Object affordances tune observers' prior expectations about tool-use behaviors

Area of Science:

• Cognitive Science
• Neuroscience
• Human Motor Control

Background:

• Observational learning of tool use relies on past experience and perceptual cues called affordances.
• Affordances provide biomechanical priors, guiding potential motor actions for tool use.
• The interaction between biomechanical priors and higher-order expectations from experience in observational learning is not fully understood.

Purpose of the Study:

• To investigate how biomechanical priors (affordances) interact with prior experience (probabilistic expectations) in observational tool-use learning.
• To determine the influence of visual uncertainty on inferring demonstrator intentions during tool use.

Main Methods:

• An action observation task was designed where participants inferred demonstrator intentions under visual uncertainty.
• The study manipulated the probability of observing specific tool functions and the biomechanical optimality of movements.
• Participants' predictions were measured based on varying levels of probabilistic and biomechanical information.

Main Results:

• Biomechanical priors significantly modulate the influence of probabilistic expectations on participants' predictions.
• When biomechanical and probabilistic priors align (converge), they have a cumulative effect on learning.
• When biomechanical and probabilistic priors conflict (diverge), they exhibit a mutually inhibitory effect.

Conclusions:

• Both biomechanical priors and probabilistic expectations are crucial for observational tool-use learning.
• The interplay between affordances and experience is complex, with effects depending on their congruence.
• Understanding this interaction provides insights into the mechanisms of human motor learning and adaptation.