You are here

Paul F.M.J. Verschure

La Nau building - 51.113
93.5421 372

Dr. Paul F.M.J. Verschure (1962)

Research professor  Catalan Institute of Advanced Research (ICREA),
Professor of Cognitive science and Neurorobotics at the Technology Department, Universitat Pompeu Fabra
Director of the Center of Autonomous Systems and Neurorobotics (NRAS)
Scientific Director of the Master in Cognitive Systems and Interactive Media (CSIM)

Paul received both his Ma. and PhD in psychology. His scientific goal is to find a unified theory of mind, brain and body through the use of synthetic methods and to apply such a theory to the development of novel cognitive technologies. Paul has pursued his research at different institutes in the US (Neurosciences Institute and The Salk Institute, both in San Diego) and Europe (University of Amsterdam, University of Zurich and the Swiss Federal Institute of Technology-ETH and Universitat Pompeu Fabra in Barcelona).

Paul works on biologically constrained models of perception, learning, behavior and problem solving that are applied to wheeled and flying robots, interactive spaces and avatars. The results of these projects have been published in leading scientific journals including Nature, Science, PLoS and PNAS. In addition to his basic research, he applies concepts and methods from the study of natural perception, cognition and behavior to the development of interactive creative installations and intelligent immersive spaces. Since 1998, he has, together with his collaborators, generated a series 25 public exhibits of which the most ambitious was the exhibit "Ada: Intelligent space" for the Swiss national exhibition Expo.02, that was visited by 560000 people. The most recent one was the Multimodal Brain Orchestra that premiered in the closing ceremony of the EC Future and Emerging Technologies conference in Prague in April 2009. 

Paul leads SPECS, a multidisciplinary group of over 30 pre-doctoral, doctoral and post-doctoral researchers that include physicists, psychologists, biologists, engineers and computer scientists supported by his own technical and administrative staff.

Paul spends his spare time with his family and training for another Ironman triathlon or sailing.

Some of the ideas he likes (of course from the SPECS group own work):

- Distributed Adaptive Control: An architecture of perception, cognition and behavior (started: 1990)

DAC is a theory of the Mind-Brain-Behavior Nexus that has been developing over the last 20 years

- Predictive Hebbian Learning (started: 1992)

Starting with an observation using robot conditioning experiments in 1993 it became clear that learning in the service of behavior must be based on prediction. Later together with Armin Duff this was formulated in a more general theoretical framework that defines the role of prediction in perceptual and behavioral learning.

- RoBoser (started 1998)

Meeting up with Jonatas Manzolli the link between robots and art - in particular computer music systems, seemed a good idea. It lay the foundation for a sequence of about 25 public exhibitions and performances.

- Temporal Population Code - TPC (started: 2001)

Starting with a suggestion in a paper by Buanomono and Merzenich with Reto Wyss we developed the idea (2003) that the spatially averaged activity of neuronal populations can provide a compact and high-capacity code of complex stimuli. By now it is it has found a lot of physiological support and applications in complex tasks such as real-time face recognition.

- Self-stabilizing negative feedback learning system of the cerebellum (started: 2000)

Following the neural substrate of classical conditioning and prediction to the cerebellum. Having delayed feedback from the output nuclei of  the cerebellum (deep nucleus) to one of its input (inferior olive) seemed curious. These robot studies showed that it is a repdictive signal that stabilizes the learning process.

- Rehabilitation Gaming Station - RGS (started: 2004)

After ADA, "the building that believes it is an organism" exhibition (2002), it became clear that interative media are powerful tools for neurorehabilitation. With RGS we have generalized this to the rehabilitation of stroke. RGS is a unique system that has shown to be more effective than standard treatment for both acute and chronic patients. 

- Consciousness solves the H5W problem (started 2010)

The next challenge for the DAC theory: consciousness.