We studied the accuracy of human subjects in perceiving the direction of self-motion from optic flow, over a range of directions contained in a 45 deg cone whose vertex was at the viewpoint. Translational optic flow fields were generated by displaying brief sequences (< 1.0 sec) of randomly positioned dots expanding in a radial fashion. Subjects were asked to indicate the direction of perceived self-motion at the end of the display. The data were analyzed by factoring out the constant component of the error by means of a linear regression analysis performed on the azimuthal and elevational components of the settings. The analysis of the variable error revealed that: a) the variance of the settings is 3-45% greater along elevation than azimuth for five observers; b) azimuth and elevation correspond, on average, to the principal components of the error in the settings; c) there are differences in the variances of azimuthal and elevational errors between upper and lower visual fields. Moreover, the distribution of the errors for azimuth and elevation in the upper and lower hemifields is not the same. All of the above evidence supports the hypothesis that heading information is represented centrally in terms of its azimuthal and elevational components.
D’Avossa, G., & Kersten, D. (1996). Evidence in human subjects for independent coding of azimuth and elevation for direction of heading from optic flow. Vision Research, 36(18), 2915–2924. https://doi.org/10.1016/0042-6989(96)00010-7