Why we chose the four Physical Sciences Themes

MEMORY, REASONING AND LEARNING:
What can computers learn from how the brain works?

Today's computer memory is a straightforward store of information. But human memory is more than a mere record of the past. It exists as a subtle relationship between past events, the current context, and the mechanisms of recall. New understanding of memory, reasoning and learning in living systems could lead to new approaches to these processes in computing.

Kieron O'Hara, Wendy Hall, and Nigel Shadbolt, Department of Electronics and Computer Science, University of Southampton
Keith van Rijsbergen, Computing Science Department, University of Glasgow

SENSORY PROCESSING:
Can computers make sense of the world?

Speech recognition, computer vision, even senses such as smell, are making rapid strides as the cost of computer power declines and as research delivers more powerful algorithms and new hardware. Research in 'artificial sensors' could benefit from improved understanding of how the human brain performs these functions. In particular, artificial systems cannot even begin to match living organisms in handling data from several senses. For example, they cannot even handle several aspects of the same task, tracking a moving object as it rotates.

Lionel Tarassenko, Department of Engineering Science, University of Oxford

SENSORY PROCESSING - A NEUROCOMPUTATIONAL PERSPECTIVE:
Lessons from living systems

For computers, vision, speech recognition and other 'sensing' is now achieved by throwing computational power at a series of complex algorithms (equations) that use probability to predict what is happening. Biological sensory systems do things differently. For example, the brain deploys a flexible array of different systems that work in parallel and adapt according to the circumstances. Recent research has begun to characterise how living systems sense their surroundings. This could lead to new computational architectures and biologically inspired artificial sensory systems.

Mike Denham, Plymouth Institute of Neuroscience and Centre for Neural and Adaptive Systems, University of Plymouth

LARGE SCALE/SMALL SCALE ISSUES IN COGNITIVE SYSTEMS RESEARCH:
Can computers think complicated?

Computers are exceedingly simple devices in comparison with the human brain. To build IT systems with any claim to real artificial intelligence we may have to build systems with the complexity of the brain. The brain may offer us ideas on how to build massively parallel systems that, like the brain, can adapt to handle new problems, repair themselves, run on little power and operate quickly with slow components. But first we have to understand how the brain works. This means looking at the workings of the brain from the 'component' level of its micro-architecture, through to the larger scale system level.

Jim Austin, Flaviu Adrian Marginean, Department of Computer Science, University of York
Andy Wright, BAE Systems, Bristol

INTERACTIONS:
Can computers leap the communication barrier?

Information technology is moving from standalone systems to a regime of connected computers. This raises issues not just of how computers communicate with one another but of human computer interaction (HCI), how people and computers interact. Computers also have an increasing impact on how people interact with one another, through email and text messaging for example. Study of how the brain manages the flow of information could help us to understand the growing complexity of the interactions between computers. Research can also help us to understand, and improve, the way in which information is stored and processed for human consumption.

Mark J. Weal, Danius T. Michaelides, Intelligence, Agents, Multimedia, University of Southampton

PLANNING: PERSPECTIVES AND PROMISE:
Can computers plan it better?

Individuals and organisations increasingly use computers to plan everything from their personal finances through to major projects. Research in artificial intelligence (AI) has begun to make an impact on automatic planning, the use of computers to generate instructions indicating who does what, when and how. Automatic planning depends on the ability to describe problems in such a way that AI systems can construct meaningful plans. Research in this area can draw inspiration from the work of cognitive and neuroscientists in understanding the role of planning in human interaction with the world.

Derek Long and Maria Fox, Department of Computer Science, University of Durham

MOTIVATION IN COMPUTING AND A.I. :
How do you motivate a computer?

Motivation clearly plays a part in human behaviour. Even in computing you could describe as motivation the ability of a robot to know when its batteries are running low enough for it to need a recharge to keep it 'alive'. Studying motivation, in humans and computers, can encourage new ways of thinking about control.

Michael Luck and Steve Munroe, Department of Electronics and Computer Science, University of Southampton