Manuel Glez Bedia

Quantifying long-range correlations and 1/f patterns in a minimal experiment of social interaction.

In recent years, researchers in social cognition have found the “perceptual crossing paradigm” to be both a theoretical and practical advance toward meeting particular challenges. This paradigm has been used to analyze the type of interactive processes that emerge in minimal interactions and it has allowed progress toward understanding of the principles of social cognition processes. In this paper, we analyze whether some critical aspects of these interactions could not have been observed by previous studies. We consider alternative indicators that could complete, or even lead us to rethink, the current interpretation of the results obtained from both experimental and simulated modeling in the fields of social interactions and minimal perceptual crossing. In particular, we discuss the possibility that previous experiments have been analytically constrained to a short-term dynamic type of player response. Additionally, we propose the possibility of considering these experiments from a more suitable framework based on the use and analysis of long-range correlations and fractal dynamics. We will also reveal evidence supporting the idea that social interactions are deployed along many scales of activity. Specifically, we propose that the fractal structure of the interactions could be a more adequate framework to understand the type of social interaction patterns generated in a social engagement.

Sensorimotor coordination and metastability in a situated HKB model

Oscillatory phenomena are ubiquitous in nature and have become particularly relevant for the study of brain and behaviour. One of the simplest, yet explanatorily powerful, models of oscillatory Coordination Dynamics is the Haken–Kelso–Bunz (HKB) model. The metastable regime described by the HKB equation has been hypothesised to be the signature of brain oscillatory dynamics underlying sensorimotor coordination. Despite evidence supporting such a hypothesis, to our knowledge, there are still very few models (if any) where the HKB equation generates spatially situated behaviour and, at the same time, has its dynamics modulated by the behaviour it generates (by means of the sensory feedback resulting from body movement). This work presents a computational model where the HKB equation controls an agent performing a simple gradient climbing task and shows (i) how different metastable dynamical patterns in the HKB equation are generated and sustained by the continuous interaction between the agent and its environment; and (ii) how the emergence of functional metastable patterns in the HKB equation – i.e. patterns that generate gradient climbing behaviour – depends not only on the structure of the agents sensory input but also on the coordinated coupling of the agents motor–sensory dynamics. This work contributes to Kelsos theoretical framework and also to the understanding of neural oscillations and sensorimotor coordination.

Case-based reasoning and real-time systems: Exploiting successfully poorer solutions

In the literature of real-time software applications, case-based reasoning (CBR) techniques have been successfully used in order to develop systems able to carry on with their temporal restrictions. This paper presents a mathematical technique for modelling the generation of solutions by a RealTime (RT) system employing a CBR that allows their response times to be bounded. Speaking in general, a system that tries to be adapted to highly dynamic environment needs an efficient integration of high-level processes (deliberative and time-costly, but close-fitting) within low-level (reactive, faster but poorer in quality) processes is necessary. The most relevant aspect of our current approach is that, unexpectedly, the performance of the system do not get worse any time that it retrieves worse cases in situations even when it has enough time to generate better solutions. We concentrate on formal aspects of the proposed integrated CBR-RT system without establishing which should be the most adequate procedure in a subsequent implementation stage. The advantage of the presented scheme is that it does not depend on neither the particular problem nor a concrete environment. It consists in a formal approach that only requires, on one hand, local information about the averaged-time spent by the system in obtaining a solution and, on the other hand, an estimation about their temporal restrictions.

Criticality as it could be: Organizational invariance as self-organized criticality in embodied agents.

This paper outlines a methodological approach for designing adaptive agents driving themselves near points of criticality. Using a synthetic approach we construct a conceptual model that, instead of specifying mechanistic requirements to generate criticality, exploits the maintenance of an organizational structure capable of reproducing critical behavior. Our approach exploits the well-known principle of universality, which classifies critical phenomena inside a few universality classes of systems independently of their specific mechanisms or topologies. In particular, we implement an artificial embodied agent controlled by a neural network maintaining a correlation structure randomly sampled from a lattice Ising model at a critical point. We evaluate the agent in two classical reinforcement learning scenarios: the Mountain Car benchmark and the Acrobot double pendulum, finding that in both cases the neural controller reaches a point of criticality, which coincides with a transition point between two regimes of the agents behaviour, maximizing the mutual information between neurons and sensorimotor patterns. Finally, we discuss the possible applications of this synthetic approach to the comprehension of deeper principles connected to the pervasive presence of criticality in biological and cognitive systems.

Collective Strategies for Virtual Environments: Modelling Ways to Deal with the Uncertainty

Videogames offer new challenges and excellent application domains for AI technology and research. This paper shows a work on ant algorithms optimizing of collective strategies in a videogame environment. In general, ant algorithms are those that taking inspiration from the observation of ant colonies foraging behavior have developed optimization meta-heuristics. In the first part of the paper, we relate the strengths of ant strategies with the well-known exploration-exploitation problem. In the second part, a particular model on how to optimize group strategies in collective tasks is analyzed in light of the ideas previously examined. Finally, we will show how these ideas can constitute an improved stage in the problem of designing non-player characters in future videogames.

Designing Strategies for Improving the Performance of Groups in Collective Environments

Capture the Flag is a well-known game mode that appears in numerous gaming platforms. It consists in a turn-based strategy game where players compete to capture the other team’s flag and return it to their base. In order to win, competitive teams must use a great deal of teamplay to generate a successful strategy. Both teams must defend the own base from incoming attackers and get into the other team’s base, then take the flag and go back home. The strategy will be a particular case of the well-known “exploration vs. exploitation dilemma” – a recurrent paradox that emerge in all systems that try to get a balance between two types of incompatible behaviors. In this paper, we will show how to apply a ”group strategy”, based on the “exploration vs. exploitation dilemma” that improves the behavior of a teamplay in a videogame platform.

Psychometric Tests in the Field of Drawing Based in Timing Measurements

In the Art Education studies, one of the most interesting trends - in terms of its theoretical foundations- is the phenomena of measuring aesthetic experiences [1],[2]. However, traditionally, the teaching of art has been focused into the technical dimension of the drawings (proportionality, composition, etc.) or in cultural aspects (symbolic role, social criticism, etc.) but the analysis of aesthetic experiences have not acquired such an important status to be scientific analyzed and included in a pedagogical domain. In this paper we propose that, through metrics able to detect the ‘immersion” or “sensitivity” of an artist drawing, it would be possible to get methodological tools capable of measuring aesthetic experiences. In order to explore our hypothesis, we have analyzed the work of students of art in several experiments and we have recorded the traces of the their drawings. The traces have been evaluated and we have tested if fractal timing is found (fractal timing is a ubiquitous clue that indicates that the among the components of a cognitive system exist a coupling). The results of the experiments seem to provide a way to improve some aspects in the daily practice of art teaching and a new way to describe psicometrical parameters.

Intermittent animal behavior: The adjustment-deployment dilemma

Intermittency is ubiquitous in animal behavior. We depict a coordination problem that is part of the more general structure of intermittent adaptation: the adjustment-deployment dilemma. It captures the intricate compromise between the time spent in adjusting a response and the time used to deploy it: The adjustment process improves fitness with time, but during deployment fitness of the solution decays as environmental conditions change. We provide a formal characterization of the dilemma, and solve it using computational methods. We find that the optimal solution always results in a high intermittency between adjustment and deployment around a non-maximal fitness value. Furthermore we show that this non-maximal fitness value is directly determined by the ratio between the exponential coefficient of the fitness increase during adjustment and that of its decay coefficient during deployment. We compare the model results with experimental data obtained from observation and measurement of intermittent behavior in animals. Among other phenomena, the model is able to predict the uneven distribution of average duration of search and motion phases found among various species such as fishes, birds, and lizards. Despite the complexity of the problem, it can be shown to be solved by relatively simple mechanisms. We find that a model of a single continuous-time recurrent neuron, with the same parametric configuration, is capable of solving the dilemma for a wide set of conditions. We finally hypothesize that many of the different patterns of intermittent behavior found in nature might respond to optimal solutions of complexified versions of the adjustment-deployment dilemma under different constraints.

Associative Learning for Enhancing Autonomous Bots in Videogame Design

The Today’s video games are highly technologically advanced, giving users the ability to step into virtual realities and play games from the viewpoint of highly complex characters. Most of the current efforts in the development of believable bots in videogames — bots that behave like human players — are based on classical AI techniques. Specifically, we design virtual bots using Continuous-Time Recurrent Neural Network (CTRNNs) as the controllers of the non-player characters, and we add a learning module to make an agent be capable of re-learning during its lifetime. Agents controlled by CTRNNs are evolved to search for the base camp and the enemy’s camp and associate them with one of two different altitudes depending on experience.We analyze the best-evolved agent’s behavior and explain how it arises from the dynamics of the coupled agent-environment system. The ultimate goal of the contest would be to develop a computer game bot able to behave the same way humans do.

Grid Computing and CBR Deployment: Monitoring Principles for a Suitable Engagement

This paper presents a mathematical technique for modeling the generation of Grid-solutions employing a Case based reasoning system (CBR). Roughly speaking, an intelligent system that tries to be adapted to highly dynamic environment needs an efficient integration of high-level processes (deliberative and time-costly) within low-level (reactive, faster but poorer in quality) processes. The most relevant aspect of our current approach is that, unexpectedly, the performance of the CBR-system do not get worse any time that it retrieves worse cases in situations even when it has enough time to generate better solutions. We concentrate on formal aspects of the proposed Grid-CBR system without establishing which should be the most adequate procedure in a subsequent implementation stage. The advantage of the presented scheme is that it does not depend on neither the particular problem nor a concrete environment. It consists in a formal approach that only requires, on one hand, local information about the averaged-time spent by the system in obtaining a solution and, on the other hand, an estimation about their temporal restrictions. The potential use of industry standard technologies to implement such a Grid-enabled CBR system is discussed here too.