The human temporal delay function

Two major aspects of temporal processing are temporal blur (or smear), which determines sensitivity, and temporal delay, which sets the effective latency of processing. Temporal sensitivity is well-known to improve with eccentricity and follow an inverted U-shaped function with spatial frequency. Much less is known about temporal processing latencies. Here, with data having sub-millisecond resolution, we present the human temporal delay function, which shows how processing latencies change with spatial frequency across the visual field. We developed stimuli that leverage a visual illusion called the Pulfrich effect. While subjects fixated, stimulus-induced interocular delays caused the rotating stimulus to appear tilted top-back or bottom-back relative to the screen. The task was to report the apparent tilt. Like Kelly (1984), we used different stimulus-types to probe processing: (i) a rotating annulus of eight radial Gabors, (ii) a rotating ring textured with bandpass-filtered 1/f noise, with passbands matched to the Gabors, and (iii) a set of Gabors with static envelopes and drifting carriers. (Different stimuli can identify processing (in)efficiencies and support disparate experimental goals.) We measured processing delays across a 4.5-octave spatial frequency range (0.33-8cpd), at eccentricities ranging from 1.0deg to 6.0deg. The speed of rotation (2-12deg/sec) and the stimulus width(s) (20-120arcmin) scaled with eccentricity. Processing delays decrease dramatically with eccentricity, change systematically with spatial frequency, and are consistent across stimulus-types. Further, sensitivity is described by inverted U-shaped functions similar to those characterizing previous work. Unlike response-time measures—which are noisy, often influenced by decision biases, and always impacted by motor contributions—the current paradigm has the advantage that temporal delays manifest as stereo-depth effects, so subjects need not make explicit estimates of any temporal aspect of the stimulus. The current results show how processing delay changes with spatial frequency across the visual field, a fundamental but understudied aspect of visual processing.