Computational Neuroscience

© Boström

To gain insights into the functioning of the neural control system, we develop computational models for nociceptive and proprioceptive information processing, as well as for motor control of human movements. We make use of state-of-the-art artificial neural networks based on reservoir computing, and are working on a reformulation of the reafference principle of von Holst and Mittelstedt.
We once started these kind of studies with simple models involving only two or  four neurons. These models were used to understand the postural reflexive responses, as well as the complex activation patterns of shank muscles during walking. Now we arrived at recurrent networks with several thousands of neurons, each interconnected with each other and trained with sensory input to learn motor patterns or to reorganize themselves into a neural map.

PIs: H. Wagner, K. Boström