Abstract Detail


Talk Sessions

Poster Sessions

All Presentations


The temporal efficiency function of the energy-based and feature tracking motion systems

21.15, Saturday, 16-May, 8:15 am - 9:45 am, Talk Room 1
Session: Motion Perception

Remy Allard1,2,3, Angelo Arleo1,2,3; 1INSERM, U968, Paris, F-75012, France, 2Sorbonne Universit├ęs, UPMC Univ Paris 06, UMR_S 968, Institut de la Vision, Paris, F-75012, France, 3CNRS, UMR_7210, Paris, F-75012, France

The low-level, energy-based motion system and the high-level, feature tracking motion system have different temporal characteristics: the sensitivity function of the energy-based system is bandpass peaking around 10 Hz and the feature tracking system is lowpass with a cut-off frequency around 3 Hz. For the energy-based motion system, the greater sensitivity to middle frequencies could be due to less internal noise or more efficient processing (i.e., requiring lower signal-to-noise ratios) at these temporal frequencies. Equivalently, for the feature tracking motion system, the sensitivity drop with temporal frequency could be due to an increase in internal noise or a decrease in processing efficiency. To investigate which underlying factor is responsible for the shape of the temporal sensitivity function for each motion system, an external noise paradigm was used to decompose the sensitivity into internal equivalent noise and calculation efficiency over a wide range of temporal frequencies. To examine the processing of the energy-based system, observers were asked to discriminate the rotating direction of a sine wave grating, which provided strong local luminance drifting cues. For the feature tracking system, the same stimulus was used except that the phase of the sine wave grating was randomized at each frame so the stimulus was drift-balanced. Results showed that the greater sensitivity of energy-based processing to middle frequencies was due to less internal noise and the calculation efficiency remained relatively constant over a wide range of temporal frequencies. Conversely, the sensitivity drop of the feature tracking system with temporal frequency was due to both a reduction in the calculation efficiency and an increase in internal equivalent noise. The shape of the temporal sensitivity function therefore reveals fundamentally different properties for the two motion systems: internal noise variation for the energy-based system and both internal noise and processing efficiency for the feature tracking system.

< Back