1. Title: PSYCHOPATHOLOGY AND COGNITIVE SCIENCE

2. Author: Henrique Schützer Del Nero

Postal Address: Rua Rubens do Amaral, 346. ZIP: 05653-010. São Paulo-SP-BRAZIL

Phone numbers: (55)(11) 8431109. E.mail: hdelnero@usp.br

3. ABSTRACT

The authors try to investigate the underpinnings of modern psychopathology as it has been used by the psychiatric community, suggesting that there is a focus of incompatibility between the way Psychiatry deals with mental diseases and the theoretical foundations of Cognitive Science. Both Traditional Artificial Intelligence and Connectionist Artificial Intelligence are based upon what we call semantical priority. Despite the differences between these two models, mental terms are supposed to be non-reducible to brain states. Psychiatry, dealing at the same time with drugs that act at the quantitative-implemental level, and symptoms that are qualitative reports that come from the patient's mind using language as a tool, represent a concrete problem that doesn't allow functionalist positions to be plainly accepted. If reduction is impossible at the level of mental terms, and if the reduction of theories is still impossible, because there are no mature theories of brain and mind, how could Psychiatry receive the benefits of such a broad science of mind like Cognitive Science? The authors propose that a third kind of models, based upon the neuron dynamics could allow an intermediate level of reduction, not so precise as terms, not so broad as theories. The level that could be reducible would be that of syndromes, sets of symptoms and signals that scaffold medical diagnosis.

4. key words: psychiatry, syndrome-reduction, consciousness, mental treatment, diagnosis

5. Talk or poster presentation. Talk preferred.

Psychiatry (P) can be seen as one of the biggest challenges to Cognitive Science (CS). It brings forth, in a crude way, the problems that concern the dichotomy mind/brain, software/hardware, acquired/hereditarial, reactive-developmental/endogenous, interpreted/non-interpreted, implementation/ algorithm-computation, etc. Far from being a mere list of references that crowd the cognitive literature from the very beginning, the above dichotomies reflect the daily problem of any psychiatrist or neurologist that has to decide whether something is to be treated with drugs or psychoteraphy or both. From a different perspective, while using drugs, what one sees is something that acts at the implemental-brain level in a quantitative way, changing dramatically the qualitative character of qualia and of inner feelings, thoughts and will.

Emotions-feelings, thoughts-cognition and will-conation are the three elements that characterize the mind (Ryle, 1949). Thence any science that deals with mental phenomena, be it the transdisciplinary CS, be it P, must share a body of common concepts regarding their foundations. The reader might reply saying that Psychology and the Neurosciences are already present, which could include psychiatric problems.

We'll try to explain, in this brief article, that there are problems concerning the foundations of P that relate to CS and have not been stressed as urgent methodological questions that ought to be presented to the psychiatric community. On the other hand, the CS community sometimes disregards the practical problems that functionalistic approaches (typical of Traditional Artificial Intelligence-TAI) cause upon the scientific approach of today's P.

P is a very broad discipline. Falsely, it includes or presupposes psychodynamical treatments like Psychoanalysis. On the contrary, it is deeply grounded in the medical tradition and has to do with pathology, as every other branch of Medicine. However, the common couple that scaffolds the medical diagnosis-- signals and symptoms-- is more complex than in other medical specialities. Signals are objective marks that can be seen by the physician without being put forth by the patient. Symptoms are subjective reports given by the patient. Fever is a sign. Pain is a symptom. Relatively invariant sets of symptoms and signs constitute syndromes. Syndromes are quasi-sets of symptoms and signs that imply a large number of potential diseases that may cause that syndrome. Water in the lungs (sign), swollen legs (sign), increased liver size (sign), breathlessness (symptom) may characterize a syndrome called congestive cardiac insufficiency (CCI). There are many heart diseases that cause CCI. That's why, concerning mental pathology, psychiatric disorders and cognitive-conative-affective abnormalities is a very difficult problem. If one considers the medical style of approaching signs, symptoms, syndromes and diseases, one can see the layman-style people use to talk about the mind and its anomalies. CS as a legitimate effort to launch the basis of a genuine science of the mental must face the problem of explaining anomalies.

1.The relation between mind states, brain states and linguistic tools.

Signs are tied to a behavioristic way to consider high mental functions. If a psychiatrist says that somebody has a depressive facies (face), this is embedded in a body of knowledge that pursues the diagnosis of a depressive syndrome. It is tied to the behavioristic tradition of considering objective-external marks as the substratum of behavior and its anomalies. The mind is precluded to exist by the behavioristic credo, something that was strongly denied and combated by early CS. However, the physician seeks other information to make the diagnosis of a depressive symptom. He/she, asks about the character of the mood, e.g. if the patient feels anxiety or not, lust or not, motivation, etc. The myriad of symptoms that can be linguistically reported by the patient (some of them may be checked indirectly) involves two major problems of the philosophy of mind, and thence of the foundations of CS: are there internal representations-sensations, etc.? And are they compatible in terms of reference with the linguistic tool that the patient uses? In other words, are we dealing with the same internal representations and feelings when we use the term anxiety or sadness? What is the relevance from the psychiatric point of view to distinguish subtleties like restlessness, apprehension, fear, a vague discomfort regarding the future, a feeling of a narrowing-horizon, a senselessness for small things, a strong liaison to the past, a feeling of overvaluating everything that has passed, a little subjective report of changing the character of minimum experiences like tasting or smelling? Are these reports linguistic variations on the same theme that urderlie poetry but don't refer to different brain-psychic states? P deals with signs that are insufficient to make a diagnosis and with verbal reports that are under suspicion if one makes the correlation between verbal report, neural substratum and mental substratum. One might say that the linguistic entities refer to psychic referents but these referents don't have brain referents: one might say that restlessness is different from apprehension, and that they have different mind substrata, but at the same time that both have only one brain substratum or referent, e.g. hyperactivity in certain attentional areas. That is why, from the psychiatric point of view, depression relates to a mood disorder and from the neurophysological point of view, many researchers prefer to talk about attention and its disorders. Depression would be therefore a disorder of attention, since attention is closer to what is known and replicated in animal experiments, because the neural substratum is more known and the external signs are easier to be observed.

This first list of problems tied to mental referents, brain referents and linguistic tools is sufficient to show that the problem the psychiatrist deals with is more complex. Chest pain is more objective than disappointment. Ask a father if he has lumbar pain and you will have a more objective answer than if you ask him if he got disappointed because his son didn't call him up. Both are symptoms, from the formal point of view described above, both are statements about inner, internal, exclusive, subjective experiences that can be linguistically communicated. Inner sensations regarding the whole body tend to be more objectively and plainly handled than those that are related to his or her own life. Life here means not only a chronological sequence of body events and their mental representants (one could say that referring to a lumbar pain is the same as referring to a feeling of self-ruin) but a structure where values, expectations, social pressure, habits, etc. tend to display a crucial importance. That's why if it is difficult from the philosophical point of view to build doctrines about the relation between language, brain states and lumbar pain, it is much more difficult to build doctrines that relate brain states, disappointment, its confession, and the linguistic tools that allow the communication between patient and doctor.

The first problem with high order mental objects, those that are related and constitute psychiatric symptoms, is that there might be a hierarchy between mental states like pain and mental states like self-esteem. Apart from the linguistic subtleties that surround more complicated mind reports, our supposition is that mental objects, although being from the same class, and hence imposing the same philosophical problems tied to the philosophy of mind, have a hierarchy of complexity. Mental symptoms, linguistically communicated, that refer to the liver are more objective and transparent than those that refer to the inner self image, located in a physical, historical and ethical world.

CS must pay attention to this virtual hierarchy and the possible candidates to be the underpinnings of a mental structure that has degrees of complexity relating to a brain that stays underneath and to language that carries it outward.

2.Functionalism x reductionism: drugs and rapport

P pays a large tribute to the tradition of Descriptive Psychology that comes from the XIXth century (Jaspers, 1920). Notions such as intentionality that are found in Brentano and in the first Husserl (1900) (Phenomenology) largely impressed the way of describing large number of inner representations and their relation. For example, the concept of catathymic delusion is interesting because it is a kind of delusional structure that has a depressive mood underneath, being a depressive syndrome (hence a mood disorder) nevertheless looking like a primary thought-disturbance. The character of the delusion, its contents, and a series of other signals and symptoms constitute this concept-syndrome. Descriptive Psychology and Psychopathology try to describe mental blocks and their flow building thoughts, scenarios, types, etc. Categories like personality, temperament, humor, thinking, mood, affectivity, judgement, critics, perception, consciousness, orientation, etc. that constitute the typical elements of a psychiatric examination are descendants of the XIX century when mental faculties proliferated. It was against this proliferation that behaviorism, the philosophical (from the Vienna Circle) and the psychological (from Pavlov, Skinner among others), appeared. CS as a reaction against behaviorism didn't resurrect these old styles because it laid its foundations in the concept of mental representation and of thinking as computability. Introspection gave way to formal analysis of speech reports (protocol analysis). Old P with its glamour and numerous failings gave way to a purged psychopathology well represented by the sevaral editions of the DSM (Diagnostic and Statistical Manual of Mental Disorders 1994), a list of symptoms and signs that are grouped forming syndromes and diseases without theoretical underpinnings (the idealizers are proud to say that DSM is atheoretical!). Drugs advanced in terms of treating major psychiatric disorders and psychotherapy abandonned the realm of Psychoanalysis and became "cognitive". In other words, P advanced in the comprehension of the brain structures that urderlie mental disorders, creating better drugs. Psychotherapy became more tied to a behavioristic 'retrainment'. Psychophathology became a list of symptoms and signs without strong stylistic roman-driven relations. Hermeneutics and the art of interpreting semantic similarities, analogies, became discredited. But P doesn't have at this moment a body of paradigms that can explain the coexistence between drugs that act at the synaptic-implemental level, promoting quantitative changes in the firing rate of neurons and qualitative improvement at the mental level. Psychiatrists still know that talk-therapy is recommended, but they cannot explain how they decide to give one (drugs), another (talk- therapy) or both. Ecumenical approaches might say that both are better, acting in synergy. It is wrong, and science pursues boundaries and structure. One has to explain why, how, when, etc. Top-down psychotherapy and bottom-up psychopharmacology, without any paradigm that unifies both (but the empirical in a very poor sense scientifically speaking) are the very signs that P embraces functionalistic approaches regarding talk-empathy-therapy and reductionist approaches regarding drug prescription. It is serious because they are not rival fields like Traditional AI and Connectionist AI. They live and act together. For practical purposes it works, but from the scientific point of view it is necessary to make a choice and to explain things in order to scaffold the basis of a strong P, as it is desirable to have a strong Psychology. As one can see it is not only a matter of credo but it is a matter of health, its criteria, boudaries, etc.

The argument of multiple instantiability that supports functionalism, i.e. the same program can be run in different architectures, thence knowledge of the brain mechanisms is not necessary to comprehend the logic of the level of algorithm and of computation. To understand thinking is a matter of defining the atoms and the rules of inference that allow strings to be made, therefore deduction, induction and even abduction. Functionalism is a very strong position and our supposition is that there is no reply for it. Fodor (1975) states it clearly, and even Hooker (1981) doesn't destroy completely the strength of this argument. The price to pay for abandonning the functionalist view is to propose a kind of eliminative materialism that destroys common ordinary language, the language that scaffolds the report of symptoms a psychiatrist needs to make the diagnosis of a psychiatric syndrome. But there are more problems with the functionalist view regarding daily psychiatric practice. If there is genuine dissociation between the laws of the brain and the laws of the mind, what is the function of drugs in the treatment, if they cause a quantitative modification at the implemental brain level with qualitative changes in the algorithmic-computation-mind level? Of course, the functionalist opponent could reply that if there is dissociation it doesn't follow that there is no relation, e.g. token-identity. Functionalism would in this case deny only that there is type-identity between brain states and mind states, precluding a radical translation or reduction from one vocabulary unto another. Token-dentity, qua monistic from the essence point of view, insofar dualist from the predicate point of view, might accept that given the human brain there might be a relation that endorses the use of drugs at the quantitative-brain level with modifications at the qualitative-mind level. Functionalism, we reckon, would not accept that the laws governing human psychopharmacology would be equivalent to the laws governing machine psychopharmacology, i.e. the principles that govern the human brain level and may cause a delusion at the mind level are not the same as the principles governing a chip-silica substratum that may cause the same delusion in a machine. The delusion is the same from the structural point of view, since it occurs at the mind level of the human and of the machine, but their substrata, thence their treatments, are different. This could be compatible with functionalistic multiple instatiability arguments and with token-identity, but it would follow that the study of the human brain and psychopharmacology are things that interest Medicine and Psychiatry and not a broad Science of the mind and its pathologies. To study a general pathology, useful to humans and machines would imply to study things at the mental level, their structure, atoms, relations, maybe in a logiscistic way (e.g. Predicate Calculus). But this would turn psychopathology back to the style of the XIX century, when Descriptive Psychology tried to grasp the structure of mental illness.

A problem arises if one considers the fact that there is a mutual interrelation between mental states and brain states. Depressive states may be caused by environmental events or by cerebral events. There is a slight difference between reactive depression (caused by environmental contextual events) and endogenous depression (caused by brain altered events), but for practical purposes one may say that they can be indistinguishable, since: a) there is almost always an environmental event that preceeds an endogenous depression; b) there is the possibility that a reactive depression assumes the shape of an endogenous depression (what Psychiatry pre-DSM called vitalized depression). Then, it is possible that the relation of token-identity and of "mutual causation" (if it makes sense to talk about causation between states-predicates of the same substance) renders crucial the study of the human brain subserving psychic disorders and if there are disorders that can begin by the mind pole and reshape cerebral-endogenous depression, there might be something stronger that correlates mental structure, accessible from the linguistic-mental point of view, and brain underlying states. If a loss or a disappointement can cause a depression, first a reactive depression then a vitalized depression, one that begins as reactive but follows with an endogenous character, then there might be a map from the structure of depressions in the mental mode (the mode that could urderlie a general theory of psychic disorders for humans and machines) and in the brain mode. If this is true, the knowledge about the brain can be more general than only a study for human purposes. It may be a fundamental source to understanding a general theory, for men and machines, of psychic-mental disorders. If this hypothesis doesn't demolish functionalism it might cause it trouble, since it ventures that there might be a map from the mental structure that underlies depressive syndromes to the cerebral structures that urderlie them, and vice versa. This is true from the psychiatric point of view considering the example of two types of depression. From the external point of view it is extremely difficult to make a differential diagnosis between them, as it is in a large number of psychiatric disorders, where criteria are not met and test and response tends to be the only way to discriminate if the disturbance is primary (brain, bottom-up caused) or contextual-reactive-historical-biographical (top-down caused). If they resembled one another and if the final diagnosis comes with test and response, this might suggest that token-identity and the non-equivalence between brains and machines concerning the way they promote mental illness is a questionable position, in spite of being a difficult thesis to demolish, but with eliminative materialism, which we conclude is not compatible with P in the today's mode.

This second problem elicits a position that I would say is important to CS and to a scientific P. Maybe there is map from syndromes from the mental point of view and syndromes from the brain point of view. If reduction is impossible concerning terms because there are problems like intentional idioms, cross-classification, etc., there might be a reduction concerning syndromes that are an intermediate degree between terms and theories. If the medical style seeks syndromes and P makes onle syndromic diagnosis. The evolution of the disturbance confirms if it was primary or not, i..e. endogenous or reactive. If there is a strong resemblance, We'd say a homeomorphism between top-down caused (reactive) and bottom up caused (endogenous) syndromes, then there must be a type-identification between mental syndromes and brain syndromes, which would imply that knowing the human brain and its psychopharmacology might be essential to understanding general mental pathology in men and machines.

But what could be the shape of this general psychopathology that accepts functionalistic arguments at the level of mental terms and that seeks reductionism at the syndromic level, something that could be important from the fundamental point of view to P to Neurpsychology and to CS?

3. Neural networks, consciousness, will and synchronic oscillations

This third problem deserves a full article. Briefly, I will propose an alternative tentative hypothesis that could pave the way to a reductionistic general theory of psychopathology based on a homeomorphic map betwen syndromes at the mental level and their correlates at the brain level., something that could be interesting as a foundational stone to CS and to P. The fundamental point is that the following strategy could avoid a strong position like eliminative materialism since language in a certain sense has its place as a tool to communicate symptoms and it is not so strong as functionalism, since the general brain mechanisms that underlie psychiatric disorders could be homeomorphic to the mechanisms that might underlie artificial-machine psychpathology.

Consciousness can be considered as the arena where every mental disorder occurs. If every psychiatric syndrome is not necessarily a primary affection of consciousness, for example, imagine something that begins at subcortical areas that in principle could not be "conscious", the end product is always conscious. Why? Because if the triad -- signals, symptoms and syndromes -- is the way one has to access mental disorders, and if symptoms deserve language to be the medium of communication, thence consciousness is required to comunicate, elaborate and to seek the referents to be described as symptoms in ordinary language. In a certain sense I would say that consciousnes is what characterizes the mental, although representation, intentionality, maps, shemes, etc. can be run without consciousness. As it is matter of controversy, let us accept as a postulate that consciousness is the essence of the mental. And if it is difficult to define consciousness, at least one of its main predicates, we think, is essential: the will or the capacity of purposeful acts and thoughts. If the will is difficult to define at least one of its predicates is easier to be described : voluntary control and its opposite -- automatic control. The number of examples that might be given from this dychotomy (voluntary x automatic control) are extensive and well known. I'll not put them here. I'll just say that voluntary control can be a subset of conscious states that has a relatively well known substratum (the frontal lobes), as automatic control has its substratum -- the cerebellum. One might say that if every mental disorder involves in a large degree a disturbance of consciousness and if consciousness can be represented by the will and this by voluntary control that is mainly coordinated by the frontal lobes, then a good candidate to be the target for a syndrome-type-reduction (and it is really strongly involved in psychiatric syndromes) is the frontal lobe. The hypothesis that synchronization, mainly in the 40 Hz frequency, among certain populations of neurons coud be the substratum of consciousness. But synchronization is a syntactical way to describe the physical way informatiom might be encoded in the Central Nervous System. Early, the digital hypothesis posited that neurons could be all or none, 0 or 1, implementers. Nowadays, the approach is more analogical, ranging from a spectrum of values let us say from - m to + m and the codification of information might be coordinated through the frequency of action potentials at the synaptic level. Neurons can be described mathematically as oscillators. Assemblies of neurons oscillating can synchronize or not. There is a way to describe populations of neurons oscillating through ordinary differential equations. The frequency code, that can be the way to codify objects (or syndromes if one thinks in a scheme of type-syndrome-reduction from the mental to the physical), can be analized through topological tools and through the Theory of Dynamical Systems (TDS). The notion of structural stability and of non-stability for certain parameter values (ordinary parameter values for structural stability and bifurcation for non-stability) are very good ways to understand the qualitative topological change at the solutions domain (or at the space of states of a system). Quantitative changes in a parameter near bifurcation values may cause qualitative radical modifications in the behavior of the system (accessible through topology). This is a brief overview of modern TDS that became famous mainly because of the virtual appearance of chaos in certain situations (Some authors suppose that there is chaos in the brain-signal domain and this could be the way that explains such a diversity of states and so for).

If synchronicity depends on parameters and if it stays under consciousness, and if consciousness can be seen as a special part of psychiatric syndromes, one could reckon that this is a beginning to trying to launch a scientific reductionistic approach to type-syndrome reduction. Connectionist Artificial Intelligence (CAI) began to ruin the build of a strong gap between mind and brain. Abandoning the logicistic approach of TAI, it inserted, from the processing point of view, new ways to manipulate objects. Be it A and B mental objects. TAI supposes that the rules that relate A and B are logical. CAI supposes they can be dynamical, not known, minimizing errors and "energy", seeking local minima and so forth. But CAI, if it assumed a reductionistic flavor regarding the rules of connection, it maintained the semantical priority, or functional emergence of elements like A and B regarding their brain substrata. The interpretation for a node, for a basin of attraction in CAI is still a mental block. Then one can say that if TAI was functionalistic regarding the mental blocks and their rules, CAI is functionalistic regarding the entities, the mental blocks, the As and Bs, but reductionistic regarding a presumed brain style of relating As to Bs. For our purposes As and Bs are not mental atoms (since we've conceded that they cannot be radically translated into a brain language) but they are syndromes, firmly tied to consciousness, will, voluntary control and the way the CNS manipulates and recognizes information in order to decide wheter processing it consciously or automatically. This could be a radical reduction because A and B, qua syndromes, might be syntactically explained through codification via frequencies. Thence entities and rules of connecting them, being the entities not terms, but syndromes, would allow to map homemorphisms from the brain to the mental, endorsing a biologically strong Psychiatry that doesn't have to abandon the boat of ordinary language nor of old psychopathology (or if one prefers of DSM). The invariances that characterize syndromes could be mapped unto brain invariances, and the interpretation of oscillations and frequencies might be a way to endorse a genuine reduction, not so radical as to translate language, but sufficiently radical to explain the nature of a concomitant bottom-up and top-down process that happens in psychiatric syndromes.

We suggest a radical inversion in the use of tools, different from every other effort (e.g. Stein & Young 1992) that relates Psychiatry to Cognitive Science.

REFERENCES

American Psychiatric Association (1994) Diagnostic and Statistical Manual of Mental Disorders- DSM-IV. APA. Washington.

Fodor, J. (1975): The Language of Thought. Harvard University Press

Hooker, C. (1981): "Towards a General Theory of Reduction" in Dialogue, vol.XX, 1-3

Husserl, E. (1900): Logische Untersuchungen. Max Niemeyer. Max Niemeyer Verlag

Jaspers, K. (1920): Allgemeine Psychopathologie. Verlag von Julius Springer. Berlin.

Ryle, G. (1949): The Concept of Mind. Hutchinson.

Stein, D. & Young, J.(ed) (1992): Cognitive Science and Clinical Disorders. Academic Press.