@article {1806, title = {A modeling framework for adaptive lifelong learning with transfer and savings through gating in the prefrontal cortex}, journal = {Proceedings of the National Academy of Sciences}, volume = {117}, year = {2020}, pages = {29872{\textendash}29882}, abstract = {

The prefrontal cortex (PFC) enables humans{\textquoteright} ability to flexibly adapt to new environments and circumstances. Disruption of this ability is often a hallmark of prefrontal disease. Neural network models have provided tools to study how the PFC stores and uses information, yet the mechanisms underlying how the PFC is able to adapt and learn about new situations without disrupting preexisting knowledge remain unknown. We use a neural network architecture to show how hierarchical gating can naturally support adaptive learning while preserving memories from prior experience. Furthermore, we show how damage to our network model recapitulates disorders of the human PFC.The prefrontal cortex encodes and stores numerous, often disparate, schemas and flexibly switches between them. Recent research on artificial neural networks trained by reinforcement learning has made it possible to model fundamental processes underlying schema encoding and storage. Yet how the brain is able to create new schemas while preserving and utilizing old schemas remains unclear. Here we propose a simple neural network framework that incorporates hierarchical gating to model the prefrontal cortex{\textquoteright}s ability to flexibly encode and use multiple disparate schemas. We show how gating naturally leads to transfer learning and robust memory savings. We then show how neuropsychological impairments observed in patients with prefrontal damage are mimicked by lesions of our network. Our architecture, which we call DynaMoE, provides a fundamental framework for how the prefrontal cortex may handle the abundance of schemas necessary to navigate the real world.The code used to generate the data in this work is available at https://github.com/tsudacode/DynaMoE.

}, issn = {0027-8424}, doi = {10.1073/pnas.2009591117}, url = {https://www.pnas.org/content/117/47/29872}, author = {Tsuda, Ben and Tye, Kay M. and Siegelmann, Hava T. and Sejnowski, Terrence J.} }