There is an implicit assumption in most models of movement direction p
erception that sensitivity to motion is directionally isotropic, i.e.
equal for movement in all directions. However, most previous research
on directional effects on human motion perception have measured respon
ses (e.g. manual reaction times) to suprathreshold stimuli rather than
motion sensitivity. I investigated this possibility by measuring mono
cular sensitivity to motion coherence in small field random-dot kinema
tograms with global leftwards or rightwards motion as a function of ec
centricity along with the horizontal meridian in each eye of six obser
vers. Although foveal motion sensitivity was isotropic, small bu signi
ficant differences (about 0.1 log units) in sensitivity in favour of c
entripetal motion were observed at eccentricities between 5.0 and 12.5
deg. Motion anisotropy was significantly larger in the temporal than
the nasal visual hemifield. In a second experiment, motion coherence t
hresholds for upwards and downwards movement were measured foveally an
d at +/- 5 deg on the vertical meridian. At the fovea, no difference i
n sensitivity to upwards vs downwards motion was observed but these ve
rtical axis thresholds were significantly higher than foveal threshold
s for movement along the horizontal axis. Directional anisotrophy in f
avour of centripetal motion was present in the inferior visual field b
ut directional isotrophy was found for the superior visual field. Perh
aps this generally heightened sensitivity to centripetal directions ma
y function normally to facilitate figure/ground segmentation in the pr
esence of centrifugally-moving background contours that typically acco
mpany forward locomotion.