Processing of Refutation Texts

• In Event: 52.031 - Relational Reasoning in STEM Domains: What Empirical Research Can Contribute to the National Dialogue

Abstract

Purpose/Theoretical Framework

Effective STEM learning requires knowledge revision that can be accomplished with refutation texts. Refutation texts acknowledge misconceptions, directly refute them, and provide a satisfactory explanation
One hypothesis for the effectiveness of refutation texts has been proposed in the context of the five principles comprising the Knowledge Revision Components framework (KReC; Kendeou & O’Brien, in press). According to KReC, once information becomes part of memory, it cannot be erased, and always has the potential to be reactivated (Encoding Principle) via passive activation processes (Passive Activation Principle). Refutation texts enable the co-activation of previously-acquired and newly-encoded information, a necessary condition for knowledge revision (Co-activation Principle). Co-activation enables the integration of the correct and incorrect information into a single network (Integration Principle). As the amount of correct information is increased, it begins to dominate the network, draw increasing amounts of activation to itself, and away from the incorrect information (Competing Activation Principle). This competition is driven by the explanation and, ultimately, determines whether knowledge revision will be successful or not.

Objectives

In two experiments, we aim to provide direct evidence for the actual cognitive processes evoked by refutation texts in the context of KReC.

Methods and Data Sources

Participants were graduate students who read a series of texts that discussed Newtonian concepts (10 experimental, 10 filler). Each experimental text appeared in one of two conditions: refutation or control. Texts included a target sentence that stated the correct conception. In Experiment 1, 28 participants read the texts while their eye-movements were recorded using an Eyelink 2K. In Experiment 2, 30 participants read the texts while performing a think-aloud. After reading, all participants completed a 10-item post-test.

Results and Discussion

In Experiment 1, readers made longer forward, F (1, 26) = 7.03, p = .013, = .22, and look-back fixations, F (1, 26) = 3.96, p = .052, = .13 on the explanations in the refutation than in the control texts, suggesting deep processing of the explanations. This processing facilitated revision, as indicated by faster forward fixations on the correct conception sentence in the refutation than the control texts, F (1, 26) = 7.03, p = .013, = .22.
In Experiment 2, the refutation texts elicited more coherence-based processes than the control texts. Specifically, more associations to prior knowledge, F (1, 28) = 10.95, p = .003, = .28; cognitive conflict, F (1, 28) = 17.35 p = .000, = .38; correct inferences, F (1, 28) = 6.82, p = .014, = .19; and, comprehension monitoring, F (1, 28) = 22.57, p = .000, = .45.
Post-test scores were higher in the refutation than in the control conditions (Experiment 1: F (1, 26) = 48.97, p = .000, = .65; Experiment 2: F (1, 28) = 20.99, p = .000, = .43), suggesting successful revision of misconceptions after reading the refutation texts.

Scientific Significance

Taken together, these findings advance our understanding of the kind of reasoning that unfolds during reading of refutation texts. Findings are consistent with the KReC framework.

Authors

• Panayiota Kendeou, University of Minnesota

• Edward J O'Brien, University of New Hampshire