Influence and Representation of Task Elements on Neural Networks

Hosted by Greg

External speaker

Nicholas Menghi

Position: Postdoc, Doeller Lab at Max Planck Institute for Human Cognitive and Brain Science, Leipzig

Title: The effect of task similarity during representation learning in brains and neural networks

 

Local Berlin speaker

Claudia Böhm

Position: Group leader, Neural Networks Lab, Charité

Title: Organization of task elements as functional modules in medial prefrontal cortex

Abstract:

In our talks, we explore how neural systems represent task-relevant features to support learning and adaptive behavior across species. In rodents, neurons in the medial prefrontal cortex encode elements of a navigation task—such as key locations, actions, and task-phase-dependent movement directions—in patterns that reflect their functional role within the task. These structured population codes reveal how abstract features of a behavioral task are organized in the brain, suggesting a functional motifs that may support flexible behavior. In humans, we look at the development of task representations sharing or not a common structures. Using neuroimaging, we show that neural systems dynamically adjust representational geometry to resolve interference and support generalization. Finally, we compare our findings to highlight a common principles of neural representation across rats, humans, and machines.

 

 

-----------------------------

 

This event is part of the PostDoc Network Speaker Series.

Guests are welcome!