COSSAC

COGENT focuses on new techniques and methods for cognitive modelling. COGENT is a flexible computational modelling system within which to develop and explore symbolic and hybrid models of cognitive processes.

A collaboration between Birkbeck College, London and the Imperial Cancer Research Fund (now Cancer Research UK); supported by the Engineering and Physical Sciences Research Council (1996-2001).

The Cogent cognitive modelling system

Overview

In 1990 the MRC/EPSRC JRC initiative in cognitive science funded work on the Sceptic computer language (Cooper et al. 1996) which was developed into a powerful simulation tool and rigourous method for computational modelling of cognitive and neurological systems. Continuing funding by the EPSRC lead to the development of a powerful modelling environment, COGENT (Cooper & Fox, 1998), which allows models to be constructed and tested using a mixed graphical and rule-based system, together with extensive support for combined programmes of experimental, theoretical and model-based research (e.g. documenting developing theories over time, archiving of theoretical and experimental results, and direct comparison of predicted and actual data, sharing of models and data via the internet).

COGENT takes an object-oriented approach to modelling. This has two important motivations:

* Objects in a model are encapsulated. That is, their internal structure is defined separately from their interaction with other objects. This allows the overall structure of a simulation to be developed independently of the detailed operation of its components. Models may be hierarchically refined by working first on the most general structure then refining the operation of components, and the implications of different lower-level assumptions within the same high-level structure may easily be assessed.
* Generalised object classes may be defined, with the objects that comprise a particular model being specialised instances of them. Hierarchical specialisation may be used. For example (in the field of cognitive neuroscience) a general "information storage buffer" class may be specialised into a "memory buffer" class, which may be further refined into a "short-term, limited capacity memory buffer" object.

COGENT has been successful, leading to publications in the technical literature (Cooper & Fox, 1998) and in basic research. A book on the software and its use in cognitive science has been published. More than 1000 copies of the system have been downloaded from the internet.

Link:

Main COGENT website: Birkbeck, University of London

Publications:

COGENT-related COSSAC publications

COSSAC Contact

David Glasspool: dglasspo@inf.ed.ac.uk