The Verbal Process Creates the Illusion of Consciousness

by Harland Harrison

"If flesh creates spirit it is the marvel of marvels" -the Gospel of Thomas

Presented as a poster at "Tucson III: Toward a Science of Consciousness" Tucson AZ, 1998

Email: Harland Harrison (HarlandH@harlandh.cnc.net)

Abstract

While a philosopher's zombie, (an android without consciousness), could accomplish the essential tasks for human survival, people nonetheless report subjective experience, and exhibit behavior offering no selective advantage. I theorize that a species of mammal which recently evolved speech, would inherently report the common attributes of consciousness, including qualia, a sense of "self", free will, certain altered states (ASCs), identification with groups, an interest in abstract truth, and a potential to perceive the supernatural. The combination of a computational Verbal Process, (VP), evolved to run simultaneously in the brain with an older, representational, model of reality, a Simulation of the Environment, (SE), produce these phenomena, and create the illusion of subjective consciousness.

If an individual's fitness increases by understanding speech, words must at times be modifying her behavior, and hence must be modifying her SE. Since not all speech is reliable, she must judge and select what to believe. Her choices and methods of judgement alone suffice to account for many aspects of consciousness. Thus, a quale appears as the force to believe a perception; the surpernatural arises as the ability to override perception; group identities establish trusted sources of information; truth seeking surmounts dependence on group identity. Unconscious mammalian behavior combines with VP effects to explain more phenomena identified with consciousness.

Computer science and neuroscience are consistant with the VP model. Servo-control-system theory proves the necessity of a world model in a pre-conscious mind. A Hebbian organization of cerebral cortex and associative memory implies parallel processing of this model. By contrast, standard compiler parsing techniques applied to a rule-based description of language (such as that proposed by Noam Chomsky), require a stack-like, sequential VP. This difference indicates the processes are separate. Speech is generally confined to one cerebral hemisphere, implying it may surplant other functions. Lesion and imaging studies suggest possible neuroanatomy of the VP components.

Presenting consciousness as an illusion within a computational model of reality compliments the Hindu view of reality as the illusion of maya, and the Buddhist view of the self as an illusion. It cuts through the "hard problem" like the proverbial Gordian Knot. The VP concept sheds light on magic, totemism, grave goods, ASCs, brain-washing, tribal dancing, mythologies, meditational practices, and some observations of Sigmund Freud and Carl Jung.

Introduction

To explain how the faculty of language could arise by evolution, I outlined a simple, computational model of the mind. It started with the perceptual process of a mammal, from which the brain creates a model of the world around it, the Simulation of the Environment (SE). On top of this computational process, evolution added a Verbal Process (VP), also computational. Surprisingly, this resultant model predicted that the animal would now report several, if not all, of the subjective attributes of consciousness, such as qualia and free will! The model also accounted for some puzzling aspects of human behavior, including the tendency to believe in the supernatural.

Possessing the power of language gave Homo Sapiens a great selective advantage. The magnitude of that advantage could allow many epiphenomena to occur along with the development of speech. Some of those effects could even be disadvantageous. Natural selection would not remove characteristics linked to such a valuable trait as communication. I submit that the attributes usually associated with subjective consciousness actually arise from these computational processes. Thus, consciousness needs no further explanation than that of a computational process which happens to create the illusion of subjective experience.

The body of the paper is divided into four sections:

Structure details the VP theory of the mind.

Effects describes the effects and epiphenomena consistant with the VP structure

Neural Correlates discusses possible neural correlates and implications for physiology

Conclusion includes some meaning for philosophy, and a summary


Structure

Simulation of the Environment

The brain of a complex animal, such as any mammal, must contain some model of the environment in which it lives. An accurate model will also approximate the behavior of the environment, becoming a simulation of the environment, (SE). The brain of a mammal must include an SE, because many of the tasks a mammal can accomplish require the equivalent of highly accurate control systems. These tasks include body motion, but can be generalized to include higher level goals as well (Powers) . A mammal must use both open-loop and closed-loop control systems. (See Appendix I for basic information on Control Systems) It uses a closed loop, with feedback, for an unfamiliar action. After learning a behavior well, it can switch to open loop control, by just repeating the pattern generated in the learning phase. Setting up and maintaining the control systems, requires measuring, recording, and simulating the controled variables. All the parameters which the animal can control, considered together, form a model of the environment which the brain must possess. That model includes not only how things are, but how they will change. The SE models the world for the animal.

Figure 1, The Simulation of the Environment (SE)

In Figure 1, the SE receives input from the senses. It performs "feature extraction", recognizing patterns to identify objects. The pattern-recognition process continues to higher levels of abstraction, ultimating creating "concepts" from the incoming "percepts". The SE can also regenerate the input from the higher level concepts. Planning, imagining, and dreaming, reverse the process, creating lower level data from higher level abstractions.

Different formats of computer images can illustrate the distinction drawn between percepts and concepts, and the pattern-recognition process. A pixel-map image, a series of unconnected dots forming a picture, exemplifies a "percept", the raw sense data. A vector-set image, instructions for drawing an object, exemplifies a "concept", such as an identified object. Pattern recognition, in this case, consists of tracing the outline of the object in the pixel map and creating a vector set which represents it. Drawing the vector set into a pixel map regenerates the image. The brain also represents incoming visual data in pixels, in retinotopic maps such as V1, from which the brain can extract features and objects.

The behavior of a non-speaking mammal depends entirely on conditioning associated with the SE. The SE incorporates all of the animal's knowledge about the world. Conditioned responses associate stimuli from pleasure and pain receptors with different elements in the SE, giving valence to objects and circumstances. The control systems which avoid pain and seek pleasure can also operate on the associated elements as well. Thus the SE drives behavior, making the animal forage, fight, or run in fear, depending on the state of the animal's model of the world.

The simulation of the SE depends on implicit memory. The SE forms from the moment of birth, by recording and simulating the behavior of objects observed in the world. Associative memory links the simultaneous percepts of an object together (probably by a Hebbian mechanism of synaptic potentiation). A subsequent appearance of some of those percepts causes the entire set to repeat spontaneously as the recognition of the object. Events which have occurred consistantly, tend to repeat spontaneously as well. Sequential patterns can record the behavior of objects, and repeat that behavior spontaneously. Thus the SE can advance, registering events from sketchy perceptions, and predicting what will happen in the future. Because the SE can advance,the animal can make decisions which show planning. Because the SE can acquire new implicit memory, the animal can make decisions which show learning.

The SE as described is a purely computational process. Simulations and control systems run on computers, by calculation alone, without requiring any consciousness. Identifying objects from features, and, inversely, recreating features for a given object type, are standard database functions (See Appendix II for basic information on databases) which computers implement today for tasks such as sorting hand-addressed mail. Some computer programs can also adapt and aquire knowledge as they operate. Nothing about the description of the SE alone involves subjective consciousness; the SE requires only computational processes.

Verbal Process

Homo Sapiens added verbal functions to the SE model. Verbal comprehension must have started gradually, and continued to evolve, because it gave individuals an ever-increasing selective advantage. Their behavior began to depend on words instead of just on personal experience. Perception and communication began to share control of behavior. The VP gradually assumed a major share of control because it improved behavior for survival.

To improve on the conditioned responses of the SE already in place, the VP had to use the same data as the SE. Limited speech could not create an independent database equal to an animal's life experience. Instead, the VP had to start with the knowledge already contained in the SE, and improve upon it. Philosophically, it could solve the "symbol grounding" problem by attaching symbols to experience.

Figure 2, The Verbal Process (VP)

In Figure 2, the VP manipulates data in the SE. This data gives words their semantic meaning. As the simulation runs it provides words to the VP, by the path marked "Sequence by Semantics" in the figure. As the VP parses syntax it provides words to the SE, by the path marked "Sequence by Syntax" in the figure. Thus words take meaning by arranging elements in the SE and then permitting the simulation to run.

The meaning of words depends on semantics. The meaning of complete phrases depends not only on semantics, but on syntax as well. Word order and grammatical structure may vary substantially, yet produce the same meaning. A single "not" can completely change the meaning of a sentence. The two processes, SE and VP, providing semantics and syntax respectively, create meaningful symbolic representation.

Parsing syntax most likely involves some rule-based manipulation such as the transformational-generative linguistics proposed by Noam Chomsky. Such rules do not allow handling words in the exact order of a sentence. The parser must rearrange the words and phrases to deduce the meaning. Computers perform a similar task in parsing many kinds of algebraic expressions, including the predicate calculus which is a kind of language. Computers generally parse expressions with a stack (Bauer). The VP probably does not literally contain a stack, however, because semantics clearly helps parse grammatically complex but unambiguous phrases (Pinker).

Speech presumedly evolved gradually by natural selection. Some authorities do believe however that language arose first for thinking, and only later did communication appear (Pinker). Even in that case, internal comprehension, just as understanding the words of others, would have arisen gradually by natural selection because it enhanced behavior. Since language evolved slowly the changes to the brain would have depended on the SE provided the semantics for the new verbal process, regardless of whether thinking or speaking evolveded first.

Equalization of Salience

Language is not just Pavlovian reflexes. Words sometimes carry a high emotional value, and sometimes they do not. Oliver Wendell Holmes recognized this fact by finding that freedom of speech does not confer the right to shout "Fire" in a crowded theater (Schenck vs United States). After that shout, further discourse may become impossible.

The valence of a phrase also depends on syntax. The phrase may be conditional or negated, reducing or even inverting the valence of the phrase. Phrases using words like "not", "if", and "unless" alter or completely invert the emotional value of key words. The actual valence appears only after parsing the complete syntax.

The VP must control the emotional value of words in order to insert them into the SE. First, it must coerce the words into the SE, which if it continued to process a simulation, would produce its own, different, result. Then, the words may need to be manipulated as symbols as Noam Chomsky has suggested. The VP must not cause an emotional reaction which the syntax does not support throughout this process. Finally, after completing the syntax, the words must create the proper reaction intended.

So, to control the emotional value of words, the VP temporarily suppresses the valence of symbols in the SE. The grey line marked "Equalization of Salience" arbitrarily provides this function by switching off any conditioned responses. When suppressed, the VP constructs a step in a scenario in the SE using the "Sequence by Syntax" path. When not suppressed, the valence drives the simulation forward, supplying words back to the VP through the "Sequence by Semantics" path. When the valence is suppressed, any symbol is nearly as likely to occur in the SE as any other. The order depends on the stack-like process of syntax. For this reason, I call suppressing the valence, "equalization of salience" .

The equalization of salience occurs periodically. Syntax alternates with semantics when parsing phrases, as data passes from the VP into the SE and from the SE to the VP. The notion of a switch that inhibits emotions is an arbitrary one. The VP need not directly control the equalization of salience itself. It might use some natural oscillation which already occurs for other reasons. Equalization of salience only requires some mechanism by which the VP can manipulate symbols without regard for their valence.

Credence

A person cannot believe everything she hears. Not all speech is reliable. Deception has an evolutionary basis, and has always been a potential human trait. Without mentioning any profession, I will assert that deception remains a major problem in modern society. Nature must have evolved ways to deal with it.

In any species, varieties can arise which exploit the weaknesses of other varieties (Dawkins). In the case of evolving speech, a variety that learned to fabricate information could easily dominate a naive variety. Nature is full of mimicry, angler fish, and fish with eye spots on their tails, milk snakes that look like coral snakes, and edible butterflies that resemble toxic monarchs. A variety that learned to repeat without veracity would be trading false information for truth, and would gain advantage quickly. This could potentially bring the evolution of speech to an end.

To protect themselves from deception people generally use two strategies which I will call "source evaluation" and "content evaluation". People can evaluate a message they hear or read by its source, that is by considering who says it or where it comes from. They can also evaluate the message by its content, how it fits in with what they already know. They dismiss any messages that seem unreliable by these tests.

Certainly we apply these two strategies logically, but I believe that instinctive forms of these strategies evolved first. We tend to automatically reject information that conflicts with what we already believe. We tend to believe what trusted people say, much more than people we suspect. People base these choices on feelings, rather than logic. Evolution built the source evaluation and content evaluation strategies into the human brain.

Source evaluation starts with kinship and then extends to non-relatives. An individual can always trust relatives not to betray her (Dawkins). Believing unrelated individuals requires mutual altruism. Individuals can communicate to the benefit of both, provided they can trust and believe each other. Mammals practice mutual altruism in such phenomena as the warning calls of prairie dogs. People have extended mutual altruism to allow trusting and communicating with unrelated people as well as kin.

Content evaluation works by testing information against an existing base of reliable information. To test new information requires keeping a series of accepted propositions readily available. At the same time, information can best be evaluated in large portions. To support content evaluation, people develop narratives, descriptions about the way things are. The narratives contain their knowledge about the world. Maintaining long narratives allows better content evaluation.

Body and Self

The SE must contain an image of the body in the environment since the animal manipulates the body in the world. That image coordinates proprioception, the sensation of the position of the musclature, with visual and tactile reconstructions of space. A great many conditioned responses must also attach to the body image. All original pleasure and pain associates with it. Conditioned responses based on experience attach to the body image to avoid injury and maintain health.

Mammals care not only for their own bodies, but also for their young. To use the lessons of experience contained in the body image, the conditioned responses must transfer to the offspring. Thus the mother "instinctively" protects her children from danger, by protecting them from things she has learned pose a danger to herself.

The transferable body image creates a sense of value beyond the body. A mammal must feel that the importance of her body extends to her offspring. In many species males help guard and rear the young. Some species will defend their mates. Some will bond with non-relatives in mutual altruism. Transfering the body image solves all of these situations. The animal will feel about others bodies in the same way she feels about her own.

Humans go beyond other mammals in their extent of transference. They form friends in all situations. They help strangers altruistically. They often adopt unrelated children, and even care for animals of other species with no intended gain to themselves. People have heroes, adore celebrities, and identify with sports teams. They can extend feelings usually reserved only for kin in other species, to other people, to groups, and even to abstractions.

VP theory explains such behavior as a consequence of speech evolution. Source evaluation required establishing mutual altruism between communicating individuals. (See Credence) The importance of trust in communciation lead to exaggerated and mobile feelings of affinity. The abilities to trust, cooperate, and communicate, synergistically co-evolved. Their advantages far outweighed any slight disadvantages of the epiphenomenal behavior.

Extensive transference gives a person a sense of "self". Affinity starts with the body image, but for a human affinity moves easily. The body image, no longer confined to the physical body, or to the children, becomes a separate entity, the "self". The person experiences the self as her sense of importance, as what she values and protects. The self moves with narratives as credence develops for different people or abstractions. The person may call the self, her "mind", or her "spirit", and come to believe that she is a spirit, instead of an actual body.


Effects

Qualia

Perceptions have particular realistic qualities or "qualia". Seeing something red for example, we sense the quale of redness. One can know all about the color red, and even remember it, without the quale occurring. VP theory accounts for qualia because it says the brain uses the same model of any object both for perception and for speech. The difference between symbolic activation and perceptual activation of the same model creates qualia in the mind.

For reasons of survival, perception must have certain credibility, while thinking must remain relatively speculative. An animal would not survive long if it tried to make danger or hunger disappear by thinking it away. Perception thus becomes immediate and irrevocable. You cannot stop what you perceive. You can easily change your thoughts, but you cannot easily change what you see. However, VP theory asserts that perception activates the same neural patterns as speech. Perception just activates them in a way that thoughts cannot inhibit. Perhaps the same neurons fire more frequently. (See Topological vs Temporal Coding) This stronger, immutable, activation forms the basis of qualia.

Perceptions appear to have "qualia" because of their direct connections into the logical symbols in the SE. The VP can manipulate the abstractions from these percepts, but cannot change the data as it enters. The qualia appear at these entry points as data which does not change. This implies each quale should correspond to a type of perceptual attribute in the cortex. There are different retino-topic maps for brightness in V1, and for color in V4, so people perceive qualia corresponding to brightness and hue. Qualia occur where perceptual attributes make direct connection to the brain.

Free Will

Free will is the ability to act arbitrarily. Just doing what we want does not prove free will, because that still leaves us at the mercy of our desires. The VP theory accounts for the appearance of free will.

The verbal process creates the illusion of free will because it can alter the emotional value of actions according to a verbal narrative. The narrative makes an action more desirable or less desirable, and thereby may cause it to be executed or not. By imagining a positive value for taking a certain action, the VP leads the rest of the brain to take that action. For an insignificant action, any reason for taking the action may suffice. We can even perform some actions just to "prove" we have the free will to do them. By following a verbal the plan, the VP creates the illusion of free will.

The VP controls behavior indirectly, although sometimes quite strongly. Behavior seems to follow the instruction of the VP explicitly, rather than the dictates of emotion, but emotion always remains in final control. The conditioned responses actually take the action, although the VP may initiate it. A gedanken experiment may illustrate.

Consider the first sky-diving lesson for a physicist. He knows more than most anyone how the parachute works and why such an adventure is safe. Still, his heart races as he prepares to jump. He will have as much trouble as any other novice making that first jump. His extensive knowledge of aerodynamics does not control his emotions. After he has jumped a few times his response will change. Now his heart rate remains steady as he exits the plane. It accelerates as he nears the ground. Yet, from the very first jump, he knew the probability of suffering injury upon landing.

Legend says that Muceus Scaevola held his hand in a flame to impress his captors. Few people could match that feat. Certainly nobody could do it arbitrarily. Scaevola was acting for the gloria of Rome, a group endeavor apparently more important to him than life, since he had just risked his life in an assasination attempt. Such acts of courage may not reflect free will as much as devotion; though noble, the emotions remain in charge.

Truth Searching

Some people have a special interest in finding things out. Many people devote their lives unselfishly to science, to the benefit of us all. Scholarship for its own sake, rather than money-making, characterizes the academic community.

Not everyone cares about such things. Other people delight more in family and friends than facts or the thrill of discovery. This deserves no moral judgement, but perhaps some explanation.

As mentioned above, (See Credence) people use at least two different strategies to decide if they should believe what someone tells them. If people just learn these strategies intellectually, it means little. But if these strategies are instinctual as suggested, if they are built into our emotions by evolution, then they could cause differences in values. People who rely on source evaluation would be interested in tracking friends and family. Whereas people who rely on content evaluation would have a greater interest in finding consistancy. They would become the scientists and truth-seekers.

Hypnosis

The phenomenon of hypnosis has been well studied (Kappas). The credence function of the VP explains its effects. Belief normally depends on both source and content evaluation. The content evaluation keeps the bystander from accepting the statements of the hypnotist. The hypnotized subject however, has accepted the hypnotist as a totally credible source. She uses almost no content evaluation because she trusts the source. She accepts what the hynotist describes. She may perceive what she hears because she has reduced alertness in the hypnotic trance, which approaches a state of dreaming.

The subject will accept stipulations contrary to her perceptions, based on her high source-rating of the hypnotist. She will also accept intentions for her future free-will actions, because those depend only on arousing emotion for them. Her memory can be apparantly controlled by telling her not to remember certain events when she awakens. She follows this command by not talking about or concentrating on her recollections of those events. She appears to have amnesia. However, under a later trance, the prohibition can be removed. Awakened the second she can "remember" again. The memories were still there, but were just not being accessed during the hypnotically induced amnesia.

This explanation predicts that if something should interfere with the source evaluation of the hypnotist, it would break the trance. A hypnotized subject cannot be induced to do anything she considers strongly immoral.

An immoral suggestion interferes with her high evaluation of the hypnotist. She cannot simultaneously trust the source and believe that he wants her to do it. She must either ignore the command or wake up. By contrast, some studies have shown hypnotized subjects attempting quite dangerous acts, presumedly trusting the operator's judgement (Tart).

Subjects often describe the experience of hypnosis as seeing the objects of directed attention lit up, with the rest of the field of vision in dim light or darkness. Meditating people report a similar visual effect; the object of concentration appearing to glow.(Tart) A VP model accounts for this aspect of trance as an increase in the salience of the perception. Objects in the field of concentration have a greater representation in the VP, than objects outside the field. Since the VP uses the visual system for semantic meaning, it increases the neural activity associated with the represented objects. More detail becomes visible, and the subject can ignore the field of concentration less easily. Subjectively, field appears more brightly lit because of the greater detail observed.

Dreaming

Almost all mammals exhibit REM sleep and presumedly dream. Such universality suggests that dreaming serves some purpose, a purpose which it can only accomplish during sleep. Whether dreaming serves a purpose or not, it illustrates that the mind has only one instantiation of the simulation of reality. We do not dream when awake, and we dream only one dream at a time.

Dreams may function as a kind of off-line processing for the brain. Dreams may subject experience to additional feature extraction which could not occur during real time. Experience records vvents and objects without full comprehension. Dreaming later could make sense of the patterns. For example, an animal might be surprised by prey or a predator hidden in vegetation. At that moment he must act immediately, and cannot stop to analyze the situation. If he stopped to observe, however, he could learn something important about camouflage techniques. Dreaming would give him a chance to evaluate his experience later, without losing real-time opportunity.

Lucid Dreaming

Lucid dreamers, "oneironauts" (Laberge), learn to recognize that they are dreaming. Then they easily influence the events in their dreams by conscious thought. Techniques of recognition of the dream state show that the mind functions the same way when dreaming as when awake. The ease with which subjects take control of their dreams, effectively altering their reality at the moment, supports the thesis that thought could impinge on waking perception as well.

A subject learns to recognize when she is dreaming by forming the habit of asking herself, while she is awake, if she is in a dream. The technique LaBerge calls Mnemonic Induction of Lucid Dreaming, (MILD), consists of remembering throughout the waking day to pose the question whenever certain events occur. When the same events occur later in a dream, the subject asks herself if she is dreaming. Then she can determine the answer by attempting acts of magic, such as flying!

Oneironauts can plan their dreams. They can remember questions to ask. Their intentions, verbally expressed while awake, control their dreaming experience.

Stephen Laberge considers the dream world to be a simulation of an environment: "[D]reams are simulations of the world created by our perceptual systems" If we assume that waking experience employs the same perceptual systems, then verbal thought could influence waking experience as well.

The MILD technique shows mental behaviors learned while awake, carry over into dreams. Consciousness during dreams seems to be no different from consciousness when awake. During lucid dreams however, consciousness can control experience.

Lucid dreaming clearly demonstrates omnipotence of thought. The dreamer can control his "world" by thought during a dream. Dream experience may be one source of the belief in omnipotence of thought, either as implicit knowledge, or by actual confusion of the dream and waking states. However, meditation techniques also establish control of perception, in states considered different than sleep.

Complex

A complex appears in a word association test as a cluster of words about a particular topic with a slow, interconnected, emotional response. VP theory can account for the observation of complexes.

The word association test requires the subject to respond with a word which a stimulus word suggests. In the VP model, the stimulus word enters by the Syntax path and the response comes from the SE by the Semantic path. For the process to function requires equalization of salience, which amounts to temporary suppression of emotion. In the case of an emotional topic, almost by definition, equalization of salience will be more difficult and take longer.

Sigmund Freud and others claimed success in treating patients by psychoanalysis, the talking cure. The effectiveness of psychoanalysis lies beyond the scope of this discussion. But if the standard of mental health is the word association test, then just giving the patient practice talking about the topics might bring about improvement. A neural function, such as the postulated function of equalization of salience, can improve with repetition.

Poetry

The emotional quality of poetry distinguishes it from prose. VP theory accounts for the emotional impact of poetry.

Parsing speech requires equalizing the salience of a stream of words. The listener must ignore irrelevant associations that do not contribute to the train of thought. When understanding prose, the VP must suppress these associations strongly, which momentarily delays and weakens their intended, emotional, impact.

Poets choose words for reasons other than syntax and semantics, such as meter, rhyme, or even the appearance of the words on the page. The words and phrases in a poem suggest the words that follow because of those patterns of meter, rhyme etc. that form the structure of the poem. The VP uses less suppression to parse poetry because these associations, which would be irrelevant distractions in prose, happen to suggest the correct words in poetry. Less suppression of association in support of syntax, allows poetry to evoke more emotion than prose.

Subject and Object

A person may normally consider herself to be something quite different from ordinary objects in a way that does not seem to depend on the body image nor on emotions. (See Body and Self) She pictures herself as the observer, while everything else is something observed. She is the subject, while every other mental item represents only an object. As the one who does the experiencing, she considers herself separate and distinct from every experience, and from every object an experience may be about.

Although the subject-object division may seem permanent and fundamental, meditational practices can change this appearance. In the Tibetan Buddhist practice of Samadhi the mind may merge with the environment (Gyatso 1966). In a state called "cittopasthapana" the mind enters the object of meditation, seemingly reversing the subject-object relationship. Meditation can continue to achieve the Summit of Becoming, a state of "neither perception nor non-perception", where no objects exist. The subject and object, the things experienced and the one who experiences, which normally appear quite distinct and separate, can intermingle and disappear during meditation.

The VP theory explains the subject-object distinction as a matter of directed attention. Perceptions certainly draw attention to themselves. They have an intrinsic salience, but the verbal process can alter that salience. The VP can direct attention, making observation seem like an active process instead of a passive one. A person cannot perceive the VP, but she can perceive its effect. The VP, directing attention toward one abject or away from another, seems like a kind of invisible agent, like the subject which the person claims to be.

By this reasoning, the states of Samadhi occur by varying the way the VP redirects attention. Focusing attention on one object to the exclusion of the self causes cittopasthapana described above. Not directing attention at all, extinguishes not just thoughts, but the senses of qualia and self. Accordingly, instructions on meditation frequently stress calming the mind, to stop internally verbalizing, and to learn not to cling to ideas or feelings. Attachment to ideas means increasing their salience, a mechanism required by the verbal process. A continuous narrative therefore makes non-attachment impossible. The states of Samadhi come about by controling and limiting the redirection of attention, an effect of the VP which creates the subject-object division.

Omnipotence of Thoughts

Sigmund Freud (Freud 1950, Totem and Taboo) took the term Omnipotence of Thoughts (OT) from an obsessional patient to describe the magical thinking Freud found both in his patients and in primitive cutures. People consistantly fell into the belief that their thoughts could control the physical world, or at least predict future events.

VP theory explains the occurance of the OT belief as the observation of a subjective effect, the actual effect of words overriding perceptions. Narratives must have frequently modified the SE during evolution, or speech could never have affected critical behavior. (See Verbal Process) The OT belief occurs because people realize that words still can, and do, modify the SE. Things they come to believe can seem just as real to them as what they actually experienced. The erroneous conclusion, that belief actually changes reality, follows easily. The actual fact, that belief can falsify subjective experience, requires greater objectivity to reach.

Perception normally corresponds to reality, and the mind normally cannot control perception. However, students of meditation can learn to visualize with great clarity. The Dalai Lama suggests meditating with eyes open because the visualizations will have more clarity than with the eyes closed (Gyatso 1991). His students can extinguish normal perception and replace it with images developed from concepts.

A Western intellectual may never experience any OT effects. He may not understand the superstitions of others, who claim first-person experience as validation. Descriptions like "self delusion" or "suggestibility" refer to OT effects, but do not explain them. He views the idea of thoughts altering reality with suspicion, (and rightly so!), without knowing why others reason differently.

VP theory predicts the lack of OT effects among Western thinkers. They rely on content evaluation for what they believe, testing concepts in a narrative built up from long periods of education and inculturation. (See Credence) That narrative generally maintains consistancy with physical reality. (This is, after all, the goal of science.) Since they do not often believe thoughts that conflict with physical reality, they have few thoughts which could contradict their perceptions.

In the exceptional case of certain religions, however, Western culture may require a belief that contradicts the evidence. Thus a certain scientist can pursue his own field objectively, while maintaining a belief in young-earth Creationism which confounds geologists. His belief in science represents only part of the narrative he uses for content evaluation. He remains unconvincable on some points, despite his acceptance of the scientific method in other fields.

Magic and Taboo

Sir James George Frazer describes magic and taboo in his epic work, The Golden Bough, (Frazer), as twin aspects of spirtual but natural forces. The sorceror believes he manipulates spirtual forces by natural laws, laws which also apply without his intervention. Homeopathic or sympathetic magic depends on the Law of Similarity which asserts that objects are connected supernaturally to what they resemble. Contagious magic depends on the Law of Contact which asserts that objects once in physical contact remain connected in a spiritual way. Taboos are just the prohibitions necessary to avoid undo consequences of these same forces.

VP theory explains the belief in homeopathic and contagious forces. While the VP can modify perceptions, (See Omnipotence of Thoughts) it does not have an unlimited ability to do so. Ordinarily, people lack the concentration to acheive the OT effect. Events and memories that help concentrate, even force them to think about the subject, makes the OT effect stronger. Believers thinking the physical world has changed, conclude that the mental aids have magic powers.

Emile Durkheim studied totemism (Durkheim) which he considered to be the earliest religion. Among all the peoples he studied, he found a common belief in a division between the sacred and the profane. This division extends to everything. Throughout the world, there are sacred places, sacred objects of all kinds, and sacred days when no profane work can be done.

A spiritual force inhabiting certain objects expresses the same concept as sacredness. Joseph Campbell (Campbell 1964) describes the spiritual force known by different names throughout the world, "Melanesian mana, = wakon of Dakota, = orenda of Iroquois, = manitu of Algonquins, all related to the Roman concept of numen, an 'immanent magical force infecting certain phenomena' " Durkheim concluded, from his study of totemism, that a totem only functioned like a word that identified the clan. The animal or object itself may be quite insignificant, but as a symbol, it contained the magic attributed to it.

The VP theory explains mana, the spiritual force, as the symbolism that aids the OT effect. The totem helps recall the mental state of belief which can change perception. Mana erroneously describes that power of mental association as something within the physical object. Sacred items aid in OT, while profane items pull back toward the physical world. The division between the VP and perception projects into the world as the sacred and the profane. The OT effect, the potential of the VP to intrude into perception, projects into the world as mana.

VP theory predicts that two sources of magic should, in general, aid each other. The credence for magical beliefs depends mainly on source evaluation, rather than content evaluation, becauses they both must actually conflict with experience in the physical world. They both contribute to OT, and since consistancy does not matter, they can always work together. This describes the effect of natural syncretism, the combination of different spiritual beliefs. Beliefs often combine and assist each other, even when they have contradictory mythologies. An example is Santeria where people who consider themselves to be Christians, combine pagan gods with Catholic saints. Santeria flourishes in Cuba and Haiti today, even though the Roman Catholic church does not approve.

Bonding and Ritual

W. W. Sargant studied 'crisis' changes in conditioned behavior, during war, (Sargant 1957) near drowning, and ecstatic rituals like tribal dancing and evangelistic religious services (Sargant 1974). He found that these events made people hyper-suggestible. In that state, they could aquire new values. Typically a ritual would foster "bonding" making the participants feel connected. The ritual may also reinforce their beliefs, or lead to conversion.

Sargant gave an example of a warriors' dance in an African tribe. All young men served a period as warriors in this tribe. As warriors they lived stoic lives, while the elders enjoyed the wealth of the society and married the young woman of the tribe. The warriors danced together until they collapsed. Women were forbidden to dance with them. Sargant reported seeing woman start to dance with the warriors until the older members of the tribe stopped them. Interviews with warriors implied that they felt their society was unfair to them as young men. The warriors consciously believed that the dance helped them to conform to the societical norm. It seemed that the women were not allowed to dance with the warriors to prevent any identification with them as lovers, friends, or even brothers.

Although the rituals could foster allegience, Sargant argued that the suggestability induced in the subjects was open to ideas beyond that of the group running the event. In The Mind Possessed he discussed a young man who frequented the charismatic kind of Christian church services, even though he was a non-believer, because he found he could often seduce young women immediately after their ecstatic religious experiences!

Human behavior includes much cooperation. Most species employ mutual altrusim, cooperation between unrelated individuals, sparingly because of the danger of betrayal (Dawkins). The bonding experiences Sargant describes aim at establishing mutual altruistic behavior.

Walter J. Freeman (Freeman 1995) suggests that bonding may employ the same neurochemical mechanisms as sexual or parental bonding. Indeed, pagan rituals may include sexual orgies, and temple prostitutes (Frazer). Although mutual altruism does not have to use the same mechanism as kin selection, an adaptation for mutual altruism based on the familial altruistic mechanism would have a selective advantage over a separate mechanism. Nature often tests mutual altruism by rewarding betrayal, in situations like the famous "prisoner's dilemma" (Dawkins). A variety which could betray a partner could not arise as easily in a species with a common mechanism, since the ability to betray an unrelated partner might allow betraying kin as well. Mutual altrusim could evolve to use the same physiological mechanisms as affinity for kin with fewer problems of betrayal.

The VP theory predicts that the bonding experiences Sargant described should lead not only to cooperation, but to belief. As mentioned above, (See Credence) people test communications by source evaluation and content evaluation. The bonding ritual improves only source evaluation of the group. The emotional attachment engendered by the bonding ritual, attachment to individuals and to the group as a whole, will make a participant trust them as sources of information, and believe what they say. The rituals appear similar, regardless of the creed of the group, because the rituals themselves do not actually convey the message. A ritual may even help subvert the creed, in the cases mentioned above, because the ritual fosters source evaluation, making the actual participants believe each other, rather than making them believe any particular idea.

Sacrifice

Sacrifice is the most uniquely human, and the most evolutionarily inexplicable behavior. All over the world people traditionally destroy human life, food, and property for religious reasons. "Human sacrifice was THE religious experience in the iron age" (Tierney). What can explain sacrifice, in the light of evolution? Why did natural selection not remove from the gene pool of our ancestors, any variety of homonid that followed such a wasteful practice?

One answer attaches sacrifice to another, overwhelmingly desirable, characteristic. The practice of sacrifice occurs as an epiphenomenon to a characteristic with a selective advantage. Nature cannot delete one without the other. The VP theory identifies that characteristic as language. The power of speech causes the error of omnipotence of thought, which leads man into magical thinking. (See Omnipotence of Thoughts) Magical thinking works best in moments of powerful emotion. (See Bonding and Ritual) Magical acts first surrounded natural death. Then, magicians began to cause death to reap the magic.

Marvin Harris argues, somewhat differently, in Cannibals and Kings (Harris) that sacrificial customs conform to ecological needs. For example, population pressure in a jungle leads to head-hunting and infanticide. A shortage of protein lead the Aztecs to sanction cannabalism by instituting large-scale human sacrifices. This explains very well why sacrificial customs vary from one culture to another. It does not explain why people believe in sacrifice in the first place.

The early prevalence of regicide further confounds attempts to explain sacrifice. Royal sacrifice was a periodic affair in some places, sometimes including all the king's family, servants, and property. Society evolved slowly from this inconvenience. Sir James George Frazer (Frazer) noted customs all over the world which involve a substitute for sacrificing the king. A tendency to kill the unneeded might confer a selective advantage, but killing the leader, and burying the wealth he has gathered, does not.

The religions themselves can always explain their sacrifices. The immortality of the soul or spirit of the sacrifice pervades these explanations. The sacred force, the mana, (See Magic and Taboo) assumes a new mission, after destroying the physical parts releases it. Religions build spritual constructions from the mana of the sacrifice. They do not consider the sacrifice as destruction.

Emile Durkheim (Durkheim) found the divine nature of the sacrifice to override considerations of status or moral condition. Rendering something sacred inherently purified it. Sacrifice even purified the offerings of expiatory ceremonies. A community would first invest an offering with impurities, so as to rid themselves of the sins, and afterward eat the same meat, thinking it spiritually enriching. Even today Roman Catholics still recite:

Agnus Dei, qui tollis peccata mundi, miserere nobis.

Lamb of God, who takes away the sins of the world, have mercy on us. (Vatican)

The sacrifice cannot accomplish such transformation simply by destruction. Even postulating the accumulation of mana left from the flames, does not fully explain it. Instead, I suggest that sacrifice adds to the spiritual picture because it strengthens the omnipotence of thoughts. Practitioners ascribe the omnipotence of thought to an increase in spiritual power, and continue the practice believing that their magic will more than compensate their loss at the altar.

Shamans may know that their practices work through the mind rather than the physical elements. Although exposed to modern religions they may not turn exclusively to prayer or meditation. Patrick Tierney, in The Highest Altar (Tierney), quotes Eduardo Calderón, a Peruvian shaman:

"A [positive] thought-form like Jesus or Buddha can go on for eternity, for as long as people think about him... [forgotten] he'll be dissolved. ...[Sacrifice victim] Clemente Limachi is a thought-form that people are charging up... By praying to him they give him greater potential... He's been crystallized into a deity, a high power... When an animal is sacrificed ... people pray 'Let us have bigger flocks' But it's not the smoke from the animal... it is the united thoughts that produce results."

So "united thoughts" give a sacrifice its power. The effect of emotion on belief, explored by Sargant, explains why a sacrifice can unite people's thoughts. The phenomenon of omnipotence of thoughts, discovered by Sigmund Freud, explains why the practice may seem efficacious. VP theory explains, in turn, omnipotence of thoughts. The practice of sacrifice is an ultimate consequence of the evolution of speech, a trait so valuable that nature selected for it, regardless of the consequences.

Monotheism

Some people consider monotheism an important cultural invention of the Western world. However, Mircea Eliade (Eliade 1959) showed how monotheism tended to arise among many primitive peoples whenever they experienced catastrophe. I will show how the VP theory accounts for the monotheistic effect of catastrophe which Eliade described.

Mircea Eliade identified neglected creator gods among primitive people . Although they may have created the universe, these creator gods normally receive little devotion. The people worship more relevent, though less powerful, dieties of fertility or agriculture in normal times. During a crisis, such as a plague, a drought, or a flood, the people turn to the creator gods in desperation.

Eliade saw the same effect among the ancient Hebrews. In good times they welcomed the gods of their neighbors, Astarte and Baal, (the Golden Calf). During repeated invasions and enslavements, the Hebrews always resorted to their creator god, Yahweh. Eventually Hebrew monotheism became codified into the Levantine religions, Judism, Christianity, and Islam.

The VP theory accounts for this change from pantheon to monotheism during catastrophes. A pantheon naturally arises during prosperous times as people seek specific cures for specific problems. The various deities offer myth and ritual aimed at common desires, modifying the SE of believers efficiently by concentrating on narrow, isolated, concepts. One god cannot work as well on all desires as a pantheon, because a pantheon provides a specific symbol tailored for each need. Totems and myths accumulating around different gods, acheive control over all aspects of life. The individual gods work efficiently, but each works only in his respective sphere.

During a crisis the pantheon fails. The emotions of recent experience have risen, making the SE harder to modify. The individual gods of the pantheon can no longer overcome the reality of perception. The gods fail because each works only in a limited sphere. The symbolism of all of the gods collectively would have enough credence to modify the SE, but the gods do not all work together. Because of the separate myths, totems, and rituals, a believer can summon only a small part of the total symbolism of his pantheon at once.

An omnipotent creator god provides a panacea for catastrophes. Reducing the number of gods allows concentrating more symbolism on the crisis. The believer can attach enough credence to one single myth to overcome the powerful emotion of the crisis. The one god appears as the creator of the universe, because he opposes the nature of the entire universe in the SE. He recreates the animals and plants during a drought; he cures the sick and raises the dead in an epidemic.

Theology

Both Frazer and Durkheim separate magic and religion, but others consider this separation untenable (Lowie). Elements suggestive of religion occur in the "rudest cultures". Indeed the theistic confession of sins, asking forgiveness from a supreme being, occurs among the Sanpoli and Nespelen (Salish Indians) whom Wilhelm Schmidt ranked as Urvölker, or oldest peoples. (Eliade 1987)

Yet, modern religions have moved beyond magic and sacrifice, each using its own schema. While Christianity has sought to replace paganism, the Roman Catholic tradition incorporated some of the old goddess images into the Virgin Mary (Frazer). Tibetan Buddhism incorporated dieties and customs of the old Bon tradition to spread its message. The need for advanced theologies and the reason for their success is obvious from the practices they replace. Religion fills a real human need, a need for magic, sacrifice, and explanation for the apparent omnipotence of thoughts. To tame these phenomena they have replaced totem, magic, and sacrifice, with Communion, Kosher food, circumcision, myth, and prayer. Modern religions control the human tendencies to magic and sacrifice, and the continuous success of religions as social endeavors, proves the power of those tendencies toward magical thought.


Neural Correlates

Single Instantiation of SE

The structure presented here assumes that there is only one SE, that the simulation has only one instantiation. Subjectively, this seems correct; we only experience one world at a time. Although we dream, probably for a reason, we never have several simulataneous dreams while leading one wakeful life. Some solid evidence indicates a single instantiation as well.

The brain may use a topological coding, one neuron may represent one object or one location. (See Topological vs Temporal Coding) To reinstantiate an existing data structure requires duplicating the initial data which in this case represents any relevant knowledge of the world. A spatial representation would make duplicating large blocks of data difficult. Even if the data is in temporal codes, the structures controling the duplication should be visible, like the bus lines in a computer.

The cerebrum has two hemispheres which can operate in parallel, and do basically the same tasks. Generally, only one, dominant, hemisphere will speak and have full command of language, but either hemisphere can develop the ability. Hemispheric specialization did not evolve expressly for speech, because it has been found in other animals (Hamilton and Vermeire). Studies on post-comissurectomy patients, split-brain patients in whom the corpus callosum has been cut, have shown that both hemispheres can read and respond to written or heard instructions. Young children who have had one hemisphere removed because of epilepsy can learn to talk again with the other hemisphere.

The fact that only one hemisphere masters speech, while either one could, indicates that speech detracts from other processing. The speech functions would have global impact if, as suggested above, the verbal process uses the one and only instantiation of the simulation for semantic meaning.

Topological vs Temporal Coding

The way the brain encodes information is still disputed. The brain may use a temporal orspatio-temporal coding, or a topological coding. In a temporal coding the timing of a signal identifies its content. By contrast, a topological coding assigns a meaning to a particular neuron or synapse. Morse code is a temporal coding, while checkboxes on a paper form are a topological coding. Recent research indicates the brain uses topological encoding. (See Lesion and Imaging Studies) Empirical reasoning supports this conclusion as well.

Topological coding is consistant with the VP theory. VP theory proposes a single instantiation of the SE, the simulation of reality; the theory draws conclusions about a brain which simulates only one world at any instant. If the brain used only temporal coding, it might create multiple instantiations which exchange information. The various world simulations could copy each others signals, and so obey the same rules and contain similar objects. However, if particular neurons and synapses, "grandmother cells", represent the features of the world, the brain could not arbitrarily copy those features elsewhere to create similar world simulations. A topological code limits the possibilty of multiple instantiations of the SE, consistant with VP theory.

Empirical reasoning based both on the functioning of synapses and the bandwidth of neurological circuits, suggests the brain uses topological encoding. The accepted model for memory, first described by Donald Hebb, depends on varying the strength of synapses. When the presynaptic neuron successfully activates the postsynaptic neuron, the synapse becomes stronger, able to repeat the transfer with less stimulation. Synapse strength varies with both temporary, working storage activation and long-term potentiation (LTP) for permanent memory. Since the conductivity across a synapse varies, it cannot reproduce an amplitude pattern nor a temporal pattern.

Computers, by comparison, do use temporal coding, but the frequencies in electronic circuits are thousands of times higher. A computer bus signal typically consists of 32 signal lines carrying 10 nanosecond pulses, each line fanning out to perhaps a dozen destinations. With each pulse, the destination and meaning of the signal changes. The brain operates much more slowly, using wider signal paths. The spike trains of neurons are measured in milliseconds, but the billions of cortical neurons can have thousands of synapses. The computer uses temporal coding because it has many cycles and few signal paths. The brain, by contrast, has many signal paths, and very few cycles. The brain can most efficiently encode data topologically, dividing the message over the many signal paths, rather than dividing the already very low frequency.

In the computer, control signals are carried on separate lines. The data is time-division multiplexed into a high-frequency stream. If the brain uses massively parallel, separate lines for its data, control signals would be temporally encoded. The nature of the spike train code is therefore process-control signals, eg activate a neuron, write the synapse, read the synapse back. Note that in this view synapse strength does not direct information like a switchboard, but rather the synapse connections themselves contain the knowledge of the world.

Topological coding in the cerebral cortex does not rule out temporal coding in peripheral nerves. Sequential patterns can be detected and generated by delay lines made of binary elements. Electronic signal processing circuits convert serial patterns to parallel and vice versa using shift registers, assemblies of flip-flops or one cycle delay lines.

Arnold Trehub, in The Cognitive Brain, shows how a spatial pattern can be learned and recreated by Hebbian action on synapses. In his model, Hebbian cell assemblies create bidirectional links between features and abstractions. These links implement the database-like functions of the brain's associative memory. (See Appendix II for basic information on databases) A Detection Matrix activates Class cells, one of which learns each new stimulus pattern presented by an Imaging Matrix. The Class cells then feed back to the Imaging Matrix, fanning out over the cells to make connections to the active ones during learning. Later, the Imaging Matrix can activate the best match of Class cell to recognize a pattern, or a Class cell for a given object can re-create features into the Imaging Matrix.

Lesion and Imaging Studies

Research on monkeys by Michael Graziano and co-workers (Graziano) supports the existence of a model of reality in the brain. They identified neurons in the ventral premotor cortex which correspond to a monkey believing an object is in a particular place, relative to the head or to a hand. The activity reflected the animal's belief that the target was there, not just its perception. The neurons responded to seeing a target close to a spot on the body, feeling a touch there, and for some neurons even hearing a nearby sound. For example, some neurons increased their firing rate when the experimenter touched the monkey's face, or brought a visible target near to the face. If the head was turned while the target was visible, the rate reverted to its baseline value, until the head was turned back. The experimenters then shut off the lights, and observed that neuron kept responding in the dark, even after the target was removed. Only when they turned the lights back on, so the animal could see the target was gone, did the firing rate return to baseline.

Graziano's research proved the existence of a model of objects in space, in this case in a coordinate system centered around the body. It also showed that the model of space is topologically encoded, with neurons representing points containing objects.

Hannah and Antonio Damasio and co-workers studied lesions causing selective aphasia, the inability to retrieve certain words (Damasio 1996). The Damasio's studied a very large database of lesions to demonstrate that in selective aphasia, the semantic class of words affected reflects the location of the lesion. They went on to show by a PET activation experiment that normal subjects use the corresponding areas to generate the same classes of words. Thus words for tools appear to reside in the posterior section of the left inferotemporal (IT) region, words for animals in the anterior IT, and words for people in the anterior temporal pole.

Topological distribution of word generation ability suggests topological representation of words and concepts as well. Synapses would connect the cell that represents each word to cells representing the semantic meaning of the word. As the brain learned to speak, each new word required selecting a cell to identify the word. A cell in general proximity to cells representing the concepts of a word may already have some innate connectivity to those cells and so become the best choice to represent that word. Word cells would tend to cluster by their semantic class, and in regions near to the basic brain functions associated with their class.

Research has shown some topological encoding in all animal brains. For example, olfaction, discussed by K.M. Dorries in Neuron, Jan 1, 1998 (Dorries) can function on a topological pattern. Discrimination between closely related smells does require intact spatio-temporal patterns. But Walter J Freeman (Freeman 1991) has shown that such patterns vary whenever an animal learns any new stimulus. The temporal activity includes far more than the odor itself. Perhaps a sequential series of comparisons to identify the odor presented creates a consistant pattern, a pattern which varies whenever learning new stimuli adds new possibilities to the search.


Conclusion

The Hard Problem As the Gordian Knot

Like the mythological Gordian Knot, the Hard Problem preserves a vestige of ancient mysticism. A prophacy made Gordias the king of Phrygia. An oracle then foretold that whoever untied the knot on King Gordias oxcart would rule all of Asia, conforming to a principle of magic that powerful spirits can be restrained by cords and knots (Frazer) . Alexander the Great cut the Gordian Knot and ruled by profane force in lands where sacred prophecy had reigned before. Today technology advances by sharp reality into provinces once held only by the human spirit. The Hard Problem offers a refuge from the onslaught of computation.

Some people may not welcome a suggestion that computational processes create subjective experience. Computational theories cut the Hard Problem profanely, instead of unraveling a spiritual mystery. They explain consciousness as an epiphenomena, turning the soul into an illusion. Yet the spiritual traditions pose similar enigmas. Hinduism teaches that reality itself is only the illusion of maya, which imprisons the atman, or spirit. Buddhism teaches a doctrine of "anatman" or "no soul", in which, although the spirit may be born again and again, the continuous self is only an illusion created by attachment to this sorrowful world.

The mystical and physical theories form a duality; each views the opposite realm as somewhat of an illusion. It requires no less faith to accept that the physical world is an illusion, than it does to conceive of the mind as a computation. The Bhagavad Gita maintains that the worlds of matter and consciousness are grounded in the same spiritual reality. The physicalist, scientific, viewpoint must, in turn, explain not only why people experience consciousness, but also, why people often believe that spirit creates the physical world.

Summary

The Verbal Process theory depicts a model of the human mind which accounts for subjective consciousness, and which explains why people believe in the supernatural. The model presupposes that all mammals use one Simulation of the Environment (SE), a mental reproduction of reality, for experiencing, dreaming, remembering, and planning. Adaptation of the human brain for speech created a Verbal Process which manipulates symbols in the same SE. The effects of sharing perception with verbal processing creates the appearance of qualia, free will, a sense of self, and the omnipotence of thoughts (OT), the illusion that mind can change reality. The OT effect suffices to explain the common belief in magic and spirits. Thus the theory not only explains how subjective consciousness can arise from computational processes, but explains why many people insist on dualistic or spiritual explanations for consciousness instead.

References

Bauer, F.L. and Eickel,J. (1974) Compiler Construction (NewYork: Springer-Verlag)

Campbell, Joseph, (1949) The Hero with a Thousand Faces (Princeton, N.J. : Princeton University Press)

Campbell, Joseph, (1964) Occidental Mythology (New York : Penguin Books)

Colby, Carol L. (1998) 'Action-Oriented Spatial Reference Frames in Cortex', Neuron, Vol 20(1), Jan 98 (Cambridge, MA : Cell Press)

Damasio, Antonio R. (1994) Descartes' Error : Emotion, Reason, and the Human Brain (New York : G.P. Putnam)

Damasio, Hannah et al (1996) 'A neural basis for lexical retrieval', Nature, Vol 380 , 11 Apr 96

Dawkins, Richard, (1989) The Selfish Gene (New York : Oxford University Press)

Dorries, K. M. (1998) 'Olfactory Coding: Time in a Model', Neuron, Vol 20(1), Jan 98 (Cambridge, MA : Cell Press)

Durkheim, Emile (1915) The Elementary Forms of the Religious Life / Translated by Joseph Ward Swain (New York: Free Press)

Eliade, Mircea (1959) The Sacred and the Profane; the Nature of Religion / Translated by Willard R. Trask. (New York, Harcourt, Brace)

Eliade, Mircea (1987) The Encyclopedia of Religion / Edited by Mircea Eliade (New York, Macmillan)

Frazer, James George, Sir, (1913) The Golden Bough; a Study in Magic and Religion. (New York : Macmillan)

Freeman, Walter J. (1991) 'The Physiology of Perception' Scientific American, Vol 264, (2) Pgs. 78-85, Feb 91

Freeman, Walter J (1995) Societies of Brains: A study in the Neuroscience of Love and Hate (Hillsdale, NJ : Lawerence Erlbaum)

Freud, Sigmund (1939) Moses and Monotheism / Translated by Katherine Jones (New York: A.A.Knopf)

Freud, Sigmund, (1950) Totem and Taboo: Resemblances Between the Psychic Lives of Savages and Neurotics / Translated by A. A. Brill (New York: Dodd)

Freud, Sigmund (1960) The Ego and the Id / Translated by Jon Riviere (New York : W.W. Norton)

Gyatso, Tenzin, H.H. the Dalai Lama (1991) Path to Bliss: A Practical Guide to the Stages of Meditation / Translated by Geshe Thubten Jinpa (Ithaca, NY : Snow Lion Pub)

Gyatso, Tenzin, H.H. the Dalai Lama (1966) The Opening of the Wisdom Eye (Wheaton, Ill : Theosophical Pub)

Graziano, Michael S.A.; Hu, Xin Tian; Gross, Charles G.(1997) 'Coding the Locations of Objects in the Dark' Science vol 277 , 11 Jul 97

Hamilton, Charles R; and Vermeire, Betty A. (1988) 'Complementary Hemispheric Specialization in Monkeys' Science, Vol 242, 23 Dec 88

Harris, Marvin (1977) Cannibals and Kings : the Origins of Cultures (New York : Vintage Books)

Jaynes, Julian (1976) The 0rigin of Consciousness in the Breakdown of the Bicameral Mind (Boston : Houghton Mifflin )

Kappas, John G. (1987) Professional Hypnotism Manual (Van Nuys, CA : Panorama Pub.)

Knuth, Donald E. (1973) The Art of Computer Programming (Reading, MA, Addison-Wesley)

LaBerge, Stephen (1990) Exploring the World of Lucid Dreaming (New York : Ballantine Books)

Lowie, Robert A. (1924) Primitive Religion (New York : Boni & Liveright)

Paivio, Allan (1971) Imagery and Verbal Processes (New York: Holt, Rinehart and Winston)

Pinker, Steven (1995) The Language Instinct: How the Mind Creates Language (New York : Harper Collins)

Powers, W.T. (1973). Behavior: The Control of Perception. New York: Aldine Publishing Company.

Sargant, William Walters (1957) Battle for the Mind, a Physiology of Conversion and Brain-washing. (Garden City, NY: Doubleday)

Sargant, William Walters (1974) The Mind Possessed; a Physiology of Possession, Mysticism, and Faith Healing (Philadelphia,PA: Lippincott)

Tart, Charles T. (1969) Altered States of Consciousness; a Book of Readings. (New York: Wiley)

Tierney, Patrick (1989) The Highest Altar : the Story of Human Sacrifice (New York : Viking/Penguin)

Trehub, Arnold (1991) The Cognitive Brain (Cambridge, MA : MIT Press)

Vatican (1969) Missale Romanum

Winckelgren, Ingrid (1992) 'How the Brain 'Sees' Borders where There are None' Science, Vol 256, 12 Jun 92

Appendix I - Control Systems

There are two types of control systems, called open-loop and closed loop systems. An open-loop system relies on "ded-reckoning"; it computes how to reach a destination from its previous location. The open-loop process is completely computational, feed-forward only. A closed-loop system uses feedback. It continuously estimates a course to take it from the curent location to the target. Closed-loop control systems must measure the controled variable, and open-loop systems must model the way the controled variable changes.

Equation (1) defines a general open-loop system, showing that the path of motion is fixed by the starting point and the destination:

(1) x = G(g, x0, t)

where

x is the controled variable

g is the goal

x0 is the starting point

G is a precise function of time and goal

Equation (2) defines a general closed-loop system. It shows that the speed depends on the distance to the target.

(2) dx/dt = H(g-x)

where

x is the controled variable

g is the goal

H is a function of distance

In the closed loop, the function H does not require much accuracy. It must simply point in the right direction, and equal zero at the target:

(3) H(0) = 0,

H(x) > 0 for x > 0

H(x) < 0 for x < 0

The closed-loop compensates for variations in the function H() by feedback; it continuously compares the current location to the target. In the open loop, by contrast, the function G() must be extremely accurate or the motion does not achieve the exact goal.

A complex animal, such as any mammal, must use some closed-loop systems. Engineers can design machines entirely around open loops, but they do so by "calibration". They carefully adjust each unit during manufacture so that G() is precisely what it should be. A developing embryo does not have that luxury. In fact all systems must be initially closed-loop because, unlike a machine, the components of an animal, muscles, tendons, bones, and nerves, are not precise nor constant enough to hold a calibration.

Once closed-loop motion exists, the animal can begin to use open-loop systems. She can do this by memorizing an open-loop function G() which mimics the closed-loop behavior. She first makes the motion with a closed loop system, and remembers the motions she made. Later she can repeat exactly the same motions again, without monitoring, to achieve the same result. Thus actions become automatic by practicing.

So, to move the body, or work towards any goal, an animal must first create a closed-loop system. A closed-loop system requires expressing the goal, g, in the same terms as the physical state, x, so they can be compared by subtraction. The value of any physical state which can be controled, must therefore be represented in the brain.

What is the form of the representation ? It can differ somewhat from reality but Equations (3) constrain it to approximate the physical state. Let x´ stand for the internal representation of x . For example , x´ represents what a distance looks like to the animal, in contrast to x which means what a ruler would show. Since the animal perceives x´, rather than x, goals should be expressed in terms of x´ as well. Using g´ for the internal representation of a goal, Equation (1) becomes:

(4) dx/dt = H´(g´-x´)

For H´() to meet the restrictions of Equations (3) imposes a simple condition on x´. The condition is that x´ must be "monotonic" with respect to x. This means that as x increases, x´ also increases. If this were untrue at some location, attempts to move to that point would fail, because the direction of motion would change when nearing that point.

So control systems require some representation, x´, of the controled variable, x. Although open-loop systems may function automatically without containing x´ at the moment, setting up the open loop required x´ initially. While x´ many not be exactly the same as x, the two are closely related.

Appendix II - Databases

The associative memory of the brain resembles a computer database. For example, if you think of a tomato you can easily name some of its attributes: smooth, red, round, edible vegetable. Given these qualities the mind can identify the original object, a tomato. In fact a subset of qualities can usually suggest a list of items that fulfill the partial description; we can easily name a few red and edible objects, such as rhubarb, apples, peppers and tomatoes.

A database stores information about an entry, such as a prototypical tomato, in a "record" divided into fields:

Name: Tomato

Color: Red

Shape: Round

Surface: Smooth

Edible: Yes

There will be a large number of such records, for example records for apples, oranges, tomatillos and chirimoyas. The problem of database management involves quickly finding the records containing some attribute such as "Color:red."

Computers connect records by "links." A link just shows you how to move from one field to another. The most common type of link in computer systems is an address pointer, a field containing the address of another field. An arrow designates a link:

Red Vegetables -> Tomato

It should be pointed out that the fields of a record, such as the description of a tomato above, are actually connected by some kind of links. The computer engineer will say they are "sequential", one field right after another, and so he does not need to add any pointers. But in fact they are still linked by the "addressability" of the machine. Each field has a numerical address, and the next field in that record starts at the next available address. ( Note that in 'C' "number++" just adds one, whereas "address++" calculates the next address, which may require a greater increment.)

Ordinarily links go in only one direction. They do not automatically go back. Thus "Red Vegetables" above leads to "Tomato" but "Tomato" does not lead to "Red Vegetables". To go both ways, combine two links, in opposite directions. This is called a "bidirectional link" and written with a double arrow:

Love Apple <-> Tomato

Computer databases can use a variety of techniques to tie records together. Most of these are not appropriate for a brain to use. Circular linked lists use just one pointer to the next record, but if you go far enough you can reach the previous record:

Apple -> Pear -> Orange -> Apple

where "Apple", appearing twice, denotes the same entry.

Index tables use a sorted list of entries, just like an alphabetic index in a book.

Apple -> Red

Berry -> Blue, Black

Cherry -> Red

To employ such a list efficiently, a programmer uses a "binary search"; stepping through each entry to find the desired target becomes too slow as the list approaches capacity. The binary search starts by looking in the middle of the list. Since the list is in order one check tells whether the target is in the first half or the second half of the list. The search continuous by looking in the middle of the appropriate half. Repeating this process narrows the search down to one entry in short order (in a time proportional to the logrithm of the number of entries).

A binary tree resembles an index list with the searching function already built it. Each entry has exactly two links:

Date -> Berry, Fig

Berry -> Apple, Cherry

Fig -> Eggplant, Goober

You search the binary tree by starting from the first entry, choosing it, or following one of the two links depending on how the target compares to the entry, and continuing until a match is found.

There are more data structures used by computers, but they all take advantage of characteristics lacking in the brain. Circular linked lists are a conceivable static structure, but they pose difficulty when elements must be added to a list. Somehow a link between two elements must be broken, the ends remembered, and then hooked together through the new element. The index list has the same problem adding an element because there are implied links between the entries of the list, elements which computer engineers store sequentially. Further the binary search counts on being able to jump ahead a number of entries. In computers this is done by address arithmetic, an impossible feat when there are no address numbers! Finally the binary tree looks more promising but it has problems also when the list grows. A link has to be erased to add an entry, but a worse problem can occur. Suppose a small category suddenly becomes much larger, then the tree becomes "unbalanced" If for example we add:

berry,goose

berry,elder

berry,blue

berry,black

all the new entries can wind up strung out in a line in front of Cherry. For statistical reasons, unbalanced trees usually become only a few layers deeper (Knuth). This often poses no problem in computers which process the entries sequentially, but in a brain, where neuron speed is critical, accessing through any extra layers would be a handicap.

Because computers have many features unavailable to the brain, features such as computed addressing, unlimited temporary storage, easy erasure and rewriting, and rapid pointer accesses, machines can use many techniques to create database structures. Bidirectional links are the most likely structure available to the brain.

Conclusion


Please Email any comments to: Harland Harrison (HarlandH@HarlandH.cnc.net)