The dynamics of active categorical perception in an evolved model agent

Notions of embodiment, situatedness, and dynamics are increasingly being debated in cognitive science. However, these debates are often carried out in the absence of concrete examples. In order to build intuition, this paper explores a model agent to illustrate how the perspective and tools of dynamical systems theory can be applied to the analysis of situated, embodied agents capable of minimally cognitive behavior. Specifically, we study a model agent whose “nervous system” was evolved using a genetic algorithm to catch circular objects and to avoid diamond-shaped ones. After characterizing the performance, behavioral strategy and psychophysics of the best-evolved agent, its dynamics are analyzed in some detail at three different levels: (1) the entire coupled brain/body/environment system; (2) the interaction between agent and environment that generates the observed coupled dynamics; (3) the underlying neuronal properties responsible for the agent dynamics. This analysis offers both explanatory insight and testable predictions. The paper concludes with discussions of the overall picture that emerges from this analysis, the challenges this picture poses to traditional notions of representation, and the utility of a research methodology involving the analysis of simpler idealized models of complete brain/body/environment systems.

Beer, R.D. (2003). The dynamics of active categorical perception in an evolved model agent (with commentary and response). Adaptive Behavior 11(4):209-243.

Advertisements

,

Leave a comment

Evolving Communication without Dedicated Communication Channels

Artificial Life models have consistently implemented communication as an exchange of signals over dedicated and functionally isolated channels. I argue that such a feature prevents models from providing a satisfactory account of the origins of communication and present a model in which there are no dedicated channels. Agents controlled by neural networks and equipped with proximity sensors and wheels are presented with a co-ordinated movement task. It is observed that functional, but non-communicative, behaviours which evolve in the early stages of the simulation both make possible, and form the basis of, the communicative behaviour which subsequently evolves.

Quinn, M. (2001) Evolving Communication without Dedicated Communication Channels. In J. Kelemen and P. Sosík, editors, ECAL01, pages 357–366. Prague: Springer.

, ,

Leave a comment

Toward minimally social behavior: Social psychology meets evolutionary robotics

We report on a set of minimalist modeling experiments that extends previous work on the dynamics of social interaction. We used an evolutionary robotics approach to fine-tune the design of a recent psychological experiment, as well as to synthesize a solution that gives clues about how humans might perform under these novel conditions. In this manner we were able to generate a number of hypotheses that are open to verification by future experiments in social psychology. In particular, the results indicate some of the advantages and disadvantages of relying on social factors for solving behavioral tasks.

Froese, T. and Di Paolo, E. A. (2009) Toward minimally social behavior: Social psychology meets evolutionary robotics, in Kampis, G., Karsai, I and Szathmary, E. (eds) Advances in Artificial Life Proceedings of the 10th European Conference on Artificial Life, ECAL09, Budapest, September 13-16, 2009, LNAI 5777, Springer Verlag, pp. 420 – 427.

, , , ,

Leave a comment

Stability of coordination requires mutuality of interaction in a model of embodied agents

We used an evolutionary robotics methodology to generate pairs of simulated agents capable of reliably establishing and maintaining a coordination pattern under noisy conditions. Unlike previous related work, agents were only evolved for this ability and not for their capacity to discriminate social contingency (i.e., a live responsive partner) from noncontingent engagements (i.e, a recording). However, when they were made to interact with a recording of their partner made during a successful previous interaction, the coordination pattern could not be established. An analysis of the system’s underlying dynamics revealed (i) that stability of the coordination pattern requires ongoing mutuality of interaction, and (ii) that the interaction  process is not only constituted by, but also constitutive of, individual behavior. We suggest that this stability of coordination is a general property of  a certain class of interactively coupled dynamical systems, and conclude that psychological explanations of an individual’s sensitivity to social contingency need to take into account the role of the interaction process.

Froese, T. and Di Paolo, E. A. (2008). Stability of coordination requires mutuality of interaction in a model of embodied agents. In From Animats to Animals 10, The Tenth International Conference on the Simulation of Adaptive Behavior, Osaka, Japan, July 7-10, 2008.

, , ,

Leave a comment

Monostable controllers for adaptive behaviour

Recent artificial neural networks for machine learning have exploited transient dynamics around globally stable attractors, inspired by the properties of cortical microcolumns. Here we explore whether similarly constrained neural network controllers can be exploited for embodied, situated adaptive behaviour. We demonstrate that it is possible to evolve globally stable neurocontrollers containing a single basin of attraction, which nevertheless sustain multiple modes of behaviour. This is achieved by exploiting interaction between environmental input and transient dynamics. We present results that suggest that this globally stable regime may constitute an evolvable and dynamically rich subset of recurrent neural network configurations, especially in larger networks. We discuss the issue of scalability and the possibility that there may be alternative adaptive behaviour tasks that are more ‘attractor hungry’.

Buckley, C., Fine, P. Bullock, S. and Di Paolo, E. A. (2008). Monostable controllers for adaptive behaviour. In From Animats to Animals 10, The Tenth International Conference on the Simulation of Adaptive Behavior, Osaka, Japan, July 7-10, 2008.

, , , , , ,

Leave a comment

Extended homeostatic adaptation: Improving the link between internal and behavioural stability

This study presents an extended model of homeostatic adaptation designed to exploit the internal dynamics of a neural network in the absence of sensory input. In order to avoid typical convergence to asymptotic states under these conditions plastic changes in the network are induced in evolved neurocontrollers leading to a renewal of dynamics that may favour sensorimotor adaptation. Other measures are taken to avoid loss of internal variability (as caused, for instance, by synaptic strength saturation). The method allows the generation of reliable adaptation to morphological disruptions in a simple simulated vehicle using a homeostatic neurocontroller that has been selected to behave homeostatically while performing the desired behaviour but non-homeostatically in other circumstances. The performance is compared with simple homeostatic neural controllers that have only been selected for a positive link between internal and behavioural stability. The extended homeostatic networks perform much better and are more adaptive to morphological disruptions that have never been experienced before by the agents.

Iizuka, H. and Di Paolo, E. A. (2008). Extended homeostatic adaptation: Improving the link between internal and behavioural stability. In From Animats to Animals 10, The Tenth International Conference on the Simulation of Adaptive Behavior, Osaka, Japan, July 7-10, 2008

, ,

Leave a comment

New models for old questions: Evolutionary robotics and the ‘A not B’ error

In psychology the ‘A not B’ error, whereby infants perseverate in reaching to the location where a toy was previously hidden after it has been moved to a new location, has been the subject of fifty years research since it was first identified by Piaget [1]. This paper describes a novel implementation of the ‘A not B’ error paradigm which is used to test the notion that minimal systems evolutionary robotics modelling can be used to explore developmental process and to generate new hypotheses for test in natural experimental populations. The model demonstrates that agents controlled by plastic continuous time recurrent neural networks can perform the ‘A not B’ task and that homeostatic mediation of plasticity can produce perseverative error patterns similar to those observed in human infants. In addition, the model shows a developmental trend for the production of perseverative errors to reduce during development.

Wood, R. and Di Paolo, E. A. (2007). New models for old questions: Evolutionary robotics and the ‘A not B’ error. In Proceedings of the 9th European Conference on Artificial life ECAL 2007. Springer-Verlag.

, ,

Leave a comment

Minimal agency detection of embodied agents

Agency detection is studied in a simple simulated model with embodied agents. Psychological experiments such as double TV-monitor experiments and perceptual crossing show the central role of dynamic mutuality and contingency in social interactions. This paper explores the ongoing dynamical aspects of minimal agency detection in terms of the mutuality and contingency. It is investigated how the embodied agents can establish a live interaction and discriminate this from interactions from recorded motions that are identical to the live interaction but cannot react contingently. Our results suggest that the recognition of the presence of another’s agency need not lie on complex cognitive individual mechanisms able to integrate past information, but rather on the situated ongoingness of the interaction process itself, on its dynamic properties, and its robustness to noise.

Iizuka, H. and Di Paolo, E. A. (2007). Minimal agency detection of embodied agents. Proceedings of the 9th European Conference on Artificial life ECAL 2007. Springer-Verlag.

, , ,

Leave a comment

Adaptation to sensory delays. An evolutionary robotics model of an empirical study

Evolutionary robotics simulations can serve as a tool to clarify counterintuitive or dynamically complex aspects of sensorimotor behaviour. We present a series of simulations that has been conducted in order to aid the interpretation of ambiguous empirical data on human adaptation to delayed tactile feedback. Agents have been evolved to catch objects falling at different velocities to investigate the behavioural impact that lengthening or shortening of sensory delays has on the strategies evolved. A detailed analysis of the evolved model agents leads to a number of hypotheses for the quantification of the existing data, as well as to ideas for possible further empirical experiments. This study confirms the utility of evolutionary robotics simulation in this kind of interdisciplinary endeavour.

Rohde, M. and Di Paolo, E. A. (2007). Adaptation to sensory delays. An evolutionary robotics model of an empirical study. Proceedings of the 9th European Conference on Artificial life ECAL 2007. Springer-Verlag.

, , , ,

Leave a comment

Adapting to your body

This paper investigates the processes used by an evolved, embodied simulated agent to adapt to large disruptive changes in its sensor morphology, whilst maintaining performance in a phototaxis task. By avoiding the imposition of separate mechanisms for the fast sensorimotor dynamics and the relatively slow adaptive processes, we are able to comment on the forms of adaptivity which emerge within our Evolutionary Robotics framework. This brings about interesting notions regarding the relationship between different timescales. We examine the dynamics of the network and find different reactive behaviours depending on the agent’s current sensor configuration, but are only able to begin to explain the dynamics of the transitions between these states with reference to variables which exist in the agent’s environment, as well as within its neural network ‘brain’.

Fine, P., Di Paolo, E. A., Izquierdo, E. (2007). Adapting to your body. Proceedings of the 9th European Conference on Artificial life ECAL 2007. Springer-Verlag.

, , ,

Leave a comment