How Working Memory Works: Brain Circuits & Reverse Engineering | Dr. Julio Martinez-Trujillo

Added:

"Activity-Silent" Working Memory and Short-Term Synaptic Plasticity

Traditional models of working memory rely heavily on "persistent activity"—the continuous, energy-expensive firing of neurons within recurrent circuits to maintain information. However, a major alternative perspective known as "activity-silent" working memory challenges this classic model. Proponents of this theory argue that working memory can be maintained during temporary gaps in active neural firing. Instead of continuous electrical activity, information is encoded through transient changes in the strength of synaptic connections (short-term synaptic plasticity). When the information needs to be retrieved, a general neural impulse reactivates the silent network to read out the stored state. This counterpoint suggests that working memory is far more energy-efficient than recurrent network models imply. For students and researchers, this shifts the focus of reverse-engineering and neural technology from tracking continuous firing rates to decoding dynamic, short-term synaptic modifications.

The cognitive psychological definition of working memory, including its distinction from short-term and long-term memory systems.
Basic neurobiology, specifically how neurons communicate via action potentials and synapses to form functional brain circuits.
The fundamental role of the Prefrontal Cortex (PFC) in executive function, attention, and sensory information integration.
The basic computational concept of Recurrent Neural Networks (RNNs) and how feedback loops allow systems to maintain state over time.
The development of Brain-Computer Interfaces (BCIs) that decode cognitive states and intention from population-level neural activity.
Neuromorphic computing paradigms that design hardware architectures based on biological working memory and recurrent neural circuits.
Clinical applications in computational psychiatry, exploring how microcircuit disruptions lead to working memory deficits in schizophrenia and ADHD.
The scientific debate surrounding working memory mechanisms, specifically comparing persistent activity models with activity-silent (synaptic) working memory.
117 views0likes1:18:25@BiomedicalEngineeringatFIUOriginal Release: 2023-04-01

Working memory encoding occurs in association areas of the brain (such as the prefrontal cortex) through persistent neural firing maintained by recurrent network dynamics, rather than being stored in early sensory areas; this is supported by research using virtual reality tasks, optogenetic manipulation, and multi-electrode recordings showing that neurons in the medial temporal lobe and prefrontal cortex encode contextual and episodic information through coordinated sequential firing patterns, with NMDA receptors playing a crucial role in maintaining these memory representations.