When people with upper-extremity amputations were interviewed at least 1 year after amputation, they indicated a strong desire for functional prosthetic training, leading to the recommendation that training should be provided to such persons as soon as possible after the prosthesis was provided.1 This idea received subsequent support when a survey by Durance and O'Shea2 of people with upper-extremity amputations indicated that 90% of the persons who received training were able to functionally use their prosthetic device at the time of the survey compared with only 50% of those who did not receive training. Together, these findings suggest that prosthetic training influences the overall acceptance and subsequent functional use of the prosthesis. Even though quite some time has passed since these studies were conducted, controlled empirical studies comparing various means of organizing practice for learning to use an upper-extremity prosthesis are scarce, despite the recent infusion of motor learning principles into other rehabilitation environments.3
Only a few studies have attempted to apply principles derived from motor learning research to determine effective training methods for instruction in the use of an upperextremity prosthesis.4,5 A contemporary perspective in motor learning research suggests that practice should be designed to promote thorough processing of task-relevant information, forcing the learner to “solve” the motor problem associated with the task. This perspective has placed emphasis on the cognitive effort that the learner expends as a result of the practice schedule, such that practice conditions that increase cognitive effort result in greater retention of skill over time and the capability to transfer learning to new performance contexts.6 This is in contrast to practice conditions that may temporarily benefit acquisition performance by minimizing cognitive effort but that often degrade retention and transfer performance.
A motor learning principle that is well-known to heighten cognitive processing in early practice is the principle of variability of practice. In essence, training programs that reduce the variability of experiences by having the learner repetitively perform multiple trials of the same task under the same context may minimize cognitive effort. On the other hand, a practice context that is structured to increase the number of task variations experienced during a practice session enhances cognitive processing.
One way to further enhance cognitive effort within a practice session is to structure variability of practice in such a way that it elicits a learning phenomenon referred to as contextual interference (CI). This learning phenomenon, first introduced by Battig7 in 1979, is related to the order in which multiple tasks or variations of a task are performed during practice. Battig's original notion was that the difficulty of the practice context affects the strength and flexibility of the memory representations for the tasks being learned. Consistent with the notion of cognitive effort, Battig proposed that structuring practice to encourage interference among task variations forced the learner to process the tasks more deeply and more elaborately, leading to memory representations that were more easily retrieved and more resistant to forgetting.
Thus, the practice context can be arranged to minimize or maximize interference among tasks. A low CI condition is created when one task is practiced for repeated trials and is uninterrupted by practice of any other tasks. This type of practice schedule, typically referred to as a blocked practice schedule, allows the learner to concentrate on refining one particular task before moving on to another task. In contrast, under conditions of high CI, tasks are experienced in a randomized sequence within the practice session. The learner rotates practice on the tasks in such a way that the same task may only rarely be experienced on two consecutive trials. This is generally referred to as a random practice schedule.
The typical finding in the literature is that blocked practice leads to faster improvements and superior performance at the end of the acquisition phase. Paradoxically, however, random practice conditions produce significantly better performance during retention and transfer tests of learning.8 In other words, there is a reversal in performance superiority from acquisition to retention and transfer. This pattern of results has been shown for the learning of laboratory tasks9-13 and for learning various sports skills.14-18 Thus, structuring the practice of multiple tasks to induce high CI can promote motor skill learning.
Although there is remarkable consistency in the motor learning literature favoring high CI, the effect has not been examined for activities of daily living that might be practiced in a therapeutic environment. Given the limited time that a therapist typically has with a patient, as well as the limited number of tasks and task variations that can be experienced in a given rehabilitation setting, practice sessions that promote retention of the gains that are made with the therapist and that facilitate transfer to other contexts away from the clinic are highly desirable. Understanding practice contexts that promote motor learning clearly has important implications for encouraging persons with amputations to accept an upper-extremity prosthesis. The purpose of the present study was to examine the acquisition, retention, and transfer of selected prehension skills under conditions of low and high CI. To have sufficient statistical power to determine the most effective practice context, the training was given to ablebodied participants who wore a simulator that mimicked an actual upper-extremity prosthesis in form and function (e.g., an adequately-sized population of recent upper extremity amputees was not available for a study of this nature). We have used the simulator successfully in previous research to gain insights into the unique challenges associated with learning to use an upper-extremity prosthesis.4
Participants were randomly assigned to either a blocked acquisition group or a random acquisition group. Both groups practiced three different tasks across 2 days of acquisition. Within a day, the blocked group completed all practice attempts on a particular task before engaging in practice on the next task. The random group practiced the three tasks interchangeably, such that the task to be performed on the ensuing attempt was not predictable.
To determine the practice schedule that elicited the greatest degree of motor learning, a retention test and an intertask transfer test were administered 24 hours after the second day of acquisition. In the retention test, participants performed the tasks practiced in the acquisition phase. These tests were conducted to determine the relatively permanent changes in skilled behavior that resulted from practice. In the intertask transfer test, participants performed three new tasks that had not been practiced but shared some common elements with the acquisition tasks. It was hypothesized that: 1) the random acquisition practice schedule would lead to superior retention performance and intertask transfer than the blocked practice schedule, and 2) no significant gender differences would be found.
MATERIALS AND METHODS
Forty-eight college-aged, right-handed persons (24 men and 24 women) were recruited as participants. All participants were screened for eligibility. The inclusion criteria consisted of: 1) no acute medical incidents within the previous six months; 2) no uncorrected vision or hearing difficulties; 3) no evidence of excessive postural deviation as assessed by Kendall's postural screen19; and 4) freedom from neurological or upper-extremity musculoskeletal problems that might influence performance. After participants signed an institutionally- approved informed consent form and were randomly assigned to one of two acquisition groups (12 men and 12 women per group): a random acquisition group or a blocked acquisition group.
The prosthetic simulator was designed to mimic a prosthesis for a below-elbow amputation with considerable residual limb length that allowed forearm supination/pronation (Figure 1). It used a figure-9 harness fitted around the shoulder contralateral to the prosthesis. The harness was attached to a cable that ran across the back and upper arm ipsilateral to the simulator and into a leather humeral cuff on the proximal portion of the simulator. From the humeral cuff, the cable ran the length of the simulator to interface with a Grip3 voluntary-closing prehensor identical to a regular terminal device (model STG300; TRS Inc., Boulder, CO). With the simulator donned and placed atop a table in a position that allowed the anterior aspect of the forearm to be in contact with the tabletop, the terminal device (TD) was oriented in the horizontal plane with the tips of the TD parallel to the tabletop. Participants opened the TD by adjusting with relative motions of the torso, shoulders, and arm the tension of the cable that was fit with the simulator. A lightweight aluminum frame was fabricated to represent the forearm portion of the prosthesis. To don the simulator, the participant first placed the contralateral limb through the harness and then slipped the forearm into a stretchable nylon glove mounted within the aluminum frame, placing the hand just proximal to the TD. Placement of the hand in the glove disallowed motion of the wrist to contribute to simulator movement. Thus, the simulator was designed to resemble, as closely as possible, an actual prosthesis in both fitting and function.
A task board was fabricated to allow performance of a variety of prehension tasks. The task board was secured to a table at which participants were seated. All tasks started and finished from a common start/stop button located at midline and 15 cm in front of the body.
TASKS AND PROCEDURES
Before being seated at the task table, participants were provided with assistance donning the prosthesis. Once donned, harness tension was adjusted until the distance between the tips of the TD prehensors was 6 cm with the simulator held at the participant's side in the anatomical position. While standing, participants viewed a model on videotape wearing the simulator. The model demonstrated several basic control motions used to operate the cable to open and close the TD. The demonstrated motions included humeral flexion combined with an elbow extension motion, a bilateral shoulder shrug motion, and an elbow flexion motion with the upper arm internally rotated and fixated. Participants passively watched the model perform the motions one time, followed by watching the model a second time while concurrently imitating the motions. Further practice of the control motions was disallowed.
Participants then were seated at the task table and instructed that they were not allowed to lift themselves from the chair, but could lean forward or side-to-side to perform the tasks with the simulator. The height of the chair was adjusted so that the top surface of the task board was at waist height.
Each participant performed each task. Before the first trial of a given task, the experimenter verbally explained the task. To initiate a trial, participants were cued to depress the start/stop button with the prosthesis. The experimenter then indicated the task to be performed. After a 3-second period, the experimenter said, “Ready.” After a random period of 1 to 4 seconds, participants were prompted to initiate the trial with an auditory tone. The intertrial interval was approximately 10 seconds.
Although not cued to initiate the task as soon as possible, participants were instructed to perform as rapidly as possible while maintaining accuracy. Initiation time (time from presentation of the auditory tone until release of the start/stop button) and movement time (time from release of the button until return to the button to terminate the trial) were recorded via two clocks (accurate to the millisecond) interfaced with the start/stop button. Initiation time was an indicator of response-planning capability for performing a particular task. Movement time reflected the temporal summation of several task-related factors, including opening the TD, grasping the task object, manipulating the object, releasing the object, and returning the prosthesis to the start/stop position. A low movement time indicated that each of these factors was performed more effectively.
In the acquisition phase, 15 trials on each of three different prehension tasks were performed on 2 consecutive days, for a total of 90 acquisition trials. Order of practice on the three tasks within a day was either blocked or random. One of the acquisition tasks was derived from the Upper Extremity Function Test.20,21 The other two tasks were designed based on pilot testing. The tasks were representative of a range of upper-limb functional prehension tasks that involve manipulation of objects at a variety of locations in the workspace.
In the Pipe Transfer Task (derived from the Upper Extremity Function Test), a steel pipe (7.5 cm high and 2 cm in diameter) was placed over a wooden dowel (2.5 cm high and 1.5 cm in diameter) located at the midline on a shelf elevated 25 cm above the table surface. The participant made a 40-cm forward reach with the prosthesis to grasp the pipe, removed it from the dowel, and transported the pipe to place it over a second dowel mounted in the table surface at a distance 20 cm from the edge of the table. After release of the pipe from the TD, the participant returned to depress the start/stop button with the prosthesis.
The Rod and Nut Task involved removal of a 2-cm diameter nut from a “question-mark”-shaped unthreaded rod mounted vertically in a block of wood (15 × 10 × 8 cm). The nut was large enough to be slipped over the rod without making contact. The end of the curved rod faced medially. The block was located at the midline, 15 cm from the edge of the table. The task was bilateral in that the anatomical hand stabilized the block on the table surface while the TD slipped the nut off the rod. After removing the nut, the participant maintained the nut in the TD and returned to depress the start/stop button with the prosthesis.
In the Electrical Plug Insert Task, the participant initiated the task with a three-prong electrical plug grasped by the prosthesis. On the cue to move, the participant used the anatomical hand to grasp a three-prong outlet anchored to the shelf, while using the prosthesis to simultaneously transport the plug and place it in the outlet. The edges of the plug and outlet were required to meet for the task to be performed correctly. After inserting the plug, the participant released it and returned to depress the start/stop button with the prosthesis.
To determine the degree of learning as a result of the acquisition experiences, all participants returned 24 hours after the second day of acquisition to perform two tests of learning: a retention test and an intertask transfer test, administered in that order. In the 24-hour retention test, each of the acquisition groups was randomly divided into halves, with half of each group performing the retention tests under its acquisition training schedule while the other half performed under the opposite acquisition training schedule. This was done to determine the acquisition condition that best prepared participants for the opposite performance context. Five trials of each task were performed in the retention tests.
In the 24-hour intertask transfer test, participants performed five trials of three new tasks, with one of the tasks derived from the Jebsen Test of Hand Function.22 The three transfer tasks were chosen as analogs to the three tasks practiced in acquisition. These tasks were performed in the same manner indicated for the retention test.
In the Simulated Feeding Task (derived from the Jebsen Test of Hand Function), a kidney bean was placed on the task board at midline and against a 5-cm upright board, which was 48 cm in length and secured parallel to and 15 cm from the front edge of the table. The trial was initiated with a regular teaspoon in the TD. The participant manipulated the bean onto the teaspoon and transported it to a can 7.5 cm high and 5 cm in diameter that was held at midline by the anatomical hand. The bean was dropped from the spoon into the can, after which the participant returned to depress the start/stop button with the prosthesis. This task was the analog to the Rod and Nut task because of the bilateral nature of the task and the use of pronation and supination motions to operate the simulator.
In the Sphere Transfer Task, a spherical wooden object, 5.5 cm in diameter, was stationed on the wooden dowel located at midline on the task board surface (see Figure 1). From the start button, the participant reached forward to grasp the sphere with the TD, removed it from the dowel, and transported it to place it over the second dowel mounted on the shelf above the task board surface. After release of the sphere, the participant returned to depress the start/stop button with the prosthesis. This task was the analog to the Pipe Transfer task because it required grasp and manipulation of a curved object and the ability to use fine aiming skill to manipulate objects onto a target.
In the Padlock Opening Task, the participant initiated a trial with a padlock key grasped by the TD. On the cue to move, the participant used the anatomical hand to grasp a padlock attached via an eyebolt to the shelf above the task board, while simultaneously transporting the key with the prosthesis to the padlock to insert it in the lock. After doing so, the participant opened the padlock, released the key, and returned to depress the start/stop button with the prosthesis. This task was the analog to the Electrical Plug Insert task because it was bilateral in nature and involved fine manipulation skill to insert an object into a target.
Initiation time (IT) and movement time (MT) were measured per trial in milliseconds. To compare the influence of acquisition schedules on performance of the different tasks within the same analysis, scores for individual trials were converted to a standard scale of measurement, the z-scale. Conversion of raw score ITs and MTs to a standard scale was necessitated by the different time bases for IT and MT for each task. The z-scale has a mean equal to zero and a standard deviation of 1, such that raw scores are expressed in standard deviation units from the mean. Thus, z-scores are analogous to effect sizes. Negative z-values indicate a raw score value that fell below the mean of the distribution. Because IT and MT are time-scales, a positive z-score indicates a raw score time that was longer than the mean IT or MT, whereas a negative z-value indicates a raw score time that was shorter than the mean IT or MT. Trials were then grouped into blocks of five, with a mean z-score calculated for each trial block. Analysis of the z-scores allowed exploration of the group by task interaction in each analysis. In all analyses, a type I error rate of p ≤ .05 was used. When the Mauchly's test indicated a violation of the assumption of sphericity for repeated measures factors and associated interactions, the degrees of freedom were adjusted based on the lower-bound estimate. Reported F-ratios that used the adjusted degrees of freedom are denoted by an asterisk after the F. Significant main effects were explored with multiple comparisons; significant interactions were explored with simple main effects tests. The Bonferroni correction was used in all post hoc tests.
In separate analyses for IT and MT, the z-scores were subjected to group by gender by task by day by block (2 × 2 × 3 × 2 × 3) analyses of variance (ANOVAs) with repeated measures on task, day, and block.
Retention Test Analyses
In separate analyses for IT and MT in retention, the z-scores were subjected to a group (blocked-blocked, blocked-random, random-random, random-blocked) by gender by task (4 × 2 × 3) ANOVA with repeated measures on task.
Intertask Transfer Analyses
In separate analyses for IT and MT, the intertask transfer test z-scores were subjected to a group (blocked-blocked, blockedrandom, random-random, random-blocked) by gender by task (4 × 2 × 3) ANOVA with repeated measures on task.
INITIATION TIME IN ACQUISITION
The day main effect, F(1,44) = 119.41 (p < .001), block main effect, F*(1,44) = 105.05 (p < .001), and the day by block interaction, F*(1,44) = 65.33 (p < .001), indicated a significant degree of improvement across blocks on days 1 and 2 of acquisition. The lack of interactions with the task factor indicated a similar degree of improvement across all tasks. Thus, regardless of the practice schedule, both groups significantly improved their ability to respond as a function of acquisition practice.
Significant differences between the groups were detected among blocks across days in the day by block by group interaction, F*(1,44) = 15.65 (p < .001) (Figure 2). Simple main effect tests indicated that the blocked acquisition group responded significantly more slowly in block 1 of day 2 than the random group. This indicated that the improvement in performance resulting from acquisition day 1 was sustained for the random group; however, gains in response speed dissipated after day 1 for the blocked group. No other block means differed among groups in acquisition. The gender main effect was also significant, F(1,44) = 4.49 (p = .04); the mean IT for women was significantly longer than for men [mean z-scores (SE) = 0.311 (0.205) and −0.313 (0.203) for women and men, respectively].
INITIATION TIME IN RETENTION
No significant group differences were detected in retention (p = .08). Thus, with respect to speed of responding, both acquisition conditions provided sufficient stimulus to promote retention of the criterion tasks. A significant gender main effect was detected in retention, F(1,40) = 4.35 (p = .043); the mean IT for women was significantly longer than for the men [mean z-scores (SE) = −0.323 (.215) and −0.961 (.203) for women and men, respectively]. The failure to detect task or group by task effects in retention indicated that relative performance was statistically equated across tasks.
INITIATION TIME IN INTERTASK TRANSFER
A significant group main effect was detected, F(3,40) = 3.23 (p = .032). Multiple comparison tests indicated that the random-blocked group responded significantly more quickly than the blocked-random group. Although other group differences were not detected, initiation time means plotted in Figure 2 indicate a general trend for the groups who experienced random practice in acquisition to respond more rapidly in novel performance contexts than the groups who experienced blocked practice in acquisition.
A significant gender main effect was detected, F(1,40) = 5.09 (p = .030). The mean IT z-scores (SE) were 0.461 (0.27) and -0.460 (0.29) for women and men, respectively. Relative performance across the three novel tasks was equated within groups because of the absence of task or group by task effects.
MOVEMENT TIME IN ACQUISITION
The day main effect, F(1,44) = 125.25 (p < .001), block main effect, F*(1,44) = 102.47 (p < .001), and the day by block interaction, F*(1,44) = 38.80 (p < .001), indicated a significant degree of improvement in speed of performance across blocks on days 1 and 2 of acquisition. The lack of interactions with the task factor indicated this degree of improvement to occur similarly across all tasks. Thus, regardless of the practice schedule, both groups significantly improved their ability to perform the tasks rapidly as a function of acquisition practice.
Significant differences between the groups were detected among blocks across days in the day by block by group interaction, F*(1,44) = 5.38 (p = .025) (Figure 3). Simple main effect tests indicated that the blocked acquisition group moved significantly more slowly in block 3 of day 1 than the random acquisition group. This indicated that the random acquisition condition did not hamper acquisition performance, as has been reported in previous literature. No other block means differed among groups in acquisition.
The gender main effect was also significant, F(1,44) = 4.90 (p = .032). The mean MT for women was significantly longer than for men [mean z-scores (SE) = 0.305 (0.190) and -0.306 (0.198) for women and men, respectively].
MOVEMENT TIME IN RETENTION
No significant group differences were detected in retention (p = .22). Thus, with respect to speed of movement, both acquisition conditions engendered movement proficiency among the groups on the criterion tasks. Although not significant, it indicates that the greatest loss in movement proficiency was by the group transferred from blocked to random conditions (Figure 3). A significant gender main effect was detected in retention, F(1,40) = 6.511 (p = .015). The mean MT z-scores (SE) were -0.477 (0.15) and -0.928 (0.10) for women and men, respectively. The absence of task or group by task effects in retention indicated that relative performance was statistically equated across tasks.
MOVEMENT TIME IN INTERTASK TRANSFER
A significant group main effect was detected, F(3,40) = 6.40 (p = .001). Multiple comparison tests indicated that the random-random group moved significantly more rapidly than the blocked-random and blocked-blocked groups. In addition, the random-blocked group moved significantly more rapidly than the blocked-random group. Movement time means plotted on Figure 3 thus indicated that random practice in acquisition resulted in an advantage in movement proficiency in novel performance contexts compared with blocked practice.
A significant gender main effect was detected, F(1,40) = 28.97 (p < .001). In addition, the group by gender interaction was significant, F(3,40) = 3.04 (p = .04). Simple main effect tests indicated that the women in the blocked-blocked and blocked-random groups performed significantly more slowly than the men in these groups. Thus, blocked practice was more detrimental to the proficiency of women when faced with novel tasks to perform. Relative performance across the three novel tasks was equated within groups because of the absence of task or group by task effects.
The purpose of this study was to examine the effects of CI on learning to use a simulator that mimicked an upper extremity prosthesis. CI can be manipulated by adjusting the order in which multiple tasks are practiced within a given session. When the practice order is nonrepetitive, as in a random schedule, the potential for interference among the various tasks is heightened. By contrast, CI is minimized by blocked practice, where all trials are completed on one task before practice on the subsequent task commences. When learning is assessed at a later time (in retention and transfer tests), a learning paradox arises in which participants who practiced under a random acquisition context outperform those who practiced under a blocked practice context. The traditional explanation for this paradox is that participants are able to remember the tasks more effectively after random practice because they have been forced to process the tasks in a more effortful and thorough manner, leading to more robust and flexible memory representations.
The CI effect has been reported many times in the literature for able-bodied persons.8 The present study provided evidence that a high CI practice schedule could also be used successfully for training persons to use an upper-extremity prosthesis to perform functional prehension tasks. Acquisition results indicated that blocked and random conditions promoted improvement in a relatively short period of practice. Thus, structured practice, regardless of degree of CI inherent in the schedule, was useful for developing functional use of the prosthetic device.
However, tests of learning indicated important intergroup differences. In particular, evidence was provided that the initial learning context was shown to influence performance in intertask transfer. Blocked practice may have forced participants to automate their responses such that a rather rigid strategy for movement was established. Although this rigid strategy supported retention performance, its strength was detrimental to performing novel skills. This is supported by noting that the group who transferred from blocked to random conditions in intertask transfer had the greatest loss in speed of responding (IT) and movement proficiency (MT) when performing the new tasks. In contrast, random practice in acquisition seems to have resulted in a more generalizable capability to move proficiently, as evidenced by the superior performance on the novel skills. Stated otherwise, learning was less dependent on the original acquisition context after random practice compared with the more repetitive and less effortful blocked practice. Thus, heightening CI in acquisition enhanced transfer of skill beyond the original practice session.
The previous finding is relevant to the goals of rehabilitation because random practice mimics real-world contexts in at least two important ways: First, as in a random schedule, real-world skills are typically performed interchangeably. Second, the exact tasks practiced in the clinic are rarely performed in real-world contexts. Therefore, random practice seems ideally suited to promoting learning in the contexts that are most likely to be experienced by the patient.
The lack of a significant difference between blocked and random groups in retention indicated that both practice schedules were effective for promoting skill retention when the tasks performed were the same tasks practiced in acquisition. The lack of a CI effect in retention has been reported previously.23,24 For example, Albaret and Thon23 suggested that highly complex tasks could engage a learner practicing under blocked conditions in sufficient problem-solving effort to obscure the traditionally reported benefit of random over blocked practice. One suggestion, then, is that learning to use the prosthetic simulator may have resulted in sufficient task complexity and effortful problem solving (regardless of the practice schedule) to attenuate the CI effect. A second factor that might help to explain the lack of group differences in retention is the nature of the blocked schedule used in the present experiment. In the majority of CI studies, all of the practice is provided on a single day. As such, participants in the blocked group complete all of their practice trials on a given task before moving on to practice the next task and subsequently do not have the opportunity to “revisit” the tasks before retention and transfer. However, in the present study, participants did have an opportunity on the second day to revisit the tasks practiced on the previous day. Thus, the second day of practice perhaps generated enough interference (or sufficient opportunity to retrieve task relevant information from memory) to facilitate retention performance despite the blocked nature of practice within each day. The superiority of the random group for IT on the first block of the second day of practice lends some support to this contention.
Regardless of the reasons for the lack of group differences in retention, there are important practical reasons for considering random practice a better alternative to blocked practice in the clinical setting. For example, it is rare that the identical tasks practiced in therapy will be performed in the world outside the clinic. In addition, few daily activities are performed in a blocked manner; therefore, the blocked practice schedule is least similar to the performance context one would typically encounter. With that point in mind, it is important to note that the group who practiced under blocked conditions, and subsequently performed tests of learning under a random schedule, demonstrated the least amount of retention compared with the other groups. Therefore, it is suggested that to obtain the greatest amount of long-term retention, an person learning to use an upperextremity prosthesis should practice tasks in a therapeutic setting using a random schedule.
An unexpected result was the significant gender differences found in retention for MT and in intertask transfer for IT and MT. It was consistently found that the male participants exhibited significantly faster initiation and movement times in the tests of learning. This may indicate a general tendency to move more rapidly because of the greater amount of contractile force typically generated by men. As for IT, gender differences surfaced only when novel tasks had to be performed in the intertask transfer tests. However, when the tasks performed were well practiced, as in retention, no difference in movement planning time was detected. In a rehabilitation setting, this finding suggests that female clients may need more practice trials than men to reach a criterion level of proficiency with a prosthesis.
Evidence exists that the immediacy with which a person with a recent amputation begins to practice with the prosthetic device is paramount in the overall outcome for using the prosthesis functionally.1,25 The period in which inflammation of the stump subsides coincides with the period in which the prosthesis is most readily accepted. In turn, training with a prosthetic simulator during this time may help promote acceptance of the soon-to-be-fitted prosthetic device. In an earlier study, we showed that training with a simulator on one arm can lead to positive skill transfer to the contralateral arm.4 Thus, it is suggested that a skill training program can consist of two stages: initially, functional skills can be practiced with a simulator on the intact limb before the actual prosthesis is available. The second stage coincides with the provision of the prosthesis. At this time, the wearer can engage in practicing with the actual prosthesis under a practice regimen in which CI is heightened. This two-stage training program would take advantage of two well-founded motor learning principles: bilateral transfer and variability of practice.
Although we suggest a two-stage rehabilitation program after amputation, it must be kept in mind that the investigations on which this idea was derived used simulated amputees. Therefore, an assumption is made that transfer of motor learning principles from the sample population to an amputee population will occur. Further investigations with persons who have sustained recent amputations are necessary to empirically validate the two-stage rehabilitation program.
In summary, this study has provided initial insight into a question that has not been addressed in the rehabilitation literature—how can practice be organized to facilitate the learning of control over an upper-extremity prosthesis? Because it is impossible for an individual with a recently acquired prosthesis to practice all of the tasks necessary for daily living in therapy, the early practice context for learning to use the prosthesis functionally should focus on a few representative tasks that are practiced in a random order. In addition to permitting gains in performance and allowing such gains to be maintained over time, such a schedule has the advantage of promoting performance on novel task variations. Because the client may not grasp the logic in performing various skills under practice conditions that promote interference, it is important for the rehabilitation specialist to educate the client about the expectations for learning derived from practicing under a difficult acquisition context.