Interaction of first- and second-order direction in motion-defined motion.

Area of Science:

• Visual perception
• Motion processing
• Computational neuroscience

Background:

• Motion-defined objects require specialized processing beyond simple nonlinearities.
• Understanding the link between object-surface motion and emergent object motion is crucial.
• Previous research indicated some cooperativity between first- and second-order motion information.

Purpose of the Study:

• To investigate how motion direction is estimated for motion-defined objects.
• To analyze the integration of first- and second-order motion cues in direction perception.
• To explore the underlying mechanisms of motion processing networks.

Main Methods:

• Human observers reported perceived motion direction of objects defined by luminance contrast or random-dot kinematograms.
• Stimuli varied in the spatiotemporal properties of object region versus background.
• Conditions included coherent object motion (Fourier motion) and static or diagonal dot motion within the object (drift-balanced and theta motion).

Main Results:

• Direction sensitivity for Fourier motion matched luminance-defined objects.
• Performance decreased with static dots (drift-balanced motion).
• Accuracy further declined in theta motion, with perceived direction being an intermediate blend of object and dot motion, indicating velocity vector pooling.

Conclusions:

• First- and second-order directional information for motion-defined objects are not separable.
• This suggests motion processing subsystems do not produce independent percepts.
• Findings support a two-layer motion-processing network model where velocity vectors are pooled.