Abstract
The fast, reflexive Ocular Following Response to large field motion (OFR) is thought to be driven by low level motion energy computation. When moving images contain broadband spatial frequencies, localised interactions between them can result in features like edges. Spatial phase invariant motion energy models should remain agnostic to such features. Here, we tested that assumption considering OFR as highly sensitive to contrast and total energy. We sought to create well controlled phase-varied stimuli to probe differences in the eye speed of human volunteers recorded with a video eye tracker over a 250ms task epoch from stimulus onset. We exploited Phase Congruency (PC): a dimensionless measure of the localised alignment of sinusoidal luminance components at different scales. PC is proportional to local luminance energy, normalised by the sum of the local amplitudes of the separate composite frequencies, making it awkward to compute and susceptible to noise. We used dynamic luminance noise textures as stimuli and exploited published PC estimation methods implemented with appropriately adjusted filters and sensitivity parameters (Morrone & Burr, 1988, Proc.Roy.Soc.B 235:221-245; Kovesi, 2000 Psych. Res 64:136-148). We ranked 250 stimulus cases based on PC repeating this to build a test bank of 2-6 octave bandwidth stimuli. Parameterised movies running from low to high phase coherence were used in an OFR task with trials containing motion at 24deg/s preceded by a centralising saccade. For the narrower (2 octave) bandwidth there was no difference in eye traces over the 250ms. Stronger responses for high-PC stimuli emerged after 140ms for images with over 3 octaves of bandwidth, relatively late in a computation with an 80ms latency from onset. The early 60ms is consistent with motion energy computation; phase sensitivity emerges later possibly from a dynamic accumulation of broadband signals necessary for the neural implementation of a separate motion feature sensitive process.
Meeting abstract presented at VSS 2015