot of that work has been largely descriptive.

Recall that I went down the cognitive path because I’d found what looked like the traces of computational operations in “Kubla Khan”. Those traces haven’t disappeared simply because I’d been unable to create a model that would explain them. They’re still there, and I’ve gone on to describe other such structures in other texts.

* * * * *

When I set out to explain cognitive science in that 1978 dissertation I felt that I had to define cognitive science. The term itself had only been around a few years and my literary readers were not familiar with the literature. Well, as you know, cognitive science has never been strictly defined. It’s always been a loose collection of themes, ideas and models, and the same is true of evolutionary psychology. So, anyone who wants to define such a thing has quite a bit of leeway in doing it.

I defined cognitive science as investigating a five-way correspondence between: 1) behavior, 2) computation, 3) computational geometry (that is, neuro-anatomy and physiology), 4) ontogeny (human development), and 5) phylogeny (human evolution). That’s a rather generous definition, even by the loose standards of “cognitive science” and, if you think about it, you’ll see that it pretty much covers whatever you’d want from a psychology, unless you want the social dimension explicitly spelled out (but that is at least implicit in behavior).

No doubt it was crazy to declare such a scope in a dissertation and you might think its crazy to declare such a scope at all. No doubt it is. But let me repeat Richardson’s list of David Herman’s disciplinary run down: “neurophenomenology […], as well as social psychology, cognitive linguistics, discourse analysis, ecological psychology, emotion theory, philosophy of mind, conceptual integration theory, cognitive ethology, and theories of extended or distributed cognition, among others” (p. 366). Is my list any crazier than that? Here’s Joseph Carroll’s list:

The sciences most relevant to literary study form an integrated complex summarized by the term “biocultural theory.” Those sciences include evolutionary biology and the evolutionary social sciences: anthropology, human life history theory, evolutionary social and political theory, affective and cognitive neuroscience, personality psychology, and linguistics. [8]

Literature is complicated. To a first approximation the literary mind is the human mind. Hence an account of the literary mind is going to have to cover pretty near the entire range of disciplines concerned about the human mind.

That’s crazy? Necessary perhaps, but crazy. Those of us who have succumbed to this craziness have done so by reading extensively in many disciplines. But it’s going to take a lot more than broad reading to pull it off.

It seems to me that the most useful contribution literary scholars can make to that effort is to provide meticulous descriptions of the (computational) properties of literary texts. I’ve said quite a bit about that in my article on literary morphology [9], and I hope to say a bit about that in a later post in this series. I didn’t have all that theoretical work under my belt when I did that original work on “Kubla Khan” and I don’t call on it in the analytic and descriptive work I’ve done in the last decade. It is something you can do without having to spend a couple of years reading in half a dozen or more disciplines.

And isn’t that why one gets a degree in literature, to read and work on literary texts?

References

[1] On the Poverty of Literary Cognitivism 1: Alan Richardson Makes a Case, August 23, 2015. http://new-savanna.blogspot.com/2015/08/on-poverty-of-literary-cognitivism-1.html

[2] I first wrote up those results in a 1972 MA thesis, but I didn’t publish them until the mid-1980s. Early in this millennium I revised and expanded that work.

Articulate Vision: A Structuralist Reading of “Kubla Khan.” Language and Style 18: 3 - 29, 1985. Downlaod: http://www.academia.edu/8155602/Articulate_Vision_A_Structuralist_Reading_of_Kubla_Khan_

“Kubla Khan” and the Embodied Mind, PsyArt: A Hyperlink Journal for the Psychological Study of the Arts, November 29, 2003. Download: https://www.academia.edu/8810242/_Kubla_Khan_and_the_Embodied_Mind

[3] David G. Hays. Wikipedia. Accessed Aug. 24, 2015. URL: https://en.wikipedia.org/wiki/David_G._Hays

[4] Cognitive Networks and Literary Semantics. MLN 91: 952-982, 1976. Download: https://www.academia.edu/235111/Cognitive_Networks_and_Literary_Semantics

[5] Furugori, Teiji, A Memory Model and Simulation of Memory Processes for Driving a Car. Ph. D. Diss. State University of New York at Buffalo, 1974.

Phillips, Brian, Topic Analysis. Ph. D. Diss. State University of New York at Buffalo, 1975.

Phillips, Brian, "Judging the Coherency of Discourse," American Journal of Computational Linguistics, Microfiche 35, frames 36-49, 1975.

[6] John Sowa, for example, finds it in Aristotle, Review of Philosophy in the Flesh: The embodied mind and its challenge to Western thought. Computational Linguistics, vol. 25, no. 4, December 1999, URL: http://www.jfsowa.com/pubs/lakoff.htm

[7] See discussion of the hypothetical Prospero system in William Benzon and David Hays, Computational Linguistics and the Humanist. Computers and the Humanities 10: 265-274, 1976. Download: https://www.academia.edu/1334653/Computational_Linguistics_and_the_Humanist

[8] Joseph C. Carroll, Biocultural Theory and the Study of Literature. Comparative Literature 67:1, pp. 21-28, 2015. DOI 10.1215/00104124-2861969

[9] Literary Morphology: Nine Propositions in a Naturalist Theory of Form. PsyArt: An Online Journal for the Psychological Study of the Arts, August 2005, Article 060608. Download: https://www.academia.edu/235110/Literary_Morphology_Nine_Propositions_in_a_Naturalist_Theory_of_Form

2 comments: