Motion direction is represented as a bimodal probability distribution in the human visual cortex.

Area of Science:

• Neuroscience
• Computational Vision
• Psychophysics

Background:

• Human motion perception relies on noisy sensory data.
• The visual system may integrate multiple cues for accurate motion direction inference.

Purpose of the Study:

• To investigate if spatial orientation ('streak') signals, alongside velocity, enhance motion direction perception.
• To model and experimentally validate the use of these dual signals in the human visual system.

Main Methods:

• Developed a Bayesian model incorporating velocity and spatial orientation signals.
• Used psychophysics and functional magnetic resonance imaging (fMRI) to test model predictions.
• Applied probabilistic pattern-based analysis to decode motion direction representations from fMRI data.

Main Results:

• Decoding of motion direction from visual cortex activity revealed bimodal probability distributions.
• These distributions' peaks accurately predicted behavioral errors in motion direction judgment.
• Similar bimodality was observed in the distribution of behavioral responses across trials.

Conclusions:

• The human visual system integrates spatial orientation signals with velocity for motion direction estimation.
• Cortical representations of motion direction can combine distinct signal types.
• This suggests a more complex neural basis for low-level visual feature processing than previously thought.