Structure learning in action

Abstract

‘Learning to learn’ phenomena have been widely investigated in cognition, perception and more recently also in action. During concept learning tasks, for example, it has been suggested that characteristic features are abstracted from a set of examples with the consequence that learning of similar tasks is facilitated—a process termed ‘learning to learn’. From a computational point of view such an extraction of invariants can be regarded as learning of an underlying structure. Here we review the evidence for structure learning as a ‘learning to learn’ mechanism, especially in sensorimotor control where the motor system has to adapt to variable environments. We review studies demonstrating that common features of variable environments are extracted during sensorimotor learning and exploited for efficient adaptation in novel tasks. We conclude that structure learning plays a fundamental role in skill learning and may underlie the unsurpassed flexibility and adaptability of the motor system.

Minimally invasive surgery training using multiple port sites to improve performance

“…Our results show, however, that performance at the novel port site was best after training using multiple port sites. This result is predicted by structural learning theory, which states that motor task variation can allow the central nervous system to learn general rules about how task parameters co-vary [2].…”

“…Minimally invasive surgery (MIS) is challenging because the brain must control complex movements using extremely limited sensory information obtained from a rapidly changing environment. Recent advances in psychology, neuroscience and machine learning have started to explain the amazing capacity that humans show for learning to move within sparse environments [1][2][3]. Here we demonstrate how our theoretical understanding of motor learning makes predictions that can be usefully exploited to inform surgical training and safety.…”

“…While the beneficial properties of motor task variation have been known for some time [10,17], structural learning was only recently proposed as a mechanism by which "learning to learn" could occur in the motor system [2]. Several studies have now demonstrated the ability of the central nervous system to learn task structures which facilitates the acquisition of new but similar skills [3, 7-9, 18, 19].…”

“…If surgeons are able to learn general rules about how port properties vary this may alleviate the negative consequences associated with port switching. Experimental findings show that training regimes in which task parameters are randomly or gradually varied provide support for the extraction of structural rules [2,[7][8][9]. Even when the structural rules are not extracted, motor task variation can improve future performance through other mechanisms [10][11][12][13].…”

“…While a novice driver may find it difficult to switch from the car they have always driven (and that they used to pass their test), an experienced driver who has driven many vehicles adapts to a new car quickly and easily. This phenomenon can be explained by structural learning theory, which suggests that experience can provide the human brain with exposure to a wide variety of conditions so that the underlying structure of the task can be learnt [2]. A structure is a set of rules which describes how task parameters co-vary (i.e.…”