Skip to main content

PARALLEL PROCESSING IN NEURONAL NETWORKS

Objective


Aspects of the operation of neuronal networks in the brain have been studied following several interlinked lines of research using complementary experimental, analytical and simulation techniques.

The role of the hippocampus in memory was studied using network models of the hippocampus. Realistic mathematical models for long term memory are being developed and analysed and then applied to theories of hippocampal function. Experimental investigations have taken place in the rat involving neuroanatomical research on the intrinsic circuitry of the hippocampus. Results showed that there maybe more extensive intrahippocampal and extrahippocampal projections from the CA1 region of the hippocampus than generally recognized. Behavioural and electrophysiological work on the role of N-methyl-D-aspartate (NMDA) receptors in learning has shown that blocking these receptors with the selective antagonist D-2-amino-5-phosphonopentanoic acid (D-AP5) at intercerebral concentrations blocks hippocampal long term potential (LTP) resulting in a deficit of new learning but no impairment in the retrieval of previously established information. It was also found that the same drug had little or no effect on the short term storage of spatial information.

Experimental work has continued examining whether saturation of hippocampal LTP by repeated trains of high frequency stimulation can cause an anterograde amnesia for certain kinds of hippocampal dependant learning. Work is also underway examining the phenomenon of mossy fibre LTP in vivo whose characteristics have important computational implications.

General properties of neural network models in relation to memory have been studied involving approaches to biological realism, anatomical organization, learning rules, information encodings in the brain, network simulations on parallel computers, models of brain damage, neuronal network properties and their physical analogues, and experimental studies of neuronal networks.

Models o f information processing in vision have also been developed involving neural representations invariant with respect to motion, models for information processing of optical flow, Hebbian rules in the visual system and image restoration.
Mathematical models and computer simulations using transputers of the types of processing performed in neuronal networks by biological information processing systems will be made, in order to provide simulations of computer processes which might be useful in pattern recognition and memory, and in order to understand better how biological information processing and storage works computationally. Part of the work will focus on how the hippocampus is involved in memory using neuronal network models of it based on neuroanatomical and neurophysiological investigations, part of the work will focus on models of visual information processing, and these related investigations will be complemented by theoretical investigations of information storage in networks of neurones by D. Wallace, G. Parisi and B. Derrida.. In addition, it is proposed that Prof. Wallace's group at Edinburgh provide transputer-based neuronal network simulation support for the EEC BRAIN stimulation initiative.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

THE CHANCELLOR, MASTERS AND SCHOLARS OF THE UNIVERSITY OF OXFORD
Address
University Offices, Wellington Square
Oxford
United Kingdom

Participants (4)

Commissariat à l'Energie Atomique (CEA)
France
Address
Centre D'études De Saclay
91191 Gif-sur-yvette
Consorzio Genova Ricerche
Italy
Address
Via Dell'acciaio 139
16152 Genova
ISTITUTO NAZIONALE DI FISICA NUCLEARE
Italy
Address
Via Enrico Fermi 40
00044 Frascati
THE CHANCELLOR, MASTERS AND SCHOLARS OF THE UNIVERSITY OF CAMBRIDGE
United Kingdom
Address
Trinity Lane, The Old Schools,
CB2 1TN Cambridge