Abstract
Accurate information about three-dimensional (3D) motion is essential for interception. Being able to detect changes in the speed of motion is potentially important, as approaching objects are unlikely to maintain constant velocity either by intent, or because of the force of gravity or friction. However, evidence from the interception literature shows that acceleration is not taken into account when judging time-to-contact from looming (i.e. retinal expansion). These data may reflect a curious insensitivity to 3D acceleration, a possibility that has received little empirical attention. As a first step towards a better understanding of this apparent lack of sensitivity, we assessed discrimination thresholds for 3D velocity changes. Observers viewed animations of an approaching object undergoing an increase (acceleration) or decrease (deceleration) in its simulated approach speed over the trial. The stimulus was a thin outline disk that was viewed monocularly, such that looming was the only available cue to motion in depth. On each trial, observers discriminated acceleration sign. We measured psychometric functions for three interleaved average speeds. To discourage observers from using non-relevant cues (e.g. due to regularities in the stimulus and correlations between variables) we randomized the simulated starting and ending distance. Our results show that observers were able to detect acceleration in depth, but their thresholds were very high (about a 25-33% velocity change). While precision did not depend on average velocity, there was a velocity-dependent bias: observers were more likely to report that the object accelerated for higher average approach speeds and vice versa. Thus, observers were sensitive to the acceleration of an approaching object under minimal cue conditions, but they could not completely dissociate speed and acceleration. We will discuss which signals could support monocular discrimination of 3D acceleration and produce the bias we found. Furthermore, we will extend these experiments to consider stereoscopic 3D acceleration.
Meeting abstract presented at VSS 2013