16 January 1998

Can you remember where you parked your car this morning? Not where you usually park your car, but where you parked it today?

If, like most people, you sometimes forget where you left your vehicle, you could have experienced a deficit in episodic memory.

University of Queensland Psychology Department researchers Professor Michael Humphreys and Dr Simon Dennis often use the car park analogy to discuss their ground-breaking work on episodic memory.

Professor Humphreys and Dr Dennis are developing new mathematical and computerised models of episodic memory to improve knowledge of the brain's workings. Their study also has potential in improved diagnosis and treatment of memory problems from head injury and advanced age, and even in human-computer interfaces.

Project findings could also be important in safety-critical applications such as refining learned appropriate or safe responses by key public safety professionals, such as ambulance officers and air traffic controllers.

The University of Queensland episodic memory research group, which has attracted more than $800,000 in Australian Research Council large grants since 1984, is an international leader in the field of mathematical modelling of episodic memory, along with groups at the Universities of Toronto, Indiana, Carnegie-Mellon and Oregon.

'Episodic memory is what allows you to remember what you had for breakfast this morning, rather than what you normally have for breakfast,' Professor Humphreys said.

'It allows you to remember the significance of that note you scrawled in your diary or to remember roughly what date someone might have sent you an email.

'Episodic memory allows people to remember specific events through a match with the context.

'The assumption has been that it is important to understand and reinstate the context of an event - its time, place and circumstances - when trying to retrieve that memory.

'Part of the trouble in understanding how context is used to retrieve episodic memories is that sometimes these contextual manipulations will work only on one kind of task, and not another.

'Our research is targeting episodic memory and the blending of memories. Improved understanding might unlock some of the secrets of other kinds of memories, for example in eyewitness memory and repressed memories as a result of traumatic incidents.'

Professor Humphreys and Dr Dennis are developing a new mathematical model to understand and map memory performance changes which occur between events.

Their work concentrates on the processes which help differentiate one memory from another, using ideas from neurophysiology and neuropsychology.

Professor Humphreys said the hippocampus was known to be the brain area crucially involved in episodic memory. However, science had been able to unravel only a few details of its structure to improve understanding of how memories are stored or how the appropriate memory was retrieved when two or more memories or a blend of memories occurred.

He said problems with episodic memory were associated with head injury or ageing, as episodic memory appeared to be more sensitive to these changes than other types of memories. Improved understanding of these memories could be useful for diagnosing the severity of head injuries, and for recovery.

'It is also known in amnesia that some memory abilities are relatively preserved while others are impaired,' he said. 'If we can understand one aspect of the situation we can perhaps understand other aspects of human memory such as problem-solving, and attention.'

Dr Dennis said the work also was important in the human-interface with computers.

'Human memory requirements are not conducive to learning high levels of skill in computers,' he said.

'If you need to find a file in a computer you have to search through a hierarchy of directories. In essence, to access a computer memory you need to know where that file is in the structure.

'However, to retrieve something from human memory, one needs to know the context. Instead of requiring to know where things are located, it should be possible to re-design computers around the document context.

'This work could assist to improve designs for compatible interfaces with human memory systems and in using clues from understanding of how human memory works to rethink how to construct computers.'

Dr Dennis, a postdoctoral research fellow whose PhD on the nature of computation was supervised by Dr Janet Wiles, has been seconded to the Co-operative Research Centre for Distributed Systems Technology to work on the human-computer interface in the Centre's Resource Discovery Unit. His work is concentrating on search engine technology on the World Wide Web.

Professor Humphreys, who has been studying context and the role of episodic memory since 1968, was co-author on a landmark 1989 mathematical model of episodic memory published in Psychology Review with Dr John Bain (now of Griffith University) and retired academic Dr Ray Pike.

This led to a paper examining the computational requirements of human memory in Behavioural and Brain Sciences co-authored with senior lecturer in psychology and computer science Dr Wiles, and Dr Dennis, and another mathematical model of episodic memory co-authored with Dr Mark Chappell in Psychology Review in 1994.

Professor Humphreys has written more than 70 scientific papers and book chapters.

Media: For further information, contact Professor Humphreys, telephone 07 3365 6670 email: mh@psy.uq.edu.au, or Dr Dennis, telephone 07 3365 6778, email: mav@psy.uq.edu.au