On November 29, 2022 Prof. C. Green Shawn presented several research lines from his lab that address the challenge of producing more generalizable learning gains through behavioral training. The discussion begun with both theoretical and empirical work focused on the characteristics of behavioral training paradigms that most strongly impact whether eventual learning is task-specific or more generalizable. The presenter outlined the need to differentiate between two fundamentally different forms of generalization: (1) generalization as “an immediately enhanced ability to perform new tasks” (i.e., immediate transfer) and (2) generalization as “an enhanced ability to learn to perform new tasks” (i.e., learning to learn).

Taking these functional forms of generalization into account necessitates a shift in analytical approach, as most standard analyses inherently confound these forms of generalization, particularly analyses that rely on aggregate measures of performance such as average accuracy or average reaction time across blocks of trials. Finally, the presenter discussed how many characteristics of behavioral training that tend to strongly produce generalization of learning, such as high degrees of stimulus variability, also tend to slow the learning of those tasks. This can significantly impact participants’ motivation to engage with the training. The talk concluded with techniques, some drawn from game studies, to maintain high motivation even in the face of difficult training.

Prof. C. Shawn Green received his Ph.D. in Brain and Cognitive Sciences from the University of Rochester. He then completed a post-doctoral fellowship at the University of Minnesota focused on machine learning and computer vision before joining the faculty in the Psychology Department at the University of Wisconsin-Madison in 2011. His research program focuses broadly on learning in the perceptual and cognitive domains. In particular, his work seeks to examine the characteristics of behavioral experiences that impact how quickly new tasks are learned, how deeply new tasks are learned, and whether training on one task then generalizes to new tasks. In this, he uses a range of methods from standard basic lab tasks such as those involving black-and-white gratings or complex working memory span tasks to modern forms of entertainment, such as video games and virtual reality. He has published over 80 peer-reviewed papers that together have been cited more than 16,000 times, including in high-ranking journals such as Nature, the Proceedings of the National Academy of Sciences (3x), Current Biology (4x), Trends in Cognitive Sciences, Nature Reviews Neuroscience, Psychological Science (2x), and Neuron. And he has received funding from a variety of sources including the National Eye Institute, the National Institute on Aging, the National Institute of Child Health and Human Development, the National Institute of Mental Health, the National Science Foundation, and the Office of Naval Research.