DISCUSSION PERIOD

A Model of the Text Generator, Michael W. Mair

A "Model of the Text Generator" purports to be an Organic Basis for Consciousness. The Project thus represents an ultimate in theoretical ambition, and the validity of even attempting such a thing is much challenged. These challenges take two forms.

The first comes from those, usually working in some aspect of brain science, who say that our present state of knowledge is such that we cannot even imagine what such a model would be like. They denigrate all such attempts to the status of a kind of after-dinner entertainment, not for serious consideration. These critics may well be right. But it is the view of this author that we do not know even that which we do not believe until we state it, that there are many stages in the evolution of a true theory, as Marcel Kinsbourne has emphasized. This paper is an attempt at a synthesis, certainly not an assertion of "how things are" in the brain and behaviour. It seeks to render visible and explicit some trends in the literature.

The second challenge is more radical, and more ancient. Basically it comes from those who insist that thought and experience have to be unamenable to accounting for by any mechanism. It is the ambition of this study to contribute to the emergence of a theory whereby the brain might be rendered "transparent," that is, completely understood in principle if not in detail. There may still remain, when and if this is done, aspects of experience which do not seem to derive from "in here" at all. This point of view will emerge during my paper. I am reminded of a recent conversation I had with a woman psychologist at a conference who, on learning what it was that I considered to be the legitimate goal of the enquiry, said, "Well., I hope you never succeed!" She never stayed to hear the argument. If she had, she might have found the conclusion a little too unphysical.

But this second challenge has been well worked over by philosophers. For example, J. D. Searle (1979) speaking at a recent symposium on Brain and mind, said of the theory that philosophy is not continuous with the empirical disciplines:

I myself think that this theory is not refuted, but just became irrelevant by the march of events. Philosophy is much more interesting today than it was twenty years ago simply because we no longer want to make a distinction between philosophical questions and other kinds. If this means that the empirical researchers are marching in on our territory, so much the better, because if they look behind them, they will see that we are marching in on their territory too.

But enough of the apologia; Kenneth Craik, writing in the 1940s, well defined the problem, and if in attempting to formulate a stage in the solution I have looked extensively to the work of Karl Pribram, it is not only because I reviewed the work of the latter for this Institute, but because truly his contribution has made of these topics a "Pribram world."

From Craik (1966) :

The adequacy of our examples and explanations of animal mechanisms and behaviour will be largely governed by the general view we take of animal life and its function. There are various theories--such as the stimulus-response theory, the theory of instincts and drives, and the theory of conditioned reflexes and modifiable responses. None of these seems to me to put the emphasis in the right place; the nature of the animal and human mind seems rather to be to copy its environment within itself in an active dynamic model which is capable of realising tendencies and

possibilities of that environment which are obstructed in this outer, world by the separation of the wrong energy-relations among the parts of that environment; and once these possibilities have been mentally realised they can often be brought to pass physically through the motor mechanisms of man and animal. This notion of life bringing to fruition the possibilities ofthings and acting as a mode of communication for the phenomena of inanimate nature seems to fit, though perhaps in a somewhat romantic way, most of the phenomena. It entails, of course, many actions which seem to be just responses to stimuli and also involves certain tendencies of inner origin which suggest instincts and drives; but if the explanation is framed in terms of these the emphasis will, I think, prove to be in the wrong place: we shall find our postulated mechanism grossly inadequate to explain these phenomena.

I will start with an assertion. "The brain generates text." What is text? Whole courses at the summer institute were devoted to approaching this question. Figure 1 is a graphical transcription of certain instrumentally derived parameters of the speech- with-movement stream with which two participants to a dialogue patterned their space-time during a 3.7-second fragment of a half-hour-long conversation recorded four years ago. The methods used to obtain both the data and the transcriptions of this and other fragments have been described elsewhere (Mair, 1977), but in brief, a "four eyes" split-screen video presentation of their faces was obtained using one video-camera and recorder, while the subjects were in elective eye-contact. Then a three-dimensional nose movement plot ("x" and "y" co-ordinates against time) was achieved by taking the position of each nose on the "x" and "y" axis of the video monitor for each one-tenth of a second time interval for each fragment. The head movement plots were drawn by the computer graphics programme "Picasso," in a format designed to be visuallypleasing. The effect is of a garden fence viewed from a first floor window, which has height ("y" axis), recedes from and approaches the observer ("x" axis), and the long axis of which is time, from left to right. Naturally, heads move in four dimensions ("x," "y," "z," "t") but this seems near impossible to transcribe or represent on a two-dimensional surface such as Figure 1. That this is the case has a resonance which transcends the merely technical aspect, and which I think of as "The Dimension Problem."

The voice transcription was obtained by using two "Laryngographs" (Fourcin, 1974), a device which delivers a plot of fundamental frequency against time from an impedence measure takenacross the larynx. My two young human subjects were thus captives, not only in the sense that they were constrained in this highly structured environment, but also in that they were physically wired up to the machines. Nevertheless, with wine, tact, and time, they would at times become quite enrapt in each other's performance, and I have concerned myself with characteristics of the "joint production" at these times. It is possible to point to certain characteristics of shared text which may occur during such states. It must be admitted, however, that an important criterion in the selection of fragments such as this for transcription was they were aesthetically pleasing. speech and movement plots are drawn to the same scale within and between interactants, and precisely aligned by visual inspection of plosives and visualisable auditory events such as the striking of an object in the view of the camera.

This is then a fragment of "The Text of Everyday Life" in Paul Bouissacl's sense (Bouissac, 1977) delivered as a voice and movement choreography. It is rather like concrete poetry. Indeed, this fragment once formed the frontispiece of a poetry magazine.

Linguists seldom attend to the nuances conveyed by speech melody. When they do, the result is scarcely if ever convincing. Similarly semioticians and others, when they analyse poetry, generate many words. Monique Leon, in a recent study of a fragment of French colloquial speech, attempted to categorise all the things one can tell, or at least thinks one can tell with high inter-observer agreement about people from the way they speak. The striking thing that a text

FIGURE 1

fragment demonstrates when plotted in time like this is that although such nuances are in a sense there (I could write, I think, for ever about the contextual surround of this fragment which feels to imbue it with such meaning), it all happened very quickly. I try to work out how the brain has done this. My method is that of the "temporal microscope." As Reng Thom has pointed out (in a lecture at this institute),there are two ways of studying a phenomenon: to collect many examples and generalise--study the "species" characteristics, or to take just one example and analyse, and then see if what one has discovered holds true for other examples. Another way of putting it would be to say that I try to put the brain back in its natural habitat (time) to see how it turns out this tremendously detailed object, the text, its natural product. From this and a very few other fragments I make my as yet very incomplete theory. To the future belongs the task of seeing whether it will work or keep on working globally.

In many words, around the time of this fragment, the male voice speaker was trying to tell his partner the location of the technical college in Cambridge. He's already told her it was near the station. She's said, "Well that's... well..." with high and hesitant pitch. He then said, definitively, to clinch it, "You know where the swimming pool is." Her answering, "No," was coquettish, to me, as if to say, "So what if I don't." At that point, he gave up this line of attack and turned towards me (I was standing around nearby) and said (still to her): "Perhaps HE can tell you where the swimming pool is." But first he had plonked the wine bottle which he happened to be holding in his hand onto the table, and then they had both laughed. This fragment is also analysed in Mair, 1977.

What is the brain doing? If one wishes to know how something works, one must know what it does. But that has always been the problem of behavioural science, its methodology, to know what data is. As Craik has said: "The adequacy of our examples and explanations . . . will be largely governed by the general view we take 11 (1966). The fragment of data that I have installed in this paper is just and only an instrumentally derived extraction from what was the case during those 3.7 seconds. If the nuances of an extended poetical description of it, even such as I have given, were "in it" rather than imposed on it (e.g., "he said, definitively, to clinch it...," or "her answering 'No' was coquettish..."), and I think the latter is at least half the reality, then they had to be active in the very limited time span during which that text was generated. Because it was all so quick, we can look to known temporal processes in the brain to see whether or if their time forms intersect with the time forms in our text fragment. Truly, we might discover "how the brain did this" (p. 43). But for them and for us, the fragment was in context, both past and future. The text actualises the moment in the context (Benson & Greaves, 1979), and as we discuss it now, or at some other time, we re-embed it in contexts of our own, some of which we try to infer as theirs; and so we generate our extended poetical descriptions.

Even when I try to describe the physical characteristics of the wave forms and timings in the fragment, I abuse the reality. I believe that one of the brain's own properties gets in the way of felicitous description. The brain makes categories out of continuities, it "isolates out" discrete entities in what would otherwise have been a continuum. So if I make a list of what the brain has done in this fragment, the elements in the list are some-what arbitrarily demarcated. If I do not create such isolates, then I cannot talk to you about it at all! Such are brains.

The Brain is: 1) Segmenting Time. This fragment has a kind of chronometer ticking in it, or more accurately, two chronometers. One is of roughly syllable duration (Allen, 1973; Mehler, 1980), and the other is of suprasegmental, or prosodic span. The rhythms of the second, for this fragment, are also fitting into the rhythm of the "plonk" of the wine bottle, and of the synchrony of the laugh. There is a supra-individual beat running through the fragment.

2) A supra-individual beat implies one timing device for two people, two brains. That this might indeed be an appropriate interpretation is seen at the temporal synchrony. Brains can worksynchronously. 3) There is melody, that is to say, fundamental frequency varies in time. The melody looks like a trajectory. Its form has a continuity through the unvoiced phonetic patterning. 4) There are phonetic micro-patternings. These form clusters of syllable length. Other work (Lieberman, 1972) has demonstrated that there are formants, i.e., bi-modal discontinuities in theidentification of phonemes with continuous change within their component frequencies. 5) There are non-articular movements, in this case, head movements as transcribed by the movements of the nose. These are absolutely integrated with the temporal process of speech. They constitute one patterning programme, one trajectory of speech with movement (Kendon, 1973). 6) There are Plans, Percepts and Cognitive Change Points. We know that there are Plans, in the sense of motor programmes, because the time forms of the speech/movement trajectories have clearly demarcated boundaries and singular shapes. Thus they are manifested plans, and must have been present in the brain prior to their engagement in real time. But also, prior to plan, there must be the percept of which the plan is a transformation. The text is an ordered concatenation of plans and percepts. It tells a story. There must be and is relevance in the ordering of this concatenation.

The sequence Percept/Plan/Manifestation (motor action) does not just occur at any old time, or in any old order. It is possible to place in time, by this technique, where the Percept/Plan transition must have come, within a very few tenths of a second, and occasionally, in other examples, within about one-tenth of a second.

All these arbitrarily demarcated parameters of text have been fitted into 3.7 seconds in this fragments. The brains did this. HOW?

Figure 2 is a three-dimensional model of the brain. It is a caricature, and like all caricatures, its construction reflects the preconceptions of the artist. It represents a kind of "conceptual anatomy," not of course the real thing--a pedagogical device for explaining my argument. It is just the sort of thing that most neuro-physiologists abhor. An extended description is in order.

We see three paired plates, forming sections of the side walls of an oval box. The two front (frontal lobes) and the two back (occipital lobes) are converging. The two side ones (an imaginary compaction of parietal and temporal lobes) are parallel. This brain, like real brains, is bilaterally symmetrical.

There is a bridge joining the two temporo-parietal plates. In real brains, it joins the paired frontal and occipital lobes to each other as well. In fact, the identification of these three paired entities is as arbitrary in conventional anatomy as it is in this caricature. The brain has no names. Here, to add insult to injury, the atrophied remnant of that magnificent structure which has excited the popular imagination so much because of the consequences of severing it has been torn away altogether so that we can "see in" to the deeper structures. That transverse connecting band is the corpus callosum.

The deeper structures, the core brain, were hard to draw. They are very complex, and poorly understood. I have a colleague who spends her working life staring down a microscope at sections of the core brain to try and have structure stand out from the tangle. Even those structures that are larger and well-established as entities in the literature, are not immune from occasional redefinition. Such a revolution in conceptual anatomy may be upon us in such matters as the supposed separate anatomical and functional reality of the corpus striatum and the limbic system.

FIGURE 2

What is worse, without even making decisions about anatomical and functional identity, just drawing the substance is hard. The thing bends at its top end; as the spinal cord goes through its transitions in becoming the core brain, it kinks forward so that all one's carefully understood topographical relations in, say, the developing story of spinal cord and brain stem, must be bent as they change to become the poorly understood structures of the core brain.

I have represented this forward kinking in my diagram. For neuro-anatomists and anyone else who cares, the hind brain (fourth ventricle, floor of same, with associated cranial nerve nuclei)might be just below the kink. The cerebellum is not there at all! It would stick out of the back of the brain stem as it approaches the kink. I have not put it in because it seems that it may just be a very sophisticated timing device for making trajectories smooth, a kind of calculus machine (Braitenberg, 1977). If this were so, then of course the absence of or damage to the cerebellum would be (and IS) disastrous for the entrainment of motor programmes. But all the same, it seems to be somewhat of a "dumb box," and can be dispensed with in our conceptual anatomy.

The aqueduct and mid-brain are located at and forward from the kink. Students of "blind sight" (Weiskrants, warrington, Sanders, Marshall, 1974) should note that the tectum is here. We areapproaching the "business" end of this business. The Edinger-Westphal nucleus, which controls the focus of the eye is around here. But if "focus" cannot focus (as in the blind) wherewith shall it be focused? Focus "in mind" and focus "in sight" are tantalisingly close concepts. Meanwhile, the cranial nerves continue to sprout from the sides of the mid-brain.

Just a little further forwards, the great tracts carrying sensation from all parts of the body get to their great relay station, the thalamus. From there, they project somatotopically to the cortical plates. This is not shown in this diagram. Nor are the great descending tracts in the "internal capsule," which go straight into the spinal cord and thus to effector organs, or to cerebellum, or both, or other. They would fit, however, sprouting out sideways from thalamus to cortex, and plunging down on either side of the thalamus into the cord, between the central kinked stem (bifid up front), and the paired "ears." How strange to caricature the brain and just leave out the fibre tracts of the internal capsule! There is an embryological justification and a new interpretation of conceptual anatomy which justifies this omission. The paired thalami, the relay stations, are here imagined to be in each side of the bifid front portion of the stem.

Now what of these ears? "Up Front," here, we have a bewildering array of clumps of cells and fibre tracts, with their inter-connections which have been conceptually demarcated into such entities as the Basal Ganglia (Caudate, Putamen, Globus Pallidus, perhaps red nucleus and substantia nigra, perhaps including even the subthalamic nucleus, the one to which damage appears to give the wild yet formed movements of hemiballismus and chorea), the Hypothalamus which is the supposed "head ganglion," of the autonomic system yet whose lateral nuclei are involved with such motor processes as the satiation of hunger and thirst; and the Limbic system, this controversial set of entities, sometimes thought of as a circuit, which are thought by some to constitute a "visceral" (MacLean, 1978), i.e., base functions brain, but to which some others, like Karl Pribram, and the author of this paper following him, seek to ascribe such elevated tasks as the generation of the text itself. This supposed system includes the hippocampus, of "memory" fame, the amygdala, the pillars of the fornix, the lateral hypothalamus, the mammary bodies, the septal nuclei, and the nucleus accumbens. For a thorough description of this system the reader is referred to Isaacson (1974). But it must be emphasised that exploration of these core brain structures is very much "hot topic" in contemporary neuro-physiology. What I have done is just to say that the core brain structures can be thought of as a circuit, lumped them all together, and drawn them accordingly. This renders explicit one trend in neuro-physiological thinking. It thereby negates others; but it makes our model make sense, if not the reality.

Action segments time. It also delivers a world that has been conceptually rearranged. My "C"-shaped core brain circuits are incomplete. They are closed by action on the world. As one acts (the down push on the pedal of the bicycle), one monitors what one has done, and it is this which, in association with an idea of direction, engenders the next action, and so on. So we pedal through time, in which we can only travel one way, and thus engender or have engendered for us the experience, the sort of time we are having in both conceptual and segmentary senses simultaneously. The circuit lays down a track in time, a text. Only the unfortunate with specific hippocampal damage can achieve the Zen ideal of "no track" (Leggett, 1977).

Is there any evidence that the brain works even remotely like this? Naturally, I say yes. Karl Pribram is one who may have had such an idea in mind since at least 1960 (Miller, Gallanter & Pribram, 1960). With his co-authors then, he speculated on the limbic system as the "processor," a terminology which although not that of a circuit, certainly gives it a bigger function than that of mediating those imagined entities, the "emotions." More recently, in a paper "How Is It That Sensing so Much We Can Do So Little?" (Pribram, 1974), he describes the way that the basal ganglia might play an active role in perception. in between these two papers we have many others from him, on "Attention," which subdivide that concept into registration, effort, and readiness/significance mediated by Hippocampus, Amygdala, and Basal Ganglia respectively (Pribram &McGuinness, 1980). Creuzfelt too (1979) writes warmly of the basal ganglia in this regard, and there is the work of R. Hassler (1978) that he cites in which that author says (of his work on cats): "the three parts of the Corpus Striatum control intentional actions (my emphasis) and not simple movements without conscious representation." Perhaps we might conceptualise the core-brain as a kind of "circuit of the self" and this notion becomes an effective organiser of many observations on cortical (as distinct from core-brain) pathology--the self in the sense of acting subject can be there, despite the scotomas,. specific incompetences, distortions of body schema, personality changes, and localised motor weaknesses of cortical damage and atrophy (Gardner, 1977). It is of special interest to note that babies with very little in the way of a cortex can do surprisingly well, sometimes for months, in eliciting by their relevant motor actions the caring behaviours in their custodians necessary for their physical survival. But lesions produced within the core brain itself produce specific, repeatable, and drastic consequences for behaviour. Again Isaacson chronicles these effects, but the phenomenon of supra-nuclear palsy (a slowing down without loss of competence) is well known. I have personally seen unfortunates with intact cortices who are awake, but do--nothing. They had had tumours scooped out of their hypothalamuses. Finally, it has been found possible (Robbins, Sahakian,.1979; Kelly, Seviour, Ivreson, 1975) to differentiate the polar opposites perseveration and hyperactivity (doing the same thing again and again, and doing many different things very quickly) by specifically placed "transmitter lesions," a kind of internal bombing of cells working with particular transmitters. The structures involved appear to be a component of the corpus striatum, and the nucleus accumbens. Perhaps, later in this paper, when we come to try and make a mathematical reality of the notion of "cognitive change point," we should put that part of our text generator circuit just there. After all, it is transmitter disturbance which characterises the "high" of LSD, the same transmitter (on the Dopamine series) which is crucial, up here, in the circuit of the self. The oors of perception might be cleansed thereby (Huxley, 1954), but participation in text, that is, in the concatenation of relevancies shared by others, is decidedly impaired.

Now what of the cognitive change points, which subjectively we might identify as the moments when we realise something. We might also realise Consciousness as the span of time and structure between these. I take this idea of Consciousness being none other than the dominant plan from Tim Shallice (1972, 1978). The fact that it involves spatial span as well as temporal segment is most interesting, and will be highly relevant when we come to discuss the Fourier transform analysis of sensory input that the cortex may be doing. As Shallice points out (1978), that little space called consciousness would in this model not be the end of the road in our search

for the springs of human action "in there." I avoid making a hierarchy of this, but that little space of consciousness, always and forever just too short to be caught by itself (a finite limit to introspection) is evidently but the tip of the iceberg of a vast organisation.

But as distinct from Shallice, who would seem to see a kind of competition occurring on the cortical surface for access to the "action system," between plans (and Mountcastle 1979) argues similarly when he writes of the "Command Function"), I am arguing, after Pribram, that it is the core-brain which accesses, selects, and delivers from the diversity of informations present in the parallel processing cortex (and employing and validating from the "frontal stacks") the concatenation of relevancies that constitutes the text, segmenting time as it does it. From Craik, it may be the cortex which has the "active dynamic model" of the environment, and the core-brain which realises the possibilities in the relations between parts of that environment, thus bringing them to pass physically through the motor mechanisms (Craik, 1966).

What of this parallel processing cortex, where everything is going on all at once? The reciprocal arrows from core-brain to cortex in Figure 2 are not thalamic projections or cortical projections to the spinal cord. They are meant to symbolise this reciprocity whereby the core-brain can access selectively. Accessing selectively can be interpreted as either turning off diversity or letting in singularities. Hence the reciprocity in the arrows. It is quite possible that this reciprocal relation to the cortex is not everywhere in the core-brain. My guess would be that it is the property of the Corpus Striatum.

In his book on the limbic system, Isaacson (1974) calls the cortex "The Guru" because he stresses its abilities in the manipulation of time. Certainly the human brain has most expansion in those cortical areas--frontal and temporal lobes--which deal in plan and memory, and I have suggested elsewhere (Mair, 1980) that the cardinal human innovation might be a "time trick"--the ability to spend present time organising things for another time. Although the mechanism of text generation might be the same in man and animals, animals are quite constrained in their exigeses by their immersion in their physical environments. They can go places and do things, that is all. Even where they would appear to make rudimentary tools (Van Lawick Goodall, 1971), they get thrown if the context-embedded motor sequences are broken, e.g., the phenomenon that Michael Chance has discussed as the "recombinatorial" abilities (Chance) of man as distinct from animals in the construction of motor programmes for dealing with the world. Rene Thom (1975) has christened this physical-context-free world of ours the Virtual World, but he stresses: "If man has escaped from the fascination of things through the use of language, he remains under the fascination of action incorporated in the grammar of language (a verb conjugates etc.)" (p. 311). In a sense, in the arbitrary sign system (in Saussure's sense) of language (1966), our species would appear to have turned out a brain product into the world (or rather a great number of such systems of brain products) and then continued to live within these diverse logico-semantic systems (or "Cultures"). But as far as the humble processor is concerned, this sophistication of the human virtual world and cortex which gives us the ability to recognise a sharp phenomenological difference between classes of sign (e.g. Peirce's Index., Icon, and Symbol [Peirce, 1940; Savan, 19761) might not be relevant to the mechanics of access, assembly, and delivery which constitute the cycle of the text generator in man and animals alike.

On the question of "Localisation of Function," which was the topic of Karl Pribram's address to he colloquium of the summer institute, we can see that the access problem and the localisation problem are likely to have a joint solution. Looking once more at Figure 1, the Swimming Pool Fragment, it is evident that the speaker had to find "swimming pool" as a concatenation of articulatory targets very quickly, such as to deliver it as an intonational form, a trajectory. similarly, his partner had to match this against some version of a logico-semantic system. Peter

Salus has referenced some work which suggests that a search procedure (looking for a word) is self-terminating. It's like a little quantal "pulse" of search. Then one must try again.

A hologram analogy is appealing aesthetically. It would explain why there can often be such a sparing of function with extensive cortical damage. It shows how classes of function might be lost--there can be a regionalisation of function with the hologram model (e.g., it is true, as Karl Pribram pointed out, that with loss of the "face area" in the Right Inferotemporal cortex recognition of all faces is lost, not just certain particular faces [Hecean & Ajuniaguerra, 19521). And in the matter of access, it would certainly facilitate matters to have, within a regional specialisation, all data simultaneously co-present and at the same place. But then notions of space and time do tend to become subjectively less comprehensible as one ascends each level of integration in the calculus. (Like musicians, mathematicians start young!) However, we must never lose sight of the destination of our argument--to understand the text generator. Although the model may work by differential topology--although this and other mathematical theorems may describe the time courses and the interactions of the processes involved, the product is text. Where text is incomprehensible to everyday subjective understanding, then text and generator have failed, in all senses.

In addition to these circumstantial advantages of a hologram theory of cortical processing, there is the hard data with which Pribram and others have backed up this model of cortical (not core, note) function. This evidence was convincing, but would need separate review. As Pribram put it, the battle in this argument is between the "Feature Creatures," those who think that a kind of componential analysis is performed on the world in the cortex, and that perception is a kind of reasserably, and the "Frequency Freaks" of which he is one, who think that a Fourier Transform analysis of both auditory and spatial frequency is performed on the world, and that perception is a kind of "tuning in." Personally, I have seen spatial frequency analysis demonstrated for the visual system, and am convinced that this principle plays at least some part in the operation of the text generator.

In summary of this section on the anatomy and physiology o'f the generator, we have outlined conceptually if not in nerve fibre and electricity a device which could generate text. It consists of a dynamic parallel processor (the cortex), and an assembler and projector (the core). So we swim into time, like time fish, leaving behind the debris or order of our texts. As we swim, we move in space as well, either in the architecture of the environment, or the models of the virtual world which we span in our arguments. I would like to conclude this paper with an attempt to apply Rene Thom's catastrophe theory to my model of text and generator. This is of course the most ambitious part of the enterprise --and the most speculative. But catastrophe theory is at least a very apt metaphor for the process of text elaboration that I have observed, and would seek to understand. It "looks pretty." The suspicion remains, however, and it is just a hint from the aptness of it, and the timing of it, that like Fourier transform for the cortex, and theorems of structural stability and morphogenesis may be more than a metaphor in the circuit of the core.

The Catastrophe Theory Metaphor for the Process of Text Generation: The Predator Prey Cycle

Rene Thom provides an application for his catastrophe theory in the metaphor of the cycle of predator and prey. I find, at the level of text, that I can present a model of the process of its generation in this way. First, I must present the diagram of the process of text generation which is in the final paper of Steps towards Principles of Text Regulation (Mair, 1978). The presentation is similar in format to the voices in Figure 1, but it has been both improved, and denatured. It is improved in that the female voice "NO" has been extended into an up-going

melodic form, to complement the shape of the down-going male voice trajectory. It has been denatured in that there are no phonetic micro-patternings--a "song without words."

FIGURE 3

Now that this artificial stylisation has been done, we can discuss the chief conceptual elements in the presentation without the noise of mere contingency. We see that we have a sine wave, the nodes of which are punctuated with a Percept/Organise/Plan sequence, deriving alternately from one participant and then the other. Real life dialogue is seldom like this. A musical taxonomy might best describe the diversity of overlappings, simultaneities, interruptions, and ruptures, etc., etc., each of which is the unique characteristic of a text episode.

However, when the process is stylised and rendered accessible to visual comprehension, we can see that the metaphor of capture and captured (Predator and Prey) is apt. In Tim Shallice's sense (1978), the immediate content of consciousness of the one comes to constitute the immediate content of consciousness of the other. THIS HAS TO HAPPEN IF THERE IS MUTUAL UNDERSTANDING. Of course, there can often be interaction without contentful, shared inter- change (and Roch Lecours used the example of "Glossolalia" in his presentation at the colloquium). Incidentally, songs without words need not always be empty (Bouissac, 1977).

Let us explore this metaphor more fully. Catastrophe theory is an extension of the theorems of differential calculus which have proved so manifestly successful when applied to such physical sub-strata as the orbits of planets and the trajectories of projectiles. Naturally such applications must allow for the perturbations of ideal trajectories contingent on the imperfections of mere substance--mass maldistributions, wind resistance and so on. Rene Thom is decidedly coy about the point where analogy can be homology in the application of catastrophe theory. If we have a theory which behaves in a similar way to a process in the environment, it does not follow that the environment is working off that theory. Is everything that snaps a catastrophe? An application so general must be vacuous. However, let us continue. I do not have the tools to elaborate this analogy/homology argument further. Thom's argument is that: (1) feeding implies predation, (2) predation implies capture. He represents this as one of his diagrams of an archetypal morphology, which may be more than just a picture--in some sense a transcription of a mathematical transformation within the theorem (perhaps a theorem is something proven which might not apply, a theory something which one can apply, but which has not been proven).

The idea of a diagram being a transcription from a theorem is illustrated in the top drawings of Figure 4. Admittedly, this copying is not quite faithful to Thom. In particular, the lines should splay out from the centre of the round figure, not half way up. Now the idea seems to be that these

shapes are the graphical manifestation of mathematical absolutes. So that the idea of a discontinuity in a system achieves precise expression in them.

Figure B is a projection of Figure A. Thus as one travels in some idealised mathematical realm in the circuit 1, 2, 0, 3, 4, 5, 1, 2, ... etc., one has thereby experienced/generated a discontinuity drawn by yet another mathematical transcription (a kind of taking of cross-section), and sub-figures one through five tell the story. They are sections, in a sense, of Figure A.

To understand it, it is helpful to fill the basins in the sub-figures one through five with an imagined fluid. At: 5. All is well, there it is in one place. 1. Trouble. The singularity at that one place is being distorted. Just after 1, 2. and quite definitely by 2, it is clear that there are TWO basins. It is the fact of the sudden development of this bi-modal condition which constitutes the "catastrophe," as in Figure A, when proceeding downhill, one glides, or suddenly one comes to an edge. 3. At 3 the bimodality is being distorted, so that at 4. Quite suddenly, singularity is re-established once more. All the water is 5. Back in the same pot. So we have a cycle here which generates a discontinuity, or rather two such. This is the most elementary of the elementary catastrophes, of which there are only the finite number of seven. And it's getting worse! Let's try and tell the story of Thom's metaphor in easy terms.

Imagine a Fat Cat. There it is at 5, digesting and content, until the biological cycle impels it to seek prey. Then something troubles its sight. It becomes obsessed by the idea of mice. As it approaches point 1, it in a sense has lost its identity to the very idea of mousehood. It is the prey of the rodent species. Its nervous system has become (in Thom's terms) an "Organ of Alienation" for it. Ah-ha. Then it spies a mouse. In an instant, it is itself again, in the sense of "Here am I. and there's the mouse." It has an Instantaneous Cognito experience. This occurs at point 1. It is Perception. As it leaps to engulf the poor creature, the singularity of the mouse for it, though remaining present, diminishes until the second instantaneous event, of Capture, occurring at 4. Thereafter, it resumes its career of the Fat Cat, until the instability endemic to its nature manifests once more.

I have noted that in interaction there is in some conceptual sense at least a cycle. One turn round a cycle generates a sine wave. I have illustrated this in the lower figures "C" and "D" of Figure 4. I have also noted that in speech, over the apex of the wave, a "POINT" gets made. In the terminology of the linguist Halliday (1967), tone groups are so demarcated, information points.

I give an example of this from the second paper of Steps (Figure 4D). The percipient reader may have already spotted what I am trying to do, but I am mixing the level of Analogy with the level of Homology. Speech delivers percepts. The Point of Expulsion of a "POINT" in utterance may be metaphorised by position 4 in Rene Thom's circuit. But what got predated was this consciousness content of the other person. He, like the cat, was quite alienated into the idea of getting the other person's idea, as the motor programme went across. We might identify this stage with "Listening." At 1 in their own personal cycle, however, he grasped it, got the percept, or perhaps one should say, "got the picture." The latter conception is more apt, because the moment of perception is here being imagined as the "rush" into a stable visual percept, of which the details--the rest of the tone group--are filled in later. The idea is that the speech of one does this to the consciousness of the other--delivers a vision--more or less synchronously with the actual details of the melody. This would only be so for the "enrapt" diad. We have the principle of supra-individual beat in a text episode explained as an idea of the direct manipulation of the circuit of the generator of one by the time form of the melody of the other.

FIGURE 4

I have adapted Figure 3 for this model in Figure 5. The diagrams corresponding to points 1 through 5 in the Percept/Organise/Plan/Act sequence are to be read from Figure 4. one can see that while one "Fat Cat" rests, a kind of "refractory period" in Craik's sense (and see Figure 2), the other is being entranced, grasping the percept, and then, through point "O" capturing, i.e., expelling the intrusion of the content of consciousness of the other from her own. This is her own succeeding utterance. Of course, one can be captured by another, be unable to "cap" the argument, be persuaded or charmed, and the model easily accounts for this. Finally, I am suggesting that the form of the melody of the other drives one's own core-brain processor. So that's how we get to hypnotism. when you control the circuit of the self, YOU "have" the other person completely.

FIGURE 5

We have, then, a model of text and generator, which perhaps does justice at least in its scope to its subject. As Professor Stephen Hawking (1980) said in his inaugural lecture to the Chair of Physics at Cambridge, "However it is too difficult to think up a complete theory of everything all at one go (although this does not seem to stop some people. . . )." Yet there is a fundamental schism in the human sciences between those who believe in the possibility of a general theory at all, however remote its achievement might be, and those to whom the very idea is anathema. Of what are they so afraid? Whether this theory be right or wrong, or just naive and disorganised, the attempt was there, it seemed worth trying. What are they, who deny the right to try, so afraid of? They need not have worried. Hawking (1980) remains convinced that space-time is four-dimensional. It seems likely that our text generator is for processing and generating four-dimensional shapes. That is the experience, and there is nothing in this model in more than four dimensions. But ultimately, the model itself, even if right does not generate human reality. For that, as it was for the original limbic time fish, one must Project (Urion, 1978).

Addendum

It has been pointed out to me that one cannot claim any sort of application of Catastrophe theory unless one can label axes on a graph of one of the elementary catastrophes themselves. In this addendum, I shall attempt to do this for the application of the cusp catastrophe in my model of the process of text generation. In so doing, it has been gratifying to find how the model describing the discontinuities in consciousness can be found embedded in the wider cultural matrix from which convictions derive. Also, the model predicts, when so labelled, that text generation is only possible within a certain physiological range (of values in x, y). Some of the organic brain states can be seen as being outside this physiological range, and mimic the outcomes of interaction when these too deliver the individual to states outside the range within which text generation can occur.

I have termed these finite number of discrete motor patterns (such as laughter, tears, violence, embarrassment) "Tropisms" else-where, to emphasise that when the shared text has entered in on one of them, they tend to proceed to completion. It is as if the configuration of elements which constitutes each immediate shared state of play has a certain internal tension which resolves into the next state of play, which is then established by motor action either of speech or movement direct action, in both cases, but on a model which is virtual, or physical, or an intermingling of these analytic levels, as is usually the case in human text). This is the concatenation of relevancies, but if any one of these states of play transgresses certain limits of instability or over- stability, there is no more concatenation but--one of the "Tropisms" ensues. I represented this diagrammatically in Steps Towards Principles of Text Regulation (Mair, 1977, p. 28).

An axis vertical to the paper = Time ie, continuing argument (between Tropisms)

N.B The identification and demarcation of "Tropisms" is at present a little haphazard

There was no discussion of pitch range in this "model of the text generator" paper. There were suggestions about how pitch change creates rhythm, about how the "point" of an utterance comes

with the going over an apex of pitch wave (usually), and that how the identity of rhythm and creation of the discontinuities in states of play in the mind of the other might together approach an explanation of such phenomena as "charm" (in its performance aspect), or even hypnotism. But it is the characteristic of longer sections of text to show a variable range over which such rhythmic and syntactic functions of pitch are enacted, and in earlier papers I depicted a high pitch excursion-~a "going over the top of the melody"--as a bid for topic control (Mair, 1977, p. 42). The suggestion is that, in naive speech such as the speech of children at least, the stability or instability of a cognitive state of play--the obstacle that one must overcome in shifting it for example--does not only depend on the immediate contextual and longer-term cultural values of the elements of its constitution, but also on the range in pitch with which it is delivered. In interaction, there IS always continuity in the melody across participants, and a winning argument, unless asymmetric status relationships have skewed the bias, WILL always recruit a melody asymmetric to and which completes that of the vanquished opponent. This retrospective predictability is perhaps the most disconcerting insight gained from prolonged study of single real-time texts. It needed the concept of "Projection" to escape from that impasse, and thus, of course, opened up the epistemologically more serious problem that the theory now has with the concept of "dimesion".

The axes of the three-dimensional cusp can be labelled for dimensions of "x" (commitment), "y" (pitch or energy), and "z" (change in state of play) to produce a syntheisis which transends these divers theoretical issues.

It has to be noted that the parallelogram x, y is the control space of the third cusp, its representation minus one dimension. Thus the dotted line circle is meant to be the cycle of normal interaction as portrayed in Figure 4, which has a fast phase (of the discontinuity) and a slow phase (of recovery, or refractory period), but his only holds for a certain range of values "x" and "y".

Suppose we now apply the suggested labels to these axes, we can consider some hypothetical existential predicaments. If "x" is of a value below the cusp point at all parts of the cycle, the

dotted line at no point generates discontinuity. There is no state of play delivered. Action depends on coherent plan. Such a cycle would not generate discrete actions at all. Behaviourally, this id "Dither". Extreme hyperactivity is like that.

If one person has a high value of "x" in a state of play, then the other, travelling in "y" with his pitch excursion, might not achieve the necessary value of "y" for that value of "x" to reach the cusp fold. Thus the state of play of the first would remain unchanged. Here our model approaches theoretical adequacy for such matters as "conviction in the voice." A higher value in "Y" (pitch) for the speaker would perform the necessary "flip" in the recipient. It must be noted that this way of argument involves a duplication within each self. Each is recipient and projector in a fashion which interacts internally, and alternately with the other; and becomes the other in turn, the internalisation of the other in the self. Such duplication is convenient for the explication of "inner speech," as well as being necessary for the theory of interaction. It is of interest that Rene Thom himself proposes a duplication of cycles (motor and sensory).

If the value in "x" is so high in one interactant that it "cannot be reached" by pitch excursion in the other ("y") such as to change the immediate state of play, then we have confrontation, and the tropism of "Violence" if topic is continued. Action continues, without change in cognitive state of play. Such a synthesis employs the same model for the textual consequences of changes in "x" and "y" as variables which are at once personal and organic.

To state the argument a different way, one can say that normal interaction can only proceed while the individual is operating within a defined range of potential commitment and energy, the actual position in this range for any immediate state of play being cultural and contextual; and that normal interaction is the experience of an individual in that range as manipulated in and by the text. The text moves the individual within the physiological range. Where the text itself moves the individual outside that range, we get Tropisms. But for an individual whose system is already outside that range, the text is one long tropism, and one has witnessed some unfortunates for whom this would be an apt description; as, for example, the extreme agitated schizophrenic (low "x," high "y"), the depressive (low "x," low "y"), the psychopath (high "x," high "y"), and the perseverating Parkinsonian (high "x,'. low "y"). Doubtless, this is all too neat, there are other dimensions and variables involved. But this paper is only an exploration! The author has already exceeded the limits of his competence.

Note

I am grateful to Gereth Leng for some discussion of these ideas. He cannot be held responsible, however, for the mathematical naivete which might remain. I have also benefited greatly from discussion with Jane Adam and Solome Fester. I am grateful to Eileen Ghosh for her sympathetic typing.

Cambridge, November 1980

References

Allen, C. I. "Segmental Timing and Control in Speech Production." Journal of Phonetics, 1 (1973), 219-37.

Benson, James D., and William S. Greaves. "Field of Discourse: Theory and Application." Paper delivered to the 6th Systemic Workshop, Cardiff, September 1979--York University, Toronto.

Bouissac, Paul. Circus and Culture. Bloomington: Indiana University Press, 1977. (See pp. 81, 90, 196 for definitions of "text.)

------ "The Timeless Tools of Time." Paper presented at Wener-Grenn Symposium--Vienna, 1977.

Braitenberg, Valentino. On the Texture of Brains. Berlin and New York: Springer Verlag, 1977. (See p. 76.)

Chance, M. R. A. Personal communication: "Combinatorial Competence." Sub-department of Ethology, University of Birmingham, UK.

Craik, Kenneth. "The Nature of Animal Behaviour." In The Nature of Psychology. Ed. S. L. Sherwood. Cambridge: Cambridge University Press, 1966.

Creutzfeldt, 0. D. "Neurophysiological Mechanisms and Consciousness." In Brain and Mind. Ciba Symposium, 1979.

Fourcin, A. J. "The Laryngograph." In Ventilatory and Phonological Control Systems. Ed. B. Wyke. London: Oxford University Press, 1974.

Gardner, Howard. The Shattered Mind. London: Routledge and Kegan Paul, 1977.

Halliday, M. A. K. Intonation and Grammar in British English. The Hague: Jan.ua Linguarum, Mouton, 1967.

Hassler, R. "Striatal Control of Locomotion. Intentional Actions and Integrating and Perceptive Activity." Journal of Neurological Science, 36, 187-224.

Hawking, S. "Is the End in Sight for Theoretical Physics?" Inaugural Lecture, Chair of Physics--University of Cambridge, April 1980.

Hecean, H., and J. de Ajuniaguerra, et al. "Le Probleme de 1'agnosie des physionomies encigphales." 4 (1952), 322-55.

Huxley, A. L. The Doors of Perception. London: Chatto and Windus, 1954.

Isaacson, R. L. The Limbic System. New York: Plenum Press, 1974.

Kelly, P. H., P. W. Seviour, and D. Ivreson. "Amphetamine and Apo-morphine Responses in the Rat following 6-OHDA Lesions of the Nucleus Acumbens Septi and Corpus Striatum." Brain Research, 94 (1975). (See p. 507.)

Kendon, Adam. "Studies in Dyadic Communication." In Social Communication and Movement. Ed. M. von Cranach and I. Vine. European Monographs in Social Psychology series, 1973.

Lecours, Andrg Roch. "Glossolalia." Paper presented at the colloquium, "The Neurological Basis of Signs in Communications Processes," Victoria University-~Toronto, June 1980.

Leggett, Trevor. "Translations of Japanese Zen Texts." In The Tiger's Lair. London: Routledge and Kegan Paul, 1977. (See p. 31, final paragraph.)

Leon, Monique. "Semantique et intonation." Unpublished MS. Victoria University, Toronto.

Lieberman, A. M. "Perception of the Speech Code." In Human Communication: A Unified View. Ed. E. David and P. Denes. New York: McGraw Hill, 1972.

MacLean, P. D. Article in Journal of Nervous and Mental Diseases, 1 (1958), 1-11.

----------"A Mind of Three Minds: Educating the Triune Brain." In Seventy-seventh Yearbook. National Society of Education: University of Chicago Press, 1978, pp. 308-42.

Mair, M. W. "The Melody of the Text." Review of Circus and Culture by Paul Bouissac--Indiana University Press. Semiotica 1980 (forthcoming).

--------- Steps towards Principles of Text Regulation. Working papers and prepublications--Toronto Semiotic Circle, Victoria College, 1978.

Mehler, J. "Language Disposition in the Infant." Paper delivered at the colloquium, "The Neurological Basis of Signs in Communication Processes," Victoria University--Toronto, June 1980.

Miller, G. A., E. Gallanter, and K. Pribram. Plans and the Structure of Behaviour. New York: Holt, 1960.

Mounteastle, Vernon B. "An Organising Principle for Cerebral Function." In The Mindful Brain. Ed. G. M. Edelman and V. B. Mountcastle. Cambridge: M.I.T. Press, 1979.

Peirce, C. S. Philosophical Writings of Peirce. New York: Harcourt, Brace, 1940.

Pribram, K. H. "How Is It that Sensing so Much We Can Do So Little?" In The Neurosciences Study Programme III. Ed. F. 0. Schmitt and F. G. Worden. Cambridge: MIT Press, 1974.

--------- and D. McGuinness. "The Neuropsychology of Attention." In The Brain and Educational Psychology. Ed. M. C. Wittrock. New York: Academic Press, 1980.

Robbins, T. W., and B. J. Sahakian. "'Paradoxical' Effect of Psychosomatic Stimulant Drugs in Hyperactive Children from the Standpoint of Behavioural Pharmacology." Neuropharmacology, 18 (1979) , 9 31-50.

Saussure, F. D. A General Course in Linguistics. Ed. C. Bally, A. Sechehaye, A. Reidlinger. Trans. Wade Baskin. New York: McGraw Hill Paperbacks, 1966.

Savan, D. An Introduction to Peircian Semiotics. Working papers and prepublications--Toronto Semiotic Circle, Victoria College, 19 76.

Searle, J. D. Discussion Secti'on: "The Historical Background." In Brain and Mind. Ciba Foundation Symposium 69 (new series). Excerpta Medica, 1979. (See p. 39.)

Shallice, T. "The Dual Function of Consciousness." Psychological Review, 1979, pp. 383-93.

-----------"The Stream of Consciousness--Scientific Investigations into the Flow of Human Experience. In Information-Processing Approach to Consciousness. Ed. K. S. Pope and J. L. Singer. New York: Wiley, 1978, pp. 117-57. (See esp. pp. 133, 147.)

Thom, R. Structural Stability and Alorphogenesis. New York: Benjamin, 1975. (See pp. 297 et seq., also 311.)

Urion, Carl. "Control of Topic in a Bilingual Speech Event." Diss. Alberta 1978.

Van Lawick Goodall, Jane. In the Shadow of Man. London: Collins, 1971.

Weiskrants, L., E. K. Warrington, M. D. Sanders, and J. Marshall. "Visual Capacity in the Heinianopic Field following a Restricted Occipital Ablation." Brain, 97 (1974), 709-28.