Neural circuit connectivity and dynamics for structured memory and computations
Brains have a remarkable capacity to store and use information to perform computations, often sequentially or with few stimulus presentations. Understanding how they achieve this is key for neuroscience and machine learning applications, and requires identifying basic principles behind memory and computation. Although single-neuron and population-level activities are often well understood, the backbone of circuits - their connectivity - rarely is. In this talk, I first describe models performing sequential computations to show structured connectivity motifs that give rise to activity consistent with experiments and to propose that networks can learn to structure their connections to perform such computations. Then, I examine how neural circuits can use synaptic plasticity to rapidly encode information in an organized manner that supports flexible computations. Together, these results suggest mechanistic principles for memory and computation in biological and artificial neural networks.
Related publication: Kalel L. Rossi and Aneta Koseska, Structured connectivity for structured sequential computations, https://doi.org/10.64898/2026.08.03.742515
This IFISC Seminar will be broadcasted in the following zoom link: https://us06web.zoom.us/j/89466064429?pwd=po9p99eAEYVPaNI8xIIGoOIz0hOqaF.1
Coffee and cookies will be served 15 minutes before the start of the seminar
Detalles de contacto:
Raul de Palma Contact form