Evidence of a role for such mappings demonstrates the importance of scene and image statistics to the processes underpinning cue integration and the perception of motion-in-depth. These results suggest that cue integration is affected by both the uncertainty of sensory signals and the mapping of those signals to real-world properties. Models that took these factors into account predicted human performance in speed-in-depth and cue conflict discrimination tasks, where standard linear integration models could not.
We show that motion-in-depth cue integration depends upon sensitivity to the joint probabilities of the scene parameters determining these signals, and on the probability of CD and CS signals co-occurring. The interpretation and integration of these cues depends upon multiple scene parameters, such as distance moved, object size and viewing distance, posing a significant computational challenge.
For humans, important visual cues for motion-in-depth include changing disparity (CD) and changing image size (CS). Motion-in-depth perception is critical in enabling animals to avoid hazards and respond to potential threats.
