The online processing of dynamics

Area of Science:

• Visual perception
• Cognitive neuroscience
• Physics perception

Background:

• Estimating object mass and velocity is crucial for navigating dynamic environments.
• Previous models often treat mass and velocity estimation separately, lacking a unified mechanism.

Purpose of the Study:

• Introduce and validate the Online Processing of Dynamics Model (OPoD) for visual mass estimation.
• Investigate the joint and reciprocal influence of mass and velocity perception.

Main Methods:

• Developed the OPoD model, positing joint encoding of mass and velocity in a shared representation.
• Conducted three experiments to test OPoD's predictions regarding mutual biases between mass and velocity perception.
• Compared OPoD's performance against an adapted noisy Newton model.

Main Results:

• Empirically confirmed that initial mass impressions bias perceived post-collision velocity.
• Demonstrated that velocity biases subsequently influence relative mass judgments.
• OPoD provided superior quantitative fits compared to the noisy Newton model.

Conclusions:

• Mass and velocity are jointly encoded and mutually constraining within a momentum-like representation.
• OPoD offers a mechanistic explanation for integrated physical expectations in dynamic scene perception.
• The model provides an ecologically grounded complement to probabilistic frameworks.