The asymmetry system properties of the visual system: Evidence from serial face processing

Area of Science:

• Cognitive Neuroscience
• Computational Vision
• Psychophysics

Background:

• Visual perception is a continuous process, but its system properties remain incompletely understood.
• Visual serial dependence, where current perception is biased by prior stimuli, suggests potential Markov-like processing.
• Investigating serial dependence offers insights into how the visual system integrates past and present information.

Purpose of the Study:

• To investigate the system properties of visual serial dependence.
• To determine if visual processing adheres to Markovian principles (memoryless property).
• To examine how prior input (face) and prior output (perception) influence current judgments of facial traits.

Main Methods:

• Participants rated faces on attractiveness, trustworthiness, and dominance in randomized sequences.
• Experimental design manipulated whether consecutive ratings involved the same or different traits.
• Statistical analyses included derivative of Gaussian, Markov chain, and linear mixed-effects modeling.

Main Results:

• Serial dependence was disrupted when alternating between attractiveness and dominance traits.
• Both memoryless and Markov assumptions were violated under these specific alternating conditions.
• No disruption or violation of assumptions was observed under other trait-alternating conditions.

Conclusions:

• Different aspects of visual computation can exhibit asymmetrical system properties.
• Serial dependence is a valuable tool for probing the rules of visual information integration.
• The visual system's integration of past and present information is more complex than simple Markovian models suggest.