Internal Models and Predictive Coding from a Robotics and Cognitive Science Perspective: Prof. Bruno Lara and Dr. Alejandra Ciria

Prof. Bruno Lara (Cognitive Robotics, UAEM, Mexico) and Dr. Alejandra Ciria (Cognitive Psychology, UNAM, Mexico) will give a presentation on their research on internal models and predictive coding from a Robotics and Cognitive Science perspective. They are currently visiting researchers at the Adaptive Systems Group at HU Berlin with an Alexander-von-Humboldt Fellowship.

Abstract:

A cognitive system can be conceived as one which fulfills its goals anticipating the causes of its sensations by containing a predictive model of itself and its environment to select and guide action. During the last years, our research has focused on the issues of Internal Models, anticipation and multimodal representations within this framework. We have developed a number of systems that successfully make use of these concepts in the production of coherent behavior. Now, our main research interest is the study of predictions and how these can be learned by an agent taking into account the specific dynamics of its internal states. Special interest lays in the impact these dynamics bring into the learning capabilities and the interactions with the world of an agent. How does an artificial agent decide what is the relevant information to learn during its interactions with the world? Predicting future states accurately while seeking unanticipated novel states are competing pressures within an agent. We believe that these competing pressures should be resolved by tracking the prediction error dynamics of its internal states. An artificial agent endowed with this mechanism should be able to decide and plan its future actions in accordance with the proper type of novelty given its current knowledge and capabilities. Therefore, an artificial agent should be intrinsically motivated to select actions given its epistemic value providing the capability of open-ended learning.

Bruno Lara is a Professor at the Universidad Autónoma del Estado de Morelos (UAEM) and the Head of the Cognitive Robotics Lab at the Science Research Center at the UAEM since 2005. He holds a PhD in Mechatronics from King’s
College London. He did a postdoc in the TheoriLabor in the University of Jena, working on evolutionary robotics, and then a postdoc at the Max Planck Institute for Psychology Research in Munich, focusing on research on Cognitive
Robotics. In 2011, he spent a sabbatical stay in the Cognitive Robotics Lab at the Humboldt- Universitat zu Berlin. He is an Alexander von Humboldt Fellow.
His research interests include internal models, prediction, sensorimotor representations, and evolutionary robotics.

Alejandra Ciria is a postdoctoral researcher at the Humboldt-Universität zu Berlin, Department of Computer Science in the Adaptive Systems Group, Germany, Berlin (Alexander von Humboldt Foundation, Special Alumni Sponsorship, 2019).
She obtained a Master’s Degree in Cognitive Sciences in 2013 at the UAEM, Morelos, México.
She then studied a Ph.D. in Experimental Psychology at the Faculty of Psychology, UNAM, México, graduating in 2019.
Since 2011, she is an active member of the Cognitive Robotics Lab at the CInC-UAEM, México.
Her research focuses on computational modeling under the predictive processing framework, as well as the study of prediction error dynamics as a self-regulating mechanism.
Alejandra ́s research interests are centered in Psychophysics, Experimental Psychology, Cognitive Robotics, and Cognitive Sciences.