Abstract
Perception of current stimuli can be affected by recently seen stimuli, known as serial dependence. In studies of self-motion (heading) perception, previous work shows perceived direction is biased towards the center of the display. Here we examined serial dependence in heading perception after removing center bias. In Experiment 1, the display (80° x 80°) simulated observer translation (3 m/s) in a 3D random-dot cloud (depth range: 0.565-2.0 m) consisting of 200 dots. On each trial, heading direction was randomly chosen (±32°, ±16°, ±8°, ±4°, ±2° or 0°) and presented for 0.5 s. Participants indicated perceived heading with a mouse-controlled probe. Experiment 2 varied signal-to-noise ratio by replacing 0%, 25%, 50%, or 75% dots with random-motion vectors. Experiment 3 increased dot number to 500 to increase motion signal strength. To evaluate center bias, we performed a linear regression between perceived and actual heading. To evaluate serial dependence, after subtracting the center-bias, we performed another linear regression between the heading bias (the residual difference between perceived and predicted heading on the current trial) and relative heading offset (difference between perceived heading of the previous trial and actual heading of current trial). We found: (1) a repulsive serial dependence in heading perception from optic flow; (2) increased center bias and serial dependence effects with decreasing signal-to-noise ratio; (3) both effects decreased with the increased number of motion signals in the flow field. Our study is the first to evaluate serial dependence in heading perception from optic flow and finds a negative (repulsive) effect. We show that signal strength in the flow field affects heading judgments, with lower signal strength increasing center bias as well as the reliance on judgments in previous trials.
Meeting abstract presented at VSS 2018