Motor Learning of Compatible and Incompatible Visuomotor Maps Scott T. Grafton1, Joanna Salidis2, and Daniel B. Willingham2 Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/13/2/217/1759041/089892901564270.pdf by guest on 18 May 2021 Abstract & Brain imaging studies demonstrate increasing activity in occurring with respect to target location. Positron emission limb motor areas during early motor skill learning, consistent tomography (PET) functional imaging of compatible learning with functional reorganization occurring at the motor output identified increasing activity throughout the precentral gyrus, level. Nevertheless, behavioral studies reveal that visually maximal in the arm area. Incompatible learning also led to guided skills can also be learned with respect to target location increasing activity in the precentral gyrus, maximal in the or possibly eye movements. The current experiments exam- putative frontal eye fields. When the incompatible task was ined motor learning under compatible and incompatible switched to a compatible response and the previously learned perceptual/motor conditions to identify brain areas involved sequence was reintroduced, there was an increase in arm in different perceptual±motor transformations. Subjects motor cortex. The results show that learning-related increases tracked a continuously moving target with a joystick-controlled of brain activity are dynamic, with recruitment of multiple cursor. The target moved in a repeating sequence embedded motor output areas, contingent on task demands. Visually within random movements to block sequence awareness. guided motor sequences can be linked to either oculomotor or Psychophysical studies of behavioral transfer from incompa- arm motor areas. Rather than identifying changes of motor tible (joystick and cursor moving in opposite directions) to output maps, the data from imaging experiments may better compatible tracking established that incompatible learning was reflect modulation of inputs to multiple motor areas. & INTRODUCTION mentary motor area. With additional practice on a Humans can acquire new skills in a matter of minutes to second day there is evidence for a further increase of hours. With sufficient practice, motor skills such as activity in contralateral putamen (Grafton, Woods, & typing or sports are retained lifelong. This ease is Tyszka, 1994). This basic observation of enhanced activ- remarkable given the complexity involved in the timing ity in the motor circuit with practice has been observed and generation of new muscle synergies. A persistent with many other procedural and implicit sequence question for functional mapping is where a new motor learning tasks when studied in the initial period of skill skill is represented in the brain. Converging evidence learning (Hazeltine, Grafton, & Ivry, 1997; Jueptner et al., from invasive and noninvasive brain mapping techniques 1997; Shadmehr & Holcomb, 1997; Doyon, Owen, Pet- have begun to identify a set of sensory and motor areas rides, Sziklas, & Evans, 1996; Flament, Ellermann, Kim, where learning-related changes might occur. Ugurbil,&Ebner,1996;Karnietal.,1995,1998;Rauchetal., Previous imaging studies of procedural motor learn- 1995; Jenkins, Brooks, Nixon, Frackowiak, & Passing- ing from our laboratory examined the pursuit rotor task ham, 1994; Schlaug, Knorr, & Seitz, 1994; Seitz et al., (Grafton et al., 1992). This is a prototypical procedural 1994; Grafton et al., 1992, 1994; Grafton, Hazeltine, & task in which subjects hold a stylus and chase a small Ivry, 1995; Haier et al., 1992; Roland, Gulyas, & Seitz, target positioned on the outer edge of a rapidly rotating 1991; Lang et al., 1988; Saint-Cyr, Taylor, & Lang, 1988). disc (Ammons, 1947). Over the course of 20±40 min the Complementary methods, such as transcranial magnetic time on target, a measure of accuracy and learning, stimulation, have identified analogous learning-related increases dramatically. The skill can be retained over increases of the relevant digit representations in human many years (Ammons et al., 1958). As subjects learn this motor cortex during implicit sequence learning (Clas- skill, imaging studies identify progressive increases of sen, Liepert, Wise, Hallett, & Cohen, 1998; Pascual- activity in contralateral motor cortex and the supple- Leone, Grafman, & Hallett, 1994). Interpretation of motor learning-related changes of brain activity requires a thorough identification of what 1 Emory University School of Medicine, 2 University of Virginia skill or knowledge is actually acquired by a subject D 2001 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 13:2, pp. 217±231 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892901564270 by guest on 02 October 2021 (Willingham, 1998, 1999). For procedural tasks such as RESULTS the pursuit rotor task, subjects could be learning any Experiment 1a: Psychophysics of Spatially combination of the following factors: (1) spatial location Compatible Procedural Learning of the target (2); eye movements; (3) specific sequential arm movements; and (4) an internal model of abstract This experiment validated a procedure that leads to motor goals such as limb direction or gestures. There is implicit motor skill learning under spatially compatible behavioral evidence that in the pursuit rotor task more conditions with respect to limb, target, and cursor than one of these factors may contribute to performance movement directions. The task, a modification of the improvements (Adams, 1987). For example, eye tracking pursuit tracking task originally developed by Pew (1974), is more kinematically constrained than other learning without arm movements can lead to subsequent perfor- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/13/2/217/1759041/089892901564270.pdf by guest on 18 May 2021 mance enhancement, although it is of lesser magnitude procedures such as mirror writing or pursuit rotor. than performance gains achieved by limb movement. Unlike pursuit rotor, there is no explicit representation Although procedural learning tasks such as pursuit rotor of the spatial sequence that could define movement. are typically described as implicit tasks, it is clear that From observations made during pilot experiments with rate of learning can be strongly modulated by conscious videotaping it is apparent that subjects consistently strategies (Rawlings & Rawlings, 1974; Rosenquist, follow the moving target with the eyes. Thus, there is 1965). For example, in the pursuit rotor task the target a general congruency between eye movement and target moves in a circle. Subjects could use the mental repre- movement in this task. sentation of ``circle'' to generate an internal model of Subjects tracked a target that moved across the screen the task requirements, reducing the need for visual in the horizontal direction only. There were two kinds of guidance. This effect might recruit frontal areas known trials, sequence and random. In random trials, the target to be involved in motor imagery such as the SMA or reversed direction at randomly generated screen posi- parietal cortex. Thus, the effects of mental rehearsal, tions. In sequence trials, the target reversed at random imagery, and strategy may complicate interpretation of positions for the first last and middle 10 sec of each trial; functional mapping results. during the remaining 60 sec the target turned at points The goal of the present set of experiments was to determined by a sequence. Figure 1A shows the hor- behaviorally isolate and then identify functional ana- izontal position of the target through the 90-sec trial and tomic substrates for some of these factors involved in Figure 1B shows a repeating 10-sec movement pattern. the acquisition of new visuomotor skills. We present In each trial throughout the experiment, the same 10- two experiments. In Experiment 1a we validate a continuous visually guided tracking task that can be used to assess procedural learning. The task is a major improvement over the pursuit rotor task for control- ling variability of kinematics, precision, and mental strategies. The psychophysical data establish that learning is sequence specific and implicit. In Experi- ment 1b we use this task with brain imaging to examine the functional anatomy of implicit procedural learning under spatially compatible conditions. The results show an increase of activity predominately in the limb motor cortex, in agreement with previous imaging studies of procedural learning. In Experiment 2a this tracking task is validated behaviorally under a spatially incompatible condition. The technique of behavioral transfer was then used. After learning a sequence of movements, subjects switched to a com- patible tracking condition. Based on performance im- provements after the sequence was reintroduced, the knowledge associated with this task is shown to be linked to target locations. In Experiment 2b the func- tional anatomy of implicit procedural learning under spatially incompatible conditions, followed by beha- vioral transfer, is examined. In the incompatible case, Figure 1. Schematic of the motor learning paradigm used in learning-related increases develop predominately in Experiment 1. Subjects tracked a target that moved horizontally on a computer screen for 90 sec (A). A repeating pattern was intermixed oculomotor areas rather than limb areas. With transfer with brief random movements present in the initial, middle, and final to the compatible condition