|
Chaotic Logic -- Copyright Plenum Press © 1994 |
I have analyzed the mind as a collection of interconnected, intercreating processes; and I have proposed that the overall connectivity structure of this collection is that of a dual network. The dual network structure, however, is extremely flexible; it encompasses many possible patterns of connectivity. One parameter which varies widely among these possible connectivity patterns is the degree of modularity.
Fodor (1987) has argued that human perceptual processes are strongly modularized, in the sense that most vision processes need connect only with other vision processes, most hearing processes need connect only with other hearing processes, and so forth. The origin of this modularity is as yet unknown -- some of it probably results from straightforward genetic programming, but the greater part of it may well self-organize as a part of the infant brain's growth process. Neural Darwinism suggests that, if there did arise significant connections between low-level vision processes and low-level hearing processes, these connections would quickly disappear due to lack of utility.
In a similar way, it is quite possible for higher levels of the dual network to become modularized. In this chapter I will use the word dissociation to refer to modularization which occurs as a result of childhood or adult mental dynamics, as opposed to modularization which is present in the brain at birth. For instance, I will talk about personality dissociation -- dissociation involving subnetworks involving significant portions of the self/reality belief system.
This is an abstraction and generalization of the revolutionary concept of "dissociation" introduced by psychologist Pierre Janet around the turn of the century. For nearly three quarters of a century Janet's work was ignored, assumed to have been superseded by psychoanalytic ideas. In 1970, however, Ellenberger's Discovery of the Unconscious rescued Janet's work from obscurity and brought it to the attention of the psychological community. Since that time, dissociative phenomena have been studied with increasing vigor, mainly in the context of multiple personality and post-traumatic stress disorder; and they have been related with modern brain science in a satisfying way (Modell, 1980; Mitchell, Osborne and O'Boyle, 1985).
Here I will integrate the classical concept of dissociation with the dual network model and the cognitive equation, thus arriving at a striking new framework for understanding mentality, one which synthesizes and (hopefully) clarifies all the ideas of the previous chapters. The central claim of this framework is that partial personality dissociation is central to the formation of structurally conspiratorial belief systems; which are in turn essential to productive, creative logical thought. Or, in a formula: no powerful intelligence without strong internal conflict. What might at first seem an obstruction to logic, is in fact necessary to the evolution of useful logic-guiding systems within the mental network.
This synthetic framework serves to bring the abstract psychology of the previous chapters closer to everyday human life. For this reason, it should be of interest not only to psychologists, but to anyone concerned with better understanding their own mentality. It gives, for perhaps the first time, a sensible idea of how the creative diversity and perversity of human personality might emerge from the evolutionary dynamics of neural pathways.
And in the final section, I will argue that this framework is pregnant with implications, not only for human psychology, but for engineering. I will propose a new kind of computer science called A-IS, or artificial intersubjectivity. A-IS centers around the idea of programs which socially interact with one another and hence develop interrelated, dissociated personality structures. Only in this way, I contend, could computers ever simulate or supersede the wonderfully chaotic reasoning of the human brain.
12.1. MULTI-MENTALITY
'The World is One!' -- the formula may have become a sort of number-worship. 'Three' and 'seven' have, it is true, been reckoned as sacred numbers; but abstractly taken, why is 'one' any more excellent than 'forty-three,' or than 'two million and ten'?
-- William James
This quote is humorous, but at the same time it makes a very serious point. "Uni-" means one, so that the very word universe conceals a philosophical presupposition. Why should unity be a fundamental character of the world? Who says the world doesn't have diversity, rather than unity, at its core? Why not a multiverse, rather than a universe?
What form might a multiversal world take? William James was interested in subjective realities -- in the "semi-real" realities perceived by individual minds. He wanted to understand the world as an indeterminately large group of interacting, intersecting subjective realities. And he wanted to tie this in with the pragmatist idea that only observable properties are real. But he was disappointingly vague on the details.
By replacing the word "World" with the word "Mind" in James's quote, one obtains an equally valid bon mot:
'The Mind is One!' -- the formula may have become a sort of number-worship ... why is 'one' any more excellent than 'forty-three' or 'two million and ten'?
William James broke new ground with his theory of the "stream of consciousness"; he was also one of the first to seriously question the unity of mental experience. In the "stream of consciousness" metaphor, he did not rule out the possibility of rocks or islands in the stream, breaking up the flow into several distinct pieces.
Perhaps the deepest-ever insight into the fundamental diversity of the psyche was achieved by the novelist Fyodor Dostoevsky. In The Idiot, for example, the angelic but tragically unstable Prince Myshkin represents an aspect of Dostoyevsky's own consciousness. Myshkin thinks only good of other people; his only desire is to help. Now this certainly does not describe the man Fyodor Dostoevsky. But Dostoevsky felt and acted thisway at certain times; Myshkin was one of his subpersonalities.
And in The Brothers Karamazov, the four brothers Ivan, Dmitri, Alyosha and Smerdyakov may be understood to represent separate "voices" in Dostoyevsky's mind, independent "streamlets" of Dostoevsky's consciousness. Alyosha is a less pathological Myshkin, the Myshkin sub-personality tempered by the realism of the rest of Dostoevsky's mind. Dmitri is a sensualist, a confused womanizer and gambler; Dostoevsky, under the influence of his Dmitri aspect, gambled his savings away many times. Ivan is a writer and philosopher, tirelessly agonizing over the problem of God in the modern world. Finally, the half-brother Smerdyakov represents the "worst of Dostoevsky," the evil, petty, vindictive, cunning sub-personality that all of us possess to some degree. As his diaries suggest, Dostoevsky viewed his own life as a constant struggle between these various sub-personalities, these competing modes of consciousness.
In recent years, psychologists have rediscovered this Dostoevskyan notion of multi-consciousness. Multiple personality patients like Sybil and Billy Milligan are virtually household names. And several psychological theorists have proposed that the kind of "dissociation" apparent in multiple personality is different in extent rather than kind from the mental dissociation observable in the ordinary person.
This is what Somerset Maugham meant when he wrote
There are times when I recognize that I am made up of several persons and that the person that at that moment has the upper hand will inevitably give place to another.
Maugham did not have multiple personality disorder -- each of the "several persons" making up his psyche was aware that its name was Somerset Maugham, and was aware of most if not all of the experiences had by the other "persons." But Maugham, like Dostoevsky, was a good enough self-observer to recognize that his mind was to some degree dissociated; that it consisted of several largely disconnected "functional personality units."
Ronald Fischer (see McKellar, 1979) reports experiments in which people are asked to memorize material under the influence of alcohol. When sober they exhibit poor recall -- but when given alcohol again theirmemory improves. This illustrates the phenomenon of "state-dependent memory." John does not have one unified memory -- drunken-John has his own memory, as does sober-John. The experience reported by Somerset Maugham is one step beyond this -- several personality units, each possessing its own "unit-dependent memory" as well as access to a shared memory store. And multiple personality disorder is but one step further: an amazingly large part of the shared memory store is divvied up among the various independent personality units.
The fundamental multiplicity of mind and world was expressed beautifully by the Russian philosopher Mikhail Bakhtin in his masterwork Problems of Dostoevsky's Poetics:
It should be pointed out that the single and unified consciousness is by no means an inevitable consequence of the concept of a unified truth. It is quite possible to imagine and postulate a unified truth that requires a plurality of consciousnesses, one that cannot in principle be fitted into the bounds of a single consciousness, one that is, so to speak, by its nature full of event potential and is born at a point of contact among various consciousnesses. ...
Not a single objective world ... a plurality of consciousnesses, with equal rights and each with its own world, combine but are not merged in the unity of the event.
12.2. DISSOCIATION AND THE DUAL NETWORK
As we have seen, the concepts of multi-reality and multi-consciousness are far from novel; they date back at least a century, to Janet, Dostoevsky and James. Up to this point, however, these ideas have not received a systematic theoretical analysis. I suggest that the dual network model provides the key to understanding dissociative psychological phenomena.
Recall that the dual network model analyzes mind in terms of two semi-autonomously functioning networks: an associative memory network, which self-organizes itself according to the principle that related entities should be stored "near" each other; and a perceptual-motor hierarchy, which operates according to the multi-levellogic of a flexible command structure. And it makes the central hypothesis that these two networks are superposed.
This superposition implies a roughly "fractal" structure for the associative memory network. And, more to the point, it implies that, if a section of memory is somehow split off or "dissociated" from the rest of memory, then a section of the mind's control network is also split off, as an automatic consequence. This explains, in one immensely simple step, how the attempt to suppress unpleasant memories can lead to the creation of an autonomously acting and remembering psychological unit. In other words, as will be shown in detail below, it explains the basic phenomenon of traumatic memory and post-traumatic stress syndrome.
Multiple personality is a little more complex: it has to do with the self, an intricate self-referential construction and a complex belief system. However, we may make a few simple observations. Post-traumatic stress syndrome is often a consequence of a single painful event -- e.g. watching a close friend die a bloody death. Multiple personality, on the other hand, is generally a consequence of repeated painful events, usually beginning in early childhood. Very often these events are incestual rape, or severe child abuse.
In post-traumatic stress syndrome, the painful event usually occurs after the person's self is formed. The person already has a unified self-image, so if his mind wants to shut off offending memories, it has to shut them away from the well-formed self. In multiple personality, though, the painful events occur while the person's self is still forming. Therefore, the "split off" memories are subjected to the self-formation process, just as much as the rest of the dual network. While not a complete explanation, this gives some idea of why multiple personality disorder should exist, and why different types of traumas should give rise to different psychological problems.
12.2.1. Dissociation and the World
On a more philosophical level, the dual network perspective makes clear that there is not so much difference between
1) the various personalities of a person suffering from multiple personality disorder (MPD)
2) the various personalities which exist in the world
3) the various sub-personalities of a normal person
Just as MPD results from the splitting-up of a single person's "dual network," so do individual personalities result from the splitting-up of the universal dual network. This idea unifies Dostoevsky's psychological idea of multi-consciousness with James' philosophical idea of a multiversal world. It is a dramatic conclusion -- but at the same time it is a new beginning. For it opens up a whole new way of looking at the mind and world: as multiple phenomena.
Far from being isolated pathologies, dissociative mental disorders are natural and necessary features of mental life. In other words, all mental action is a kind of interplay between different "personalities" -- different semi-autonomous agents, which help to mold one another's reality, which possess individual "senses of identity," which partially share the same memory, and which compete with one another for attention.
Aside from traumatic experiences, what might cause a section of the dual network to split off and become semi-autonomous? The answer to this is surprisingly simple. Two things only are required. In order to split off and survive on its own, a subnetwork must first be complete in itself, in the sense of being a strong attractor of the cognitive equation. And second, it must have relatively few connections with the other parts of the mental networks -- otherwise its autonomy would not last.
From this description it should be obvious that a dissociated subnetwork has something in common with a structurally conspiratorial belief system. The difference is not absolute, it is one of degree. A "subnetwork" is expected to have a more marked dual network structure than a belief system, which may contain few levels and display the dual network structure only to a small degree.
One way to distinguish the two is with depth-to-breadth ratio. A belief system tends to involve a sizeable collection of beliefs on roughly the same level of abstraction -- the same level of the hierarchical mental network. It is "shallow" but "broad." On the other hand, a dissociated subnetwork like a subpersonality tends to span a fairly large number of different levels of the hierarchical network; its depth exceeds its breadth. In other words, a dissociatedsubnetwork contains all the levels needed to do, whereas a belief system only guides other systems in doing.
12.2.2. The Subtlety of Personality
A dissociated personality subnetwork or subpersonality is centrally concerned with two things:
1) constructing the reality perceived by the mind, and
2) constructing the self-image "perceived" by the mind
Separate personality subnetworks are interconnected in the sense that they access, to a great extent, the same memory store. And they also have in common certain parts of the self/reality system, particularly the lower and more basic levels.
What makes human beings so interesting is that, by altering the common aspects of the self/reality system, and by altering the associative memory structure, each subpersonality affects the environment in which the other subpersonalities live. Thus, relations between subpersonalities of a mind are somewhat more intense than relations between people in the physical world. Perhaps the best physical-world analogy for the subpersonalities of a single mind is a community of psychokinetics, each one living a normal life, but also continually altering the physical world in response to the alterations made by the others. In such a community, one could never be sure what was "objectively there," and what was merely placed there by somebody else for some particular purpose. This is precisely the situation with which subpersonalities are presented.
12.3. TRAUMATIC MEMORIES
Evolutionary psychology reveals that partial personality dissociation is not only normal but necessary for efficient mental functioning. In the history of psychology, however, the main role of the concept of dissociation has been in the characterization of various pathological mental conditions. To help bridge the gap between these two perspectives, in this section I will discuss perhaps the simplest form of pathological dissociation: traumatic memory, and the related "post-traumatic stress syndrome."
Rape and violent wartime combat might seem to be rather memorable occurences. But sometimes traumatic experiences such as these are not stored in a person's memory in the ordinary way. Instead, they seem to enter the mind and disappear; they are shut off from conscious memory and reflection, until in certain situations, they pop up intensely and unexpectedly, rendering the "rememberer" mentally dysfunctional.
In the words of van der Kolk and van der Hart (1991),
Lack of proper integration of intensely emotional arousing experiences into the memory system results in dissociation and the formation of traumatic memories. Janet called these new cores of consciousness "subconscious fixed ideas." [T]raumatic memories of the arousing events may return as physical sensations, horrific images or nightmares, behavioral re-enactments or a combination of these. Since fixed ideas have their origin in a failure to make sense of a past experience, they fulfill no further useful function and lack continued adaptive value.
Janet's term "fixed ideas" is reminiscent of the dynamical term "fixed point." It is suggestive of the idea that traumatic memory systems, like structurally conspiratorial belief systems, are attractors to the cognitive equation.
All in all, the phenomenon of traumatic memory fits in well with the dual network view. Why do the traumatic memories "split off" and become autonomous? Because, it seems, certain experiences are simply difficult to connect with the remainder of the mental network. The mind tirelessly seeks to improve its organization, to cut-and-paste parts of the traumatic-memory subnetwork with elements from the rest of the mind. But these attempts fail; they lead only to nightmares, re-enactments of the traumatic experience, and so forth.
And why does the mind fail in its attempts to re-organize and integrate the traumatic experiences? Not, as one might think, primarily because there may be few connections to be drawn, but rather because those connections that could be drawn would be painful ones. When reorganization hits on a real connection, this connection itself causes severe unfulfillment of expectations, which is the definition of strong emotion. Moreover, the specific nature of this unfulfillment is a feeling of decreasing order -- a feeling of disruption of previously coherent thought systems. This is precisely, according to Paulhan (1880) and SI, the definition of unhappiness.
But when "correct" reorganizations are continually rejected because of induced unhappiness, the very reorganization processes become confused. They futilely seek to adjust and improve their algorithms and strategies. The behavioral result is that traumatized individuals react to stressful situations with irrelevant movements, emotions and thoughts that represent fragments of their traumatic memories. As Janet (1904) put it, it is "as if their personality development has stopped at a certain point and cannot expand any more by the addition or assimilation of new elements." In the most extreme case, there is the phenomenon of re-enactment. A person may repeatedly go through the exact words and physical motions of a traumatic experience, yet still be unable to answer simple questions regarding these words and motions. This implies that the traumatic memories are not integrated with the higher-level verbal and cognitive sections of the dual network.
Normally we retain high-level patterns in our experiences, and very little of the experiences themselves. But the situation with traumatic memories is just the opposite. They have not been subjected to the usual rearrangement-based pattern recognition processes, because these processes proved too painful. Instead, they have been retained as a full-fledged subnetwork of perceptions and actions, untouched by rearrangement. In Janet's words,
The person must not only know how to do it, but must also know how to associate the happening with the other events of his life, how to put it in its place in that life-history which each of us is perpetually building up and which for each of us is an essential element of his personality.
What sort of therapy helps people suffering from traumatic memories? What is needed is to get the relevant rearrangement processes back to their prior state of productivity. One useful strategy is to introduce ideas which are related to the traumatic memories, but easier to integrate into the remainder ofthe memory. For instance, many women stigmatized by rape have been helped by imagining that they have all the power in the world, and are applying it to the perpetrator. This allows the specific memories of the rape to be cut-and-pasted with other elements of memory, in a less painful way.
So, in sum, what differentiates traumatic dissociation from healthy dissociation? Traumatic memories are a case of forced dissociation. They represent combat between the hierarchical and heterarchical structures of the dual network. Integrative rearrangements of the traumatic memories are "successful" by the standards of the associative memory network; they lead to common pattern. But they are rejected by the control network due to the unhappiness they generate, the abundance of unfulfilled commands. Successful, healthy dissociation, on the other hand, is harmonious with the entire dynamic of the dual network: it involves a division into successfully functioning parallel subnetworks, which deal with different, relatively unrelated problems. For this very reason, healthy dissociated subnetworks are able to deal with the common segment of memory without fear of wreaking havoc.
A traumatic memory subnetwork must isolate itself from the rest of the memory, or else risk causing distracting, troublesome pain. Thus traumatic subnetworks can never truly be functional. No subnetwork of such small size can be truly complete in itself -- the task of intelligence is too difficult for that.
12.4. DISSOCIATION AND THE STRUCTURE OF BELIEF
I have said that the self/reality belief system is a tool for guiding the construction of other belief systems. Boundary-setting, discussed in Section 12.2.6, is one example of this "guiding" dynamic at work. Another example, I suggest, is the topic of this chapter: dissociation. A small child learns dissociation in the context of her self/reality belief system. This dissociated structure, then, serves to create other autonomous mental structures -- in particular, structurally conspiratorial belief systems, which are crucial for the production of creative logical reasoning.
To explain why this should be true, let us begin chronologically. Once a child learns that she must act different ways in different situations, then she will inevitably develop relatively autonomous personality subnetworks corresponding to the different situations. These subnetworks will not achieve the degree of separation observed in multiple personality patients, but they may well have different likes and dislikes, and different ways of responding to the same stimulation.
This process may also be looked at linguistically, in terms of the theory of language given above. As a child learns that the same words have significantly different meanings in different situations, she will develop a semantic system with distinct subsystems, and these subsystems will take the form of semi-autonomous subnetworks of her dual network. And the same thing that happens with spoken language, will also happen with the language of behaviors (as discussed earlier) -- thus resulting in semiautonomous personality subnetworks.
Now: these different sub-personalities, though they may have arisen in specific social situations, may well emerge on cue in situations different from those which elicited them. The way a person deals with any given issue may be determined by different sub-personalities at different times. Thus there is a kind of evolutionary competition among subpersonalities.
The result of this competition, I suggest, is that a sub-personality will flourish to the extent that it can create belief systems which
a) support its interests, and
b) stand little chance of being destroyed by other sub-personalities
Quite clearly, the best way to achieve (b) is to create structurally conspiratorial belief systems. If a belief system depends on outside factors for its survival, these factors may well shift when the controlling subpersonality shifts. But if a belief system can survive on its own, then it has a much better chance of "waiting out" an hostile environment.
To see the importance of this, recall the conclusion reached in Chapter Ten, that productive belief systems tend to be those that receive significant support both externally and conspiratorially. External needs are too strongly fluctuating to be relied upon as a sole source of support.
But how do these structurally conspiratorial belief systems develop in the first place? Yes, they areattractors of the cognitive equation, so they may be arrived at by "accidental iteration." But, to use evolutionary terminology, how much better to have a force explicitly selecting for structural conspiracy! This is exactly what dissociated personality networks do. Each of the competing subnetworks specifically reinforces related subnetworks that operate by structural conspiracy, and are hence not easily disrupted by competitors.
My contention is that this specific selective pressure is mentally necessary. It is not necessary for the maintainance of structural conspiracies, which by definition maintain themselves. Rather, it is useful for the maintainance of belief systems that, while close to being structurally conspiratorial, are not yet truly self-supporting. The iteration of the cognitive equation is mind-wide; it is not restricted to the individual subnetworks that happen to be converging to attractors on their own. It will tend to mix up subnetworks even if they are somewhat close to being autonomous. The extra push toward autonomy may often be needed; and personality dissociation may thus be a crucial part of the development of effective thinking and acting.
12.4.1. Dissociation and Logic
The social uses of dissociation are obvious. In today's society, it rarely pays to have the same personality at work and at home. But what I am claiming here is something much stronger and more radical. I am claiming that partial personality dissociation is not only socially but cognitively necessary. By biasing the selection of belief systems toward the structurally conspiratorial, it also biases the selection of belief systems toward the productive. Or in other words: no dissociated personality, little chance of systematically creative belief production.
And this brings us back toward logic and reasoning. Logic, if you recall, requires a semantic, analogical system to guide it. And the quality of a chain of logical reasoning depends at least as much on the productivity of this system as on the cleverness of the deductive rules. The conclusion? Without dissociation, ideological, paranoid and otherwise pathologically conspiratorial belief systems would be rare. But so would be productive belief systems; and hence, so would be creative logical thought.
This, finally, is the true meaning of the phrase chaotic logic. Dissociated personality networks, and the structurally conspiratorial belief systems which they encourage, are attractors of the cognitive equation, supporting apparently chaotic dynamics. But without these strange attractors, the rich reserve of analogies required for deductive logic would never be created. Logic thrives on chaos. And, conversely, logic itself is a crucial tool of these belief systems and sub-personalities; it aids them in maintaining their attractor status ... chaos thrives on logic.
12.4.2. The Meta-Dynamics of Paranoid Belief
A complete treatment of the practical psychological implications of abstract "dissociative dynamics" would be out of place here. However, it seems worthwhile to give at least a few hints. Toward this end I will now briefly return to Jane's paranoid belief system, discussed extensively in earlier chapters.
Now I will be able to say a little more about the possible origin of this paranoid system. But as before, I must emphasize that this analysis is not intended as a definite diagnosis of Jane's specific problems, but only as an illustration of certain general principles.
Jane demonstrates many, many different dissociated subpersonalities. Chief among these, however, are: 1) an obsessive subpersonality, in which the world is perceived as hostile and in need of constant mocking scrutiny; and 2) a happy-go-lucky subpersonality, in which she makes an excellent impression on others, and is good-natured almost to the point of being giddy. These are not full personalities; they share most of the same memories. But on the other hand, they are not merely moods either; they are alternate systems of perceiving and classifying data.
The alternation between these two subpersonalities might perhaps be characterized as "manic depression." But obviously there is more to it than that. It would seem that, at very least, there is an unusually complex form of manic depression at work here.
In the obsessive subpersonality, Jane is overly attentive to facial expressions, the colors of clothing, the letters on license plates, and so forth; she is constantly categorizing things in unusual ways. She demonstrates perceptual patterns that might be called "compulsive," and her behavior tends toward the unusual and offensive. She will often act out specifically toshock people; cursing, flashing, making faces, and so forth.
In the happy-go-lucky subpersonality, on the other hand, Jane is open-minded and accepting toward other people's ideas. She tends not to notice details of her surroundings, and her behavior is generally quite unexceptional, except for perhaps a slight overexuberance. She is a pleasant companion and a good conversationalist.
The worst of Jane's depressed moods seem to occur when she is in her obsessive subpersonality, and she is unable to find an external source to blame for her problems (most of which are caused, of course, by the paranoid behavior of the obsessive subpersonality). The happy-go-lucky subpersonality is not so concerned about these problems, and thus is not worried about where to place the blame. But every time the obsessive subpersonality comes back again, it needs to once again begin its quest for an external source to blame.
Therefore, obviously, it is in the interest of the obsessive subpersonality to create a blame-placing belief system which will persist even when the happy-go-lucky subpersonality is in charge. How can this be done? One way, of course, is to create a structurally conspiratorial blame-placing belief system; a system that will maintain itself indefinitely, that will keep itself going even when the reigning subpersonality has no use for it. Perhaps the obsessive subpersonality will experiment with many different strategies for apportioning blame; but those which are less conspiratorial will be less likely to survive the fluctuations of control. Personality dissociation provides a selective force in favor of structural conspiracies -- such as Jane's paranoid belief system.
12.4.1.1. A More Detailed Model
In slightly more detail, one may say that the obsessive subpersonality contains the following beliefs:
D0 = I am unloved
D1 = I am good and lovable
D2 = They are bad
This system is not in itself an attractor for the cognitive equation; it is partially self-supporting, but it also relies on the remainder of the mind.
The dynamics here are simple enough. D0 chips away at D1; but D1, acting on D0, helps to produce D2. And D2, acting on D1 and D0 collectively, helps to produce D1, thus counteracting the effect of D0 (if one is not loved by bad people, that increases rather than decreases one's goodness).
But the problem is that D0 is a self-reproducing belief: it is a pattern in the behavior which it produces. It would seem that perhaps D0, and the behavior systems to which it is connected, are in themselves an attractor for the cognitive equation. For the behavior system is produced by D0 and its own internal dynamics; and D0 is produced by the behavior system.
The effect of D0 on D1 is so strong that D1 is powerless to counteract it, even via D2. So what could be more natural than to counteract D0 by making D2 self-perpetuating -- by making it a structural conspiracy. This is what is accomplished by the conspiratorial belief system described earlier. This entire belief system, with all its complex dynamics, is merely a way of making D2 as strong as possible.
This, on a deeper level, is the meaning of Jane's refusal to take blame. Taking blame for anything subtracts from D1, which is already in serious trouble. But the conspiratorial belief system within D2 works along with D1 to counteract the powerful effect of the self-reproducing belief D0 -- which is, most likely, the root of the whole problem.
This is still a very partial, sketchy analysis of Jane's situation. But it does serve to illustrate the perverse complexity of the mind. One sees belief-system attractors grow within subpersonality attractors, and spawn new belief-system attractors in the common memory, generating a hierarchy of chaotic pattern dynamics -- and all to counteract the runaway self-perpetuating growth of a single belief of the utmost simplicity: "I am unloved."
12.4.3. Dissocation and Creativity
In Jane's case, dissociative dynamics led to an undesirable, overly rigid belief system. But precisely the opposite result is also possible. To give a little bit more of the flavor of the implications of dissociative dynamics, I will now discuss very briefly two famous thinkers, and comment on the role dissociativedynamics played in the development of their thought. These two thinkers, Jung and Nietzsche, are extreme cases; they were more dissociated than most. But they provide an excellent illustration of how belief systems, once they have been made conspiratorial by dissociative dynamics, may also benefit from dissociation in more complex ways.
12.4.3.1. Carl Jung
In his autobiography, Carl Jung analyzed his life work as a result of cooperation and competition between two subpersonalities, whom he called "Number One" and "Number Two." Number One was scientific and practical; Number Two was spiritual and cared little for the material world. Each of the personalities erected its own belief systems: Number One a rational, objectivist view of the world, and Number Two a mystical perspective.
And each of these belief systems turned out to be strong enough to withstand those times when the non-supporting subpersonality was in control. The result was a mental network capable of incredibly powerful, uniquely creative logical reasoning. The competition between the two subpersonalities necessitated the development of much more robust belief systems than would otherwise have been necessary. And the robustness, the structural conspiracy of these belief systems, was crucial in providing analogies to guide Jung's masterful trains of thought.
For a simple example, consider Jung's concept of an "archetype" -- an abstract concept-structure or meta-idea which appears in myths, thoughts and dreams. A simple example is the "resurrection" theme of "the hero and rescuer who, although he has been devoured by a monster, appears again in a miraculous way having overcome whatever monster it was that swallowed him." This archetype may be found in a rather high percentage of movies, novels and television shows!
These archetypal images are not specific pictures -- the hero need not be big and strong, and the monster need not be a huge ugly green beast. The archetype is a structure -- in this case, it is a structure which consists of roles and types of events. Each role (hero, rescuer, monster) and each type of event (rescue, devouring, miraculous reappearance) is simply a certain collection of patterns, and each one may be fulfilled in a number of different ways. As Jung put it,
Again and again I encounter the mistaken notion that an archetype is determined in regard to its content, in other words that it is a kind of unconscious idea (if such an expression be admissible). It is necessary to point out once more that archetypes are not determined as regards their content, but only as regards their form and then only to a very limited degree. A primordial image is determined as to its content only when it has become conscious and is therefore filled out with the material of conscious experience.... The archetype in itself is empty and purely formal, nothing but a... possibility of representation which is given a priori. The representations themselves are not inherited, only the forms, and in that respect they correspond in every way to the instincts, which are also determined in form only. The existence of the instincts can no more be proved than the existence of the archetypes, so long as they do not manifest themselves concretely. (Jung, 1934)
The collection of all archetypes, Jung called the collective unconscious. It is -- or so he hypothesized -- an inherited, a priori part of every human mind. Archetypes subtly guide all our feelings and acts.
Jung did not discover the notion of "archetype" by scientific, logical analysis; he discovered it by pure intuition, by seeing the mind as an abstract structure and thus understanding its dynamics. This was clearly a Number Two process. But a pure, spiritual intuition into the mind would not survive the scrutiny of Number One. In order to keep its insight, Number Two had to form the concept of "archetype" into a powerful, self-maintaining ideational system. Once this was done, then Number One not only refrained from destroying the concept; it latched onto it, refined and improved it, yielding the scientific notion of archetype that we have today.
12.4.3.2. Friedrich Nietzsche
For a more complex example, consider the philosopher Friedrich Nietszche, on whose work I have drawn so liberally in these pages. Nietzsche demonstrated at least two prominent subpersonalities. One was a mild-mannered, friendly and quiet philologist, who hated seeing pain and avoided causing anyone offense. Theother was the brilliant, arrogant madman whom one sees in works such as Thus Spake Zarathustra and Ecce Homo. Following the example of Jung, let us call these Number One and Number Two.
Number One forced Nietszche to hide the radical nature of his philosophy from casual acquaintances. On one occasion, when he saw a horse about to be whipped by its master, it caused him to stop and vigorously hug the horse. Number Two, on the other hand, impelled Nietszche to forsake classical philology and spend his life in a passionate quest to destroy all the ideas he had been raised to believe in: religion, morality, reality, truth. As has often been observed, Nietszche's philosophy encapsulates the contradiction between these two emotional views of the world.
Number Two supported an incredibly productive belief system of mistrust and skepticism, a disputative belief system capable of seeing the holes in any argument, and of combining and manipulating abstract ideas with great dexterity. This belief system was not merely nihilistic; it consisted of a repository of clever tools for demonstrating the falsehood and vanity of any point of view. One sees this system at its best in aphoristic works such as Human, All Too Human, Dawn and The Gay Science.
Number One, on the other hand, silently upheld the values against which Nietzsche's work railed. It supported a more traditional philosophical belief system: it perceived an underlying order in the universe, it respected the difference between right and wrong, and it had a powerful sense of spirituality. This was the belief system which governed Nietzsche's personal life. Number Two wrote tirades against asceticism; but Number One was responsible for Nietzsche's own ascetic lifestyle.
Nietszche's most dramatic ideas, the eternal recurrence and the will to power, may be seen as the result of synthesizing aspects of these two conflicting belief systems. The eternal recurrence is a cynic's version of afterlife. The will to power is a will superseding all notions of "free will" -- with its militaristic "order of rank," it is a morality "beyond good and evil." Zarathustra's beautiful sermons display an atheistic spirituality beyond all traditional concepts of Godliness. Much of the strength of Nietzsche's thought results from its dual source: two productive,structurally conspiratorial belief systems, usually competing but occasionally collaborating.
12.4.3.2. Conclusion
This very cursory study of two great thinkers indicates an important aspect of dissociative dynamics: namely, the possibility of synergy between competing belief systems. Two dissociated subpersonalities need not become unified in order for their respective belief systems to combine with one another. Jung and Nietszche are two examples of creativity emerging from the synergy between the structurally conspiratorial belief systems of different subpersonalities. In neither case was a complete synthesis attained; but in both cases, the interaction and partial reconciliation of conflicting systems proved tremendously productive. The role of this sort of synergy in everyday life and thought would seem to be a very fertile area for future investigation.
12.5. DISSOCIATION IN THE UNIVERSAL NETWORK
What is the difference between the dissociated subpersonalities of a given mind, and the separate minds in the world? After all, as noted above, the different minds in the world are just semi-autonomous subnetworks of the universal network. Are we all perhaps just subpersonalities of one particularly advanced multiple personality patient?
In fact there are two main differences between a collection of subpersonalities and a collection of minds. The first is that subpersonalities are mainly conscious in sequence, not in parallel. There is certainly some parallelism going on: one sub-personality may passionately declare "I love you" while another simultaneously and silently ridicules the remark. One subpersonality may raise a gun to shoot someone, while a competing subpersonality causes the legs to buckle, preventing the murder from occurring. But these are extreme cases; there is much more parallelism among the different personalities in the real world.
And the second outstanding difference regards memory access. Dissociated subpersonalities, although largely disconnected from one another, still have access to a common memory store. In normal mental functioning, every personality has access to almost every memory in thebrain; the main thing is that different memories are more easily accessible to certain subpersonalities than to others. State-dependent memory is important but not all-pervading.
Different personalities in the world, on the other hand, do not appear to have access to a common memory store. They are connected at the bottom, via physical reality, but this would seem to be the extent of their interconnection. This feature is shared by the various personalities of multiple personality patients -- for instance, one personality may speak Italian while the other does not. But even in these exceptional cases, there is still some degree of common memory, much more than between two different people.
Rupert Sheldrake's (1981) theory of the morphogenetic field attempts to destroy this distinction, claiming that each person's memory is aphysically connected to everyone else's. So that, for instance, once a thousand people learn the formula for solving cubic equations, a small "trace" of that knowledge becomes available to everyone, thus making the process of learning that particular formula universally easier. But this dramatic prediction remains unproven.
So, in sum, I have argued that the difference between the people in the world and the subpersonalities in one mind is a matter of degree rather than absolute distinction. There are serious differences in the amount of parallel consciousness and the existence or amount of common memory. However, there is a significant simlarity in that, just as different subpersonalities collectively create their "environment," different people collectively create their reality.
12.5.1. Why Not One Mind Only?
These observations lead to some rather interesting philosophical ideas regarding the nature of our collectively constructed external reality. Subpersonalities are behooved to encourage structural conspiracies, so that the processes they create will not be destroyed by other subpersonalities. And by the very same reasoning, minds will do well to create reality structures which are structurally conspiratorial, so that the reality structures they create will not be disrupted by other minds. This suggests that the immensely conspiratorial nature of the reality belief system ispartly due to its construction at the hands of competing individual consciousnesses.
In other words: a reality created by one consciousness alone would probably not be very interesting; it would have little generativity, because of the lack of structurally conspiratorial subcomponents. The competition of different minds encourages structural conspiracy and hence creativity. This is a novel, thought-provoking answer to the old philosophical puzzle of the multiplicity of consciousness. Why not, as the Buddhists would have it, one mind only? Because that path leads to a boring world. If intricate structure is a criterion of value, then multiple consciousnesses are valuable indeed.
This does not exactly give a reason for the multiplicity of consciousness. But it does give something to go on: the fact that the universal network is a multiple-consciousness attractor for the cognitive equation. If one accepts this equation, the only other thing to be taken on faith is that, starting from wherever it did, the universe eventually converged on the universal network structure. And chaos implies that there need be no real "reason" for this. Convergence to one attractor rather than another can be the result of pure chance.
12.5.2. The Future of Reality
These ideas may appear to be "out there" -- philosophical meanderings unrelated to any issues of practical substance. However, this perception is far from accurate. The ideas of this section are not merely theories about the relation between mind and reality, they are computational theories about the relation between mind and reality. And this means that they fall into the category of theoretical science, rather than philosophy. For, although current technology does not permit the relevant tests to be carried out in a reasonable time frame, these theories are in principle empirically testable.
To see this, consider the possibility of virtual reality technology, which would allow us to put our consciousnesses into simulated bodies living in simulated physical realities. Given this technology, it would be easy to experiment with different methods of collective reality construction. For instance, one could easily verify whether or not one consciousness is enough tocreate an intricate world -- whether or not, as I have claimed, defensive structural conspiracy is required.
Short of full virtual reality, it is also possible to conceive of simulated realities: collections of artificially intelligent programs that collectively construct their own simulated world. Though less dramatic, this would also permit direct empirical test of theories about the mind/reality relation.
12.7. ARTIFICIAL INTERSUBJECTIVITY
The dual network model and the cognitive equation are computational models. In this final section, I will briefly explore the possibility of using them to do practical computation: to design a computer program which displays the sensitive interplay of chaos and logic that today is only associated with human minds. I will describe a new type of algorithm, which I call an artificial intersubjectivity, or an A-IS.
12.5.1. AI and Alife
Let us approach this "new type" of program obliquely, by way of the two most exciting branches of modern computer programming, artificial intelligence and artificial life....
In artificial intelligence, first of all, one seeks to write programs that will display the full range of behaviors that humans term "intelligent." There are already programs that display many of the behaviors that we call intelligent -- doing arithmetic, algebra and calculus, flying jet planes, playing championship chess, recognizing voices, etc. But these programs are invariably narrow in focus: each one does its schtick, and is unable to generalize its intelligence to other contexts. A true artificial intelligence would be able to learn, and learn how to learn, just like a person. It would not necessarily need to know how to do long division like a pocket calculator -- but it would need to be able to learn to do long division, to recognize faces, to play new games,....
In the 1960's and early 1970's, it was widely believed that one could achieve artificial intelligence by programming a sufficiently clever "thought algorithm." Now, however, this is no longer believed to be the case. Today it seems to be a terribly long way from voicerecognition and championship chess to true intelligence. The modern AI community is torn between the "old-fashioned" programming approach and the even older, recently rediscovered "connectionist" approach, which seeks to write programs loosely modeling brain function. Connectionism has succeeded in many instances where old-fashioned programming repeatedly failed. But on the other hand, connectionism seems to be even less competent at dealing with logical reasoning and other aspects of linguistic thought.
When one writes a program to imitate the brain, on a coarse or a fine level, one is writing a program that is in a sense chaotic and unpredictable. One knows what the program does, but not how it does it. Thus, the "connectionist" approach to AI has given up on the programme of first logically understanding an action, then writing a program to simulate it, based on this understanding. This unpredictable aspect of connectionist programs leads us to our second type of programming: artificial life, or Alife. Alife seeks to take the self-organizing unpredictability of connectionism and apply it to the simulation of biochemical or ecological rather than neural systems.
For instance, several different groups of researchers have run computer simulations of self-organizing systems of enzymes, with an eye toward understanding the dynamics underlying the evolution of life. Bagley and Farmer (1992) have treated the origins of metabolism in this way; whereas Boerlijst and Hogeweg (1992) have modeled the well-known hypercycle theory of the origins of reproduction.
And, on a higher level of organization, various researchers have simulated "artificial ecosystems" from ant farms (Collins and Jefferson, 1992) to systems of coevolving parasites (Hillis, 1992). Richard Dawkins (1986) has investigated "biomorphs," artificial life-like shapes generated by a process of progressive evolution.
12.5.2. A-IS
In AI, one seeks programs that will respond "intelligently" to our world. In Alife, one seeks programs that will evolve interestingly within the context of their simulated worlds. It is, of course, not difficult to synthesize these two research programmes to obtain the idea of "artificially intelligent artificiallife" -- synthetically evolved life forms which display intelligence with respect to their simulated worlds.
But A-IS, artificial intersubjectivity, constitutes a large step beyond this hybrid concept "AI Alife." What I am suggesting is to simulate a system of intelligences collectively creating their own "virtual" reality. The universal network model gives us a blueprint for the joint construction of realities; it remains only to put this blueprint into action by making appropriate computational simplifications.
12.5.2.1. The Nature of Human Intelligence
What would be the point of this formidable programming exercise? There are at least two good reasons for pursuing A-IS. First of all, consider: what if we humans are only intelligent with respect to the reality which we have collectively created for ourselves? Setting aside the unanswerable question of the "ultimate" existence of an objective reality, what if we are only intelligent with regard to the subjective reality which we collectively, culturally construct and live within?
This proposition may be taken in two ways. First of all, if one considers intelligence as an optimization problem, as was done in Chapter Three, then the conclusion becomes almost inevitable on an evolutionary level. After all, the general problems of global optimization and pattern recognition are unsolvable. The human brain consists of some general-purpose optimization routines, plus a whole host of special case tricks tailored to the environment for which it evolved. The structure of this network of tricky processes may be universal -- but two entities with the same global structure don't necessarily have the same abilities.
But whether or not one accepts this "evolutionary" point, the same conclusion follows even more surely on a psychological level. For as we have already shown in Section Two, the adult human mind is specifically tailored to its culturally constructed collective reality. We have to learn to think -- an infant doesn't know how; and the evidence shows that a child left to mature in isolation will never adequately learn how. We learn to think by practicing on examples that have to do with the self/reality belief system; and this belief system develops properly only in a social context, i.e. only in the context of explicitly creating subjective reality jointly with other minds.
So, in sum, it seems quite certain that the process of thinking is inseparable from the process of participating in the collective construction of a reality. And this fact indicates the necessity for a new type of programming, one that might be roughly characterized as "AI + Alife + feedback between the two" -- a community of artificial intelligences, acting in an artificial collective subjective world, and simultaneously acting on that world.
One might argue that collective construction of reality is not enough -- that the "adult/child" relationship is necessary for the development of intelligence; that one cannot become a "mental adult" except under the tutelage of another "mental adult." But of course, this is a chicken/egg problem ... who was the first "mental adult"? On the other hand, the idea that collective reality construction is necessary for intelligence presents no chicken/egg problem, since there can quite well have been a first tribe, a first group of organisms biologically capable of some degree of intelligence.
Perhaps, indeed, a high degree of intelligence requires a few dozen or a few thousand generations of co-creating minds working gradually toward "mental adulthood." But even if this is true (which I rather doubt), it is not a fundamental obstacle to the concept of A-IS, of computer-simulated intersubjective reality construction. After all, in Alife one routinely simulates thousands of generations of evolution. In Theodore Sturgeon's classic story "Microcosmic God," a scientist breeds organisms called "Neoterics" which evolve so fast that they zoom beyond mankind in a matter of months. Robert L. Forward's novel Dragon's Egg pursues a similar theme, except that the rapidly evolving organisms are not human creations but the natural fauna of a neutron star. With sufficiently fast computers, this science-fictional "souped-up evolution" process could be simulated, allowing artificial intersubjectivity to evolve over numerous generations.
The universal network model gives us a handy, elegant way of achieving this type of artificially intersubjective program. Namely, simulate a collection of dual networks connected at the bottom. The bottom levels are the collective subjective reality; the upper levels are the individual thought processes of the "intelligences." Under appropriate conditions, the presence of a common subjective reality will cause thevarious networks to "converge" to a common belief system regarding their "external world." This belief system will inevitably include a role for themselves -- an "imaginary subject."
12.5.2.2. Dissociation and A-IS
And this leads us to the second good reason for pursuing A-IS: only by developing a natural self/reality dynamics can a mind develop dissociated personalities which encourage structural conspiracies. Productive structural conspiracies are necessary for systematic, clever logical thought. Therefore, by creating a community of collective-reality-constructing AI agents, we will implicitly be creating AI agents which are adept and creative at directing their logical reasoning. This creativity will not necessarily be a clone of human creativity, because the specific belief systems and dissociated subpersonalities may be different. But there is no reason that computer creativity achieved in this way could not equal or exceed human creativity in utility and power.
This is a fundamentally new approach to computer reasoning. Neither connectionism nor old-fashioned rule-based AI comes anywhere near to acknowledging the complex process dynamics of intelligence. Alife and connectionist AI may support various types of self-organization and chaotic dynamics; but only A-IS can fully manifest the systematic self-generation that is chaotic logic.
12.5.2.3. The Question of Implementation
From the point of view of current implementation, there are two serious problems with the A-IS idea: memory and speed! It would be possible to run a stripped-down version of A-IS on contemporary massively parallel supercomputers, such as the larger "Connection Machines" manufactured by Thinking Machines, Inc. But although one could surely obtain interesting results in this way, one would not be doing justice to the concept of A-IS. Agents of relatively little intelligence will be able to develop collective reality dynamics of relatively little subtlety.
Each human brain contains maybe 1011 simple numeric processors. Even with more efficient techniques at our disposal, it seems unlikely that we can get by with onlya few hundred thousand analogous processing units, which is what today's most powerful parallel computers offer. The most promising path toward developing true A-IS, I suspect, is nanotechnology (Drexler, 1986), or molecular computing. Using molecular computing techniques, it may be possible, in the not-too-distant future, to grow computers, to create computers which add onto themselves like crystals do. If this possibility should be actualized, then it will not be too long before A-IS becomes a practical science.