Visual learning at fast and slow timescales is driven by distinct plasticity rules in primate inferotemporal cortex

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Object recognition relies on familiarity learning, involving changes in the inferotemporal cortex (IT).
• Familiarity learning occurs over multiple timescales (minutes to days) and involves complex cellular, synaptic, and network alterations.

Purpose of the Study:

• To investigate the underlying plasticity mechanisms of familiarity learning in the inferotemporal cortex.
• To differentiate between short-term and long-term learning mechanisms using an integrated experimental-theoretical approach.

Main Methods:

• Recorded IT activity in two male macaques during familiarity learning tasks.
• Utilized an integrated experimental-theoretical approach combining neural recordings and computational modeling.
• Developed recurrent neural networks with experimentally inferred learning rules.

Main Results:

• Identified two distinct timescales of learning-related changes: minutes and days.
• Observed gradual response decrease across sessions, consistent with synaptic plasticity (long-term learning).
• Detected rapid response decay and increased spontaneous activity within sessions, indicative of intrinsic plasticity (short-term learning).

Conclusions:

• Familiarity learning involves both slow synaptic plasticity and fast intrinsic plasticity mechanisms.
• These distinct mechanisms operate at different timescales, contributing to the complex dynamics of learning and recognition.
• Computational models incorporating these plasticity rules successfully replicated observed learning dynamics.