Serial Pattern Learning by Event Observation: Effects of Varying Amounts of Pattern Experience

The present study was designed to replicate the findings by Howard, Mutter & Howard (1992) that (1) overt behavioral responding is not prerequisite to serial pattern learning and (2) that observation produces a qualitatively different form of representation than an overt motor response. We also sought to extend these findings by determining how much exposure to the pattern is necessary to replicate these effects and by examining the role of stimulus-to-response mapping by adding an additional response group (unmapped-response). A version of the serial learning task used by Howard et al. (1992) was used. The task consisted of two phases. During the acquisition phase , an asterisk appeared in one of four locations on a video monitor. Three groups received either 1, 2, or 3 blocks of trials in which the position of the asterisk followed a 10-trial pattern (pattern block). Subjects in each group either (1) manually responded to the asterisk location (mapped-response group), (2) simply observed the asterisk locations (observation group), or (3) made a manual response that was unrelated to the pattern (unmapped-response group). During the remaining three blocks of the acquisition phase all groups responded to the asterisk location. Of the three remaining acquisition blocks, the first and third blocks were pattern blocks, while the location of the asterisk on the second block was determined randomly (random block). The difference in response times between the random block and the preceding pattern block provided an indirect measure of pattern learning. During the prediction phase , subjects predicted the locations of the asterisk. Prediction accuracy provided a direct measure of pattern learning. Results of our indirect measure of pattern learning supported the findings by Howard et al. (1992) that overt behavioral responding is not prerequisite to serial pattern learning. In addition, the amount of training strongly influenced both procedural and declarative learning. However, we were unable to find conclusive evidence to support the proposal by Howard et al. that observation produces an advantage over response on the direct measure of pattern learning. One possible reason for this could have been low statistical power to detect group differences. Because effect sizes for group differences were small to moderate (group, r|2 = .04; group by training, r|2 = .06), and power analyses for these effects indicated that power was very low (group, power = .40; group by training, power = .45), we could not rule out this possibility. A second possible reason could have been a slight difference in methodology. While Howard et al. included only one awareness probe (following prediction), our design included two awareness probes (one following acquisition and one following prediction). It is possible that the addition of the early awareness probe obscured group differences by sensitizing subjects to the possibility of a pattern. Further research employing greater power and different methodology will be needed to resolve this issue