Beyond Synapses: Reconfigurability as a Unifying Principle of Learning.
Event Details
- Type
- Center for Studies in Physics and Biology Seminars
- Speaker(s)
-
Jennifer Schwarz, Ph.D., professor, Syracuse University
- Speaker bio(s)
-
Advances in artificial intelligence have made it clear that learning does not require a brain. This raises a deeper question: what are the physical mechanisms that enable learning across systems? In neuroscience, learning is typically associated with synaptic plasticity. However, plasticity can be viewed as one instance of a broader principle known as reconfigurability that spans neuronal, cellular, and even nonliving systems. At the neuronal level, perturbations reorganize functional connectivity and shift the stability and accessibility of network activity patterns, highlighting that changes in synaptic plasticity manifest as a reorganization of network-level behavior. Extending beyond the brain, multicellular tissues can be trained to encode stress patterns through mechanochemical feedback, where both parameter adaptation and cellular rearrangements contribute. At an even more minimal level, disordered particle packings can learn associative tasks via local driving, with memory stored in structural rearrangements rather than fixed connections. Together, these results suggest a unifying physical learning framework for living and nonliving matter.
- Open to
- Public
- Phone
- (212) 327-8636
- Sponsor
-
Melanie Lee
(212) 327-8636
leem@rockefeller.edu