Publications
Creativity assessment in science and engineering is increasingly based on both human and AI judgment, but the cognitive processes and biases behind these evaluations remain poorly understood. We conducted two experiments examining how including example solutions with ratings impact creativity evaluation, using a finegrained annotation protocol where raters were tasked with explaining their originality scores and rating for the facets of remoteness (whether the response is "far" from everyday ideas), uncommonness (whether the response is rare), and cleverness. In Study 1, we analyzed creativity ratings from 72 experts with formal science or engineering training, comparing those who received example solutions with ratings (example) to those who did not (no example). Computational text analysis revealed that, compared to experts with examples, no-example experts used more comparative language (e.g., "better/worse") and emphasized solution uncommonness, suggesting they may have relied more on memory retrieval for comparisons. In Study 2, parallel analyses with state-of-the-art LLMs revealed that models prioritized uncommonness and remoteness of ideas when rating originality, suggesting an evaluative process rooted around the semantic similarity of ideas. In the example condition, while LLM accuracy in predicting the true originality scores improved, the correlations of remoteness, uncommonness, and cleverness with originality also increased substantially -- to upwards of $0.99$ -- suggesting a homogenization in the LLMs evaluation of the individual facets. These findings highlight important implications for how humans and AI reason about creativity and suggest diverging preferences for what different populations prioritize when rating.
Abstract Background The ongoing underrepresentation of Latinx students, particularly girls, in engineering calls for new approaches to broaden participation. The present study aims to address this disparity by implementing a culturally responsive gamified engineering activity for migratory high school students, an often‐overlooked subset of Latinx youth. The activity integrated principles of culturally responsive instruction and elements of gamification to teach the engineering design process and boost students' engineering interest and self‐efficacy beliefs. Method A repeated‐measures mixed ANOVA was used to analyze pre‐ and post‐survey measures from 144 Latinx migratory high school students who participated in a summer program. Results After engaging in the activity, Latinx migratory high school students had a significantly greater interest in engineering, and their overall confidence in their ability to academically excel in engineering increased. Students felt more efficacious in their abilities to tinker with devices (i.e., tinkering self‐efficacy) and apply design principles to solve a real‐world problem after the activity (i.e., design self‐efficacy). Notably, the gender gap in general engineering self‐efficacy narrowed considerably, with Latinas demonstrating a larger increase than Latinos. The gender difference in tinkering self‐efficacy was nearly eliminated post the activity. Conclusions The study highlights the importance of creating culturally situated engineering experiences for Latinx migratory high school students that call on their desire to make a difference in their communities, while also making these opportunities accessible to students who have been invisible in the conversation to broaden participation. Approaches implemented in this study may be particularly effective in addressing gender disparities in engineering.
ABSTRACT Two interconnecting challenges contribute to a substantial gap between the supply and demand of talented STEM professionals in the United States workforce. First, is a substantial gap between the rising demographic diversity of the U.S. population and the limited diversity of Americans in STEM education pathways and careers. Second, inadequate attention connecting STEM research and teaching to socially relevant real‐world issues makes STEM education and careers less appealing to young people. Many STEM learning environments do not satisfy young people's core psychosocial needs to belong, feel competent, and engage in work that is socially relevant and personally meaningful. In this article, we leverage social psychological research to highlight evidence‐based solutions to these challenges. We detail actionable steps that can be taken by schools, universities, government agencies, and STEM employers to attract, motivate, and retain new, diverse STEM talent in the United States. By increasing the presence of successful role models, providing career development opportunities for meaningful professional interactions and relationship building, and connecting STEM content to social good, institutions can support young scientists' motivation to pursue and persist in STEM careers while at the same time address social issues of concern to the American people. Many of these evidence‐based actionable steps have stalled or are at risk because of federal actions of the current administration that hurts Americans' well‐being, national progress, and U.S. competitiveness on the global stage.
When first meeting somebody, we're faced with the challenge of “getting to know them.” Why do some questions seem to enable this better than others? In Experiment 1, participants (N=185) evaluated a large bank of conversational questions. We found that questions varied along a reliable latent dimension of interpersonal depth ranging from “small talk” to “deep” questions. In Experiment 2 (N=188), participants answered a subset of these questions along with a number of self-report personality scales. Using a language model to estimate how informative participants' free responses were, we find that individualized personality predictions were more accurate when incorporating free responses; furthermore, responses to deeper questions supported more accurate personality inferences than small talk. Taken together, results suggest not only that responses contained the statistical information necessary to make abstract social inferences, but also that people have accurate intuitions about which conversational topics enable learning about and connecting with others.
Empowering Educational Leaders for AI Integration in Rural STEM Education: Challenges and Strategies
This study explored the role of educational leaders in integrating artificial intelligence (AI) into rural K-12 STEM education. The research focused on the perceptions, strategies, and barriers that educational leaders face in navigating AI integration, particularly in the unique context of rural schools. Drawing on survey and focus group data from rural school leaders, this study examined how rural educational administrators perceive AI's potential and the challenges associated with its adoption. It provided insights into leadership strategies for addressing infrastructural limitations, fostering professional development, and ensuring equitable AI adoption. Ultimately, this research aimed to empower rural educational leaders to harness AI technologies to enhance STEM education.
Numerate adults know that when two sets are equal, they should be labeled by the same number word. We explored the development of this principle—sometimes called 'cardinal extension'—and how it relates to children’s other numerical abilities. Experiment 1 revealed that 2- to 5-year-old children who could accurately count large sets often inferred that two equal sets should be labeled with the same number word, unlike children who could not accurately count large sets. However, not all counters made this inference, suggesting that learning to construct and label large sets may be a necessary but not sufficient step in learning how numbers represent exact quantities. Experiment 2 found that children who extended labels to equal sets were not actually sensitive to exact equality and that they often assigned two sets the same label when they were approximately equal, but differed by just one item (violating one-to-one correspondence). These results suggest a gradual, stagelike, process in which children learn to accurately count, learn to extend labels to perceptually similar sets, and then eventually restrict cardinal extension to sets that are exactly equal. (PsycInfo Database Record (c) 2025 APA, all rights reserved)
We present the first physiology education study that investigates students' understanding of diffusion using a resources framework. Students frequently used "high to low" knowledge resources and rarely coordinated them with "random motion." Of the 23% that included "random motion" resources, the vast majority described random motion starting only after equilibrium. While "high to low" resources are sometimes productive, when students coordinate "random motion" resources, they have tools for a more nuanced understanding of physiological phenomena.
Existing frameworks of task values have called for greater attention to contextual factors that inform decision-making. A critique of this research is a lack of attention to the cultural and situational milieu embedded in motivational theories. Investigating the development of academic values through obligations to the family and self-perceptions of academic ability adds to our understanding of the broader factors that drive student motivation in STEM. This paper explored the roles of family-related obligations associated with the motivational utility values of college STEM majors and the mediating role of academic efficacy. College students at two large ethnically diverse public research institutions shared their experiences in an initial survey as part of a larger longitudinal study on student adversity, motivation, and persistence in STEM (N = 1571, Mage = 20.41). The results revealed that academic efficacy weakens the roles of caregiving obligations on the perceived utility of their STEM major. The role of financial obligations to the family on students' utility values operated indirectly through self-efficacy. Implications of these findings for future research are discussed.
Navigation and episodic memory are foundational cognitive processes that guide future decisions and are often linked to one another due to their behavioral and neural similarities. However, the extent and nature of their interdependence is unclear. We investigated this question using a real-world encoding experience with 8- to 13-year-old children and young adults. Participants were guided on a tour during which they learned the spatial locations of a series of objects and had various experiences with them. Factor analysis of three measures commonly used to assess spatial memory and four tests commonly used for episodic memory revealed two highly correlated latent factors. The first factor, which we termed spatiotemporal structure, included measures that require simultaneously representing all or part of the environment (finding routes, mapping the space symbolically, free recall of the experience, and spatial-temporal recognition). The second factor, which we termed perceptual/factual/locale, included perceptual and semantic recognition along with onsite pointing. These findings show that episodic memory and spatial representations are intertwined in naturalistic encoding and suggest directions for refining models of the function of these two systems.
The process of setting goals and creating plans is crucial for self-regulated learning (SRL), yet students often struggle to construct efficient plans and establish goals. Adaptive learning environments hold promise for assisting students with such processes through adaptive scaffolding. Through the examination of data collected from 144 middle school students, we present a data-driven analysis of students' explicit planning activities in Crystal Island, a narrative game-based learning environment. In this game, students are provided with a planning support tool that aids them in externalizing their science-related goals and plans before putting them into action. We extracted features from their planning tool use and connected them to several SRL processes and problem-solving outcomes. We found that students who engaged with the planning support tool were more likely to successfully complete the learning scenario. To investigate the potential for adaptive support with this tool, we also constructed a student plan recognition framework aimed at predicting students' goals and planned action sequences. This framework uses student gameplay sequences as input and student interactions with the planning tool as labels for both prediction tasks. We evaluated these tasks using six machine learning models and found that all approaches improved on the majority baseline classification performance. We then investigated additional machine-learning architectures and a technique for detecting when students enact all steps in their plans as methods for improving the framework. We demonstrated performance improvement with these enhancements. Overall, results demonstrated that the planning support tool can help students engage in SRL activities and drive adaptive support in real time.


