Interleaving exemplars from different to-be-learned categories, rather than blocking exemplars by category, often enhances the inductive learning of those categories, as measured by learners’ subsequent ability to classify new exemplars from those categories. Majority of the studies on the learning of motor skills, perceptual categories, and mathematics procedures conceptualize the interleaving benefit to be a general learning phenomenon. Results from this dissertation extend the interleaving benefit to the inductive learning of cognitive, rule-based categories (e.g., statistical concepts).
In this dissertation I examine factors that modulate this interleaving benefit, such that interleaving is more or less effective than blocking depending on whether the learning emphasis is on discriminating between categories (discriminative-contrast hypothesis) or encoding commonalities within a category (commonality-abstraction hypothesis), and depending on whether the temporal spacing between exemplars from the same category optimally promote distributed retrieval practice of critical features shared within a category (study-phase retrieval hypothesis). Thus, findings from the current dissertation offer further insight into the boundary conditions of the interleaving benefit.
Consistent with the discriminative-contrast hypothesis, an interleaving benefit was observed when between-category similarity was high and within-category similarity was low, and when there was no temporal spacing between exemplars to disrupt contrast processes critical to between-category comparisons. Consistent with the commonality- abstraction hypothesis, a blocking benefit was observed when between-category similarity was low and within-category similarity was high, and when exemplars were presented three-at-a-time instead of one-at-a-time. Consistent with the study-phase retrieval hypothesis (i.e., introducing spacing between exemplars engages retrieval processes that enhance learning), a blocking benefit was observed when there was temporal spacing between exemplars from the same category.
Moreover, the type of categories themselves and learners’ cognitive abilities drove the effects of category learning differently. Findings from the current dissertation begin to demonstrate the interactions between study schedules and perceptual-based categories (artists’ painting styles) and rule-based categories (statistical concepts). For instance, when between-category similarity was low, the interleaving benefit was eliminated for the perceptual-based categories, but no blocking benefit was obtained, contrary to our prediction. This suggests that blocked versus interleaved schedules may be more or less conducive to learning depending on the type of categories. Finally, learners with lower working memory capacities—that is, learners with cognitive limitations related to information processing and integration—benefited from schedules in which exemplars were presented three-at-a-time, and from schedules that were either temporally spaced or interleaved, but having neither or both manipulations produced sub-optimal performance. To conclude, findings from this dissertation clarify when, for whom, and with what kind of categories is interleaving beneficial. / Dissertation / Candidate in Philosophy / Interleaving exemplars from different to-be-learned categories, rather than blocking exemplars by category, often enhances the inductive learning of those categories, as measured by learners’ subsequent ability to classify new exemplars from those categories. I examined the generality of the interleaving benefit and the conditions that make interleaving more or less effective for learning than blocking. Consistent with the hypothesis that interleaving enables between-category comparisons, I demonstrate that the interleaving benefit generalizes to the learning of complex, rule-based categories and across all learners, particularly those with lower cognitive abilities. Conversely, blocking enables within-category comparisons, and thus can be as beneficial as, or even more beneficial than, interleaving under certain conditions—if exemplars from the same category are presented at three-at-a-time rather than one-at-a-time or if the categories are structured such that there is high-within and low-between category similarity. These findings highlight the need to shift focus away from examining which type of schedule— interleaving or blocking—is more effective for category learning to identifying when and why each type of schedule may be more effective.
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/18143 |
| Date | 11 1900 |
| Creators | Sana, Faria |
| Contributors | Kim, Joe, Psychology |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0022 seconds