Brain & Behavior Research Foundation Presentation: Dr Kanaka Rajan, Feb 9th, 2021.
Dr Rajan described computational models to “reverse engineer” how our brain works, which include building single region to multi-regions Recurrent Neural Networks (RNNs), multi-region RNNs that capture within and inter-area interactions, last but not least, her team is constraining the multi-region RNNs model based on collected data.
CURrent-Based Decomposition (CURBD): New framework for tracing neural paths across multiple brain regions in development by Dr Rajan and her team.
Helplessness shown to involve persistent stress in response to adversity in experimental settings (active to passive coping (learned helpless state)) – toy models, larval zebrafish, rats etc. This presentation used studies on learned helplessness, and larval zebrafish in particular as an illustration on how to computationally scale up single region RNN model into a multi-region RNN model of the neural dynamics in the larval zebrafish, in order to infer within and inter-area effects between different RNN regions. CURBD method can dis-entangle time effects, alternative to traditional functional connectivity methods.
Watching the presentation on rat and zebra fish experiments, I wondered about predictable versus unpredictable adversity and variations in response to different types of adversity, which was not the focus of the presentation.
Dr Rajan said that in their computational modeling, general behavior seem to be captured pretty well by their CURBD and that was good enough. Does it matter for computational models to be able to capture general behavior and ignore nuances? My current thoughts are that general behavior modeling is a good initial start, however computational modeling will need to move into capturing nuances to be useful to understand humans. Urie Bronfenbrenner‘s ecology of human development tells us that interactions between systems changes the system itself in the process. how can computational modeling capture this?
Visit Brain & Behavior Research Foundation to view this presentation and watch other presentations.