Refining the law of practice.

Area of Science:

• Cognitive Psychology
• Human Performance
• Skill Acquisition

Background:

• The "law of practice" quantifies skill acquisition speed-up using nonlinear functions relating response time (RT) and practice.
• Previous models, like the power law, were limited by averaging artifacts and struggled with exceptions to steadily decreasing learning rates.

Purpose of the Study:

• To propose and evaluate a novel law of practice that accommodates initial learning delays and models the entire RT distribution.
• To provide a more flexible and accurate model for skill acquisition dynamics.

Main Methods:

• Developed a new nonlinear function for the law of practice, incorporating a delayed learning component.
• Utilized hierarchical Bayesian modeling to analyze data across diverse tasks and participants, minimizing averaging biases.
• Employed inference procedures to assess the flexibility and fit of different learning laws.

Main Results:

• The proposed delayed exponential law provided a superior fit across a majority of experimental paradigms.
• This new model successfully accounted for initial delays in learning and the complete distribution of response times.
• Hierarchical Bayesian modeling confirmed the robustness of the findings by pooling participant data effectively.

Conclusions:

• A delayed exponential law offers a more comprehensive and accurate description of skill acquisition than traditional power or exponential laws.
• The findings highlight the importance of considering initial learning phases and the full RT distribution for understanding performance improvements.
• This research provides a refined tool for quantifying learning dynamics in cognitive and performance domains.