Publications
Lesbian, gay, bisexual, transgender, queer, intersex, and asexual (LGBTQIA+) students continue to face violence, exclusion, and barriers at school, including in STEM education. A key underexamined factor in diversity, equity, and inclusion (DEI) efforts is the content of the life science curriculum, which is uniquely positioned to reinforce or refute bioessentialist, binary, and heteronormative biases. Outdated science curricula not only conflict with current scientific evidence but can also perpetuate beliefs that contribute to sexism and LGBTQIA+ marginalization. To address this, we designed four gender and sexual diversity (GSD)-inclusive biology activities, aligned with NGSS standards, and informed by inclusive curriculum frameworks. Using a mixed-methods approach, we studied 127 high school students who participated in two or more inclusive biology activities. Surveys conducted before and after implementation showed significant reductions in essential, binary beliefs about sex and gender, and increases in affirming attitudes toward sex and gender diversity. Interviews conducted after implementation further revealed differences between LGBTQIA+ and straight students' conceptualizations of biological sex. Our findings demonstrate that even brief curriculum interventions can shift student attitudes, although we hope future studies will explore the impact of sustained interventions. Updating life science instruction is essential for educational equity and scientific accuracy.
Evaluating the originality of visual images poses enduring challenges for creativity assessment. Automated scoring using AI models has proven effective in the verbal domain, yet key questions remain about evaluating visual creativity and understanding how models arrive at their ratings. The present research asks whether multimodal large language models (LLMs) can serve as judges of visual creativity zero-shot (without any fine-tuning or examples of human ratings) and whether their "reasoning" output offers an interpretable window into their evaluation process. We tested six multimodal LLMs (Gemini 3 Flash, Gemma 4 31B IT, GPT-5.4 Mini, GLM-5v Turbo, Kimi K2.5, and Qwen 3.6 Plus) on 992 AI-generated images (based on human-written prompts) and 1,500 hand-drawn sketches scored for creativity by human raters. In Study 1, all models showed substantial alignment with human creativity ratings on both datasets (r = .57-.68 on AI-generated images; r = .29-68 on sketches). In Study 2, we analyzed the step-by-step reasoning processes of three LLMs evaluating the same images and drawings. Although reasoning made model evaluations interpretable -- showing what they attend to, how they balance originality vs. quality, and how they justify their ratings -- reasoning did not improve alignment with human ratings. In sum, our findings indicate that multimodal LLMs can match human judgments of visual creativity without any additional training, and that their reasoning reveals how AI models evaluate creativity. An open scoring app implementing this pipeline is available at https://review-visual-eval-scoring.hf.space.
Children's spatial skills are linked to their later achievement in science, technology, engineering, and mathematics (STEM), making this a critical area of research in cognitive and developmental science. Here, we examined whether a common measure of spatial skills should be treated as a single construct (i.e. broad spatial assembly skills) or as representing two unique constructs, 2-dimensional (2D, like arranging pieces on a flat surface) and 3-dimensional (3D, like building a block tower) spatial assembly skills, using a three-pronged approach. With a sample of 548 preschool-aged children (M = 4.32 years, SD = 0.59 years; 50% female) from three interventions in the United States, we conducted a confirmatory factor analysis of the measure, compared children's performance by trial type, and investigated relations between spatial skills and other cognitive skills. Results showed that the measure is more accurately captured by a two-factor structure that, respectively, represents the constructs of 2D and 3D spatial assembly skills. Further, children perform better on 2D spatial assembly trials than 3D trials. When controlling for performance on the other trial type, children's 2D and 3D spatial assembly performance are also uniquely related to other cognitive skills. Specifically, performance on 2D trials was uniquely related to numeracy whereas performance on 3D spatial assembly trials was uniquely related to executive function, numeracy, geometry, and math language skills. These findings suggest that 2D and 3D spatial assembly skills are unique constructs, advancing research on spatial skill development and providing guidance for measuring children's spatial skills.
ABSTRACT In this mixed methods study, we carried out quantitative analyses in order to examine how 80 beginning elementary teachers' enactment of ambitious mathematics instruction seems to be affected by their levels of mathematics teaching self‐efficacy, their perceptions of collective responsibility among teachers at their schools, and feedback they received from principals, instructional coaches, and mentor teachers. We observed each teacher teach up to three mathematics lessons during their first year of teaching. We used the Mathematics Scan (M‐Scan) classroom observation instrument to analyze observations of teachers' implementation of ambitious mathematics instruction. We found that when novice teachers perceived high levels of collective responsibility, they were more likely to receive higher M‐Scan scores for lesson coherence and representations and higher overall M‐Scan scores. In addition, when beginning teachers had higher levels of mathematics teaching self‐efficacy, they were more likely to receive higher M‐Scan scores for lesson coherence. At the same time, we found that feedback from principals, instructional coaches, and mentor teachers was negatively associated with novices' M‐Scan scores for representations and overall M‐Scan scores. The qualitative portion of this study explored reasons for the significant negative association that we found between feedback and beginning teachers' enactment of ambitious mathematics instruction.
Human behavior is fundamentally generative: People create pictures, write stories, compose music, and engage in conversation. Traditional approaches in psychology and cognitive science have not focused on this open-endedness, instead favoring more constrained task settings that admit a limited set of outcomes. Although those approaches have been fruitful, new approaches might be needed to develop a unified understanding of the generative, open-ended behaviors that are so emblematic of human cognition. This article demonstrates the value of generative behaviors as targets for cognitive modeling by providing rich behavioral data that reveal how multiple cognitive processes coordinate. Drawing production serves as a case study illustrating this approach, showing how perception, memory, social inference, and motor control coordinate flexibly on the basis of communicative context. Recent advances in generative artificial intelligence offer both new tools for modeling open-ended human behavior and new comparative targets for understanding similarities and differences between human and machine intelligence. However, applying these tools effectively might require new experimental paradigms, larger data sets, and careful consideration of what mechanistic correspondence between models and human cognition is necessary for scientific progress. Embracing the open-ended nature of human thought and behavior poses methodological challenges but offers a promising path toward understanding the most distinctive aspects of human intelligence.
We explore the use of small language models (SLMs) for automatic question generation as a complement to the prevalent use of their large counterparts in learning analytics research. We present a novel question generation pipeline that leverages both the text generation and the probabilistic reasoning abilities of SLMs to generate high-quality questions. Adopting a “generate-then-validate” strategy, our pipeline first performs expansive generation to create an abundance of candidate questions and refine them through selective validation based on novel probabilistic reasoning. We conducted two evaluation studies, one with seven human experts and the other with a large language model (LLM), to assess the quality of the generated questions. Most judges (humans or LLMs) agreed that the generated questions had clear answers and generally aligned well with the intended learning objectives. Our findings suggest that an SLM can effectively generate high-quality questions when guided by a well-designed pipeline that leverages its strengths.
Qualitative video analysis is a foundational yet labor-intensive method in the learning sciences, particularly for investigating the multimodal and collaborative dimensions of learning. This paper introduces GAIVA (Generative Artificial Intelligence for Video Analysis), a novel framework that integrates multimodal large language models (MLLMs) and large language models (LLMs) to support scalable, consistent, and theory-informed video analysis. GAIVA comprises four sequential phases: (1) video preprocessing, (2) MLLM-based behavioral transcription, (3) human-AI collaborative codebook development grounded in theoretical constructs from computer-supported collaborative learning (CSCL), and (4) LLM-based automated coding. Drawing on a case study to analyze simulation-based collaborative science learning, we demonstrate that GAIVA generates rich behavioral transcriptions, enables the construction of stable and theoretically grounded coding schemes, and achieves inter-rater reliability comparable to human annotation. Notably, findings illustrate GAIVA's capacity to support both deductive alignment and inductive discovery. The framework surfaces a behavioral construct--Peripheral Engagement--that is consistent across data batches but absent from existing frameworks. We contribute a systematic and replicable framework for integrating generative AI into qualitative video research, enabling large-scale, multimodal analysis of collaborative learning while preserving analytical rigor, theoretical coherence, and interpretive transparency.
Mobile crowdsensing (MCS) is a distributed sensing architecture that utilizes existing sensors on mobile units (MUs) to perform sensing tasks. A mobile crowdsensing platform (MCSP) publishes the sensing tasks and the MUs decide if they want to participate in their execution in exchange for money. The MCS system is characterized by its dynamic nature in which the task requirements, the MUs' availability, and their available resources change over time. The MUs aim to find an efficient task participation strategy to maximize their income while the MCSP focuses on maximizing the number of completed tasks. As optimal task participation strategies require perfect non-causal information about the MCS system, which is unavailable in realistic scenarios, the main challenge in MCS is to find an efficient task participation strategy for the MUs under incomplete information. To this aim, a novel fully decentralized federated deep reinforcement learning algorithm, termed FDRL-PPO is proposed. FDRL-PPO enables every MU to learn its own task participation strategy based on its experiences, available resources, and preferences, without relying on perfect non-causal information about the MCS system. To replenish their batteries, the MUs rely on energy harvesting. As a result, their available energy varies over time, leading to varying availability and fragmented learning experiences. To mitigate these challenges, the proposed approach leverages federated learning, enabling MUs to collaboratively improve their models without having to share private raw data like their own experiences. By exchanging only learned models, MUs collectively compensate for individual limitations, and find more scalable, robust, and efficient task participation strategies. Comprehensive evaluations on both synthetic and real-world datasets show that FDRL-PPO consistently outperforms benchmark algorithms in terms of task completion ratio, fairness in task completion, energy consumption, and number of conflicting proposals.
Using a sample of 883 Latiné engineering students, the present study explored the measurement characteristics of the Negative Outcome Expectations Scale for Engineers (NOES-E), applying advanced psychometric techniques, including bifactor and the exploratory structural equation modeling (ESEM) framework, to ascertain the scale's most suitable factor structure. Additionally, measurement invariance across Latino and Latina participants was tested to investigate if the scale operates equally across gender groups and whether the latent mean scores are invariant. The results suggested that the bi-factor ESEM provided the best model fit when contrasted with other possible factor structures. The findings generally highlighted the general factor as reflecting the negative outcome expectations, suggesting that the use of total score is advised over the use of individual subscale scores. Findings from the measurement invariance testing across men and women indicated the NOES-E score can be compared between both groups.


