Most theories in psychology try to explain a narrow range of phenomena in a particular area of mental competence, like memory, language, understanding or visual attention. Such theories can be thought of as Micro-theories in the sense that they address only parts of the full range of mental capabilities. In most practical contexts all these mental competencies work together. Although many theories have something to say about how one kind of mental capability relates to another, few do so in a balanced way - i.e. at an equivalent level of detail. Interacting Cognitive Subsystems is a form of Macro-theory which seeks to do just that. Macro-theory is aimed at understanding how all the different components of the mental mechanism are configured, what representations those components use and the overall dynamics of their interactions in real time. The aim of integrating over areas has a long history in psychology with, for example, Newell (1990) arguing for "Unified theories of Cognition." Our approach, while in the same spirit, places rather greater emphasis on the "whole system" - including a strong role for embodiment and emotion.

Note: You can download some interactive animations of ICS by scrolling to the bottom of this page.

The diagram below represents the "full" ICS architecture - the resources involved are rather complex, but when we consider the full range range of mental capabilities we possess, this is only to be expected. There are nine different types of representation in this system, but all subsystems are built out of a minimal number of elements as indicated in the key at the base of the diagram.

The theory assumes that our mental architecture is composed of nine subsystems. Each subsystem has the same internal structure. Information arrives in a subsystem, is copied into an image record and is transformed for use in another subsystem. The subsystems differ in their inputs and outputs - they each specialise in storing and processing a qualitatively different form of mental representation. ICS stresses that mental activity occurs in mulitple domains at the same time - that is concurrently.

Barnard, P.J. and Teasdale, J.D. (1991). Interacting cognitive subsystems: A systemic approach to cognitive-affective interaction and change. Cognition and Emotion, 5, 1-39.

Teasdale, J.D. and Barnard, P.J. (1993). Affect, Cognition and Change: Re-modelling Depressive Thought. Hove: Lawrence Erlbaum Associates

Teasdale, J.D. (1999). Emotional processing, three modes of mind, and the prevention of relapse in depression. Behaviour Research and Therapy, 37, S53-S77.

Barnard, P.J. (1999). Interacting Cognitive Subsystems: modelling working memory phenomena within a multi-processor architecture. In: Miyake, A & Shah, P. (Eds) Models of Working Memory: Mechanisms of Active Maintenance and Executive Control, Chapter 9, pp. 298-339, Cambridge: Cambridge University Press.

and for those who like a challenge - the following source discusses uses of macro-theory in some detail in the context of an "APU" domain of application

Barnard, P., May, J., Duke, D. & Duce, D. (2000). Systems, Interactions and Macrotheory. ACM Transactions on Human-Computer Interaction, 7, 222-262.

The idea of developing macro-theory is, like attempts in AI to build unified theories of cognition, a challenging and ambitious project. Its development and progress cannot, by its very nature, be focused on a small range of empirical paradigms. Exploring the consequences of this idea means that it must address many different "textbook" topics.

Current projects cover the development of the basic theory, empirical investigations of how the two levels of meaning in ICS interact to yield "executive" control and some studies of neural mechanisms. Those currently outlined on this website:

• Applying ICS theory in a clinical context
• A novel process account of Anorexia Nervosa
• Research on Bipolar Disorder
• SenseCam project on Alzheimer's

• ICS and the evolution of cognitive capability and refined emotions

• Mood and memory
• Meaning attention and mood
• Anxiety and attention to threat-related meanings
• Conceptual implicit memory in dysphoria
• Attention to, and memory for, images with impact
• Automatic processing in moral cognition
• Paying attention to social meanings (fMRI study)
• Embodied Cognition

The key ideas behind ICS are really quite simple. Nonetheless, with 9 subsystems, each with multiple components, its still very hard to think about how a complex system like that depicted in the figure above actually works. In our view, complexity will inevitably result when multiple entities interact over time. Two sources may help to grasp what is involved - one is to access animations (see below) that graphically illustrate "concurrent processing". The other is to consider how nine subsystems could have evolved from very much simpler ones that lack verbal and spatial imagery as well as propositional meaning.

While we argue that nine subsystems are required to account for all the intricacies of human cognition and affect, we would probably (at best) need just four to account for the mental and behavioural capabilities of most mammals. This very much simpler mental architecture is shown below. It has a single multimodal subsystem at its centre that augments more direct routes for action control on the basis of cross-modal contingencies.

In a recent paper we have proposed a detailed mechanism, an information processing analogue of cell division in biology, which could have led successive daughter subsystems to evolve along an evolutionary trajectory that culminated in the 9-subsystem architecture shown above. Thinking about how the control of action may have been successively augmented by additional types of representation and processes is one way of gradually coming to grips with the intricacies of a highly parallel dynamic architecture. The evolutionary trajectory outlined below follows a sequence which assigns a 6-subsystem architecture to modern great apes and our last common ancestor with them. The logic that follows from our proposed mechanism asserts that there must have been (now extinct) species with 7- and 8-subsystem architectures. The computational argument on which the evolutionary trajectory, shown below, is based supports speculative inferences to be made about the properties of hominin cognition as well as about their emotions.

For more on this theoretical viewpoint on how our advanced capabilities in cognition and emotion could have evolved see

Barnard, P.J., Duke, D.J., Byrne, R.W. & Davidson, (2007). Differentiation in cognitive and emotional meanings: an evolutionary analysis Cognition and Emotion, 21(6), 1155-1183.

Two further papers addressing related evolutionary issues are

Barnard, P.J. (2010). From executive mechanisms underlying perception and action to the parallel processing of meaning. Current Anthropology, 51, S1, S39-S54.

This last paper was prepared for A Wenner-Gren symposium on the evolution of working memory that took place in Cascais, Portugal in March 2008.

Some of the wider issues that need to be addressed in relating brain, cognitive and behavioural theory to interpretations of the archaeological record of tool use by early hominins are addressed in

Barnard, P.J., (2010). Current developments in inferring cognitive capabilities from the archaeological traces left by stone tools: caught between a rock and a hard inference. In A. Nowell and I. Davidson, I. (Eds). Stone Tools and the evolution of cognition, Ch 10, pp207-226, Boulder, CO: Colorado University Press.

To understand a dynamic system you need to think about two things - (1) Principles that govern information flow and the principles underlying the strucuture of mental representations; and (2) How mulitple components in a system interact in real time. Mental activity can be thought in much the same way as the behaviour of a team. The individual members of the team are each doing something different but their conjoint activity is co-ordinated and constrained by common rules.

You can explore a simple interactive animations package which shows information flow among the subsystems of ICS. This is a cut-down version of a bigger package used to give talks. This version is focused on how central executive functions can be implemented in a distributed system. It is primarily intended for people who have already read a paper or two on ICS, or have heard a talk. It isn't really intended as a stand alone tutorial. If you do look at it, be sure to go through the overview first! The full demo takes a while to explore - but you can escape at any time. You can download zipped files of a playable animation

for IBM compatibles

for Mac OSX.

The picture below shows the main menu but this is not interactive - we should have the shockwave version up on the web soon.

These animations provide dynamic illustrations of the ICS approach to "central executive mechanisms" and some illustrations of configural patterns of information flow argued to move to extremes in key psychopathologies.

There is also a tutorial document available that we used to teach people the basic properties of mental representations in ICS. This was directed at the design of graphic displays, but the principles it describes are once again general:

May, J., Scott, S. and Barnard, P. (1995) Structuring Displays: a psychological guide. Eurographics Tutorial Notes Series. PS95 TN4, ISSN 1017-4656 Geneva: European Association for Computer Graphics.