Variable Learning I
Varying learning conditions makes learning more effortful but results in enhanced long-term retrieval. In evaluating the effectiveness of instruction or training, one must consider that conditions that produce superior initial learning may not lead to better generalized learning. In fact, variable learning conditions may make initial learning more effortful, and less enjoyable, but leads to superior generalized learning (i.e., transfer). In contrast, less variable learning conditions may produce better initial performance on a task, but results in less transfer of learning. Based upon the apparent ease of performance during non-variable learning, individuals may falsely assume that they are effectively mastering new material.
Thus there is an illusion of learning that may be observed during initial training or practice, during which performance is enhanced due to less demanding learning conditions, but this apparent learning may actually be short-lived when tested for generalization to other situations. However, when learning is made more effortful via variable learning conditions, performance during initial phases may be impaired, although such learning conditions are more likely to produce better transfer, hence higher quality learning.
The effects of variable practice on learning have been demonstrated for both motor/perceptual tasks as well as higher cognitive functioning tasks. In a motor task, 8- and 12-year-olds practiced throwing a miniature bean bag at a target that varied in distance and were later tested on a criterion distance (Kerr & Booth,1978). Each age group was divided into low-variability and high-variability groups. For the low-variability group, each of the four practice throws were for a target that remained at a specific distance that served to be the criterion distance on the posttest in test trials. For the high-variability group, practice distances were varied around the criterion, but never at the criterion.
Although both variability groups showed improvement over practice, the high-variability group outperformed the low-variability group on the posttest, where the target was at the criterion distance. This finding is interesting given that the high-variability group never practiced throwing at the criterion distance, whereas the low-variability group had. Similar effects of variable practice on motor learning have been found by Shea, Kohl, and Indermill (1990) and Shea and Morgan (1979).
Schmidt and Bjork (1992) provide an empirical summary of how making training conditions more effortful, including variable learning, may impair acquisition of motor and verbal skills but improves retention and transfer of these skills beyond the initial learning situation. In the domain of problem solving and geometrical reasoning, varying learning conditions for certain types of problems may also make initial learning more difficult, yet results in better transfer (Paas & Van Merrienboer, 1994; Van Merrienboer & De Crook, 1997).
The online demonstration for this principle is currently being developed.
Educational Applications Vary learning conditions, which may make initial learning more effortful and less enjoyable but produces better retention and transfer. This may be done by presenting information in multiple contexts or formats. See Paas and Van Merrienboer (1994) for an example of how a geometrical problem may be varied.
Suggested Readings Kerr, R., & Booth, B. (1978). Specific and varied practice of motor skill. Perceptual and Motor Skills, 46, 395-401. In a motor skills task (throwing an object at a target), compared to individuals trained with non-variable practice (target at a specific, criterion, distance), individuals who had variable practice (targets were at varied distances) demonstrated better performance on a criterion distance they had never practiced throwing, but the non-variable practice individuals had.
Paas, F. G. W. C, & Van Merrienboer, J. J. G. (1994). Variability of worked examples and transfer of geometrical problem-solving skills: A cognitive-load approach. Journal of Educational Psychology, 86, 122-133. In the domain of geometrical problem solving, the effectiveness of studying from worked example problems was compared to that of studying from conventional problems, which did not present the solution method. The variability of these study problems was also varied to test performance on transfer problems. Results indicated that studying from worked problems, compared to conventional problems, required less time and produced better transfer. Additionally, variability of problems only made a difference for the worked problems, such that high variability resulted in better transfer.
Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3, 207-217. This paper discusses several training factors that have been empirically demonstrated to decrease performance on verbal and motor tasks during the initial phases of learning, but ultimately improves long-term retention. These factors include increasing variability during training, which makes acquisition of a skill or knowledge more difficult but results in better generalization of that skill or knowledge in later tasks.
Shea, C. H., Kohl, R., & Indermill, C. (1990). Contextual interference: Contributions of practice. Acta Psychologia, 73, 147-157. The study essentially extended the findings of Shea and Morgan (1979) concerning effects of variability (random, or high variability, versus blocked, or low variability, learning) on retention to a task that was more open-ended. The study also sought to determine the effect of increasing the number of acquisition trials on retention. Whereas Shea and Morgan used only 54 acquisition trials and found that random practice conditions produced better retention only when tested under random conditions, this study examined 400 acquisition trials and found that the random group performed better than the blocked group on both random and blocked retention conditions.
Shea, J., & Morgan, R. L. (1979). Contextual interference effects on the acquisition, retention, and transfer of a motor skill. Journal of Experimental Psychology: Human Learning and Memory, 5, 179-187. Variability in the form of contextual interference was manipulated using either blocked (low variability) or random (high variability) sequences during learning of a motor task (i.e., knocking down barriers in a given order). The blocked group performed significantly faster than the random group during early acquisition and made less than half as many errors. However, the random group performed better on retention and transfer measures than did the blocked group, particularly on more complex transfer tasks.
Van Merrienboer, J. J. G. , & De Crook, M. B. M. (1997). The transfer paradox: Effects of contextual interference on retention and transfer performance of a complex cognitive skill. Perceptual and Motor Skills, 84, 784-786. This article presents results from an exploratory study on the effects of variable practice on a complex troubleshooting task (diagnosing computer-simulated problems at a water-alcohol distillery plant). During practice, the high-variability group made more errors than did the low-variability group. On retention problems, the two group did not differ in numbers of errors made. However, on transfer problems, the high-variability group made significantly fewer errors than did the low-variability group.