This paper examines the research paradigm of neurally-driven Evolutionary Autonomous Agents (EAAs), from a neuroscience perspective. Two fundamental questions are addressed: 1. Can EAA studies shed new light on the structure and function of biological nervous systems? 2. Can these studies lead to the development of new neuroscienti c analysis tools? The value and signi cant potential of EAA modeling in both respects is demonstrated and discussed. While the study of EAAs as a neuroscience research methodology still faces dicult conceptual and technical challenges, it is a promising and timely endeavor.

This paper presents a general approach for employing lesion analysis to address the fundamental challenge of localizing functions in a neural system. We describe the Functional Contribution Analysis (FCA) which assigns contribution values to the elements of the network such that the ability to predict the network's performance in response to multi-lesions is maximized. The approach is thoroughly examined on neurocontroller networks of evolved autonomous agents. The FCA portrays a stable set of neuronal contributions and accurate multi-lesion predictions, which are significantly better than those obtained based on the classical single lesion approach. It is also utilized for a detailed synaptic analysis of the neurocontroller connectivity network, delineating its main functional backbone. The FCA provides a...

How is neural information processing to be understood? One of the difficult first challenges is to identify the roles of the network's elements, be they single neurons, neuronal assemblies or cortical regions (depending on the scale on which the system is analyzed). Even simple nervous systems are capable of performing multiple and unrelated tasks. Each task recruits some of the elements of the system, and often the same element participates in several tasks. A precise quantification of the elements contributions to the different tasks may provide insights regarding the functioning of the nervous system, raise new hypotheses and lead the way to further research. Localization of specific tasks in the nervous system is conventionally done by recording the activity of the system elements during behavior, mainly using electrical recordings and neuroimaging techniques. Using the recorded activity, the correlation between elements and some...

L'apprentissage par renforcement fournit un cadre explicatif de nombreux aspects du comportement animal, en particulier de processus sophistiqu s comme l'acquisition d'un nouveau comportement, l'adaptation du comportement l'environnement, l'acquisition de squence comportementale, la gnralisation de l'apprentissage, l'extinction, le faonnage, les interactions sociales, l'ducation des petits, ... On peut donc esprer que sa formalisation algorithmique soit utile pour rsoudre certains problmes d'informatique ou de robotique. Ce type d'apprentissage fournit des ides qui mritent d'tre tudies par les informaticiens pour qui apprentissage par renforcement rime trop souvent avec une procdure d'apprentissage par essai/erreurs. Dans cet article, nous prsentons et discutons de la formalisation algorithmique de l'apprentissage par renforcement. Nous prsentons ensuite une architecture multi-agents dont le contrle de chacun des agents est ralis par un algorithme de renforcement. Ce travail a pour objectif d'valuer les possibilits d'une approche purement renforcement, d'une part comme modle du vivant, d'autre part pour valuer les possiblits de ce type d'approche en terme d'algorithme de rsolution de problme d'informatique. Enn, nous discutons le dveloppement de ces techniques pour les systmes articiels.