From JOHN WANN
It is pleasing to see you devoting space to perceptual theory, but the
interpretation you place on a recent study (This Week, 18 June) is in itself
dangerous. The study by Doug Stewart et al., of ‘driver misjudgments’ was
a useful research contribution, but you failed to mention that the ‘computer
simulation’ was merely of white bars moving on a black computer screen.
In this context, many of the critical depth cues are absent. Linear
perspective, texture gradients and the driver’s head motion all provide
very salient information about relative depth and size, and stereo-vision
has been proposed as a means of absolute depth scaling.
All these features are present for the car driver, but were absent
in this experiment. The research really demonstrates that, if most of the
natural perceptual information is absent, then drivers will use relative
size and hence misjudge the distance of a child.
In a natural setting, however, this may only happen for a one-eyed motorist,
with a very stiff neck, driving in total darkness with no road markings.
In this respect, the Department of Transport’s view of the research as ‘interesting’
but not confirmatory is correct. What is required is similar research with
visual environments where the driver has a realistic 3D visual scene that
responds to head motion, and where the display responds to the driver’s
progressive braking action (a feature also missing from the previous experiment).
Fortunately, virtual reality systems can now provide that facility; all
that is required is the DoT, insurance companies or other agencies to provide
the £100 000 or so that would enable such research.
Advertisement
John Wann University of Edinburgh
