Publications

CALL FOR PAPERS. Special Issue of the Journal Synthese: RADICAL VIEWS ON COGNITION

Guest Editor(s): Marcos Silva (Federal University of Alagoas) and Francicleber Ferreira
(Federal University of Ceará)

Special Issue Description:

Several contemporary philosophers have been developing tenets in pragmatism (broadly construed) to motivate it as an alternative philosophical foundation for a comprehensive understanding of cognition, opposed to a far-reaching representationalist tradition.
This long-established representationalist tradition in philosophy of mind and cognitive science defends that cognition is fundamentally content-involving. On the other side, some radical contenders advocate that cognition is neither basically representational nor does it involve, as in usual internalist views, processing or manipulating informational contents. They call attention to the importance of inherited and embodied practices and social interactions in order to understand relevant topics in perception, language and the nature of intentionality. They take seriously evolving biological systems and situated individuals interacting in communities over time as preconditions of our rationality, features often dismissed as not central in the representationalist and internalist tradition.

(more…)

Advertisements

Special Issue on “Computational Models of Affordance for Robotics”. CALL FOR CONTRIBUTIONS

Gibson’s theory of affordance, in its adherence to bottom-up direct perception, is antithetical to the top-down inferential models often proposed by modern robotics research purporting to tackle it. Such research assumes internal representation to be sacrosanct, but given current developments, to what extent can this assumption now be re-examined? The recently proposed sensorimotor contingency theory furthers the theoretical argument that internal representation is unnecessary, and its proof- of-concept application in robotics as well as the subsequent explosion in deep learning methodology sheds new light on the possibility of equipping robots with the capacity for directly perceiving their environments by exploiting correlated changes in their sensory inputs triggered by executing specific motor programs. This re-examination of direct perception is only one of several issues warranting scrutiny in current robotic affordance research.

The aim of this special issue is to highlight the relevance of Gibson’s notion of affordance for developmental and cognitive robotics. The issue is focused on contributions from the current panorama of robotics with an emphasis on theories from the ecological, cognitive, developmental and sensorimotor accounts.

CALL FOR PAPERS – Adaptive Behavior Special issue on “Post-cognitivist approaches to perceptual learning”

The classical cognitivist theory in cognitive science depicts perception as the result of information processing of sense data, which is transformed into a representation of the original information to be useful for the human mind. In the same vein, perceptual learning has been understood as an enrichment of sensations by representational mechanisms. In this view, the improvement in performance must be understood as the effect of a sophistication of computational algorithms entailing a better interpretation of sensory stimuli.

At the end of the 20th century, criticism against the cognitivist framework and its ideas of perception, cognition, and representation started to arise. Some of these arguments crystallized in alternative theories of cognition that offers an innovative way to understand perception and, consequently, perceptual learning.

The aim of this special issue is to document the theories and research that highlight a “4E cognition” approach to perceptual learning. The issue is focused on contributions from the current panorama of post-cognitivism with an emphasis on theories from the ecological, enactive and sensorimotor accounts.

(more…)

CALL FOR PAPERS – Adaptive Behavior Special issue: “Spotlight on 4E Cognition research in Colombia”

The last couple of decades in cognitive science have seen an increasing interest in the philosophical and scientific study of embodied, embedded, extended, and enactive cognition – so-called “4E cognition.” By now theories of 4E cognition have matured and a lot of evidence has been collected, which consequently has reshaped our understanding of the relationship between an agent’s brain, body, and its material and sociocultural world. Despite their differences in emphasis, the various strands of 4E cognition research are united in proposing that an agent’s cognitive activity is bodily mediated, especially by the context-sensitive deployment of sensorimotor capacities.

While these interdisciplinary approaches have largely been developed in Europe, the United States, and Australia, other regions have also been influenced by this growing movement and have started to advance their own original contributions. The aim of this special issue is, therefore, to put a spotlight on 4E cognition research from one such region, Colombia. It intends to do so in two respects: first, to explore the current state and breadth of the field in Colombia; second, to critically examine questions and problems elicited by this Colombian research, focusing on open challenges, with the aim to articulate more precise arguments for and against key claims advanced by 4E cognition research.

(more…)

Sensitivity to social contingency in adults with high-functioning autism

As part of his doctoral research, Leonardo Zapata-Fonseca coordinated this analysis of embodied social interaction. Great team effort!

Sensitivity to Social Contingency in Adults with High-Functioning Autism during Computer-Mediated Embodied Interaction

Leonardo Zapata-Fonseca, Tom Froese, Leonhard Schilbach, Kai Vogeley, and Bert Timmermans

Autism Spectrum Disorder (ASD) can be understood as a social interaction disorder. This makes the emerging “second-person approach” to social cognition a more promising framework for studying ASD than classical approaches focusing on mindreading capacities in detached, observer-based arrangements. According to the second-person approach, embodied, perceptual, and embedded or interactive capabilities are also required for understanding others, and these are hypothesized to be compromised in ASD. We therefore recorded the dynamics of real-time sensorimotor interaction in pairs of control participants and participants with High-Functioning Autism (HFA), using the minimalistic human-computer interface paradigm known as “perceptual crossing” (PC). We investigated whether HFA is associated with impaired detection of social contingency, i.e., a reduced sensitivity to the other’s responsiveness to one’s own behavior. Surprisingly, our analysis reveals that, at least under the conditions of this highly simplified, computer-mediated, embodied form of social interaction, people with HFA perform equally well as controls. This finding supports the increasing use of virtual reality interfaces for helping people with ASD to better compensate for their social disabilities. Further dynamical analyses are necessary for a better understanding of the mechanisms that are leading to the somewhat surprising results here obtained.

View Full-Text

Self-modeling in Hopfield Neural Networks with Continuous Activation Function

Finally a large part of Mario’s thesis on unsupervised learning in artificial neural networks has been published and is available open access:

Self-modeling in Hopfield Neural Networks with Continuous Activation Function

Mario Zarco and Tom Froese

Hopfield networks can exhibit many different attractors of which most are local optima. It has been demonstrated that combining states randomization and Hebbian learning enlarges the basin of attraction of globally optimal attractors. The procedure is called self-modeling and it has been applied in symmetric Hopfield networks with discrete states and without self-recurrent connections. We are interested in knowing which topological constraints can be relaxed. So, the self-modeling process is tested in asymmetric Hopfield networks with continuous states and self-recurrent connections. The best results are obtained in networks with modular structure.

IEEE CEC 2018 — Special Session on Evolutionary Robotics

Rio de Janeiro, Brazil. 8-13 July, 2018

CALL FOR PAPERS

Website: http://sites.google.com/a/isd.org.br/cec18er/

Evolutionary Robotics (ER) aims to apply evolutionary computation techniques to automatically design the control and/or hardware of both real and simulated autonomous robots. Its origins date back to the beginning of the nineties and since then it has been attracting the interest of many research centres all over the world.

ER techniques are mostly inspired by existing biological architectures and Darwin’s principle of selective reproduction of the fittest. Evolution has revealed that living creatures are able to accomplish complex tasks required for their survival, thus embodying cooperative, competitive and adaptive behaviours.

Having an intrinsic interdisciplinary character, ER has been employed towards the development of many fields of research, among which we can highlight neuroscience, cognitive science, evolutionary biology and robotics. Hence, the objective of this special session is to assemble a set of high-quality original contributions that reflect and advance the state-of-the-art in the area of Evolutionary Robotics, with an emphasis on the cross-fertilization between ER and the aforementioned research areas, ranging from theoretical analysis to real-life applications.

(more…)

Mario Zarco graduates with honors!

Today Mario Zarco graduated with honors from UNAM’s Master’s degree in Computer Science and Engineering for his work on self-optimization in neural networks.

The title and extended abstract of his thesis are as follows:

􀀈􀀓􀀔􀀕􀀇􀀌􀀐􀀁􀀇􀀈􀀁􀀄􀀕􀀔􀀐􀀂􀀐􀀑􀀔􀀌􀀎􀀌􀀖􀀄􀀆􀀌􀀘􀀏􀀁􀀈􀀏􀀁􀀒􀀈􀀇􀀈􀀓􀀁Estudio de Auto-Optimización en Redes Neuronales de Hopfield
􀀏􀀈􀀕􀀒􀀐􀀏􀀄􀀍􀀈􀀓􀀁􀀇􀀈􀀁􀀋􀀐􀀑􀀉􀀌􀀈􀀍􀀇􀀁
Mario Alberto Zarco López

Las redes neuronales de Hopfield de tiempo discreto, cuya dinámica presentan múltiples atractores de punto fijo, han sido ampliamente usadas en dos casos: (1) memoria asociativa, basada en aprender un conjunto de patrones de entrenamiento los cuales son representados por atractores, y (2) optimización, basado en representar un problema de satisfaccion de restricciones con la topología de la red de tal forma que los atractores sean soluciones de ese problema. En el ultimo caso, la función de energía de la red debe tener la misma forma que la función a ser optimizada, de modo que los m´ınimos de la primera también sean mínimos de la segunda. Aunque se ha demostrado que los atractores de baja energía tienen un amplio domino de atracción, la red usualmente queda atrapada en mínimos locales. Recientemente se demostró que las redes de Hopfield de tiempo-discreto pueden converger en atractores globalmente óptimos ampliando las mejores cuencas de atracción. La red combina el aprendizaje de sus propios atractores usando aprendizaje Hebbiano y la aleatorizacion de los estados neuronales una vez que la red ha reforzada su configuración actual.
(more…)

Another Master’s thesis published

The second thesis of our group has been published. Please find the title and summary below.

Minimización de la red neuronal artificial de agentes encarnados evolucionados para comunicarse referencialmente

Jorge Iván Campos Bravo

En este proyecto realizamos una minimización de la red neuronal del modelo generado por Williams et al. (2008), en dicho modelose implementan dos agentes en un ambiente mínimoen el que pueden interactuar entre ellos, pero no poseen canales especializados para comunicarse.

Su tarea es sencilla, el transmisor necesita informar al receptor la posición de un objetivo en el ambiente y el receptor necesita llegar a la posición del objetivo.

En nuestro modelo, ambos agentes utilizan la misma copia estructural de red neuronal recurrente en tiempo continuo para controlar su sistema sensorio-motor; dicha red neuronal artificial consta de tres neuronas para ambos agentes.

Se realizaron modificaciones al sistema sensorio-motor y al ambiente original para adaptar el nuevo sistema neuronal, sin perder la esencia de la motivación principal, generar comunicación referencial entre los agentes.
(more…)

Master’s thesis published

The first thesis of our group has been published. Please find the title and summary below.

Un modelo de robótica evolutiva para el reconocimiento explícito de agencialidad

Leticia Cruz Bárcenas

El estudio de la cognición social ha sido abordado principalmente desde dos perspectivas. Por un lado tenemos, el punto de vista del individualismo ampliamente usado en la cognición social, donde se plantea que la interacción y cognición social es el resultado de capacidades cognitivas individuales. Por otro lado, tenemos la perspectiva interaccionista enfocada en que el comportamiento resultante de dos o más individuos reside en los mecanismos colectivos de la interacción dinámica. A pesar de la existencia de estos enfoques, el estudio del rol en la interacción social no ha sido prioritario en las investigaciones de cognición social. Algunas de las dificultades enfrentadas en este sentido están relacionadas con la identificación de características cualitativas y cuantitativas esenciales durante el fenómeno (Lenay & Stewart, 2012).

Con el fin de tener de mejores herramientas analíticas, Auvray et al. (2009) propuso un modelo minino de cognición social que reduce este fenómeno a sus elementos más básicos. Haciendo uso de este modelo se realizó un experimento cuyo objetivo era identificar los mecanismos subyacentes debido al reconocimiento de un sujeto con intencionalidad. Los resultados mostraron que el comportamiento de los individuos propiciaba la interacción con el otro, así como la discriminación del resto de los objetos del ambiente debido a los movimientos oscilatorios individuales.

Con el fin de continuar esta línea de investigación, el presente trabajo muestra un modelo sintético que simula los resultados obtenidos en el experimento original. Utilizando robótica evolutiva se implementó un modelo para investigar la dinámica de interacción en el reconocimiento explícito de agencialidad entre agentes artificiales. El modelo demostró que existe se preserva una interacción cuando los agentes están interactuando entre ellos a pesar de que existan otros objetos/obstáculos en el ambiente.
(more…)