The dynamic cortex in perception and learning

In this video, we first demonstrate how fast oscillations may arise in the cortex by implementing a data-driven model representing all layers and cell types in the cortical column, and by applying realistic principles for experience-dependent plasticity. We continue by showing how such plausible cortical microcircuitry can be used to enable cortical structures to perform predictive coding - to generate latent representations from which the original visual input can be reconstructed. Finally, we forge a transition from these microcircuit models to large-scale multi-area models that show how the visual cortical system, cooperating with the motor system, can segregate moving objects from the background, even in cases where the subject is moving itself.
Giulia Moreni (University of Amsterdam, HBP)
Jorge Mejias (University of Amsterdam, HBP)
Kwangjun Lee (University of Amsterdam, HBP)
Matthias Brucklacher (University of Amsterdam, HBP)
Cyriel Pennartz (University of Amsterdam, HBP)
Sander Bohte (Centrum Wiskunde & Informatica, HBP);
Shirin Dora (Loughborough University);
Timo Dickscheid (JUELICH, HBP)
Hans Ekkard Plesser (NEST Support, EBRAINS);
Evan Hancock (Digital Media Content Assistant, EBRAINS);
Allen Institute for Neuroscience (Open Data)
Angelica da Silva Lantyer (University of Amsterdam, HBP)

[@synthbrain3173]

finally Brain related artificial neural networks, thanks guys for great work

[@svegritet]

Thank you! To read out results, do you use spoken language? Or...?

[@matthiasbrucklacher]

Hi, to read out object identity, we use a standard approach of (linearly) mapping neural responses to neurons encoding object classes.

[@geraldspringer6882]

❤