Decision prioritization and causal reasoning in decision hierarchies

Area of Science:

• Cognitive Science
• Decision Making
• Human Computer Interaction

Background:

• Real-life decisions often involve complex hierarchies of sub-decisions.
• Planning over vast decision spaces is crucial for efficient problem-solving.
• Understanding how humans navigate decision hierarchies is key to cognitive research.

Purpose of the Study:

• To investigate human decision-making strategies in a novel hierarchical task.
• To explore the interplay of perceptual decision making, active sensing, and reasoning.
• To identify computational heuristics underlying efficient planning in complex tasks.

Main Methods:

• Developed a novel task combining perceptual decision making, active sensing, and hierarchical reasoning.
• Participants navigated a decision tree to find a hidden target, gathering noisy evidence.
• Computational modeling was used to analyze human behavior and identify planning strategies.

Main Results:

• Despite task complexity (107 latent states), participants demonstrated efficient planning.
• Human behavior was best explained by low-complexity heuristics, not probabilistic inference.
• Key heuristics included categorical decisions, discarding unreliable evidence, and using confidence for error correction.

Conclusions:

• Humans employ effective heuristic planning strategies for decision hierarchies, integrating sensing.
• Intermediate complexity tasks are valuable for uncovering rules of human hierarchical reasoning.
• Identified heuristics provide insights into human decision-making efficiency and adaptability.