Humans decompose tasks by trading off utility and computational cost

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Decision Science

Background:

• Human behavior relies on hierarchical task decomposition for planning.
• Understanding the principles governing task decomposition is crucial for explaining goal-directed behavior.

Purpose of the Study:

• To propose and evaluate a normative framework for task decomposition.
• To investigate how computational cost influences task structuring in human planning.

Main Methods:

• Analysis of 11,117 graph-structured planning tasks.
• A behavioral study involving 806 participants and 30 diverse graphs.

Main Results:

• The proposed framework explains existing task decomposition heuristics.
• Human task decomposition behavior is best predicted by the betweenness centrality heuristic.
• The framework's predictions were distinguishable from alternative normative accounts.

Conclusions:

• Computational cost is a key factor in structuring goal-directed behavior.
• The betweenness centrality heuristic is a computationally justified strategy for task decomposition.
• This work provides a normative basis for understanding human planning strategies.