You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster E38, Friday, November 10, 10:00 – 11:15 am, Harborview and Loch Raven Ballrooms

Changing Task Demands in a Semantic Decision Task: Evidence for the Dynamic Multilevel Reactivation Framework.

Joshua Troche1;1University of Central Florida

Theories of semantic memory often fall within a spectrum from fully embodied to disembodied theories. Fully embodied theories argue for no central convergence area for semantic representations in that features are dispersed across the modal association cortices. Disembodied theories, in contrast, argue for a central convergence zone where abstracted semantic conceptualizations reside. Both extremes of this spectrum have significant shortcoming and therefore necessitate a theory that is a hybrid middle ground between embodied and disembodied. Reilly and colleagues recently described such a middle ground in the Dynamic Multilevel Reactivation Framework. This theory argues that portions of our semantic system are both embodied and disembodied and the engagement of these systems is dependent on the demands of the task. They theorize that a high order hub in the anterior temporal pole is the location of abstracted semantic representations and that as task demands change we enrich these abstracted representations with information from heteromodal and sensory cortices. In this study, we attempted to test the plausibility of the Dynamic Multilevel Reactivation Framework. Twenty-five young adult participants were shown a set of pictures while their eyes were being tracked by a RED 120hz infrared eyetracking system. The pictures, both living and nonliving objects, were grayscaled and presented to participants in a randomized order. The participants were required to complete a semantic decision task during the experiment. The semantic decision task either required them to make superordinate distinctions (e.g., Is this a living or non-living thing?), basic distinctions (e.g., Is this a plant or an animal), or subordinate distinctions (e.g., Is this a desert plant or tropical plant). Participants would indicate their answers (98% accuracy) through button press which would immediately advance the slide. Each participant had to complete a superordinate task, a basic task, and a subordinate task in a randomized order. The stimuli were the same across all tasks. We hypothesized that participants would be able to make superordinate decisions with little need to extract additional information from the pictures shown while subordinate categories would require enrichment of the abstracted representation through extraction of sensory information from the displayed picture. We measured the following outcomes: number of fixations in AOIs, the average duration of fixation in AOIs, and revisits to AOIs. AOIs were created posthoc by analyzing the heatmaps of the semantic decisions tasks and creating AOIs around areas of significant fixation. We compared the outcome measures across the three levels of semantic decision task: superordinate, basic, subordinate. We performed a Repeated Measures ANOVA for each outcome measure. For all three outcome measures, there was a main effect and decomposition revealed that subordinate semantic decision tasks lead to a greater number of fixations, larger average fixation times, and most AOI revisits as compared to the other tasks. The results fit well with the assumptions of the Dynamic Multilevel Reactivation Framework in that subordinate decisions required enrichment of the concept through sensory information as compared to basic and superordinate tasks.

Topic Area: Meaning: Lexical Semantics

Back to Poster Schedule