Daniel T. Willingham - US grants

A commonly-appreciated characteristic of motor skill learning is that much
of it occurs outside of awareness. Particularly at high levels of skill,
we seem not to be able to articulate the exact movements that make up
performance. On the other hand, conscious strategies are clearly important
in motor skill learning; consciously understood strategies are at the
heart of coaching and instruction.

A considerable amount of research has been directed at demonstrating that the
conscious (or explicit) and unconscious (or implicit) parts of motor skill
learning are supported by different structures in the brain; thus the
emphasis has been on the separability of these two components of learning.
Less work has been directed at understanding how these two forms of
knowledge work together. The goal of this project is to examine the
relationship of implicit and explicit knowledge in motor skill learning.
The experiments are motivated by a model of motor skill learning that
proposes specific mechanisms by which both implicit and explicit knowledge
contribute to motor skill learning. There are three aims of the project:

Aim 1: What is the representation underlying sequence learning? Many motor
skills involve learning a sequence (e.g., the notes of a piano sonata).
When one learns a motor sequence, what is it a sequence of? Stimulus
positions? Response locations? Muscle movements? I will utilize transfer
designs to determine the representation underlying implicit and explicit
sequence learning.

Aim 2: What is the time course of implicit and explicit learning? Some
formal models of motor skill learning suggest that implicit knowledge is
compiled from explicit representations. Behavioral experiments suggest
that explicit and implicit knowledge can be acquired in parallel, whereas
brain imaging work suggests that only one type of representation can be
acquired at one time. This project will address this question directly.

Aim 3: What is the mechanism by which explicit and implicit knowledge
interact? This project examines the possibility that explicit knowledge
can support performance on an implicit task if three conditions are met:
the subject knows that he or she has applicable explicit knowledge,
sufficient attentional resources are available, and the explicit
representation is specific in terms of the spatial targets for movement.

These experiments will further our knowledge about the normal operation of
motor skill learning, and have particular relevance to athletics, and to
skill learning in industrial situations.

DESCRIPTION (provided by applicant): The study of skill learning has, like many other areas of memory research, been influenced by the implicit/explicit distinction. Current thinking is that implicit and explicit processes are separate, but both can contribute to skilled behavior. The implicit/explicit distinction has been useful, but it has perhaps led to an overemphasis on the separability of implicit and explicit processes. This proposal addresses the question: Do implicit and explicit processes interact in skill learning? The interaction has been hard to address because implicit learning must be assessed through performance, but explicit learning can affect performance too. In the past researchers tried to ensure that no explicit learning occurred, so that implicit learning would not be "corrupted" by the contribution of explicit processes. We have developed techniques that allow the assessment of implicit learning in isolation, even if explicit learning has taken place. We can assess implicit learning in isolation for a motor skill (sequence learning) and a cognitive skill (probabilistic categorization). Thus, we can train subjects explicitly and later examine how this training affected implicit learning. The specific questions about system interaction are inspired by four principles of system interaction in the rat. (1) There are anatomically separate systems in amygdala, hippocampus, and striatum supporting different types of learning; (2) These different types of learning occur in parallel; (3) These different systems learn different types of information; (4) These different systems can either cooperate or compete, perhaps because each system learns information that potentiates complimentary or conflicting behaviors OR because of direct anatomic connections. There is excellent evidence for separate systems in humans, but limited evidence on the other three points. The motivation of this proposal is to test system interactions using the findings from rats as a starting point. Our specific aims are to test (1) whether implicit and explicit learning occur in parallel; (2) whether implicit and explicit learning use different representations; (3) whether cooperation/competition among memory systems depends on the compatibility of what each system learns. This work has potentially important implications for rehabilitation (e.g., from stroke). Both motor loss and cognitive loss is often treated via explicit training in the hopes that the explicit training will influence implicit processes.