International Conference on Theoretical Neurobiology
New Delhi February 2003
A categorical model for cognitive systems up to consciousness
Andrée C. Ehresmann and Jean-Paul Vanbremeersch
Faculté de Mathématique et Informatique, 33 rue Saint-Leu, 80039 Amiens, France
ABSTRACT. The dynamics of an animal is modulated by the interactions with his environment. How does he learn to recognize the main features of a situation and to respond in an adequate manner? How can he acquire emergent capacities, such as higher cognitive processes up to consciousness?
These questions are analyzed in the frame of a mathematical model for natural autonomous complex systems, the Memory Evolutive Systems (MES). This model, based on Category Theory, is developed by the authors in a series of papers spanning the last 15 years, summed up in their Internet site.
Here it is applied to the case of cognitive systems. We show how the "binding problem" is translated in the MES of neurons to explain the formation of higher mental objects and cognitive processes, up to semantics and consciousness for a higher animal. In particular, the emergence of consciousness would rely on the formation of a personal 'affective' memory of the animal, his body, his experiences and interactions with his environment, called the Archetypal Core, at the basis of the notion of self.
KEYWORDS. Hierarchical system. Cognitive system. Mental object. Consciousness. Categories.
The analysis of a natural complex system, say a biological, cognitive or social system, raises the following problems: The system is organized into various complexity levels, with their own temporality, yet its interactions with its environment are coherent, the components and the organization can be partially modified while the identity of the whole system is maintained. It is autonomous with a capability to learn to recognize features of the context and to develop adapted strategies in answer; how can it memorize a hierarchy of representations which are sufficiently stable though adaptable to changing circumstances? The formation of higher organizational levels introduces new properties (example: development of higher cognitive processes up to consciousness); how do they emerge from lower levels without being directly reducible to them?
These questions are studied in the frame of the Memory Evolutive Systems (MES) that the authors have developed in a series of papers since 1987 (e.g. [Ehresmann & Vanbremeersch (or EV) (1987; 1991; 2002)]). The MES represent a mathematical model, based on Category Theory, for natural open self-organizing systems such as biological, cognitive or social systems. In this model the dynamics is modulated by the interactions between the global system and a net of internal more or less competitive regulatory modules, called CoRegulators (CR) which act in parallel at their own timescale, and with a differential access to a central hierarchical Memory to the development of which they participate.
The evolution of a MES depends upon a succession of complexification processes, in which patterns of interacting objects are bound together into new objects (represented in the categorical model by the colimit of the pattern) taking their own complex identity. It is explained how iterated complexifications can lead to the emergence of a hierarchy of objects of strictly increasing complexity orders, which have both robustness and plasticity thanks to their capacity of switching between various internal organizations.
Here the model is applied to a cognitive system, modeled by the MES of neurons, corresponding to the development and functioning of the nervous system of a higher animal. In this case the complexification process is related to the "binding problem" of neuroscience; it leads to the formation of more and more complex mental objects, and characterizes how higher cognitive or 'mental' processes may emerge from physical states of the brain, thus supporting an emergentist monism (in the sense of Mario Bunge ).
Higher animals are able not only to store representations of particular perceptions, behaviors or events, but also to classify them through the detection of specific invariances, internally reflected in a 'semantic memory' [EV (1992)]. Its formation allows for the development, from birth on, of a global invariant, the archetypal core that integrates the perceptual, behavioral and affective experiences of the animal and the basic strategies associated to them, thus giving a basis at the notion of self. Consciousness is then characterized as a 3 steps process which, relying on this archetypal core, internalizes the temporal dimensions and gives evolutive advantages by permitting more adapted responses dependent on the whole experience of the subject.
In the first part the main ideas at the root of our model are explained in non-technical terms, while Part II describes how they are translated in the mathematical model which is briefly recalled.
I. CHARACTERISTICS OF A COGNITIVE SYSTEM UP TO CONSCIOUSNESS
1. Formation of mental objects. The binding problem
An animal is able to extract features of his environment, later recognize them and react with suitable behaviors. How can he develop a flexible though sufficiently robust 'memory' of his various experiences (sensory and motor inputs, internal states, behaviors and strategies,…) and what are its characteristics?
1.1. Synchronous assemblies of neurons. The response of the neuronal system to a simple stimulus consists in the activation of a highly specialized neuron; for instance, in the visual areas, there exist 'simple cells' representing segments of a given direction, and 'complex cells' representing a particular angle. But it seems unlikely that more complex stimuli, except for some exceptions, have their own 'grand-mother neuron'. Following [Hebb (1947)] and relying on neurophysiological data, most authors agree that the response to a complex stimulus will be the activation of a synchronous assembly of interacting neurons (called neuronal groups by [Edelman (1989)]). The "binding problem" (e.g. [von der Malsburg (1995)]) examines how patterns of neurons, distributed in widely separated areas, are integrated in such assemblies,
1.2. Multifold mental objects. The memory must have some plasticity, and not be rigid as that of a computer. Indeed, the same object is recognized under different aspects (a circle may appear as an ellipse), a behavior will adjust to the circumstances, e.g. the motor neurons activated for seizing an object depend on its size. The neuronal encoding of a 'mental object' (in the sense of [Changeux (1983)]), be it a perception object, a behavior or an internal state, must be multifold; in our model this multifold representation is called a (first order) cat-neuron. It is made not of just one synchronous assembly of neurons but of a whole class of such assemblies with possible switches between them, and its later recall is done through the activation of the assembly of the class the most adapted to the present context.
Just as neurons communicate through synapses, cat-neurons can interact through specific links; in our model we describe how these links are constructed; there are 'simple links' binding local communication paths and 'complex links' representing more disseminated information. For instance, there are operative interactions between perception and action, justifying the 'active perception' approach (e.g., [Thomas (1999)]).
1.3. Higher order cat-neurons. The binding problem extends to patterns of first order cat-neurons interacting through their links, which are transformed into synchronous assemblies of cat-neurons, integrated into second order cat-neurons representing more complex mental objects. And by iteration of the process there is the progressive formation of a hierarchy of cat-neurons, corresponding to more and more widely distributed information. For instance different features of an object (color, shape,…) are processed by lower level modules (specialized brain areas) and bound together at a higher level; a complex behavior requires the successive activation of a large number of simpler ones.
The activation of a second order cat-neuron generally does not reduce to that of a (even large) assembly of neurons, but requires a 2 steps process to unfold one of the synchronous assemblies of synchronous assemblies of neurons that it subsumes. More generally, a higher order cat-neuron has several 'ramifications', each representing a synchronous assembly of assemblies of… neurons and can be recalled through the step-by-step unfolding of anyone of them down to the neuronal level, possibly later switching to another one if required by the circumstances. It is a dynamic unit, gradually adjusted through internal and/or external feedback; for instance, in time the animal will learn to recognize more subtle aspects of a prey and to respond with finer behaviors. Thus, in spite of its robustness, the memory maintains enough plasticity to take into account changes in the environment.
The animal will gain more independence from the context if he can classify its mental objects, having for instance a general notion of preys in spite of their individual differences. This relies on the formation of a semantic memory. The literature on semantics is particularly large. We adopt a perspective in which a 'concept' can be seen as an internal representation of a class of items with a "family likeness" in the sense of [Wittgenstein (1953)].
2.1. Construction of the semantic memory. We model its development as a 3 step process [EV (1992)]:
· A pragmatic classification with respect to a particular attribute is effected by lower modules; e.g. a visual area dealing with color will respond in the same way to all blue objects.
· This classification takes a 'meaning' only if it is internally detected at a higher level which reflects it by the formation of a specific cat-neuron which gives an abstract representation of an invariance class; we call it a concept (e.g. the concept 'blue'), and the mental objects pertaining to the invariance class are called its instances.
· More complex concepts are formed by binding together simpler ones, and links between them are constructed. The concepts and their links form what we call the semantic memory.
Let us note that no language is supposed here.
2.2. How do concepts operate? The recall of a concept will be done through the selection of anyone of its instances, and then of anyone of the ramifications of the cat-neuron representing this instance. Balances can occur between the instances of an activated concept, as well as switches among the ramifications of these instances. In this way, the development of a semantic memory gives a double degree of freedom to modulate the interactions with the context. For example the strategy of seizing an object will be 'abstractly' represented by a concept, but its activation will consist in the selection of one of its instances whose activation will be done through the successive activation of synchronous assemblies of cat-neurons of lower orders down to simple synchronous assemblies of neurons; the selection of the instance and of the assemblies activated at each level depends on the size and shape of the object to seize, and may be modified during the motion to ensure the object is seized.
2.3. Mental vs. physical. A synchronous assembly of neurons identifies to a physical state of the brain, so that a mental object represented by a first order cat-neuron has a multifold realization into such physical states. It is different for a higher order cat-neuron which requires the step by step unfolding of one of its ramifications into assemblies of assemblies of… neurons, down to simple synchronous assemblies of neurons; however at this last level it is realized by physical states. And a concept requires in addition the selection of one of its instances. Thus a cat-neuron or a concept has a functional identity, not a physical one, but operates through physical states of the brain. This explains in which way a mental object represented as a concept or as a cat-neuron can 'cause' a physical event, and how mental properties supervene on physical properties with multiple realizability (the various ramifications), making "mental causation" [Kim (1998)] possible.
3. The Archetypal Core as the basis for self
The existence of a semantic memory allows for the development of a personal 'affective' memory of the animal, his body, his experiences and interactions with his environment, which we call the Archetypal Core (AC).
3.1. Development of the AC. The AC is formed by mental objects (cat-neurons of various orders) activated more often and during a longer period, from birth on (for instance stable aspects of the environment in contrast to more variable ones, deep feelings,…), their links and their concepts. It develops to integrate the main sensorial, proprioceptive, motor experiences, …, with their emotional overtones and the basic strategies associated to them, and to connect them into patterns whose links are strengthened in the course of time. It may be autonomously activated, so that a whole sub-system of the AC is activated as soon as a small part of it is stimulated. For instance a cat-neuron in it (say the blue sky) is archetypally linked to other cat-neurons not only of perceptions or of motor processes but also of internal states and emotions (sun, heat, well-being, swimming,…).
3.2. Functioning of the AC. The self-activation of the AC is directed and maintained for long periods by specific bundles of strong, quickly activated links connecting each archetypal cat-neuron to some other objects of the AC. These bundles, which we call fans, act as channels (to be compared to the chreods of [Waddington (1940)]) through which the activation of the cat-neuron resonates as an echo, and is propagated first to the target concepts, then oscillates through a sequence of loops based on balances between various instances of the concepts and switches among their ramifications (neurophysiologically, it relies on the thalamo-cortical loops). The fans are gradually strengthened, leading to more and more integration of the whole AC.
The AC, as a permanent representation of the animal, his phenomenal experiences, his acquired knowledge, be it pragmatic, social or conceptual and the basic strategies associated to them could be the basis of the notion of self.
3.3. The experiential Memory. Some experiences might be sufficiently significant for their concepts to have strong links toward the AC, possibly without belonging to it; with their links these form the experiential memory Exp (to be related to the "value-category memory" of [Edelman (1989)]). The activation of an object in Exp recalls the 'archetypal' experience most closely linked to it, whose activation spreads through the fans into loops which remain self-generated for a long time and diffuse to close domains. Each cat-neuron in the memory is linked by a cluster to a pattern of related experiences in Exp; if these links become strong enough, it will become integrated in Exp, and possibly later on in the AC.
4. Conscious processes
We assume that the development of the AC is at the basis of the existence of consciousness. Recently consciousness has become a much studied topic, with often diverging viewpoints (cf. papers in the Journal of Consciousness Studies, the Tucson Conferences, and numerous books, e.g. [Baars (1997); Chalmers (1996); Changeux (1983); Crick (1993); Damasio (1999); Dennett (1991);…]). In preceding papers [EV (1992); (2002)] we have proposed to characterize consciousness as a 3 parts process which integrates the structural and temporal dimensions and leads to more adapted strategies.
4.1. The extended landscape. Conscious processes are aroused by a relevant event without an automatic response, caused either externally (unknown or threatening stimuli) or internally (related to feelings, activities or strategies in progress, e.g. 'fracture' in a higher level module). It causes an increase of awareness (by activation of the reticular formation) that starts a semiotic search in the memory to try to recognize it and respond adequately. The search is done through modules of various modalities and levels, up to the semantic memory, and heavily relies on the archetypal core which, always in the background, acts as a referent and as a filter, propagating the information through the fans. The information so gathered form a 'holist' extended landscape, whose objects are the temporal perspectives of the event and of its internal reflections (to be compared with the "Global workspace model" of [Baars (1997)]). Neurophysiologically its construction would rely on the existence of what is called the "consciousness loop" in [Edelman (1989)].
4.2. A retrospection process will be operated on this extended landscape, based on a series of loops along the fans of the AC, balances between instances of concepts and switches between the several ramifications of these instances. It uncovers lower levels of the near past to find, by abduction, the possible causes of the present event, taking into account information recalled from former experiences. For instance, the sentence "the fish attacked the man" creates a fracture in a language module; the abduction process finds that the fracture comes from the fact that a typical fish is not aggressive for man; a new abduction evokes dangerous animals related to fishes, whence the recall of a stark, which makes the sentence meaningful.
4.3. A prospection process will then select long term strategies for several steps ahead, to respond in the most adapted way. This is accomplished through the formation of virtual landscapes inside the extended landscape, with the help of the whole memory and more specially the archetypal core, in which successive strategies (selected as concepts) can be 'tried' without material cost for the system. For instance, it will be possible to plan the dynamic formation of a second order cat-neuron in two steps: the first strategy will construct first-order neurons Ni and connect them by simple or complex links; then a second strategy will integrate the pattern so formed into a second-order cat-neuron.
The prospection process may allow to preserve the continuity of a process in spite of temporary interruptions, e.g. to wait for the execution of some operations imposed by one of the strategies on a lower module (in object-oriented programming languages, this is simulated by a 'thread'; cf. [Niemeyer and Peck (1997)]; also used in [Josephson (1998)]). It also allows for the simulaneous planning of several long term strategies maintaining a continuous command of lower level effectors, but only alternatively perceived at the conscious level (as talking while driving).
II. THE MATHEMATICAL MODEL: THE MES OF NEURONS
Here we give a brief outline of a mathematical model, the MES of neurons, in which the different characteristics of a cognitive system described in Part I are taken into account. It is a particular case of the general notion of a Memory Evolutive System (MES), which we have developed in a series of papers since 1986 to study natural autonomous complex systems, and in particular analyze how higher order structures can emerge.
1. Recalls on categories and Evolutive Systems
The model is based on Category Theory, a 'relational' domain of mathematics introduced by Eilenberg and Mac Lane in the fourties, and for which we refer to [Mac Lane (1971)].
1.1. Fundamental definitions. A graph (more precisely oriented multi-graph) is the data of a set of objects and a set of arrows f: N ® N' between them. A category can be defined as a graph equipped with an internal (partial) composition law associating to the pair (f: N ®N', g: N' ® N") of 2 consecutive arrows a 'composite' arrow fg: N ® N", this law is associative and each object N has an 'identity' idN: N ® N.
A pattern of linked objects (or diagram) in the category is the data P of a family of objects Ni and arrows between them. A collective link (or cone) from P to an object N' is a family (fi: Ni ® N') of arrows compatible with the distinguished links in P, i.e., for each x: Ni ® Nj in P, we have xfj = fi.
The pattern has a colimit (or inductive limit [Kan (1958)]) if there exists a collective link (li) from P to L through which each collective link (fi) from P to any N' factors through one and only one arrow f: L ® N' (i.e., for each i, fi = lif). In this case, L will be thought of as a complex object, admitting P as a decomposition.
1.2. Evolutive Systems [EV (1987)]. A natural system will be modeled by an Evolutive System K, which consists of the following data:
· a timescale T (finite or infinite subset of the real numbers) modeling its life-time;
· a category Kt representing the state of the system at each instant t of the timescale: its objects represent the components of the system and its arrows (called links) their interactions in force around this date. A complex object is represented by the colimit (also called binding) of a pattern of linked objects figuring its internal organization.
· a (partial) transition functor K(t,t') from Kt to Kt' for each t < t' in T, representing the change of states from t to t', such that K(t,t") be the composite of K(t,t') and K(t',t") for any t < t' < t" in T.
The system is hierarchical if the objects of Kt are divided into 'complexity levels', so that an object N of level n+1 is the colimit of at least one pattern of linked objects of level less or equal to n.
1.3. The complexification process (binding). For a natural system the change of states consists in the progressive formation of new objects binding patterns P which have no colimit, suppression or decomposition of some complex objects, with preservation of some existing colimits. It is modeled by the replacement of the category by a new category called its complexification with respect to the strategy [EV (1987)], which is the category in which the objectives of the strategy S are realized in the 'most economical way' (in categorical terms : it is a solution of the universal problem of realizing S). This category is explicitly constructed in [Bastiani & Ehresmann (1972)]. In particular, the links between two objects L and M, binding P and Q respectively, are of two kinds: simple links which bind a cluster of links form P to Q (so that they depend only on the 'local' interactions between P and Q); but also complex links constructed from them, in particular by composing simple links binding non-adjacent patterns. The complex links depend on the whole structure of the initial category, and they represent properties 'emerging' in its complexification.
1.4. The Multiplicity Principle. An object N of a hierarchical system is said to be multifold if it is the colimit of at least 2 non-equivalent patterns of linked objects of strictly lower levels. When a category has multifold objects, we say that it satisfies the Multiplicity Principle. An important result is: If a category satisfies the MP, so does any of its complexifications [EV (2002)]. In this case, we have proved that an iteration of the complexification process leads to the emergence of a whole hierarchy of objects with strictly increasing complexity orders. Indeed, if some of the links of the pattern P are complex, an object binding P has emerging properties in the sense that they do not depend only on the properties of the objects of P but also on those of the intermediary objects intervening in the construction of the complex links of P, so that the whole structure of the initial category is taken into account.
2. Memory Evolutive Systems
2.1. A Memory Evolutive System (MES) consists of the following data [EV (1987); (1991)]:
· An evolutive system K,
· A hierarchical evolutive sub-system Mem of K, called the memory, whose number of levels may increase in time.
· A net of evolutive sub-systems of K, called Coregulators (or CR) which direct the dynamics of the MES. The CRs act in parallel, but each at its own discrete timescale (finite subset of T) and complexity level and with a differential access to the central memory Mem.
2.2. Dynamics of a MES. At each step of its timescale, say from t to t+1, a CR operates a 3 parts trial-and-error learning process (cf. [EV (1991)]):
· As an 'observer' it integrates the information it can receive from the system and its environment into its actual landscape Lt, which is a category whose objects are specific clusters of morphisms from an object of Kt to the CR.
· As an 'actor' it selects a strategy S on Lt taking into account the different constraints and the information in Mem from earlier similar experiences; if the strategy succeeds, the next landscape Lt+1 should be the complexification L' of Lt with respect to S.
· As an 'evaluator' at the next step it compares L' to Lt+1 by a 'comparison' functor. If some errors are so detected, one of the objectives of the next strategy will be to correct them. Another objective of this strategy will be to memorize the preceding strategy and its result to allow for better choices in similar situations later on.
2.3. The interplay among strategies. Generally there is no direct concertation between two CRs, though higher associative CRs supervise some lower level CRs. At a given time, the strategies of the various CRs are reflected to the system where a global equilibration process (called the interplay among strategies) is operated. It can lead to fractures for the CRs whose strategies cannot be coherently integrated or whose structural temporal constraints cannot be respected, that force these CRs to change their strategy.
The dynamics of the MES is modulated by a dialectics between heterogeneous CRs. Indeed, let us consider the case of two heterogeneous CRs, say a lower level (or micro) CR with short steps, and a higher level (or macro) CR with longer steps. One step of the macro CR covers many steps of the micro CR, during which it can accumulate changes which are not individually perceived in real time at the macro level because of the propagation delays, or because they do not affect the stability of higher level multifold objects. However their long term accumulation makes the unchanging landscape of the macro CR more and more unreliable, ultimely causing a fracture in it. To repair its fracture, the macro CR will have to initiate a new strategy, which may retroact sooner or later at the micro level.
3. The MES of neurons
The MES associated to a cognitive system is the MES of neurons of the animal, based on the 'category of neurons' which models the brain of the animal.
3.1. The category of neurons Neur at an instant t of the life of the animal is defined as follows (cf. [EV (1991); (1999)]):
· take the directed (multi-)graph whose vertices are the neurons and the arrows the synapses from a presynaptic to a postsynaptic neuron, this graph is labeled by the strength of the synapses (related to the efficiency and delay they propagate an influx around t);
· form the category of paths of this graph, in which the composition is the concatenation of paths, and define the strength of a synaptic path as the product of the strengths of its factors.
· Then Neur is the quotient category of this category of paths by the equivalence identifying 2 synaptic paths having the same source, target and strength.
This category satisfies the MP because it is an iterated complexification of the category of atoms which satisfies it (thanks to properties of Quantum Physics (cf. [EV 2002]).
3.2. Formation of cat-neurons. An assembly of neurons is modeled by a pattern P of linked objects in the category of neurons; its synchronization entails the strengthening of the synaptic paths between its neurons so that they all work together in a coherent way. The synchronous assembly as a whole will be represented by a higher order unit, which becomes the colimit of P in a complexification of Neur with respect to a strategy requiring the binding of P [EV (1987); (1991)]. This unit takes its own identity and, since the MP is satisfied, it can represent (the colimit of) several synchronous assemblies; it corresponds to a first order cat-neuron. The complexification also describes what are the 'good' links, namely the simple and complex links, between such cat-neurons. Thus the complexification process can be iterated to bind patterns of first order cat-neurons, leading to second order cat-neurons which cannot be reduced to first order ones if some of the links of the bound pattern are complex [EV (2002)]. Successive applications of the complexification process generate higher and higher order cat-neurons.
3.3. The MES of neurons. Its state-categories are obtained by successive complexifications of the category of neurons, so that their objects are cat-neurons of various orders, modeling a hierarchy of mental objects. The CRs of the MES model particular modules or areas of the brain. Its memory corresponds to the evolutive sub-system Mem of the MES formed by memorized cat-neurons and their links. Its plasticity and robustness come from the fact that a cat-neuron is a multifold object, with several ramifications down to the neuron level, and its later retrieval can be done through anyone of them.
The semantic memory is constructed as an evolutive sub-system Sem of Mem: The pragmatic classification associates two cat-neurons which activate the same pattern of agents of a lower level CR, so that they have the same "Borsuk shape" (cf. [Cordier & Porter, 1989]): the corresponding concept is represented by the adjunction of a (projective) limit ([Kan (1958)]) of this pattern, through the operation of a higher CR [EV (1992)]. More complex concepts are obtained as limits of more elementary ones.
The experiential memory and the archetypal core are modeled by evolutive sub-systems Exp and AC of Mem. The characteristics of these sub-systems are categorically translated as follows: Exp is a final sub-system of Mem, and it admits AC as a reflective sub-system [Mac Lane (1971)]. Moreover the distinguished fans in AC equip it with a supplementary structure, namely a Grothendieck (co)topology [Grothendieck and Verdier (1972)].
The MES of neurons gives a mathematical model, based on Category Theory, for cognitive systems. It describes how the binding process (modeled by the complexification of a category) can lead to the development of a hierarchy of cat-neurons representing 'mental objects' which are both robust and plastic. A cat-neuron has a multifold structure, and is activated through the unfolding of a synchronous assembly of assemblies of…. neurons. The model also specifies what are the interactions between cat-neurons that allow for the binding process to be iterated up to the development of higher order mental objects and cognitive processes.
The representation of a mental object by a higher order cat-neuron leads to a new approach to the mind-body problem. Indeed, the activation of a synchronous assembly of neurons 'is' a physical state of the brain. But the activation of a higher order cat-neuron, obtained through successive binding processes, requires several steps, through the intermediate levels of cat- neurons of one of its ramifications down to the 'physical' level of neurons, and at each step it may propagate through one or another assembly of cat-neurons, with possible switches between them which might be contingent on the context, random (noise), quantum based [Eccles (1986)] or internally directed. Though this process causes a well-defined physical 'event', ultimately realized by physical states of the brain, it does not identify to a physical 'state'. Thus mental states dynamically emerge from physical states without being identical to them (emergentist reductionism in the sense of [Bunge (1979)].
Higher animals are able to classify mental objects into 'concepts'. The formation of such a semantic memory can lead to the evolution of an integrated 'subjective' memory, the archetypal core, which takes into account the whole experience of the animal, be it perceptive, behavioral or (as emphasized by [Damasio (1999)]) emotional. It is at the basis of the existence of conscious processes.
We define consciousness as a 3 part process, integrating the temporal dimensions: a relevant event triggers the formation of a holistic extended landscape based on the information contained in the archetypal core; in it a retrospection process searches for the causes of the event instead of reacting only to its effects; and a prospection process leads to programming on the long term, so that more effective strategies may be devised, less constrained by the immediate concerns. Thus consciousness gives a selective advantage; it does not need language (higher animals may have such a consciousness) though language grants access to more cognitive efficient processes. And for the "hard problem", the qualia could correspond to the perspectives of objects of the archetypal core in the extended landscape.
Baars, B. J. (1997). In the theatre of consciousness: The workspace of the mind. Oxford University Press. Oxford.
Bastiani(-Ehresmann), A & Ehresmann, C. (1972). Categories of sketched structures, Cahiers Top. et Géom. Diff. XIII-2; reproduit dans "Charles Ehresmann, Oeuvres complètes et commentées", Partie IV-1 (Ed. A. Ehresmann). Amiens (1983).
Bunge, M. (1979). Treatise on Basic Philosophy, Vol. 4. Reidel Dordrecht.
Chalmers, D. (1996). The Conscious Mind. Oxford University Press. Oxford.
Changeux, J.-P. (1983). L'Homme Neuronal. Fayard. Paris.
Crick, F. (1993). The Astonishing Hypothesis. Macmillan publishing C°. New York.
Cordier, J.-M. & Porter, T. (1989). Shape Theory. Wiley.
Dennett, D. (1991). Consciousness Explained. Little, Brown & C°. Boston.
Damasio, A. (1999). The Feeling of What Happens: Body and Emotion in the Making of Consciousness. Harcourt Brace. New York.
Eccles, J.C. (1986), Do mental events cause neural events? Proc. R. Soc. Lond. B227, 411-428.
Edelman, G.M. (1989). The Remembered Present. Basic Books.
Ehresmann, A. and Vanbremeersch, J.-P. (1987). Hierarchical evolutive systems.… Bul. Math. Bio. 49 (1), 13-50.
Idem (1991). Un modèle pour des systèmes évolutifs avec mémoire. Revue Intern. Systémique 5 (1), 5-25.
Idem (1992). Semantics and Communication for MES, in Proc. 6th Intern. Conf. on Systems Research (ed. Lasker). University of Windsor.
Idem (1999). Online. URL: http://perso.wanadoo.fr/vbm-ehr.
Idem (2002). Emergence Processes up to Consciousness Using the Multiplicity Principle and Quantum Physics, A.I.P. Conference Proceedings (CASYS, 2001, ed. Dubois) 627, 221-233.
Grothendieck A. and Verdier J.I. (1972). Théorie des topos, SGA 4. Springer Lecture Notes in Math. 269-270.
Hebb, D.O. (1947). The organization of behaviour. Wiley. New York.
Josephson, B. (1998). Extendibility of activities and the design of the nervous system, in Proceedings Third Intern. Conf. on Emergence ECHO III. Helsinki. August 1998.
Kan, D. M. (1958). Adjoint Functors. Trans. Am. Math. Soc. 89, 294-329.
Kim, J. (1998). Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation. MIT Press. Cambridge, Massachusetts.
Mac Lane, S. (1971). Categories for the working mathematician. Springer.
Malsburg (von der), C. (1995). Binding in models of perception and brain function. Current Opinions in Neurobiology 5, 520-326.
Niemeyer, P. and Peck, J. (1997). Exploring Java. O'Reilly.
Thomas, N. (1999). Are theories of imagerie theories of imagination? Cognitive Science 23, 207-245.
Waddington, C.H. (1940). The strategy of the genes. Allen and Unwin. London.
Wittgenstein, L. (1953). Philosophical Investigations. Blackwell. Oxford.