Comparison Benefits can be Boosted by a Progressive Comparison Sequence

• In Event: Poster Session 12
  In Poster Session: PS 12 Topic - Cognitive Processes

Abstract

A large body of evidence suggests that conceptually-based novel name generalizations is facilitated when multiple learning stimuli rather than a single stimulus are available (Gentner & Namy, 1999). The size of the comparison effect is influenced by factors such as the number of compared stimuli (Augier & Thibaut 2013), the semantic distance between the learning items (Thibaut & Witt, 2017) and the similarity between the compared stimuli (Hammer and al., 2008, 2009). In the present study, we capitalized on the latter result. We tested whether earlier comparison trials with an easier implementation of a distinctive dimension (i.e., more distinctive stimuli) would favor later comparison with a more difficult version (i.e. less distinctive) of the stimuli. We contrasted this condition with a condition in which the same total number of trials with only the difficult trials was performed. Up to now, no study investigated the effects of comparison in a sequence of trials.
A control pre-test showed that comparison was less efficient with low-distinctive than with high-distinctive stimuli.
Eighty children were divided in two age groups, young (3-4 years) and old (5-6 years), and ascribed to one of two conditions. In the High-Low condition, the first five trials (out of ten) were performed with highly-distinctive objects and the five next trials (transfer phase) were performed with low-distinctive objects. In the second condition (Low-Low), all the trials (i.e., ten out of ten) were performed using low-distinctive objects.
An ANOVA with 2 Age (young and old) and 2 Conditions (High-Low and Low-Low) as between subject factors was performed on the percentage of correct responses during the transfer phase. There was a significant effect of Age (F(1,76) = 9.01, p = .004) and a significant effect of Condition (F(1,76) = 9.93 p = .002) showing that the High-Low condition lead to more correct responses in the transfer phase than the Low-Low condition. However, t-test, showed that this difference appeared only for younger children. (Young: High-Low, M = 71%; Low-Low, M = 46%), t(38) = 2.55, p = .015). For young children, a correlation between the first trial performance (correct or error) and the performance in the second phase revealed that these two scores are significantly associated r(18) = 0.64, p = .002). This correlation was not significant r(18) = .87, p = .43) in the Low-Low condition.
These results suggest that comparing highly distinctive objects renders the commonalities more salient, what potentiates the benefits of following comparisons with less distinctive objects based on the same relevant dimension. The correlations between the first trial and later performance are interesting because they suggest that early success is a central predictor for later performance. On the contrary, accumulating more opportunities to compare less-distinctive items did not lead to progressive learning. These results show that successive comparisons with several exemplars do not lead, cumulatively to better performance.

Authors

• Yannick Lagarrigue, Laboratory for Research on Learning and Developement
  Presenting Author

• Jean-Pierre Thibaut, Université de Bourgogne Franche-Comte
  Non-Presenting Author