Training Concurrent Multistep Procedural Tasks.
Both basic and applied research have long traditions of trying to identify effective techniques to train and transfer complex skills (for reviews, see Butterfield, 1989; Cormier & Hagman, 1987; Reder & Klatzky, 1994; Royer, 1979). For example, as early as the turn of the twentieth century, Thorndike proposed his theory of identical elements (Thorndike & Woodworth, 1901), which suggests that training will be beneficial when it reflects the stimulus and response requirements used in the transfer environment. Singley and Anderson (1989) enriched the scope of identical-elements theory by including cognitive processes.
Despite such advances, little research has explored techniques for training people to perform concurrent component steps in rapid succession, although many applied settings require various types of concurrent monitoring and responding (Damos, 1991; Lintern & Wickens, 1991). This project investigates the benefits of using a forward-chaining technique to master such tasks, which seems to be an issue that has not received much empirical attention. We also compare the benefits of the forward-chaining techniques with those acquired with traditional part- and whole-task techniques and a novel hybrid of part- and whole-task training, providing an empirical comparison of techniques that has not yet been done for this class of tasks.
TRAINING CONCURRENT TASKS
Training techniques can be roughly divided into two general categories: part- and whole-task techniques (Stammers, 1982). Part-task techniques are usually procedures that fractionate or decompose concurrent tasks into separately practiced tasks (Wightman & Lintern, 1985). For example, using a part-task technique to learn to play a song on the piano may involve practicing the music for each hand separately before trying to play both parts together. Part-task techniques have been shown to promote significant transfer when attention can be allocated to individual tasks in turn (e.g., Briggs & Naylor, 1962). Whole-task techniques are procedures in which multiple tasks (e.g., the music for both hands) are practiced together, providing opportunities to develop and practice strategies to coordinate timing and cope with interference associated with concurrent tasks (Broadbent, 1982; Navon & Miller, 1987; Schneider & Detweiler, 1988). Research has suggested that whole-task techniques promote greater transfer to concur rent-task scenarios than do part-task techniques (Damos, 1991; Detweiler & Lundy, 1995; Naylor & Briggs, 1963).
A part-task technique that reduces cognitive, monitoring, and response loads during learning (Sweller, 1988, 1993), yet provides opportunities to develop compensatory strategies for successfully coping with concurrent-task demands (Schneider & Detweiler, 1988), should promote benefits similar to those of whole-task techniques and should reduce the risk of overburdening learners (see Gopher, 1993; Gopher, Weil, & Bareket, 1994; Gopher, Weil, & Siegel, 1989). The ability to develop successful compensatory strategies seems to depend on the invariant temporal dynamics of each task (Broadbent, 1982) or on the concurrence of task-specific processes (Detweiler & Lundy, 1995; Schneider & Detweiler, 1988). Part-task techniques should capture such critical relationships to provide opportunities to develop and practice compensatory strategies.
Chaining is one part-task technique that captures such relationships by segmenting multistep sequences into component steps based on temporal or spatial dimensions. Forward chaining involves learning the sequence of component steps gradually and in their natural order, whereas backward chaining involves learning the sequence of component steps gradually but in their reverse order. For example, using a forward-chaining technique to learn to play a song on the piano may involve practicing the music (both hands) for the first measure, followed by the music for the first two measures together, then the first three measures together and so on, until the full sequence is played. Forward-chaining techniques have been used to train a variety of single, multistep tasks, such as piano playing (Ash & Holding, 1990), origami (Wilcox, 1974), object assembly (Weber, 1978), and sequential button-pressing (Watters, 1992; Watters & Scott, 1992). Backward-chaining techniques have been shown to be superior to whole-task techni ques in several multiple-task settings, such as dive-bombing (Bailey, Hughes, & Jones, 1980, in Wightman & Lintern, 1985) and simulated carrier landing tasks (Wightman & Sistrunk, 1987). Research has yet to show the value of forward-chaining techniques for concurrent multistep tasks.
FORWARD-CHAINING AND CONCURRENT-TASK PERFORMANCE
When concurrent-task demands are segmented to capture consistent temporal dynamics or concurrences of particular responses, chaining techniques that divide the concurrent task into concurrent component steps should focus learners on the specific demands associated with each step. By facilitating practice on the initial step, chaining provides opportunities to identify the step's particular demands in isolation. After some level of mastery is achieved with each subsequence, subsequent steps are added one at a time, providing more focused opportunities to identify the concurrent-task demands associated with each step in the sequence. In contrast, participants trained using whole-task techniques do not receive such segmented opportunities; rather, the entire sequence of concurrent component steps is presented throughout training, making it more difficult to identify the concurrent demands associated with each step.
Sweller (1988) proposed that learners must make a trade-off between the attention and effort devoted to performing novel tasks and those devoted to learning them. Consequently the more effort and attention the task setting requires, the less of these the learner will have for identifying and developing compensatory strategies. By gradually increasing performance demands with practice, chaining techniques provide opportunities for learners to devote more resources to learning the task, especially early in practice, when novel learning demands are highest. Whole-task techniques require learners to perform all concurrent component steps on each trial of practice, increasing the likelihood that learning and performance demands will exceed the learner's capabilities, especially early in practice, when task demands are novel.
In this study, participants were asked to manually adjust the systems of a pretend submarine. As shown in Figure 1, the navigation system was represented by cross hairs located in the center of the computer screen, and the peripheral systems consisted of six support systems (temperature, oxygen, speed, radar, torpedoes, and shield) distributed around the navigation display.
Participants were randomly assigned to one of four training conditions. Participants trained with the whole-task technique were required to adjust both systems on each trial. Participants trained with the concurrent-chaining technique were also required to adjust both systems on each trial, but the number of trials per block gradually increased with training. Participants trained using the pure part-task technique were required to adjust only one system on each trial. Participants trained using the part-task chaining technique first practiced one task alone, after which the second task was included in practice but added gradually. Following training, participants performed whole-task transfer trials.
Because part-task techniques should reduce monitoring, response, and working-memory demands during training and facilitate learning (Schneider & Detweiler, 1988; Sweller, 1988, 1993), we expected the pure part-task technique to result in the most accurate training performance. Because the whole-task and concurrent-chaining techniques place greater demands on learners by requiring them to process and respond to two sets of information and to develop strategies for coping with concurrent-task interference, we expected these techniques to result in the least accurate training performance. Because part-task chaining provided an intermediate level of concurrent-task demands with respect to the other conditions, we expected this technique to promote an intermediate level of performance.
With respect to transfer, we expected the whole-task technique to promote greater transfer accuracy than either the pure part-task or part-task chaining techniques because it provided more opportunities to practice coping with the interference associated with concurrent-task demands (Broadbent, 1982; Navon & Miller, 1987; Schneider & Detweiler, 1988). Although concurrent-chaining participants saw fewer displays and received fewer opportunities to respond than did those in the wholetask condition, we expected them to perform as well as, if not more accurately than, whole-task participants. The concurrent-chaining technique provided more isolated opportunities for learners to identify the concurrent-task demands for each step of the sequence and should have facilitated learning by reducing performance demands (Schneider & Detweiler, 1988; Sweller, 1988, 1993). We also expected the pure part-task technique to result in the least accurate transfer because it provided no opportunities to identify or practice tasks concurrently. Because the part-task chaining technique provided an intermediate number of opportunities to identify and practice tasks concurrently, we expected that it would promote an intermediate level of transfer relative to the other techniques.
Participants. Eighty students from undergraduate courses at Pennsylvania State University participated for course credit. All reported normal or corrected-to-normal vision.
Procedures and tasks. All participants performed one 50-min session in which they were instructed to monitor the navigation and peripheral system displays and to correct system errors. The navigation task required participants to steer the submarine away from obstacles, represented by three brown blocks located at the end of the cross hairs in the center of the screen. On each trial, old obstacles were replaced by three new obstacles, requiring participants to press one of four directional arrow keys located on the numeric keypad (1 = left, 2 = down, 3 = right, and 5 = up) to steer the submarine to safety. When single-task trials did not require navigation task performance, no obstacles appeared. To learn the mappings between the appropriate response keys and the navigational displays, participants were given four self-paced navigation practice trials that required them to press the correct arrow key to proceed.
The peripheral systems were described one at a time. As shown in Figure 1, information about each peripheral system appeared in a separate display window. Peripheral systems were labeled, and each was framed in a blue box surrounded by a larger green box.
As shown in Figure 2, each system was always at one of three states. When a peripheral system was overactive (i.e., if the temperature was too high), the correct response was to lower the activation state (i.e., reduce the temperature) by pressing the key labeled with a "-" ("g" on a QWERTY keyboard). When a peripheral system was underactive (i.e., if the temperature was too low), the correct response was to raise the activation state (i.e., raise the temperature) by pressing the key labeled with a "+" ("h" on a QWERTY keyboard). Because system displays required scanning, subtending a visual angle of 24.23[degrees] at a viewing distance of 40 cm, the frame of the peripheral system that required adjustment turned white to attract attention. On each successive trial, the system that previously required adjustment returned to its correct moderate state and its frame returned to its original color. When single-task trials did not require peripheral task performance, all systems remained at their correct moderate state. Participants were given explicit instructions about the activation states of each peripheral system and had to provide correct responses to continue. To ensure that participants could return each error state to its correct moderate state, they practiced correcting each error state twice more in self-paced trials before continuing.
After completing task instructions and the self-paced practice, participants received instructions specific to their training condition and performed 120 training blocks and 24 whole-task transfer blocks. Trials began with a "Get Ready" warning that appeared at the center of the computer screen for 1 s, after which the navigation and peripheral systems displays appeared. Every trial lasted for 1 s, allowing up to two responses. Each block consisted of multiple trials in rapid succession, requiring participants to produce sequences of responses across trials. The sequence of navigation task and peripheral task displays and correct responses remained the same across blocks. For example, every trial x of each block showed the same displays and required the same responses as it did on every other block, although single-task trials showed only some of this information and required one appropriate response. Each sequence was randomly selected for each participant, and sequences were constrained so participants had to steer the submarine in a different direction and adjust a different peripheral system on successive trials. At the end of each block, participants received block-level accuracy feedback for each task.
Conditions. All participants performed 12 trials per block, except those in the concurrent-chaining condition. For concurrent-chaining participants, the number of trials per block gradually increased with training. For example, the first set of 10 training blocks included 1 trial per block, the second set included 2 trials per block, and the third set included 3 trials per block. In this way novel steps were gradually added to the sequence until the complete sequence was practiced.
For participants in the concurrent-chaining and whole-task conditions, every trial was a concurrent task trial that required both a peripheral and navigation task response. For participants in the pure part-task condition, every trial was a single-task trial that required either a peripheral or a navigation task response. Practice on each task was blocked, and practice order was counterbalanced.
Participants in the part-task chaining condition practiced one task only for their first 60 blocks, after which each new set of 5 blocks required an additional response. For example, Blocks 61-65 required the participant to respond to both tasks on the first trial and to only one task on subsequent trials. The next set of 5 blocks required the participant to respond to both tasks on the first two trials and to only one task on subsequent trials. In this way novel steps were gradually added until the complete sequence was practiced. Task order was counterbalanced. A comparison of training conditions is depicted in Figure 3. Following training, all participants performed 24 transfer trials that mimicked the conditions described for the whole-task training condition.
Equipment. Displays were shown on IBM PC-compatible microcomputers equipped with color VGA monitors. The Micro Experimental Laboratory (MEL) software system (Schneider, 1988) was used to program the experiment and record responses.
Results and Discussion
Training data. Visual inspection of the data revealed that three participants in the concurrent chaining and three participants in the whole-task conditions found the speeded displays, high memory load, and concurrent response requirements too demanding and essentially stopped practicing. Application of the interquartile-range rule (see Moore & McCabe, 1993) confirmed that these participants were outliers, and they were removed from subsequent analyses.
As shown in Figure 4, participants in the pure part-task condition showed the highest overall performance (M = 89%, SD = 5.9), followed by the part-task chaining (M = 77%, SD = 10.1), the concurrent-chaining (M = 71%, SD = 12.0), and the whole-task conditions (M = 61%, SD = 11.0). An analysis of variance (ANOVA) with repeated measures on concurrent task revealed a significant effect of training condition on overall training performance, F(3, 70) = 26.07, p [less than] .0001, MSE = 99.12. As expected, Bonferroni pairwise comparisons of overall performance revealed that participants in the whole-task and concurrent chaining conditions performed reliably less accurately than did participants in the pure part-task condition, ps [less than].01. Also as expected, the whole-task condition was significantly less accurate than the part-task chaining condition (p [less than].01), which was less accurate than the pure part-task condition (p [less than].05).
With respect to component task performance, an ANOVA with repeated measures on component task (navigation/peripheral) revealed significant main effects of training condition, F(3, 70) = 26.07, p [less than].0001, MSE = 198.2, and component task, F(1, 70) = 35.60, p [less than].0001, MSE 120.4, although the Training Condition x Component Task interaction was not statistically significant. As shown in Figure 5, participants in the pure part-task group performed the navigation task best (M = 96%, SD = 4.9), followed by the part-task chaining group (M = 83%, SD = 10.8). The concurrent chaining (M = 72%, SD = 14.3) and whole-task groups (M = 66%, SD = 14.1) were least accurate. Pairwise comparisons revealed that those in the whole-task and concurrent chaining conditions performed reliably worse than the pure part-task group (ps [less than].01). Further, the whole-task group was significantly less accurate than the part-task chaining group (p [less than].01), and the part-task chaining group was less accurate than the pure part-task group (p [less than].05).
With respect to peripheral task performance, the pure part-task group performed most accurately (M = 82%, SD = 10.5), followed by the part-task chaining group (M = 72%, SD = 16.6), the concurrent chaining group (M = 70%, SD = 13.9), and the whole-task group (M = 57%, SD = 13.1). Pairwise comparisons revealed that the whole-task and concurrent chaining groups performed reliably less accurately than the pure part-task group (ps [less than] .01). The whole-task group was also significantly less accurate than the part-task chaining group (p [less than] .01).
Transfer data. Visual inspection of the data revealed that two more participants essentially stopped performing. Applying the interquartile range rule (see Moore & McCabe, 1993) confirmed that these participants were outliers, and they were removed from subsequent analyses. As shown in Figure 6, the whole-task (M = 77%, SD = 12.5) and concurrent chaining groups (M = 79%, SD = 11.0) were the most accurate with respect to overall performance. The part-task chaining group fared next best (M = 68%, SD 18.5), and the pure part-task group showed the least amount of transfer to whole-task conditions (M = 59%, SD = 17.0). It is important to note, however, that all training conditions showed some positive transfer relative to untrained participants, which we determined by assessing the performance of the whole-task training group across the first 24 trials of practice (M = 41%). An ANOVA with repeated measures on concurrent task showed that the effect of training condition on overall transfer performance was signific ant, F(3, 68) = 8.96, p [less than] .0001, MSE = 311.44. As expected, Bonferroni pairwise comparisons of overall performance revealed that the whole-task and concurrent chaining groups performed reliably better than the pure part-task group, ps [less than] .001.
An ANOVA with repeated measures on component task (navigation/peripheral) was conducted on the mean percentage of correct transfer responses. This analysis revealed significant main effects of training condition, F(3, 68) = 26.07, p [less than] .0001, MSE = 99.12, and component task, F(1, 68) = 11.02, p [less than] .005, MSE = 152.33, and a significant Training Condition x Component Task interaction, F(3, 68) = 3.82, p [less than] .05, MSE = 152.33. Separate follow-up ANOVAs were run for each component task (peripheral/navigation).
As shown in Figure 7, training condition affected performance on the navigation task. Participants in the concurrent chaining (M = 88%, SD = 9.1) and whole-task conditions (M = 84%, SD = 10.3) performed the navigation task best, followed by the part-task chaining (M = 67%, SD = 23.7), and pure part-task groups (M = 61%, SD = 10.6). Analyses also revealed that the effect of training condition on the navigation task was significant, F(3, 68) = 8.96, p [less than] .0001, MSE = 311.44. Pairwise comparisons revealed that, on the navigation task, participants in the whole-task and concurrent chaining conditions were reliably more accurate than those in the pure part-task and part-task chaining conditions (ps [less than] .05).
Also shown in Figure 7, training condition affected peripheral task performance. The concurrent chaining (M = 73%, SD = 16.1) and whole-task groups (M = 70%, SD = 18.0) performed best, followed by the part-task chaining (M = 70%, SD = 18.6) and pure part-task groups (M = 57%, SD = 17.5). Analysis revealed that the effect of training condition on the peripheral task was also significant, F(3, 68) = 3.15, p [less than] .05, MSE = 310.98.
Because participants in the whole-task condition had 1440 opportunities to respond (12 times on each of the 120 training blocks), whereas participants in the concurrent chaining condition had only 780 opportunities to respond, additional analyses were conducted equating for response opportunities. We compared the performance of the concurrent chaining group on the 24 whole-task trials with the whole-task group's performance on 24 whole-task trials after an equivalent number of response opportunities (65 trials). These analyses revealed a significant effect of training condition on overall performance: The concurrent chaining group (M = 79%, SD = 11.0) outperformed the whole-task group (M = 69%, SD = 13.9), F(1, 32) = 5.20, p [less than] .05, MSE = 157.81, suggesting a benefit of concurrent chaining practice.
As shown in Figure 8, further analyses indicated that the concurrent chaining group (M = 86%, SD = 9.1) was reliably more accurate on the navigation task than the whole-task group (M = 74%, SD = 17.4), F(1, 32) = 6.50, p [less than] .05, MSE = 193.71, although no reliable difference was found between these groups on the peripheral task (p [greater than] .05).
As expected, the pure part-task condition showed the highest training and lowest transfer performance, whereas the whole-task and concurrent-chaining conditions showed the reverse pattern. The data also show that the concurrent-chaining technique can be an efficient and potentially cost-effective training technique, promoting as much transfer as the whole-task technique and using only 54% of the number of response opportunities to complete training. Although training time differences were not explicitly investigated, differences in response opportunities caused concurrent-chaining participants to train with shorter blocks, reducing their hands-on training time by 46%. Thus the data show the efficiency benefits of the concurrent chaining technique with respect to training time. In analyses equated for number of response opportunities, concurrent chaining promoted greater transfer than whole-task training.
The results support the idea that transfer to a concurrent task setting depends on the extent to which training provides opportunities to practice critical invariant relationships (Detweiler & Lundy, 1995; Lintern, 1991). The concurrent chaining and whole-task groups performed similarly, which we take as evidence that the training technique's ability to capture the invariant temporal dynamics or concurrent task relationships necessary to develop compensatory strategies was critical for transfer. When response opportunities were considered, the superior concurrent chaining performance suggests that training techniques that help learners identify such invariant relationships and practice concurrent-task demands can be quite effective. For situations in which whole-task training may not be practical early in training or too expensive, the data begin to suggest that the part-task chaining technique may offer an alternative means of promoting higher levels of transfer than the pure part-task technique, although t his claim warrants further research.
Echoing claims made previously, the data show that concurrent skill acquisition depends more on the nature of the opportunities provided by practice than on the amount of practice (e.g., Gopher, 1993; Gopher et al., 1994, 1989; Wightman & Sistrunk, 1987) and that concurrent task performance is influenced by the amount of training with critical intratask in-variants (Brown & Carr, 1989; Wightman & Lintern, 1985). In addition, the data show that techniques that promote the best training performance do not necessarily promote the best transfer performance (see Schmidt & Bjork, 1992).
The current data do not support assertions made in classic studies that have been used to inform training program design. Naylor and Briggs (1963) argued that whole-task training should be superior to part-task training for highly organized tasks such as concurrent tasks because whole-task training provides opportunities to develop and practice organizing processes. Annett and Kay (1956) argued that whole-task training should be better than segmentation for consistently sequenced tasks in which previous responses do not influence subsequent response opportunities or displays because such training provides more opportunities to practice with the entire set of sequence consistencies. Our data do not support either of these claims. However, the data do support the rationale underlying each assertion by showing that the most efficient technique provided opportunities to learn and use both intertask and intratask consistencies as organizing factors, suggesting that trainers should identify such consistencies and the strategies they promote when designing concurrent-task training techniques.
Operators are required to perform various types of concurrent monitoring and responding (Damos, 1991; Lintern & Wickens, 1991), sometimes in situations that require the rapid processing of information and high memory loads (e.g., air traffic control). In some cases (e.g., carrier landings) operators perform sequences of steps that require concurrent monitoring and responding. At some level many tasks are structured by both intertask consistencies (i.e., temporal consistencies) and intratask consistencies (i.e., when component task responses must be made together). In these ways the current tasks capture the task demands of industrial tasks. However, the current tasks provide structures that are not common. Few tasks require people to make the same sequence of concurrent responses repeatedly. Many sequential tasks are composed of cascaded rather than encapsulated component steps. That is, for many sequential tasks, the result obtained on one step either serves as input for the subsequent step or limits the re sponse space on the subsequent step. These issues are ripe for further investigation.
Further, the current data do not address the ways in which the training techniques affected individuals differently. Wightman and Sistrunk (1987) showed that the benefits of a backward-chaining technique were greater for low-aptitude participants. Jones (1966) and Ackerman (1988) also showed that different abilities are responsible for performance at different stages of practice. Although our manipulations do not allow for meaningful analyses of the variance data, pure part-task and part-task chaining participants showed the greatest increases in variance on whole-task transfer trials, suggesting that they did not have sufficient opportunities to develop the strategies necessary for concurrent task performance. It could be that concurrent chaining or whole-task training is more generally applicable, in the sense that there is less individual variation in responding later in learning.
It is also important to note that we claim the concurrent chaining technique is more efficient than the whole-task technique, based on a failure to reject a null hypothesis (the means of the two groups are similar). We do not believe that this is a critical limitation of the research given that the data support the alternative hypothesis that the concurrent chaining technique is the more effective technique when the number of response opportunities is taken into account.
The authors thank Richard Carlson and three anonymous reviewers for providing valuable comments and suggestions.
Andrew C. Peck received a Ph.D. in cognitive psychology from Pennsylvania State University in 1998. He is a faculty member in the Department of Psychology at Pennsylvania State University.
Mark C. Detweiler received a Ph.D. in cognitive psychology from the University of Pittsburgh in 1988. He is a user interface designer at Ariba, Inc., in Mountain View, CA.
Ackerman, P. L. (1988). Determinants of individual differences during skill acquisition: Cognitive abilities and information processing. Journal of Experimental Psychology: General. 117, 288-318.
Annett, J., & Kay, H. (1956). Skilled performance. Occupational Psychology, 30. 112-117.
Ash, D. W., & Holding, D. H. (1990). Backward versus forward chaining in the acquisition of a keyboard skill. Human Factors, 32, 139-146.
Briggs, G. E., & Naylor, J. C. (1962). The relative efficiency of several training methods as a function of transfer task complexity, Journal of Experimental Psychology. 64, 505-512.
Broadbent, D. E. (1982). Task combination and selective intake of information. Acta Psychologica, 50, 253-290.
Brown, T. L., & Carr, T. H. (1989). Automaticity in skill acquisition: Mechanisms for reducing interference in concurrent performance. Journal of Experimental Psychology: Human Perception and Performance. 15, 686-700.
Butterfield, E. C. (1989). Theory and practice of teaching for transfer. Educational Technology Research and Development, 37, 5-38.
Cormier, S. M., & Hagman, J. D. (Eds.). (1987). The structural process underlying transfer of training. In S. M. Cormier & J. D. Hagman (Eds.), Transfer of learning: Contemporary research and applications (pp. 151-181). San Diego, CA: Academic.
Damos, D. (1991). Dual-task methodology: Some common problems. In D. Damos (Ed,), Multiple-task performance (pp. 101-119). Bristol, PA: Taylor & Francis.
Detweiler, M. C., & Lundy, D. H. (1995). Effects of single- and dual-task practice on acquiring dual-task skill. Human Factors, 37, 193-211.
Gopher, D. (1993). The skill of attention control: Acquisition and execution of attention strategies. In D. E. Meyer & S. Kornblum (Eds.), Attention and performance XIV (pp. 299-322). Cambridge: MIT Press.
Gopher, D., Weil, M., & Bareket, T. (1994). Transfer of skill from a computer game trainer to flight. Human Factors. 36. 387-405.
Gopher, D., Weil, M., & Siegel, D. (1989). Practice under changing priorities: An approach to the training of complex skills. Acta Psychologica, 71, 147-177.
Jones, M, B. (1966). Individual differences. In E. A. Bilodeau (Ed.), Principles of skill acquisition (pp. 109-146). New York: Academic.
Lintern, G. (1991). An informational perspective on skill transfer in human machine systems. Human Factors, 33. 251-266.
Lintern, G., & Wickens, C. D. (1991). Issues for acquisition and transfer of timesharing and dual-task skill. In D. Damos (Ed.), Multiple-task performance (pp. 123-138). Bristol, PA: Taylor & Francis.
Moore, D. S., & McCabe, G. P. (1993). Introduction to the practice of statistics. New York: W. H. Freeman.
Navon, J. C,. & Miller, J. (1987), Role of outcome conflict in dual-task interference. Journal of Experimental Psychology: Human Perception and Performance, 13, 435-448.
Naylor, J. C., & Briggs, G. E. (1963). Effects of task complexity and task organization on the relative efficiency of part and whole training methods. Journal of Experimental Psychology. 65, 217-224.
Reder, L. M., & Klatzky, R. (1994). Transfer: Training for performance. In D. Druckman & R. A. Bjork (Eds.), Learning, remembering, believing: Enhancing human performance (pp. 25-314). Washington, DC: National Academy of Sciences.
Royer, J. M. (1979). Theories of transfer of learning. Educational, Psychologist. 14, 53-69.
Schmidt, R., & Bjork, R. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3, 207-217.
Schneider, W. (1988). Micro Experimental Laboratory: An integrated system for IBM PC compatibles, Behavior Research Methods, Instruments, & Computers, 20, 206-217.
Schneider, W., & Detweiler, M. C. (1988). The role of practice in dual-task performance: Toward modeling in a connectionist/control architecture. Human Factors, 30, 539-566.
Singley, M. K., & Anderson, J. R. (1989). The transfer of cognitive skill. Cambridge, MA: Harvard University Press.
Stammers, R. B. (1982). Part and whole practice in training for procedural tasks. Human Learning. 1, 185-207.
Sweller. J. (1988), Cognitive load during problem solving: Effects on learning. Cognitive Science, 12, 257-285.
Sweller, J. (1993). Some cognitive processes and their consequences for the organization and presentation of information. Australian Journal of Psychology, 45, 1-8.
Thorndike, E. L., & Woodworth, R. S. (1901). The influence of improvement in one mental function upon the efficiency of other functions. Psychological Review, 8, 247-261.
Watters, R. G. (1992). Retention of human sequenced behavior following forward chaining, backward chaining and whole task training procedures. Journal of Human movement Studies, 22, 117-129.
Watters, R. G.. & Scott. J. M. (1992). Teaching human sequenced behavior with and without prompts: A comparison of forward chaining, backward chaining and whole task procedures. Journal of Human Movement Studies, 22, 239-249.
Weber, N. J. (1978). Chaining strategies for teaching sequenced motor tasks to mentally retarded adults. American Journal of Organizational Therapy, 32, 385-389.
Wightman, D. C., & Lintern, G. (1985). Part-task training for tracking and manual control. Human Factors, 27, 267-283.
Wightman, D.C., & Sistrunk, F. (1987). Part-task training strategies in simulated carrier landing final-approach training. Human Factors, 29, 245-254.
Wilcox, B. (1974). The teaching of serial tasks using chaining strategies. British Journal of Educational Psychology, 44. 175-183.
|Printer friendly Cite/link Email Feedback|
|Author:||Peck, Andrew C.; Detweiler, Mark C.|
|Date:||Sep 22, 2000|
|Previous Article:||Overcoming Representational Errors in Complex Environments.|
|Next Article:||Team Play with a Powerful and Independent Agent: A Full-Mission Simulation Study.|