┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ 装 ┊ ┊ ┊ ┊ ┊ 订 ┊ ┊ ┊ ┊ ┊ 线 ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊ ┊

ment at the steering-wheel rim. It was assumed that the variable with the greater correlation (i.e. either force or torque, or angle or displacement) is the most efficient engineering term to represent the data.

Over the 12 subjects, for isometric control, the correlation coefficients obtained for force were significantly higher than those obtained for torque (p<0.01, Wilcoxon matched-pairs signed-ranks test) Por isotonic control, the correlation coefficients obtained for angle were significantly higher than those obtained for displacement (p<0.01) 3.1.3 Discussion

Lines of best fit to the data had gradients of less than unity for 11 subjects. The single subject that achieved a slope greater than 1.0 did so only for angle data. The effect could have arisen from the reference being presented first (i.e. an order effect) Alternatively, it could indicate that the physical variables do not reflect the parameters adjusted by is described in terms of a'Weber fraction' or percentage. Weber proposed that the absolute difference threshold is a linear function of stimulus intensity and can therefore be described as a constant per- centage, or fraction, of the stimulus intensity. This is expressed in Wbet's law Subject rwque Farce the subjects. Regardless of the deviations of references and 'matches' from the 450 line, the Spearman correlations ranked the reference and 'match' data according to magnitude without making any ass umptions about the reset values of the reference and the 'match'.

The results suggest that with idealized isometric and isotonic controls, drivers have a better sense of steering-wheel force than steering-wheel torque and a better sense of steering wheel-angle than steeringwheel displacement. It seems that subjects used the forces in their muscles and the angles at the joints of their hands and arms to position the steeringwheels.

To judge torque, subjects would need to combine estimates of force with knowledge of the distance between their hands and the centre of the steering wheel. To judge the displacement of the steeringwheel rrn， subjects would need to combine estimates of their joint angles with the length of their limbs. The estimation of torque and distance requires

more information and greater processing than the estimation offorce and angle. Consequently,it is not surprising that torque and distance result in less accurate judgements and are not preferred or‘natural. 3.2 Difference thresholds

A difference threshold is the smallest change in a stimulus required to produce a just noticeable difference in seusation 111]. Differencecuts,山resholds can be described in absolute terms, where the threshold is described in the physical units of the variable under test, or in relative terms, where the threshold often expressed as a percentage

Difference thresholds for the perception of force are available in a variety of forms. Jones 1121 reported (7 per cent) for forces generated at the elbow flexor muscles.