Abstract
Residual Latencies (RL) are non-decisional components of response time (Luce, 1986) that can in turn be decomposed into processing and motor latencies (respectively pre- and post-decisions). Thus, when motor response is similar across different tasks, differences in RL can be attributed to differences in processing latency. Previously (Caziot et al., 2014, OSA), we used RL analysis to show that processing times for luminance and stereo signals were nearly identical. Here we extend this result to different levels of luminance contrast and slant from stereo. Observers were asked to distinguish between a dark and a bright disk (contrast task), or sign of slant in a sparse RDS (slant task). Each task had two levels of difficulty (low/high contrast, small/large slant). Observers were asked to answer before a deadline that varied block by block from 200ms to 500ms. We then fitted exponential Speed-Accuracy Tradeoff Functions (SATFs) to the data. The response time at which performance (fraction correct) deviated from chance was the RL. Increased difficulty lowered the slope of the SATFs, as expected for a lower signal-to-noise ratio. We also found that RL was slightly increased for low contrast compared to high contrast. However RL were similar for the two slant conditions, and comparable to the RL in the high contrast condition. The increase in RL at low contrast can be explained by low-level physiological properties of the visual system (i.e. prior to V1). We conclude that (1) in accordance with our prior findings, stereo is not processed more slowly than luminance, including now for slant perception; (2) increased signal strength increases the slope of the SATF and (3) processing latency for tasks that are difficult due to noise during later processing stages (e.g. stereo slant) is, in general, probably not affected by the strength of the signal.
Meeting abstract presented at VSS 2015