Hierarchical decision-making produces persistent differences in learning performance.

Area of Science:

• Organizational Behavior
• Cognitive Science
• Computational Social Science

Background:

• Human organizations often feature hierarchical structures for task division and coordinated effort.
• Strategic decisions are typically made by higher levels, guiding operational execution by lower levels, often with differing timescales and information reliance.

Purpose of the Study:

• To investigate the impact of hierarchically distributed decision-making on joint learning dynamics.
• To test the hypothesis that such distribution leads to heterogeneous and path-dependent learning patterns.

Main Methods:

• Laboratory experiments involving human dyads performing repeated joint tasks.
• Assigning distinct roles: one individual for strategy decisions, the other for operational execution.
• Developing a computational model to simulate experimental conditions and predict performance heterogeneity.

Main Results:

• Observed a bimodal performance distribution in experimental dyads: some pairs demonstrated learning, while others failed to learn.
• The computational model successfully mirrored experimental settings and predicted performance heterogeneity.
• Self-reinforcing dynamics initiated by early choices were identified as a key factor in observed performance differences.

Conclusions:

• Hierarchical distribution of strategy and operational decisions can lead to significant heterogeneity in joint learning.
• Initial choices and their self-reinforcing dynamics are sufficient to explain the divergent learning trajectories observed in human organizations.
• Understanding these dynamics is crucial for optimizing organizational learning and performance.