Oxford Centre for Computational Neuroscience

Professor Edmund T. Rolls

Discoveries on the Neuroscience of Memory, Spatial Function, and Navigation


Brain Computations


Cerebral Cortex


The Noisy Brain


Memory, Attention, and Decision-Making


Neural Networks and Brain Function

Hippocampal spatial view neurons that provide an allocentric representation of spatial locations being viewed, and that are updated by self-motion (129, 152, 202, 237, 244, 247, 256, 267, 594, B15, 633).

 

Hippocampal spatial view neurons that combine information about spatial view and the objects (130, 131, 380) or rewards (387), and are involved in recall (399), providing a basis for implementing episodic memory (539, B12, 594, B15).


A theory for how hippocampal spatial view cells are involved in memory (584, 594, 539, B12, B15) and navigation (594, B15, 633).

 

Hippocampal neurons in primates that respond to a combination of spatial view and place, or to place (202).

 

Whole-body motion neurons in the hippocampus (184), more recently termed 'speed cells'. These are relevant to hippocampal spatial representation update by self-motion, i.e. idiothetic update (633).

 

Hippocampal neurons that respond to a combination of spatial view and whole body motion (184, 202).

 

Head direction cells in the primate presubiculum (271).

 

A representation of long-term familiarity memory in the perirhinal cortex (343, 388).


A theory and model of hippocampal operation and episodic memory, including pattern separation and pattern completion (111, 125, 136, 163, 186, 200, 205, 258, 266, 268, 300, 306, 307, 309, 345, 370, 403, 411, 415, 433, 453, 479, 504, 507, 521, 527, 529, 531, 539, 545, 550, 571, 584, B12, B15).


A theory and model of the generation of time in the hippocampal memory system. Entorhinal cortex time ramping cells produce through a competitive network hippocampal time cells, providing neuronal mechanisms to encode the order of events (605). The theory shows how cells could be generated that show 'replay' and 'reverse replay' (605).


A theory and model of coordinate transforms in the dorsal visual system using a combination of gain modulation and slow or trace rule competitive learning. The theory starts with retinal position inputs gain modulated by eye position to produce a head centred representation, followed by gain modulation by head direction, followed by gain modulation by place, to produce an allocentric representation in spatial view coordinates useful for the idiothetic update of hippocampal spatial view cells (612). This is important in the theory of navigation using spatial view cells when the view details are obscured (B15, 633).


A theory of navigation in humans and other primates that utilizes hippocampal spatial view cells to navigate from landmark to landmark (B15, 633). This is an alternative to navigation involving place cells, and does not require a spatial cognitive Euclidean map. Idiothetic update by head direction and whole body motion cells is part of the theory (633). Allocentric bearing to a landmark cells may also be involved in a related type of navigation (633).


A theory of how spatial view cells and hippocampal attractor networks are involved in the art of memory (the method of loci) (571, 595).


Hypertension and impaired memory: even moderate hypertension is associated with reduced hippocampal functional connectivity and impaired memory (625).


The storage capacity of autoassociation and pattern association networks with sparse representations and diluted connectivity (150, 154, 222, 228, 515, 545, B12, B15).


Basal forebrain, probably cholinergic neurons, that project to the cortex and respond to forebrain-decoded reward, aversive, and novel stimuli (144, 145, 146, 177, B7, B11). These are thought to play a role in keeping the cerebral cortex alert to potentially important stimuli, and reducing the adaptation of cortical neurons (B12). Reduction in the performance of this system may contribute to some of the cognitive changes during aging (B8, B9, B12, 540).

 

Mechanisms involving synaptic facilitation that enable several items to be held simultaneously in short-term memory (523) and that may be useful in the syntax for language (537).

 

Information can be retrieved from biologically plausible attractor neuronal networks very rapidly (in less than 2 time constants of the synapses) (with A.Treves and colleagues) (222, 235, 294). This makes cortical computation with attractor networks possible (B8, B12, B15).