As AI chatbots become more integrated into daily life, reports of people forming meaningful relationships with them are increasing. While AI can provide emotional support and companionship, little is known about why some individuals are more likely to form connections with AI and what chatbot characteristics foster these bonds. Research suggests that conversational quality plays a greater role in social connection than partner identity, as people report feeling more connected to a supportive chatbot than an unsupportive human (Folk et al., 2024). This study examines the psychological factors that shape human-AI relationships, focusing on personality, loneliness, and prior AI experience, along with perceptions of AI warmth, competence, and discomfort. UCLA students will engage in free-flowing 10-minute conversations with GPT-4o as an AI assistant, GPT-4o as a conversation partner, and a human stranger. Pre-surveys will assess personality traits, social tendencies, and AI attitudes, while post-surveys will measure connection, empathy, and social perceptions using the Robot Social Attributes Scale (RoSAS). We hypothesize that individuals who rate GPT-4o highly in warmth and competence will show smaller differences between human and AI interactions, and those higher in loneliness will report stronger connections with GPT-4o.

Overgeneral memory is associated with depressive and trauma-related disorders. Individuals with an overgeneral memory are less specific in their response when asked to recall specific autobiographical memories. Memory specificity training is shown to be effective in improving overgeneral memory in the short term. Since depression is associated with deficits in reward processing, this project aims to analyze whether overgeneral memory is associated with the components of reward processing, including: reward valuation, reward learning, and reward responsiveness. We used baseline data (n = 64) from the OGM Study including the PVSS-21 and memory specificity scores on an autobiographical memory task. We operationalized memory specificity in two ways: percentage of memories recalled that were specific and total word count of recalled memories. The different components of reward processing are defined by the different subscales of the PVSS-21. We hypothesized that greater overgeneral memory will be associated with lower reward processing. We found that total word count is significantly associated with reward expectancy scores (p = 0.023) in that increased word count is associated with decreased reward expectancy scores. We also found that the percentage of specific memories is not significantly associated with average scores on the PVSS-21 (p = 0.54). If overgeneral memory predicts blunted reward processing, treatments for depression could be improved by targeting overgeneral memory and reward processing.

Depression remains a significant public health concern, particularly among racial/ethnic minority individuals facing compounded psychosocial stressors. Previous research has demonstrated that childhood trauma, general life stress and HIV are individually associated with increased risk of depression (Teicher & Samson, 2016; Liu & Alloy, 2010; Earnshaw et al., 2013). However, limited studies have examined the combined effects of these predictors among ethnic minority populations. Utilizing existing data from the Promoting Action Towards Health Study (N=202), which examines HIV-diagnosed men who have sex with men, this study will employ hierarchical multiple linear regression analysis to assess the extent to which childhood trauma, life stress, and HIV stigma contribute to depression severity. We will use the Traumatic Life Events, General Life Stress, HIV Stigma, and Center for Epidemiological Studies Depression Scales to assess childhood trauma, stress, and depression, respectively (Cristofaro et. al, 2013; Berger et. al, 2001; Radloff, 1977). We hypothesize that higher levels of childhood trauma, general life stress, and HIV stigma will each be independently associated with higher levels of depression. By identifying key psychological and environmental stressors linked to depression, this research will guide targeted interventions and support more effective mental health policies for racial/ethnic minority populations.

Anhedonia, a component of depression, results in decreased pleasure from enjoyable activities and difficulty engaging in them. No clear correlation exists between depression and endogenous opioid (EO) reward systems, which has analgesic effects and has been linked to social bonding. This study aims to deepen understanding about prosocial reward mechanisms and anhedonia. In a within subjects design, participants underwent pain sensitivity calibration followed by a simple game to earn money for themselves and someone they are close to. After earning each monetary reward we tested participants' pain sensitivity. We expect that after playing for someone they care about (i.e., prosocial behavior), participants' pain sensitivity will decrease due to EO release. Also, we predict that participants with depressive symptoms will report higher pain sensitivity, and that we may find depression to moderate the relationship between reward type and pain sensitivity due to the opioid reward system’s preservation in depression such that depressed people have more of a difference in pain sensitivity between prosocial and self oriented rewards. These findings will help clarify if prosocial rewards are preserved in people with depression due to the involvement of the opioid system.

The connection between the gut microbiome and the brain plays a significant role in cognitive development. Little is known about its influence on decision making particularly in the context of caregiving related early adversities (crEA). Though there is evidence linking the microbiome and the reward system in adults, the matter of how gut microbiome may influence reward learning throughout development and whether this is impacted by crEAs is yet to be explored. Thus, this study examines the relationship between gut microbiome composition and reward learning behaviors in middle childhood and adolescence. Based on prior research regarding the impact of early life adversity (ELA) on decision evaluation, we hypothesize that reduced alpha diversity of the gut microbiome will be predictive of decrease in reward learning behaviors and that this relationship will be moderated by crEA exposure. A cross-sectional design will be used to analyze alpha diversity in the gut microbiome in samples from both typically-developing and adversity-exposed participants. Furthering our understanding of this relationship between alpha diversity in gut microbiome and reward behaviors post crEA exposure could help identify at-risk youth and potentially inform interventions.