How variability shapes learning and generalization. Issue 6 (June 2022)

• Record Type: Journal Article
• Authors: Raviv, Limor; Lupyan, Gary; Green, Shawn C.
• Is Part Of: Trends in cognitive sciences. Volume 26:Issue 6(2022)
• Page Start: 462
• Page End: 483
• Publication Date: 2022-06
• Subjects: variability -- diversity -- learning -- generalization -- categorization -- language; Cognitive science -- Periodicals; Cognitive neuroscience -- Periodicals; 153.05
• DOI: 10.1016/j.tics.2022.03.007
• Languages: English
• ISSNs: 1364-6613

Abstract

Abstract : Learning is using past experiences to inform new behaviors and actions. Because all experiences are unique, learning always requires some generalization. An effective way of improving generalization is to expose learners to more variable (and thus often more representative) input. More variability tends to make initial learning more challenging, but eventually leads to more general and robust performance. This core principle has been repeatedly rediscovered and renamed in different domains (e.g., contextual diversity, desirable difficulties, variability of practice). Reviewing this basic result as it has been formulated in different domains allows us to identify key patterns, distinguish between different kinds of variability, discuss the roles of varying task-relevant versus irrelevant dimensions, and examine the effects of introducing variability at different points in training. Highlights: For the past 80 years, the relationship between variability, learning, and generalization has been studied in various domains, including motor learning, categorization, visual perception, language acquisition, and machine learning. Learning from less variable input is often fast, but may fail to generalize to new stimuli; learning with more variable input is initially slower, but typically yields better generalization. This basic observation has been repeatedly reformulated under different names in different fields, but with little synthesis of similarities and differences nor recognition of different types of variability. We highlight the complementary insights made in different domains on the role of variability in learning and integrate these insights to better understand what kinds of variability matter, when do they matter, and why.