Academic journal article Perception and Psychophysics

On the Minimization of Task Switch Costs Following Long-Term Training

Academic journal article Perception and Psychophysics

On the Minimization of Task Switch Costs Following Long-Term Training

Article excerpt

Flexible, context-dependent linkages between stimulus and response are fundamental to adaptive behavior. In the present article, we evaluate the limits of this flexibility by exploring the asymptotic efficiency of people's ability to switch between two different sensorimotor mappings. Two stimulus-response (S-R) mappings were learned, either both on the same hand (unimanual condition) or one mapping per hand (intermanual condition). The S-R mappings presented bivalent stimuli and employed the same response keys. A novel training regimen successfully reduced task-switch costs to approximately 20 msec, suggesting that residual switch costs cannot be eliminated. These costs cannot be entirely attributed to the cognitive control process of task-set reconfiguration, because they are observed over cued switch intervals of several seconds. Two additional issues in motor learning were addressed: the single or dual loci of manual motor control and the coordinate system of task representation. First, the results favored the notion of independent controllers for each hand instead of a single dominant controller, since intermanual performance was superior to unimanual performance. Second, a transfer task tested internal (egocentric) and external (allocentric) coordinate systems. Transfer was more effective using the external coordinate system, suggesting that the S-R mappings reflected the association between the bivalent stimuli and external goals (i.e., the individual keys) rather than the concrete pattern of muscle contractions (i.e., the finger pressing the key). Finally, retention tests revealed that these learned S-R associations were remarkably durable, since no decrements in fluent task-switching performance were observed after 10 months without practice.

A hallmark of human behavior is the ability to respond appropriately to dynamic changes in the environment. For example, a skilled pianist makes seamless transitions between key signatures, whereas a musician of lesser ability persists in playing the notes corresponding to the previous scale (von Kries, 1895, cited in Woodworth & Schlosberg, 1954). From the perspective of human information processing, the performance of highly skilled pianists suggests that, with sufficient practice, one stimulus- response (S-R) mapping can be replaced with another without any decrement in performance. However, beginning with the seminal studies by Jersild (1927), in investigations of performance on different tasks in succession-task switching-interference effects have been consistently observed whenever switches occur (reviewed in Monsell, 2003). One challenge in developing a theoretical account of task switching is to simultaneously account for the ability of people to exercise executive control operations to endogenously drive a task switch while explaining the performance costs when people implement the task switch.

Switch costs arise from multiple possible sources. One influential view of task switching is to consider the human operator as a finite-state machine. Accordingly, a control signal determines the state of the operator, and each discrete state selects the set of task-appropriate S-R connections (e.g., de Jong, 1995; Meiran, 1996, 2000a, 2000b; Rubinstein, Meyer, & Evans, 2001). In such a finite state, only the selected task is represented, and there is no trace of an unselected task. Thus, there should be no intertask interference once the system is selected and instantiated. Switch costs could only result in the interval between a new control signal and the subsequent transition to that state. As pointed out by Allport and Wylie (2000), this view suggests that switch costs may occur immediately following a task switch but not once the system is appropriately reconfigured. This view of task switching predicts that all switch costs-even for the trial immediately following a task switch-should be completely eliminated if the appropriate task settings are configured before a response is required. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.