Inter-trial priming of speed discrimination within modalities:
vision vs audition
We investigated whether there were any inter-trial dependencies in our
stimulus sequences. Even though our stimulus presentations were brief, a
given trial in a sequence is often primed by the preceding one such that
current percept shows a bias towards the previous one. This is a priming
or ‘positive serial dependence’ effect that occurs for a wide range of
stimuli41-44 including motion45. We
conducted a sequential dependency analysis which involved binning all
trials into two groups based on whether the speed on the preceding trial
was slow (20, 40 & 50°/s) or fast (70, 80 & 100°/s). For each group,
we then calculated the best-fitting psychometric functions and compared
the PSEs. If the previous trial has no influence on the current one,
then all trials would be independent and the psychometric functions
calculated from the ‘previous slow’ and ‘previous fast’ groups should
not differ in PSE.
Figure 3 shows the results of our serial dependence analysis. For both
audition and vision, rates of ‘proportion faster’ responses were higher
when the previous trial was fast, compared to when the previous trial
was slow (Fig. 3A). This pattern is consistent with motion priming, a
positive (or ‘attractive’) serial dependence effect often seen when
preceding motion stimuli are brief46-48 and contrasts
with motion adaptation – a negative (or ‘repulsive’) serial dependence
that requires longer adaptation periods and produces the classic motion
aftereffect49,50. A bootstrap sign-test confirmed that
PSEs were significantly lower (leftward shift of the psychometric
function) after ‘previous fast’ trials for both vision and audition
(vision: p < .0001; audition: p < .0001),
establishing that the previous trial was sufficient to prime the current
trial and that vision and audition show the same effect.