Time in causal structure learning.

Area of Science:

• Cognitive Psychology
• Causal Inference
• Decision Science

Background:

• Research on causal strength judgments often focuses on the time between events.
• Temporal information, including event order and intervals, plays a role in understanding causality.
• Previous studies have not fully explored temporal cues in multi-variable causal structure induction.

Purpose of the Study:

• To investigate the role of temporal information in causal structure induction with multiple variables.
• To differentiate the impact of event order versus temporal intervals on causal judgments.
• To examine how people integrate evidence from multiple observations and sensitivity to exact event timings.

Main Methods:

• Four experiments were conducted to explore temporal information in causal structure induction.
• Experiment 1 focused on one-shot learning.
• Experiments 2-4 examined the integration of evidence from multiple observations and sensitivity to interval variability and correlation.

Main Results:

• A Bayesian model accurately predicted participants' judgments, ruling out inconsistent causal structures and favoring those with similar intervals.
• Participants demonstrated sensitivity to event order and temporal intervals between causally connected components.
• For the first time, it was shown that interval variability alone can enable accurate causal structure judgments when order cues are ambiguous.

Conclusions:

• Temporal information, encompassing both event order and interval characteristics, is fundamental for causal structure induction.
• Bayesian models can effectively capture human causal judgments based on temporal evidence.
• Interval variability is a significant, previously underappreciated cue for inferring causal structures, especially when event order is uninformative.