Purchase this article with an account.
Kwan J. Lee, Norberto M. Grzywacz; Failure of decomposition of translation and expansion/rotation in optic-flow perception. Journal of Vision 2008;8(6):29. doi: 10.1167/8.6.29.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The natural optic flow is typically complex, containing components from both ego-and object-motion in the world. Some theoretical studies proposed that the brain might decompose the optic flow locally into fundamental components, such as translation, rotation, and expansion (Koenderink, 1976; Yuille and Grzywacz, 1998). Recent psychophysical results seem to support these theoretical frameworks for rotation and expansion (Barraza and Grzywacz, 2005). In this study, we extended the psychophysical probes, using random-dot fields that expand/rotate while translating simultaneously. If such fields were viewed instantaneously, the resultant flow field would be another perfect expansion/rotation with its center shifted to the new point of singularity. Thus, one instantaneous frame of the velocity field is insufficient for one to decompose the flow field correctly into its radial/rotary and translational components. However, if one views the optic flow over an extended period, correct decomposition is possible in principle because the motion of the singularity over time could disambiguate the translation. In this investigation, we tested psychophysically whether human subjects can perform the correct decomposition over time. Subjects were requested to perform 2AFC tasks to compare the location of the perceived center to a reference. The perceived center of the combined field was obtained from 50% threshold. Our results show that humans cannot correctly decompose the flow field into its components even with prolonged stimulus presentation. Rather, humans perceive the shifted center of each instantaneous velocity field and perceive a concatenation of these centers to give a sensation of translation. Surprisingly, even adding boundaries to the moving field, which should have thus disambiguated the translation, did not correct the misperception of the center. We conclude that, contrary to the theoretical frameworks mentioned above, humans cannot decompose translation and rotation/expansion from the combined field in the case of random dots.
This PDF is available to Subscribers Only