Validation of semantic illusions independent of anomaly detection: evidence from eye movements

Text validation: Overlooking consistency effect discrepancies

The consistency effect, prevalent in the text comprehension literature, comprises longer reading times for inconsistent than equivalent consistent text continuations. It is widely interpreted as reflecting readers' effective "validation" of text coherence. However, there are also numerous phenomena of readers' deficient validation, sometimes collectively labelled "misinformation effects." This study asked whether readers become consciously aware of the text discrepancies diagnosed by the consistency effect. Experiment 1 scrutinized (a) conspicuous conceptual inconsistencies and (b) character-trait inconsistencies. It replicated the consistency effect both at a critical target sentence and at a subsequent sentence (spillover). Experiment 2 replaced self-paced reading with consistency judgments about the target sentences. The subjects overlooked almost half of the inconsistencies, thus denying that readers reliably become aware of consistency-effect discrepancies. In Experiment 3, the former target sentences were reframed as explicit questions. Accuracy for inconsistencies was statistically indistinguishable from accuracy for consistent targets, favoring the interpretation that Experiment 2 subjects overlooked known, encoded discrepancies. The results are interpreted with reference to Kintsch's (Psychological Review, 95, 163-182, Kintsch, 1988; Comprehension. New York: Cambridge University Press, Kintsch, 1998) construction-integration model, a refinement of which adds an immediate, passive stage of validation processing to construction and integration. It is proposed that passive validation affords the detection of text inconsistencies that do not reach the level of readers' conscious awareness.

Tracking your emotions: An eye-tracking study on reader's engagement with perspective during text comprehension

An eye-tracking study explored perspective effects on eye-movements during reading. We presented texts that included either a personal perspective (you) or an onlooker perspective (he or she). We measured whether fixations on the pronouns themselves differed as a function of perspective, and whether fixations on pronouns were affected by the emotional valence of the text which was either positive or negative. It was found that early in the text, processing of you is easier than he or she. However, as the character referred to by he or she becomes more familiar, fixations on he or she decrease, specifically in negative contexts.

What Can Eye Movements Tell Us about Higher Level Comprehension?

The majority of eye tracking studies in reading are on issues dealing with word level or sentence level comprehension. By comparison, relatively few eye tracking studies of reading examine questions related to higher level comprehension in processing of longer texts. We present data from an eye tracking study of anaphor resolution in order to examine specific issues related to this discourse phenomenon and to raise more general methodological and theoretical issues in eye tracking studies of discourse processing. This includes matters related to the design of materials as well as the interpretation of measures with regard to underlying comprehension processes. In addition, we provide several examples from eye tracking studies of discourse to demonstrate the kinds of questions that may be addressed with this methodology, particularly with respect to the temporality of processing in higher level comprehension and how such questions correspond to recent theoretical arguments in the field.

Processing real-world violations embedded within a fantasy-world narrative

Fantasy-text is a genre in which events routinely violate rules we know to be true in the real world. In four experiments, we explored the inherent contradiction between unrealistic fictional events and general world knowledge (GWK) to examine these competing information sources within the context of an extended fantasy-narrative. Experiments 1a and 1b demonstrated that fantasy-unrelated inconsistencies caused disruption to comprehension despite an abundance of contextual support for real-world impossible events that violate GWK. Experiment 2a demonstrated that fantasy-related inconsistencies caused disruption when they occurred at the local level and the fantasy-context stood in direct opposition to the target sentence. However, Experiment 2b demonstrated that disruption can be initially eliminated when readers encountered fantasy-related violations at the global level, but delayed-processing difficulty occurred on the spillover sentence, downstream of the target sentence. All four experiments are discussed within the context of the RI-Val model.