AERONAUTICS, EXPERIMENTAL PSYCHOLOGY
Radical rethink on cockpit design
When we look at a natural scene, how much of it can we consciously see at once? New research suggests this seemingly simple question has a complex and unexpected answer – one that can help minimise the human factor in a variety of road- and air-traffic accidents.
|Have aircraft and vehicle makers been basing their display designs on outdated theories?|
© Image: PhotoDisc
People think they see and perceive numerous objects in a scene simultaneously, but scientists have shown that we actually take in much less information than expected. New European research suggests that the traditional wisdom behind aircraft cockpit design and other display technologies could use a radical overhaul. The work, reported in the next issue of Philosophical Transaction A, a journal of the Royal Society, challenges the previous scientific consensus and indicates that changing displays so they flicker, use one colour and contain more objects will better stimulate visual reactions in pilots than conventional multi-coloured outline displays.
The retina in the human eye can register literally thousands of pieces of information simultaneously. But the way this data is processed by the brain to pick out important features in our environment and allow us to react is a complex process. It has consequences for the design of many critical machine-man interfaces from fighter pilot cockpits to vehicle dashboards and virtual gaming environments.
Cambridge University’s Greg Davis, who is author of the paper entitled ‘Characteristics of attention and visual short-term memory: implications for visual interface design’, says that our “visual brain” has evolved to prioritise relevant features in a scene and, thus, ignore irrelevant ones. But this selectivity can lead to problems in information-rich environments, such as in cars and aircraft. Accidents occur when our vision cannot cope with the abundance of switches and commands.
A ‘heads-up’ for display designers
“These failures of vision are not simply a laboratory curio: understanding them should help to provide a new generation of tailored visual interfaces [that] can minimise human operator error,” writes Davis. But to reduce the incidence of these events, a large amount of research in psychophysics and ergonomics had to be carried out to examine how display characteristics impact our ability to process multiple features at the same time.
Davis explains that most of the paper is aimed at a general cognitive neuroscience audience and provides an outline of – and evidence for – a model of human limited-capacity vision. The final section is primarily intended for an engineering audience, where he outlines the potential implications of the model for cockpit display development, in particular “head-up or helmet-mounted displays in which display objects are superimposed on the pilot’s view of the external environment”.
The experimental psychologist found that information sources can be more efficiently processed when they belong to separate objects. “Contrary to conclusions from previous work, reducing the number of objects in a display appears to have no general effect on perceptual performance. Indeed, in many cases, this adversely affects performance,” says Davis. There are two separate pathways that process different visual information. The better known one is the parvocellular-ventral processing stream which deals with ‘within-object’ relationships between shape, colour and textural features of the same object. The latest findings show that ‘between-object’ information – relationships between features from separate, neighbouring objects – are processed via a second pathway, the magnocellular-dorsal processing stream.
“All previous experimental approaches of this type, including some of our own, have only fully stimulated the parvocellular-ventral pathway, while ignoring or under-stimulating the magnocellular-dorsal process,” says Davis. By revisiting the experimental design to stimulate equally the two pathways, Davis and his team found that multi-object displays often result in better performance.
Research Contacts page
Royal SocietyPress release (11 October 2004)