We investigated the interaction between extra-retinal rotation signals and
retinal motion signals in heading perception during pursuit eye movement. F
or limited viewing aperture, the variability in perceived heading strongly
depends on the pattern of motion directions. Heading towards a point outsid
e the aperture generates nearly parallel aperture flow. This results in low
er precision of perceived heading than heading that renders the radial patt
ern of flow visible. We ask if the precision is limited by the pattern of f
low visible on the retina or that on the screen. During fixation, the two p
atterns are identical. They are decoupled during pursuit, since pursuit cha
nges radial flow within the aperture on the screen into nearly parallel flo
w on the retina, and vice versa. The extra-retinal signal is known to reduc
e systematic errors in the direction of pursuit, thus compensating for the
rotational flow during pursuit. We now ask if the extra-retinal signal also
affects the precision of heading percepts. It might if at the spatial inte
gration stage the rotational flow has been subtracted out already. A compen
sation beyond the integration stage, however, cannot undo the change in ret
inal motion directions so that an effect of pursuit on precision cannot be
avoided. We measured the variable and systematic errors in perceived headin
g during fixation and pursuit for a frontal plane approach, while varying d
uration, dot lifetime and aperture size. We found precision is effected by
pursuit as much as predicted from the pattern of retinal flow, while compen
sation is significantly greater than zero. This means that the interaction
between the extra-retinal signal and visual motion signals takes place afte
r spatial integration of local motion signals. Furthermore, compensation in
creased significantly with longer duration (0.5-3.0 s), but not with larger
aperture size (10-50 degrees). A larger aperture size did increase the ecc
entricity of perceived heading. (C) 2001 Elsevier Science Ltd. All rights r
eserved.