Attractor Models of Cortical Computations (Working) Memory - PowerPoint PPT Presentation

Master of Integrative Biology Neuroscience – UE 5BN04 2018-19 Attractor Models of Cortical Computations (Working) Memory Gianluigi Mongillo Center for Neurophysics, Physiology and Pathology Paris Descartes University – CNRS UMR 8119

Learning/Memory Learning – process by which relatively permanent changes occur in behavioral potential as a result of experience . Memory – relatively permanent record of the experience that under- lies learning. Learning – acquiring new knowledge from experience modification neurons Internal representations synapses stabilization Memory – retention of the acquired knowledge over time

Models of Behavior How do organisms select appropriate behavior? reactive stimulus response “ In mammals even as low as the rat it has turned out to be impossible to describe behavior as an interaction directly between sensory and motor processes ” (Hebb, 1949)

Models of Behavior How do organisms select appropriate behavior? reactive stimulus response “ In mammals even as low as the rat it has turned out to be impossible to describe behavior as an interaction directly between sensory and motor processes ” (Hebb, 1949) cognitive internal stimulus response representations

The Cell Assembly Hypothesis “Let us assume that the persistence or repetition of a reverberatory activity (or 'trace') tends to induce lasting cellular changes that add to its stability...” ● Perceptive experience activates sub-populations in neuronal assemblies, by increasing/decreasing firing rates. ● Pairs of activated cells potentiate synapses between them, while synapses from activated to non-activated cells are depressed. ● The resulting cell assembly is able to sustain a pattern of activity similar to the perceptive one in absence of the eliciting stimulus. (Hebb, 1949)

The Hebb Framework association areas sensory motor

The Hebb Framework association areas sensory motor

The Hebb Framework association areas sensory motor reverberatory activity

The Hebb Framework association areas sensory motor reverberatory activity

The Hebb Framework association areas sensory motor reverberatory activity

Working Memory (WM) WM refers to the mechanism(s) underlying the maintenance of task-relevant information while performing the task. Items in WM are available in a special status, which makes them able to drive/control behavior (active maintenance). Delayed-response paradigm Object Memory test cue correct delay period error

Working Memory (WM) WM refers to the mechanism(s) underlying the maintenance of task-relevant information while performing the task. Items in WM are available in a special status, which makes them able to drive/control behavior (active maintenance). Delayed-response paradigm Spatial Memory delay period cue test correct + + + saccade + + + + + + + + +

DMS task and Delay Activity test cue correct delay period error cue delay period (adapted from Meyer et al., 2007)

DMS task and Delay Activity test cue correct delay period error cue delay period (adapted from Meyer et al., 2007)

DMS task and Delay Activity test cue correct delay period error cue delay period (adapted from Meyer et al., 2007)

ODR task and Delay Activity (adapted from Funahashi et al. , 1989)

Different Delay's Durations (adapted from Funahashi et al. , 1989)

Delay Activity During Error Trials (adapted from Funahashi et al. , 1989)

Mechanistic Accounts Active maintenance as a result of the collective network dynamics Active maintenance as a result changes in single-cell excitability Active maintenance as a result of short-term modifications of synaptic efficacies

The Cell Assembly Hypothesis 'active maintenance' through reverberatory activity spontaneous activity y activity t i v i t c e n time neuronal n external input o population c memory activity t n e r r activity u c Hebbian assembly e r time (Hebb, 1949; Amit, 1995)

Persistent Activity and Network Multi-Stability + + + + + + + + cue period delay period

Persistent Activity and Network Multi-Stability + + + + + + + + cue period delay period + + + + + + + +

A Toy Model S i ( t )= 0,1 y t i v i t I i ( t )=μ ext + σ ext ⋅η i ( t )+ ∑ j J ij S j ( t − 1 ) c e n neuronal n external input o population c 1 if I i ( t )≥ 1 t n e S i ( t )= r r u 0 otherwise c Hebbian assembly e r all-to-all connected: J ij = J / N

A Toy Model S i ( t )= 0,1 y t i v i t I i ( t )=μ ext + σ ext ⋅η i ( t )+ ∑ j J ij S j ( t − 1 ) c e n neuronal n external input o population c 1 if I i ( t )≥ 1 t n e S i ( t )= r r u 0 otherwise c Hebbian assembly e r all-to-all connected: J ij = J / N ν( t )= 1 N ∑ i S i ( t ) ∞ Φ(ν)= 1 √ 2 π ∫ 1 −μ ext − J ν − x 2 / 2 ν( t + 1 )=Φ[ν( t )] dx e with σ ext

Bi-Stability through Positive Feedback 1.0 ν( t + 1 ) 0.5 0 0 0.5 1.0 ν( t )

Bi-Stability through Positive Feedback 1.0 ν( t + 1 ) 0.5 0 0 0 0 0.5 1.0 ν( t )

Bi-Stability through Positive Feedback 1.0 ν( t + 1 ) 0.5 0 0 0 0 0.5 1.0 ν( t )

Bi-Stability through Positive Feedback 1.0 0.2 ν( t + 1 ) 0.5 0.1 0 0 0 0 0.5 1.0 0 ν( t ) 0 0.1 0.1

Realistic Models with Spiking Neurons Single-cell Dynamics Network Architecture (Amit & Brunel, 1997; Brunel , 2000)

AB Model: Population Activities sel. other sel. non sel. inhib. (Brunel , 2000)

AB Model: Single-Cell Spiking Patterns sel. other sel. non sel. inhib. (Brunel , 2000)

Through Single-Cell Properties Firing is sustained by increased membrane excitability which is refreshed through network oscillations ADP time course V ( t )= V osc ( t )+ V ADP ( t )− V inh ( t ) Acetylcholine Serotonin (adapted from Lisman & Idiart, 1995)

Through Single-Cell Properties Firing is sustained by increased membrane excitability which is refreshed through network oscillations ADP time course V ( t )= V osc ( t )+ V ADP ( t )− V inh ( t ) Acetylcholine Serotonin (adapted from Lisman & Idiart, 1995)

Through Single-Cell Properties Firing is sustained by increased membrane excitability which is refreshed through network oscillations ADP time course V ( t )= V osc ( t )+ V ADP ( t )− V inh ( t ) Acetylcholine Serotonin (adapted from Lisman & Idiart, 1995)

Through Single-Cell Properties Firing is sustained by increased membrane excitability which is refreshed through network oscillations ADP time course V ( t )= V osc ( t )+ V ADP ( t )− V inh ( t ) Acetylcholine Serotonin (adapted from Lisman & Idiart, 1995)

Through Single-Cell Properties Sternberg effect (1966) Response time increase linearly with the nr. of items in memory (adapted from Lisman & Idiart, 1995)

Through Single-Cell Properties Sternberg effect (1966) Nested oscillations (slow/fast) Observed both in cortex and hippocampus: - segment information in time Response time increase linearly - time compression (for associations) with the nr. of items in memory (adapted from Lisman & Idiart, 1995)