Yesterday, or the day before, I asserted that a lot of new intellectual tools have been created since I first set out to figure out what’s going on in “Kubla Khan” and “This Lime-Tree Bowers My Prison”, but I didn’t say much about those tools or why I believe they will prove more or less to have “turned the corner” on conceptual invention in this new intellectual continent. I want to say a few words about that now.
First of all, what was I really looking for when I set out to discover a one single “grammar” that would encompass both “Kubla Khan” and the very different Conversation Poems, such as “This Lime-Tree Bower My Prison”? Well, I want to do more than provide a verbal description of what they have in common; I did that in my 1981 note (Metaphoric and Metonymic Invariance: Two Examples from Coleridge. MLN 96: 1097-1105). The idea is to get at what’s going on behind the scenes, as it were, in the mind that produces those texts.
We need to say something that’s principled and interesting on that score. Principled means that others can understand it and work from it. Interesting is a bit trickier. But in this day and age it certainly implies that what we say should be tangibly linked to and grounded in the newer psychologies, neuro, evolutionary, and cognitive. But it implies more than that. Interesting implies that solving this problem will lead us closer to something deep and fundamental.
I’ll take THAT up in a later post. For now, let’s stick to principled and to the newer psychologies.
Roughly Three Kinds of Formal Models
Consider a paragraph from the penultimate chapter of Conceptual Spaces (2000), by Peter Gärdenfors (p. 253). He’s saying that we need different kinds of computational processes for different kinds of problems:
On the symbolic level, searching, matching, of symbol strings, and rule following are central. On the subconceptual level, pattern recognition, pattern transformation, and dynamic adaptation of values are some examples of typical computational processes. And on the intermediate conceptual level, vector calculations, coordinate transformations, as well as other geometrical operations are in focus. Of course, one type of calculation can be simulated by one of the others (for example, by symbolic methods on a Turing machine). A point that is often forgotten, however, is that the simulations will, in general be computational more complex than the process that is simulated.
The cognitive networks that I explored back in the 1970s and into the ‘80s work at the symbolic level. That’s what failed. But, if Gärdenfors is right – and I think he more or less is – then we’ve got two other families of processes to take into consideration.
Hays and I and, of course, many others, believed that there was more to the mind than symbolic processing, though a few tried to argue that symbolic processing was all. Sometime in 1973 or 1974 Hays had read a book by William Powers, Behavior: The Control of Perception (1973). Powers was cyberneticist; he thought in terms of closed loop feedback systems with analog rather than digital processing. Such systems belong in Gärdenfors’ subconceptual and perhaps intermediate levels.
One major focus of Hays’s research group at that time was to integrate Powers’ feedback model with Hays’ cognitive network. Basically, we used Powers’ model – a stack of servomechanical systems where higher level systems regulated lower level ones – for sensorimotor processing (sensory perception in all modes, and motor control) and the cognitive net to organize sensorimotor processes into large-scale patterns. Thus our scheme was embodied in the sense that term has come to have in the cognitive sciences. Cognition, thought, is grounded in, given meaning by, sensory and motor experience.
That scheme was a major focus of Hays’ 1975 manuscript, Cognitive Structures, which he then published with HRAF Press in 1981. I used that scheme in my 1976 article in MLN, Cognitive Networks and Literary Semantics, and in my 1978 dissertation, Cognitive Science and Literary Theory.
But it’s one thing to have such a scheme laid out in an abstract theory, one lacking the many details needed for a computer simulation, and quite something else to make practical use of it. We know how to draw cognitive network diagrams, for example:
And we know how to relate such diagrams to texts. That’s what I did in the MLN article, where the text was Shakespeare’s Sonnet 129. I just couldn’t make it work with my Coleridge texts, not like I wanted to, though perhaps I could have forced it.
But what’s the point?
Something interesting happens, however, when you declare the shop closed. You stop looking for new stuff and made do with what you’ve got. And sometimes you find things you hadn’t been aware of.
That’s what I’ve decided to do. I believe that the means exist to say something interesting and principled about the relationship between “Kubla Khan” and “This Lime-Tree Bower My Prison.”
Principled and Interesting
While I’ve decided that cognitive network models alone adequate are not to the job. I do think we’re going to need some kind of cognitive network, so let’s start there.
As the name suggests, a cognitive network is about cognition, not the whole of the mind. Nodes in such networks are linked to words, so it is through such linkage that we can relate them to texts, such as poems. A person one knows, such as Charles (I’m thinking of Coleridge’s friend, Charles Lamb), will be linked to a name – “Charles Lamb” – to various facts about the person and to one’s history with that person. All of that can be represented in a cognitive net.
But the person’s appearance, the sound of their voice, the way they walk, none of that makes it into cognitive. That’s all sensorimotor stuff. That’s in those subconceptual systems Gärdenfors talks about. And so is our emotional relationship with that person, for the behavioral systems mediating interpersonal interaction are basically of the sensorimotor order. They may be touched by cognition, but they are of a different order.
Now, in his conversation poems, such as “This Lime-Tree Bower”, Coleridge addresses himself to someone close to him. Charles Lamb in “This Lime-Tree Bower” (LTB), his young son Hartley in “Frost at Midnight”, his fiancé, Sara Fricker, in “The Aeolian Harp”, and the Wordsworths and Hartley in “The Nightingale”. But “Kubla Khan” (KK) is addressed to no one in particular. It is thus not linked to personal history.
Can we make anything of that that is principled? I suspect so. But to do it we’re going to have to speculate a bit, quite a lot, actually. But then, where incremental conceptualization just won’t get the job done – and in this case it surely will not – speculation is our only option. As long as we’re careful, others can follow and correct us.
KK certainly does mention individuals. There’s the title character, but there’s also the “woman wailing for her demon lover”, the “damsel with a dulcimer”, and “all who heard” – the poet’s audience. There’s no reason to think Coleridge had specific individuals in mind – though I suppose he might have. But I want to take a different tack.
In what remains one of the best essays ever written about the poem (“Kubla Khan,” Proto-Surrealist Poem in Language as Symbolic Action, 1966, pp. 201-222), Kenneth Burke suggests “that this indeterminate mixture of motion and action is in effect a poeticized psychology, detailing not what the reader is to see but what mental states he is thus empathically and sympathetically imitating as he reads” (pp. 208-209). I agree, more or less.
In that spirit let’s think of those individuals, not as poetic realizations or evocations of individuals Coleridge knew, but as mental faculties or behavioral systems. So Kubla would become, say, a figure for the Will while the wailing woman would be a figure for Desire. In this view KK becomes a vision of the mental mechanisms that enact the world depicted in LTB and his other poems including of course the conversation poems.
One question that this raises immediately is whether or not these poetic figures are expressions of an explicit theory Coleridge held or whether they are somehow – I won’t venture as guess as to how – indirectly linked to actually neuro-mental mechanisms. After all, in his discussion of meter in Chapter XVIII of the Biographia Literaria, Coleridge speaks of spontaneous impulse and voluntary purpose and of how poetry must bring about “a union; an interpenetration of passion and of will.” Could passion be the wailing woman and will be Kubla? Perhaps. The concepts of passion and will are certainly rationalizations of things we experience in/about our own mind. Is KK grounded in those rationalizations or is it more directly grounded in the experiences of which those ARE rationalizations?
It seems to me that that sort of question is up for grabs at this point. Getting a purchase on such questions would be one objective of long-term research into Coleridge, and into poetry in general.
But let’s return to Peter Gärdenfors and, more specifically, to the notion of a conceptual space he regards as being between symbolic processing and subconceptual processes (p. 5):
When attacking the problem of representing concepts, an important aspect is that concepts are not independent of each other but can be structured into domains; spatial concepts belong to one domain, concepts for colors to a different domain, kinship relations to a third, concepts for sounds to a fourth, and so on. For many modeling applications within cognitive science it will turn out to be necessary to separate the information to be represented into different domains.The key notion in the conceptual framework to be presented is that of a quality dimension...There is a tight connection between distances in a conceptual space and similarity judgments: the smaller the distances is between the representations of two objects, the more similar they are.
“Kubla Khan” seems very much to be a poem organized by domains, while “Lime-Tree Bower” is not. Yes, the action in LTB takes place in two physical settings, the bower (garden) and larger world outside that garden – not to mention the sky with the sun over both – but those are the same KIND of thing, only different in size and scale. The bower is small and near while the path taken by the poet’s friends is large and distant.
But KK is clearly demarcated into (ontologically) different domain. The last 18 lines are realized in a world of mental acts and words while the first 36 lines are realized in a physically concrete world. Of those first 36 lines, the first 12 are visual and spatial, the next 20 are aural and temporal, while the last six combine the visuo-spatial and auro-temporal qualities of the previous lines – I’ve worked this out in considerable detail in “Kubla Khan” and the Embodied Mind (PSYART: A Hyperlink Journal for the Psychological Study of the Arts, 2003).
So, a poem like LTB projects personal experience in space and time onto the poem’s structure of strings (of words) and substrings while KK projects psychological domains onto the poem’s structure of strings and substrings. In particular, I believe that the structure of domains interposed between symbolic processing (in a cognitive network) and subconceptual process takes one form in KK and a different from in LTB.
That’s still vague and abstract, though not so much as the metonymy and metaphor formulation I used over 30 years ago (Metonymic Invariance: Two Examples from Coleridge). That formulation is certainly more familiar to literary scholars, but it doesn’t have much, if any, explanatory value. This newer formulation is linked to Gärdenfors’ very sophisticated account of conceptual spaces and their relationship to symbolic processing (the actual words in the poems) and subconceptual processing (feelings of will and affect and sensory images). That conception, and much of the literature on which it draws, didn’t exist in 1981.
We’re got material to work with, new material. I don’t know where we’ll end up once we start exploring, but I’m pretty sure it will be someplace we’ve not been before. And I’m willing to be it will be interesting. At this point that’s all I’m looking for.
A Note on the Possibility of Neural Investigation
Gärdenfors’ notion of conceptual spaces (domains and their associated quality dimensions) even suggests a way of looking for neural evidence. Investigators who take a dynamical approach to the nervous system treat each neuron is one dimension in a neural space of very high dimensionality. At any given moment a neuron is in some state; that state is a value along the associated dimension. The state of the entire brain can thus be conceived of as a point in this very high dimensional space. As the mind/brain moves through time, its evolving state thus becomes a trajectory of such points.*
Now, back in the 19th and early 20th Centuries physicists figured out how to do the same with a volume of gas and so we have statistical mechanics (the neural models use math developed those phenomena). Each gas molecule was treated as a dimension in a space of very high dimensionality – actually, several dimensions for each particle, but we can ignore that. But quite a bit of useful physical can be done by projecting that high-dimensional space onto only two dimensions, which we call pressure and temperature.
Obviously we knew about and did physics in terms of pressure and temperature before statistical thermodynamics was developed. But once it was developed physicists needed a way to talk about temperature and pressure in terms of this new-fangled space of ridiculously many dimensions. Well, the notion of projection achieves that.
So, is there some remote possibility that we could do something LIKE that with the brain that’s useful in talking about, say, literary texts. Consider this email I sent to Walter Freeman, a neuroscientist at Berkeley, some years ago:
I've had another crazy idea. I've been thinking about Herman Haken's remark that the trick to dealing with dynamical systems is to find phenomena of low dimensionality in them. What I think is that that is what poetic form does for language. The meaning of any reasonable hunk of language is a trajectory in a space of very high dimensionality. form "carves out" a few dimensions of that space and makes them "sharable" so that "I" and "Thou" can meet in aesthetic contemplation.So, what does this mean? One standard analytic technique is to discover binary oppositions in the text and see how they are treated. In “Kubla Khan” Coleridge has a pile of them, human vs. natural, male vs. female, auditory vs. visual, expressive vs. volitional, etc. So, I'm thinking of making a table with one column for each line of the poem and then other columns for each of these "induced" dimensions. I then score the content of each line on each dimension, say +, - and 0. That set of scores, taken in order from first to last line, is the poem's trajectory through a low dimensional projection or compression of the brain's state space.The trick, of course, is to pull those dimensions out of the EEG data. Having a sound recording of the reading might be useful. What happens if you use the amplitude envelope of the sound recording to "filter" the EEG data?
Freeman’s response was simple:
Not crazy, Bill, but technologically challenging! Will keep on file and get back to you.
I’ve not heard from him on that matter, nor do I really expect a reply any time soon.
But if Freeman’s right, then that’s a feasible line of investigation. Someday some laboratory will undertake it. And I expect they’ll find that “Kubla Khan” and “This Lime-Tree Bower My Prison” come out very differently simply because “Kubla Khan” projects more different dimensions onto its unfolding form than does “This Lime-Tree Bower My Prison.” As for just what that difference will be, when you’re ready to run the experiment, get back to me and let’s talk.
* * * * *
*For more discussion of this way of thinking about the mind-brain, see these posts: