Killam Seminar Series: The neural basis of flexible, lifelong learning
Supported by the generosity of the Killam Trusts, The Neuro's Killam Seminar Series invites outstanding guest speakers whose research is of interest to the scientific community at The Neuro and 91丝瓜视频.
To watch online, click聽
贬辞蝉迟:听
The neural basis of flexible, lifelong learning
Abstract: I will present recent work demonstrating that the auditory cortex (AC) is necessary for sound discrimination learning but dispensable at expert levels. Rather than reflecting changes in sensory tuning or map expansion, learning and performance rely on distinct higher-order computations鈥攔eward prediction for learning and action suppression to enable stable performance鈥攊mplemented by spatially clustered neural ensembles. These clusters are orthogonal to classic stimulus-tuning, revealing an organizational principle that extends beyond sensory coding. I will close by shifting from single-task learning to our developing framework for studying lifelong, multi-task learning in mice, centered on our Continual Learning Mouse Playground: a fully automated, high-throughput home-cage environment in which mice learn a progression of perceptual, contextual, and compositional tasks over months. This platform enables us to investigate how biological brains acquire, reuse, and recombine knowledge across the lifespan, with a particular focus on the mechanisms supporting compositionality and generalization.
Kishore V. Kuchibhotla
Associate Professor at Johns Hopkins University
I will present recent work demonstrating that the auditory cortex (AC) is necessary for sound discrimination learning but dispensable at expert levels. Rather than reflecting changes in sensory tuning or map expansion, learning and performance rely on distinct higher-order computations鈥攔eward prediction for learning and action suppression to enable stable performance鈥攊mplemented by spatially clustered neural ensembles. These clusters are orthogonal to classic stimulus-tuning, revealing an organizational principle that extends beyond sensory coding. I will close by shifting from single-task learning to our developing framework for studying lifelong, multi-task learning in mice, centered on our Continual Learning Mouse Playground: a fully automated, high-throughput home-cage environment in which mice learn a progression of perceptual, contextual, and compositional tasks over months. This platform enables us to investigate how biological brains acquire, reuse, and recombine knowledge across the lifespan, with a particular focus on the mechanisms supporting compositionality and generalization.
