Cause-effect perception in an object place task.

Area of Science:

• Cognitive Science
• Human-Computer Interaction
• Virtual Reality

Background:

• Understanding human causal learning is crucial for developing intelligent systems.
• Investigating causal discovery in realistic, sensorimotor contexts enhances ecological validity.
• Few studies explore the dissociation between conscious and sensorimotor representations of causality.

Purpose of the Study:

• To investigate human causal discovery in a realistic virtual reality (VR) environment.
• To examine how causal learning is represented at conscious-cognitive and sensorimotor levels.
• To compare human causal learning with state-of-the-art causal discovery algorithms.

Main Methods:

• Exploratory study using VR with haptic feedback for realistic simulation.
• Participants performed a task requiring discovery of causal links (weight/color to breakability).
• Causal discovery algorithms (PC, FCI, FGES) analyzed sensorimotor data; questionnaires assessed conscious understanding.

Main Results:

• Participants learned the weight-breakability link (76%) and color-breakability link (43%) but struggled with causal direction.
• Sensorimotor analysis showed increasing weight-force coupling and weak, noisy color-force coupling.
• Causal discovery algorithms accurately recovered the underlying causal structure across sessions.

Conclusions:

• Humans can perceive causal structures in complex sensorimotor tasks to some extent.
• Conscious and sensorimotor representations of causality appear partially dissociated.
• VR provides a valuable tool for studying human causal learning and its computational underpinnings.