A cognitive architecture that solves a problem stated by Minsky.

We named "Minsky's problem" the challenge of building up a cognitive architecture able to perform a good diagnosis based on multiple criteria that arrive one by one as successive clues. This is a remarkable human information processing capability, and a desirable ability for an artificial expert system. We present a general cognitive design that solves Minsky's problem and a neural network implementation of it that uses distributed associative memories. The type of architecture we present is based on the interaction between an "attribute-object associator (AOA)" and an "intersection filter (IF)" of successive evoked objects, with the intermediation of a working (short-term) memory.

Memories in context.

Context-dependent associative memories are models that allow the retrieval of different vectorial responses given a same vectorial stimulus, depending on the context presented to the memory. The contextualization is obtained by doing the Kronecker product between two vectorial entries to the associative memory: the key stimulus and the context. These memories are able to display a wide variety of behaviors that range from all the basic operations of the logical calculus (including fuzzy logics) to the selective extraction of features from complex vectorial patterns. In the present contribution, we show that a context-dependent memory matrix stores a large amount of possible virtual associative memories, that awaken in the presence of a context. We show how the vectorial context allows a memory matrix to be representable in terms of its singular-value decomposition. We describe a neural interpretation of the model in which the Kronecker product is performed on the same neurons that sustain the memory. We explored, with numerical experiments, the reliability of chains of contextualized associations. In some cases, random disconnection produces the emergence of oscillatory behaviors of the system. Our results show that associative chains retain their performances for relatively large dimensions. Finally, we analyze the properties of some modules of context-dependent autoassociative memories inserted in recursive nets: the perceptual autoorganization in the presence of ambiguous inputs (e.g. the disambiguation of the Necker's cube figure), the construction of intersection filters, and the feature extraction capabilities.

Cooperativity: a unified view.

Cooperativity, the departure from hyperbolic behaviour of the fractional saturation of a receptor at equilibrium (Y) for different values of ligand concentration (L), is an essential property of many physiological mechanisms and a first clue to the existence of conformational transitions and allosteric interactions. Here we investigate the properties of a simple and sensitive procedure to test and quantify cooperative behaviour. The measure of cooperativity involved is kappa = dK(L)/dL where K(L) = (1- Y) L/Y= [free sites]L/[occupied sites] is called the 'global dissociation quotient' Cooperative behaviour appears when kappa is not equal to 0, i.e., K(L) is a function of L. We have shown, for several equilibrium models of cooperative behaviour (e.g., Monod-Wyman-Changeux and Koshland-Némethy-Filmer), that K(L) can be expressed as the weighted average of the microscopic dissociation constants (K(i)) where the weights are the corresponding fractions of occupied sites (X(i)), K(L)= sigmaK(i)X(i). As a consequence, the change in the global dissociation quotient with ligand concentration for a dimer is kappa = (K1 - K2)dX1/dL. This result shows that the quantitative importance of a cooperative behaviour in a dimer depends on two factors: (i) the difference of the microscopic dissociation constants of the sites and (ii) the change in the fraction of occupied sites with ligand concentration. We analyze the generality of this unified view concluding that it would be fulfilled by every equilibrium model where there is a one-to-one relationship between free and occupied sites.

Multiplicative contexts in associative memories.

A system of networks, consisting of a first net that constructs the Kronecker product between two vectors and then sends it to a second net that sustains a correlation memory, defines a context-dependent associative memory. In the real nervous system of higher mammals, the anatomy of the neural connections surely exhibits a considerable amount of local imprecision superimposed on a regular global layout. In order to evaluate the potentialities of the multiplicative devices to constitute plausible biological models, we analyse the performances of a context-dependent memory when the multiplicative net, responsible of the construction of the Kronecker product, presents an incomplete connectivity. Our study shows that a large dimensional system is able to support a considerable amount of incompleteness in the connectivity without a great deterioration of the memory. We establish a scaling relationship between the degree of incompleteness, the capacity of the memory, and the tolerance threshold to imperfections in the output. We then analyse some performances that show the versatility of this kind of network to represent a variety of functions. These functions include a context-modulated novelty filter, a network that computes logical modalities and an adaptive searching device.

String analysis and energy minimization in the partition of DNA sequences.

Two approaches to the understanding of biological sequences are confronted. While the recognition of particular signals in sequences relies on complex physical interactions, the problem is often analysed in terms of the presence or absence of literal motifs (strings) in the sequence. We present here a test-case for evaluating the potential of this approach. We classify DNA sequences as positive or negative depending on whether they contain a single melted domain in the middle of the sequence, which is a global physical property. Two sets of positive "biological" sequences were generated by a computer simulation of evolutionary divergence along the branches of a phylogenetic tree, under the constraint that each intermediate sequence be positive. These two sets and a set of random positive sequences were subjected to pattern analysis. The observed local patterns were used to construct expert systems to discriminate positive from negative sequences. The experts achieved 79% to 90% success on random positive sequences and up to 99% on the biological sets, while making less than 2% errors on negative sequences. Thus, the global constraints imposed on sequences by a physical process may generate local patterns that are sufficient to predict, with a reasonable probability, the behaviour of the sequences. However, rather large sets of biological sequences are required to generate patterns free of illegitimate constraints. Furthermore, depending upon the initial sequence, the sets of sequences generated on a phylogenetic tree may be amenable or refractory to string analysis, while obeying identical physical constraints. Our study clarifies the relationship between experts' errors on positive and negative sequences, and the contributions of legitimate and illegitimate patterns to these errors. The test-case appears suitable both for further investigations of problems in the theory of sequence evolution and for further testing of pattern analysis techniques.

Context-dependent associations in linear distributed memories.

In this article we present a method that allows conditioning of the response of a linear distributed memory to a variable context. This method requires a system of two neural networks. The first net constructs the Kronecker product between the vector input and the vector context, and the second net supports a linear associative memory. This system is easily adaptable for different goals. We analyse here its capacity for the conditional extraction of features from a complex perceptual input, its capacity to perform quasi-logical operations (for instance, of the kind of "exclusive-or"), and its capacity to structurate a memory for temporal sequences which access is conditioned by the context. Finally, we evaluate the potential importance of the capacity to establish arbitrary contexts, for the evolution of biological cognitive systems.

Time delays in metabolic control systems.

In this work we use mathematical models with discrete and distributed time delays to analyse the stability of metabolic pathways controlled by end product. We assume the kinetics of the intermediates of the path to be unknown, and we cover the lack of information by using a time delay. We find that above a definite substrate value, there is a critical delay Tc in which a transition from stability to instability occurs. For discrete delays, we find that even if the interaction of the end product with the first (allosteric) enzyme is not cooperative, the pathway can potentially become unstable and oscillate. We then show that the existence of cooperative inhibition extends the parametric domain of instability. The introduction of distributed delays shows, when the kernels are not monotonically decreasing, that the dispersion increases the critical delay Tc. Finally, we comment on the possibility that metabolic oscillations are physiological signals useful for triggering adaptive strategies in cell behavior.

Viscoelastic models for enzymes with multiple conformational states.

In this article we represent an enzyme capable of exhibiting more than one conformational state as a viscoelastic unit embedded in a fluid medium. We show how this viscoelastic unit is thermally activated to make transitions between equilibrium states, and propose this model as a mesoscopic representation for transitions between conformational states of an enzyme. In this representation, kinetic constants for transitions between conformations are given explicitly in terms of interactions between the enzyme and the medium. Two applications of this model are discussed: mnemonic enzymes, and co-operative enzymes exhibiting half-of-the-sites reactivity.

Energy storage during DNA girase activity.

In this work, we develop a minimal two-cycle model for the action of DNA girase. One of the cycles describes the ATP dependent chemicomechanical transduction performed by the enzyme. The other cycle describes the relaxing activity exhibited by girase on supercoiled DNA in the absence of substrates. Supercoiling of DNA is described as a random walk on a topological index. The mechanical energy storage in DNA is manifested in the fact that in general, the kinetic constants for the cycles do not satisfy Wegscheider relations. Finally, a connection is established between the degree of DNA supercoiling and the hydrolysis of ATP.

Graphical coding of nucleic acid sequences.

When, in a nucleic acid sequence, the four letters C, G, A, T (or U) are replaced by suitable graphical symbols, some patterns become immediately apparent. Two sets of symbols, constructed for the analysis of either purine/pyrimidine alternations, or of regions of complementarity within a sequence are shown. In addition, another mode of coding is presented, in which the four letters are represented by vectors. The sequence is thus transformed into a planar trajectory. We show, in the case of the gene for human beta hemoglobin, that such a coding enables an easy discrimination between introns and exons.

Biological organization, Jarkievich measures and asymmetric random walks.

In this paper we analyze the organization imposed by the energy input during the migration of enzymes on DNA. We attempt to measure that organization by means of a concept proposed by A.A. Jarkievich in 1961. We found relationships among a Jarkievich measure, the energy dissipation, and the fluctuations in the kinematic velocity of the enzyme on the DNA.

Errors in the stereoscopic separation of surfaces represented with regular textures.

Stereograms containing two similar or dissimilar linear textures, either on the same surface or at two different depths, were tested on seventy subjects. Whereas random textures usually produced correct percepts, regular textures consistently led to errors of stereoscopic interpretations, including a reversal of hollows into bumps, dissociation of single surfaces into two layers, and errors in relative positioning of two surfaces. Horizontal-vertical textures tended to be seen as flatter and further away from the observer than diagonal ones. Continuous textures tended to be seen closer than discontinuous ones. In the interpretation of the results, the possibility is raised that different textures are processed independently and that the brain has no reliable method for combining the conclusions into a rigorous global percept.

Chemicomechanical transduction performed by enzymes activated by polymers.

Kinetic models for the mode of action of processive and non-processive DNA-helicases are detailed. Fluxes at the steady state are analyzed, and the random walk of the enzymes on the DNA is studied in connection with the rate constants of the chemical reactions involved in the transformation of substrate to products. Finally, the constants of the kinetic model for the processive helicase are related to the parameters of an analogous viscoelastic model.

Teoria de la migracion de enzimas sobre polimeros lineales. / Theory of enzyme migration on linear polymers

Este trabajo asume que las transiciones conformacionales en enzimas cuya movilidad este sometida a restricciones espaciales puede generar fenomenos vectoriales. Se postula que el mismo principio explicativo que da cuenta del mantenimiento de gradientes de concentracion por transporte activo opera en la interpretacion del movimiento vectorial de ciertas ATP-asas DNA-dependientes. En este ultimo caso, las restriciones al movimiento isotropo de la enzima estan impuestas por el polimero activador. Se asume que durante la accion enzimatica, entre los sitios de la enzima y los sitios del polimero activador ocurren el mismo tipo de relaciones reciprocas que en el caso de un transporte activo ocurren entre la enzima y los iones transportados. Desarrollamos un modelo cinetico simple con el fin de analizar las consecuencias matematicas de esta idea

[Generalisation of Martini's model to study the dynamic of infectious diseases (author's transl)]. / Généralisation du modèle de Martini de la représentation de la dynamique des maladies immunisantes.

Martini's model makes it possible to study various epidemiological situations arising in the study of infectious diseases: endemicity, recurrence, presence of a reservoir of pathogenic agent and active immunization. Numerical applications to epidemiological data concerning measles in Uruguay is presented.