Abstract
The current study investigated attentional selection during bridging inference generation in wordless visual sequential narratives. Comprehension processes, such as inference generation, have shown clear effects on eye-movements in reading, and more recently modest effects in studies with video and sequential narratives. Two competing hypotheses were tested: Computational Load: During inference generation, eye-movement locations will be driven by bottom-up saliency, and fixation durations will increase due to higher computational load. Visual Search: During inference generation, eye-movement locations will be driven by search for inference-relevant information, producing more fixations. Wordless children's picture stories were used. Bridging inference generation was induced through ellipses placed in the narrative during 3-image target episodes. In the target episodes, half the participants saw the full three image sequences, while the other half, in the ellipsis condition, did not see the highly inferable action in the middle of the target episodes. Previously, we showed that, in the ellipsis condition, viewers readily inferred the action, as shown by increased viewing times and think-aloud protocols on the image immediately following the ellipsis in the sequence. The current study added eye-tracking to test the effects of inference generation on eye-movements. Results showed participants in the ellipsis condition made more fixations, but had similar fixation durations to participants in the non-ellipsis condition. Fixation density heatmaps were then used to test whether fixations in the ellipsis condition were more likely to go to locations uniquely important for generating the inference than in the non-ellipsis condition, by comparing both to inference-informative heatmaps created from ratings by a separate group of participants. Overall, ellipsis condition participants were more likely to look at areas important for generating the inference. These results support the Visual Search Hypothesis in that during inference generation, participants make eye-movements to search for information that uniquely aids in drawing the inference.
Meeting abstract presented at VSS 2017