|
Computational Single-Neuron Mechanisms Linking Perception and Memory in the Human Brain| title | Computational Single-Neuron Mechanisms Linking Perception and Memory in the Human Brain |
|---|
| start_date | 2026/06/19 |
|---|
| schedule | 11h |
|---|
| online | no |
|---|
| location_info | Amphitheater NeuroSpin & Online |
|---|
| summary | Perception allows the brain to extract structure from sensory input, while memory enables these representations to persist and guide future behavior. A central unresolved question in neuroscience is how cortical perceptual representations are transformed into stable memory representations. Drawing on human single-neuron recordings combined with advanced computational analyses, I will present evidence for a region-based feature coding mechanism in the medial temporal lobe (MTL). Rather than encoding discrete concepts, MTL neurons exhibit receptive fields within a high-level visual feature space, analogous to hippocampal place cells that represent locations in physical space. I will describe a computational pathway through which dense, feature-based representations in ventral temporal cortex evolve into sparse, memory-relevant representations in the MTL, supporting discrimination, generalization, and recognition. Learning progressively refines these neural codes, increasing representational separation between similar stimuli and stabilizing memory representations. Finally, I will show that burst dynamics in MTL neurons encode novelty, memory reinstatement, and pattern separation/completion, consistent with attractor-network mechanisms. Together, these findings reveal computational single-neuron principles linking perception, learning, and memory in the human brain. |
|---|
| responsibles | Blancho |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2026/06/17 12:41 UTC |
| |
|