Characteristics of the optic flow parsing mechanism for different simulated observer movements

We have proposed (Rushton & Warren, Current Biology, 2005) and provided evidence for (Warren & Rushton, Current Biology, 2009) the existence of a global motion processing mechanism which acts to subtract out optic flow components. We suggest that this flow parsing mechanism helps moving observers to assess scene-relative object movement. Previously we have investigated the relative contributions of global (flow parsing) and local (e.g. motion contrast) mechanisms to perceived object trajectory. Here we assess whether the relative contributions are invariant for different simulated observer movements. The basic optic flow stimuli simulated forwards translation (1.2 m/s), yaw eye rotation (4.3 deg/s), or head/eye roll about the line of sight (23.2°/s). Median onscreen speed was matched at 4.3 deg/s for all three stimuli. Participants fixated at the centre of the display and judged the direction of motion of a laterally displaced probe. We systematically restricted the portion of the flow field available to the observer. In the "local" and "global" conditions respectively, only the flow inside and outside a circle of radius 3 degrees surrounding the probe was presented. Participants indicated the perceived probe trajectory by setting the orientation of an adjustable paddle. The relative tilt effect (difference between perceived and actual trajectory) was measured. The effect in the global condition was attributed to the flow parsing mechanism whereas that seen in the local condition was attributed to motion contrast. Average relative tilt ratios (local/global) were similar across movement types: 0.81 (forwards translation), 0.83 (horizontal rotation), 0.73 (roll). We also found evidence for a strong correlation between ratios for forwards translation and yaw (R^2 = 0.55) but considerably weaker correlations between roll and either of the other two movement types (R^2 <0.11). These data suggest roll motion may involve different processing from translation and yaw movements. Meeting abstract presented at VSS 2013