Statistical learning subserves a higher purpose: Novelty detection in an information foraging system.

Area of Science:

• Cognitive Science
• Neuroscience
• Psycholinguistics

Background:

• Statistical learning (SL) is traditionally viewed as a passive mechanism for detecting environmental regularities.
• This perspective assumes a stable environment and passive learners, particularly influential in language acquisition.

Purpose of the Study:

• To challenge the traditional view of SL in stable environments.
• To propose an alternative cognitive architecture integrating SL within an information foraging (IF) system.
• To explore the dynamic nature of learning environments and active learners.

Main Methods:

• Review of existing evidence on SL and environmental dynamics.
• Theoretical re-framing of SL within an IF framework.
• Discussion of implications for cognitive neuroscience.

Main Results:

• Environments are dynamic and learners are active, not passive.
• SL is a subcomponent of a broader IF system.
• IF systems detect novel patterns deviating from randomness, with SL providing a baseline.

Conclusions:

• A shift from passive SL to active IF is necessary to understand learning in dynamic environments.
• This revised framework has significant implications for cognitive neuroscience.
• The proposed model better accounts for the interplay between learners and their changing surroundings.