Opponent, or ratio, models of movement direction perception propose th
at pairs of analysers, sensitive to opposite directions, are linked in
an opponent fashion. Alternatively, distribution models posit indepen
dence of movement direction analysers. To investigate analyser indepen
dence, a direction-selective adaptation experiment was conducted in wh
ich motion coherence thresholds for random-dot apparent motion stimuli
were measured for upward, downward, leftward and rightward motion wit
h and without prior adaptation to rightward motion. Coherence threshol
ds for rightward motion were elevated on average by 0.64 log units aft
er adaptation. Thresholds for motion in the other directions remained
unchanged. Since opponent models predict enhanced sensitivity after ad
aptation for the direction opposite that of adaptation, these results
suggest that movement direction analysers operate independently. In a
second experiment subjects adapted to a bidirectional stimulus contain
ing an equal number of leftward and rightward moving dots. Leftward an
d rightward coherence thresholds were elevated even though no motion a
ftereffects were produced. A third experiment in which coherence thres
holds for test directions within a narrow range were measured with and
without prior adaptation to motion in the direction of the range's me
an showed that threshold elevation was maximal when test and adaptatio
n direction were matched and fell off as the difference in adaptation
and test direction increased. The resulting functions indicate that th
e bandwidth of movement direction analysers is between +/- 35 and +/-
40 deg, a value consistent with reported mean directional tuning funct
ions of movement sensitive units in the middle temporal area but small
er than previously reported psychophysical values.