Visual narratives, like comics, at times show depictions of characters’ imagination, dreams, or flashbacks, which seem incongruent with the ongoing primary narrative. Such “domain constructions” thus integrate an auxiliary domain (e.g., a dream) within the primary domain (the expected, physical storyworld), and may require readers to resolve seemingly non co-referential figures as co-referential (e.g., when a character’s dream shows that character as an animal). In three self-paced reading experiments, we investigate the processing and understanding of single vs. multiple domains in sequences with co-reference issues (Exp. 1) and whether graphic cues facilitate such domain switches (Exp. 2 and 3). Domain switches incurred greater updating costs but were comprehensible, with greater similarity across panels predicting faster processing, and comic reading experience affecting viewing times. The successful integration of fantasized agents which seem to lack of co-reference implies that visual narrative comprehension goes beyond event and scene perception alone, but also involves proficiency in conventional constructions related to perspective-taking and inferencing.

Data files: Exp1_Analyses.csv → cleaned data file. Exp1_CharacterQuestion.csv → cleaned data file. Exp1_Regression.csv → cleaned data file. Exp2_Analyses.csv → cleaned data file. Exp2_CharacterQuestion.csv → cleaned data file. Exp2_Regression.csv → cleaned data file. Exp3_Analyses.csv → cleaned data file.

Supplemental material: Codebook for data files.

Method: Online self-paced reading experiments. Universe: A general sample of English-speaking people, not required to be experienced comic readers. Data sources: Images from a corpus of published Calvin & Hobbes comics.