Figure 2. Unimodal data showing velocity discrimination of a set of
velocities around a standard velocity with a geometric mean of 54.03
°/s. (A) The data points show mean ‘proportion faster’ for each level of
velocity for all observers pooled into a single super subject and error
bars show 95% confidence intervals produced by bootstrapping the data
10,000 times. The continuous lines show the best-fitting cumulative
Gaussian psychometric function. (B) Parameters from the best-fitting
psychometric functions for audition and vision. The means (left) and
bandwidth (right) are plotted with errors bars showing 95% confidence
intervals from 10,000 bootstraps. The dashed line shows the geometric
mean speed.
Figure 2b compares the PSEs and discrimination thresholds for auditory
and visual motion. The PSE effectively indicates the mean perceived
speed of the stimulus set and for comparison the dashed horizontal line
in 2b shows the geometric mean of the set of speeds used (54.03°/s).
While the arithmetic mean speed is 60°/s, velocity is represented
logarithmically in cortex (Nover, Anderson & DeAngelis, 2005) and thus
the geometric mean is more appropriate. As the confidence intervals
show, the PSE for vision (46.625°/s; CI95 =
[45.196–47.909]) was significantly lower than the mean stimulus
speed and was also significantly lower than the PSE for audition
(53.526°/s; CI95 = [52.133–54.867]). This
represents a reliable underestimation of speed in vision relative to
audition. Moreover, mean perceived speed for auditory motion was
surprisingly veridical and did not differ from the geometric mean speed.
An interesting feature of the unimodal psychometric functions is the
closely matched discrimination thresholds for auditory and visual
motion. Motion discrimination is generally considered poor for
audition40 yet here the speed discrimination
thresholds (i.e., the bandwidth or standard deviation of the
psychometric function) for vision (20.488°/s; CI95 =
[18.501–22.295]) and audition (20.823°/s; CI95 =
[19.174–22.666]) were very similar and not significantly different.