Four 2026 research papers

The CARES lab has published four papers thus far in 2026. The first paper is the product of many years of work, and we are excited to finally get this work published.

The Influence of Individual Characteristics on Children’s Learning with a Social Robot versus a Human

(Langer et al., 2026)

This paper describes the study that was the beginning of our work with children building tangram puzzles and was the basis of our CAREER proposal. The study involved 32 children interacting with a human and a social robot to complete tangram puzzles. This paper reports on how the characteristics of the children related to how they interacted with each agent type, including how long they took to complete the task and where the children fixated their gaze.

Student/Faculty Partnerships to Teach Computing Ethics Beyond the Computer Science Classroom

The paper reports on our experience using curriculum co-creation to both design a whole new computing ethics course and develop computing ethics modules being integrated into courses across the liberal arts curriculum. Student/faculty partnerships led to factuly and students learning from each other and resulted in engaging learning activities and a broadening of curricular topics.

(Muchuwa & Wilson, 2026)

Salience of Surface versus Higher-level Properties in Spatial Comparison at Different Levels of Scene Similarity

This study examined how adults reason about images of young children’s tangram puzzles. Participants were presented with an incomplete puzzle and indicated which puzzle they think the child was trying to build and why. In analyzing the partcipants’ explanations, we found that surface features like shape and color were most prominent when the puzzle was almost complete or barely started, and higher-level properties like the semantics of the puzzle were most commonly referenced in the intermediate stages of the puzzle. This suggests that the reasoning processes involved in spatial comparisons might shift depending on how similar spatial scenes are.

(Vu et al., 2026)

Describing, Adapting, and Executing Behaviors across Social Robot Platforms

This paper introduced the Social Action Description Language, which is used to translate simple textual descriptions of a robot’s social behavior into a behavior executed on a robot. We demonstrated that our behavior generation approach is able to adapt to two robot platforms, showing that our LLM-based strategy can adapt to the capabilities and affordances of a particular robot platform.

(Mealey & Wilson, 2026)