Inferring latent learning factors in large-scale cognitive training data

Area of Science:

• Cognitive science
• Psychology
• Data science

Background:

• Human cognition is characterized by the ability to learn diverse tasks.
• Understanding individual differences in learning dynamics is crucial.
• Previous research often assumes specific learning curve forms or task relationships.

Purpose of the Study:

• To analyze the latent structure of learning trajectories across numerous individuals and tasks.
• To investigate covariation across learning trajectories with minimal assumptions.
• To identify underlying factors influencing learning dynamics.

Main Methods:

• Probabilistic dimensionality reduction modeling was applied to learning data from 36,297 individuals.
• Analysis of learning trajectories across 51 different cognitive tasks.
• Data-driven approach to uncover latent structures without pre-defined assumptions on learning curves or task relationships.

Main Results:

• Substantial covariation was found across learning trajectories, enabling prediction of unobserved learning.
• A general ability factor, prominent in later practice stages, was identified.
• Task-specific factors were discovered, correlating with defined task features and domains (e.g., attention, spatial processing, language, math).

Conclusions:

• Learning trajectories across diverse tasks are not independent and exhibit predictable patterns.
• A hierarchical structure of learning exists, comprising general and specific abilities.
• The findings provide insights into individual differences in cognitive learning and task-specific skill acquisition.