Dear all A second postdoc position devoted to learning in deep cortical networks is available in our Computational Neuroscience labs at the University of Bern, Switzerland. This position is focussed on formulating error-based learning in deep cortical networks in terms of a principle of least action. We will extend this theory to natural gradient learning and to learning with spiking neurons. The research of our labs is more broadly devoted to biologically realistic models of spatio-temporal processing in cortical networks and their relation to reinforcement learning, Bayesian computing, and neuromorphic implementations. The position is part of the European Human Brain Project and is available by January 1st, 2021, for 2 years (extendable). We offer a stimulating environment with three research groups in theoretical and computational neuroscience, along with experimental neuroscience at the same Department of Physiology (physio.unibe.ch/gruppen.aspx <http://physio.unibe.ch/gruppen.aspx>), as well as close collaborations with other labs in neuroscience, artificial intelligence and neuromorphic engineering. Ideal candidates should have a strong background in computational neuroscience, machine learning, applied mathematics and/or physics. Please send your CV, publication list, letter of motivation and contact information for at least two references to Walter Senn (walter.senn@unibe.ch <mailto:walter.senn@unibe.ch>) and Mihai Petrovici (mihai.petrovici@unibe.ch <mailto:mihai.petrovici@unibe.ch>), with cc to Virginie Sabado (virginie.sabado@unibe.ch <mailto:virginie.sabado@unibe.ch>). The first evaluation round will begin on November 8, 2020. The positions will remain open until filled. With best regards, Mihai Petrovici and Walter