When

Noon – 1:30 p.m., May 2, 2025

Where

Speech, Language, and Hearing Sciences Building #205

Student Showcase:
Hosain Heshmati
Laura Baiocco Pereira

Zoom: http://arizona.zoom.us/j/82368731094
 

HOSAIN & LAURA
Laura Baiocco Pereira, Graduate Student, Psychology
Metaphor processing is influenced by stimulus emotionality and task demands: Evidence from ERPs
Abstract: Past studies showed that metaphoric expressions (e.g., “she was cold to him”) require more cognitive-neural effort than literal paraphrases (e.g., “she was indifferent to him”). In event-related potentials (ERP) studies, this was revealed as an N400, a late positivity (LP), and/or a late negativity (LN). In this presentation, I will discuss how stimulus emotionality and task demands influence these ERP correlates and metaphor processing. I will present two experiments in which participants read emotional/neutral metaphorical/literal sentences and either simply read the sentences or performed a sensicality judgement task. Our findings indicate that stimulus emotionality and task demand co-determine the extent to which emotion- and semantic- related neural resources are recruited during metaphor comprehension.

 

Hosain Heshmati, Graduate Student, Educational Psychology
AI-Generated Summaries and Retrieval Practice Increase JOLs But Not Delayed Learning Outcomes  
Emerging AI tools like ChatGPT offer learners assistance with study tasks, but their impact on metacognition and long-term retention is unclear. Across three experiments, we examined how offloading study tasks to ChatGPT influences judgments of learning (JOLs)—a critical metacognitive monitoring tool guiding study behavior and time allocation— and test performance. In Experiment 1, participants who generated summaries via Chat-GPT reported higher JOLs and scored higher in the final test compared to those who wrote their own summaries. In Experiment 2, when the final test was delayed, the performance advantage disappeared, but inflated JOLs persisted. Experiment 3 replicated this pattern in a retrieval practice setting: despite no difference in delayed test performance, AI-supported participants continued to report higher JOLs. These findings suggest that AI assistance inflates JOLs through processing fluency without improving long-term retention. 
 

Contacts

Jessica Andrews-Hanna
jandrewshanna@arizona.edu