Student Research in Psychology

Research Abstracts

• Judgments of agency and culpability for AI use in college: You might not have written it—but it’s still your fault
  • Presenter: Thomas Konnick
  • Faculty mentor: Dr. Mansfield and Dr. Potts

  Judgments of agency and culpability for AI use in college: You might not have written it—but it’s still your fault

  Student use of artificial intelligence (AI) in academic work is rapidly increasing, ranging from minor grammatical assistance to full generation of written content. Because these scenarios reflect varying degrees of human-AI contribution, they raise important questions about students’ sense of agency over AI-assisted work. Specifically, do students feel agency over content generated by AI? In the present study, participants read a series of short scenarios describing different ways a student used AI to complete an essay, ranging from no AI input to fully AI-generated work. Participants rated the extent to which the student possessed agency over the essay. Results indicated that perceived agency scaled with the level of student involvement, such that greater human contribution was associated with higher agency ratings. However, frequent AI users attributed greater agency to the student across the scenarios (F_{2, 68}=6.89, p<.001, η²=.17). Notably, a dissociation emerged between agency and culpability: participants indicated that the student is accountable for any errors in the essay, even when it was largely or entirely AI-generated. Taken together, these findings have implications for how responsibility and authorship are conceptualized in educational settings that increasingly involve AI tools.
• Improved Emotion Recognition in ASD
  • Presenter: Nilah Lafontant
  • Faculty mentor: Dr. Braje

  Improved Emotion Recognition in ASD

  Research has found that, in comparison to their neurotypical counterparts, individuals with autism spectrum disorder (ASD) showed a decrease in the ability to recognize emotion in faces that were wearing face masks during the coronavirus pandemic (Ventura et. al., 2023; Ramachandran, 2021). We examined whether the common usage of face masks post-pandemic has increased the ability of those with autistic traits to recognize emotion. We measured the relationship between scores on the Baron-Cohen RMET (a test of emotion recognition) and the AQ-10 (a brief assessment of autistic traits). Here we report data collected from 27 students in the PSY105 class at SUNY Plattsburgh, and we plan to collect additional data in the Fall of 2026. In accordance with our hypothesis, all participants scored within the typical range of the RMET (22-30). The 4 participants who scored ≥6 on the AQ-10 (indicating possible autistic traits) had an average RMET score of 27.5, while the remaining participants had an average RMET score of 25.8 (t(15.36)=1.63, p=.123). There was a non-significant but positive correlation between scores on the two tests (r(25)=.377, p=.053). These results indicate that individuals with autistic traits may have improved their ability to read emotions.

• What can time tell us about agency? Evaluating action duration estimates as a measure of control. 
  • Presenters: Meghan Cox, Teddy Defayette, Myles Maldonado-McCabe, Sean Noworyta
  • Faculty mentor: Dr. Potts

  What can time tell us about agency? Evaluating action duration estimates as a measure of control.

  The sense of agency—the feeling of control over one’s actions and their outcomes—is a growing area of interest with implications for clinical conditions, design, and our legal system. Research in this area has relied heavily on self-report measures, which are vulnerable to bias. Efforts to identify more objective measures have produced mixed results, focusing largely on temporal binding—the subjective compression of action–outcome intervals. Here, we examined a novel candidate measure: distortions in perceived action duration for actions of varied difficulty. Across two experiments, participants clicked on targets of varying aiming difficulty and either (1) estimated the duration of their actions or (2) reported their sense of control using a sliding scale. In Experiment 2, we introduced passive trials in which participants had no control over cursor movement to further isolate agency-related effects. Results indicated that action duration estimates were driven primarily by regression to the mean, with some evidence for increased temporal distortion when participants controlled the cursor. Together, these findings challenge proposed time-based measures of agency and suggest that the search for a reliable objective measure remains unresolved.

• Student ratings of AI-based counseling 
  • Presenter: Erkam Arklan
  • Faculty mentor: Dr. Mansfield

  Student ratings of AI-based counseling 

  The growing demand for accessible mental health services has led to considerable interest in AI-based counseling tools. However, this interest is accompanied by concerns about data privacy, client safety in crisis situations, and the potential discomfort of receiving therapy from a non-human agent. To explore this unease, we compared student ratings of AI-generated therapy with therapy based on real counseling materials. Two five-minute counseling sessions were created: one generated by ChatGPT and the other adapted from real counseling training materials. In both sessions, the therapist used cognitive-behavioral strategies to help a client with anxiety. Participants were randomly assigned to two groups: in one group, participants were told which session was AI-generated, while in the other group, they were not. After each session, students rated the therapist’s effectiveness, and at the end of the study, they indicated which therapist they preferred. Surprisingly, participants preferred the AI-generated therapist, even when they knew it was AI-generated. These findings suggest that AI-based therapy may be more broadly accepted than previously assumed.
• In the mood for music: Exploring how making & listening to music affect wellbeing
  • Presenter: Ngozi Permaul
  • Faculty mentor: Dr. Christy

  In the mood for music: Exploring how making & listening to music affect wellbeing

  Our study sought to discover the impact of music on the mental wellbeing of SUNY Plattsburgh’s students. We investigated the relationship between the students’ recent moods, their music listening habits, and their musical background in either singing or playing instruments to see if being exposed to music was associated with the way in which students feel. The first part of the experiment was administered through a Google Forms survey, in which we had 100+ students respond to self-report measures. Upon completion of the survey students were prompted to sign up for the second part of the experiment, a music-listening or music-playing activity. The students assigned to the music-making condition replicated the kick drum rhythm played in a musical arrangement while the students assigned to the music-listening condition listened to the arrangement mindfully. Data analysis is ongoing, but preliminary results of the post-intervention survey indicated that participation in the music activity contributed to a positive mood in the majority of respondents. Responses for the ‘self-conscious’ measure were the most varied of the survey, with 40% of students reporting that they experienced high degrees of self consciousness. Students also reported feeling a neutral or weak connection to the people around them during the music activity. 85% of the students were also able to recognize and correctly identify the arrangement played in the study. Further analyses will compare these experiences between the two music-activity groups, and explore associations between students’ musical background, personality variables, and wellbeing.