📚 Research Digest

Finding: Curiosity predicts both proactive information seeking (like question asking) and attentive information seeking (like focused listening) in Nepalese first-graders, but these relationships are moderated by students' social confidence and executive function skills.

Why it matters: This reveals how individual differences in cognitive and social skills shape learning behaviors, providing insights for designing educational interventions and understanding cultural variations in classroom engagement.

Method: Measured curiosity through willingness-to-pay and exploration breadth/depth scores in 701 Grade 1 students, examining how social confidence and executive function moderate the curiosity-information seeking relationship.

Jargon: Executive function = cognitive skills like working memory, attention control, and cognitive flexibility that help manage thoughts and actions.

Finding: Researchers are using Bayesian optimization in an adaptive experiment to design autonomous vehicle deceleration profiles that communicate yielding intent to pedestrians, treating vehicle motion as an implicit human-computer interface at crosswalks.

Why it matters: This demonstrates how to design implicit communication channels in human-AI systems and provides a methodological framework for optimizing continuous interaction behaviors through human-in-the-loop experimentation.

Method: Uses a minimum-jerk model to parameterize deceleration with only 2 parameters, then employs Bayesian optimization to sequentially propose speed profiles in an immersive pedestrian simulator based on safety and efficiency ratings.

Jargon: Minimum-jerk model (MJM) - mathematical model that generates smooth motion profiles by minimizing jerk (rate of change of acceleration); Bayesian optimization - sequential optimization method that uses probabilistic models to efficiently find optimal solutions.

Finding: Stress increases alcohol choice by ~10 percentage points in daily life by selectively biasing evidence accumulation toward alcohol, but contextual factors (presence of strangers, time-of-day, alcohol cue visibility) gate whether this bias translates into actual drinking behavior.

Why it matters: Demonstrates how cognitive biases in decision-making interact with environmental context to influence behavior, providing a framework for understanding when psychological tendencies translate into real-world actions.

Method: Used ecological momentary assessment with 250 at-risk drinkers over 14 days, combining self-reports with a cognitive choice task at each prompt, then applied drift diffusion modeling to decompose decision parameters.

Jargon: Drift diffusion modeling - computational technique that breaks down decision-making into components like evidence accumulation speed and decision thresholds; Ecological momentary assessment (EMA) - research method capturing real-time data in participants' natural environments.

Finding: A smartphone app called HippoCamera that helps users create and review multimodal memory cues significantly improved episodic memory recall and psychological well-being in a person with early dementia, demonstrating that digital interventions can effectively support recent memory formation.

Why it matters: This shows how thoughtful HCI design can create meaningful therapeutic interventions, illustrating principles for designing accessible technology that supports cognitive function and demonstrates the potential for AI-assisted memory tools.

Method: Single-case study with 11-week intervention comparing recalled detail between reviewed vs. non-reviewed recorded events, plus pre/post psychological assessments and qualitative feedback.

Jargon: HippoCamera - smartphone app creating multimodal memory cues; episodic memory - memory for specific personal experiences with contextual details.

Finding: NeuroGym is an open-source Python platform that provides standardized neuroscience tasks for training and testing artificial neural networks, with a shared interface that supports both reinforcement learning and supervised learning approaches.

Why it matters: This tool could facilitate research on AI-human collaboration and decision-making by providing standardized cognitive tasks that allow researchers to compare artificial and human performance on the same behavioral paradigms.

Jargon: Gymnasium - a standard API for reinforcement learning environments; ANN - Artificial Neural Network.

Finding: FalseResMem is the first neural network designed to predict which images will trigger false memories in visual recognition tasks, achieving consistent performance (r=0.48) and generalizing across different image categories including objects, scenes, and faces.

Why it matters: Understanding systematic biases in visual memory has direct applications for HCI design, particularly in interface elements that rely on visual recognition, and provides insights into cognitive biases that affect human-computer interaction.

Method: Combines ImageNet-pretrained ResNet50 features with a retrained AlexNet-like architecture, trained on large-scale object memory datasets and validated through 10-fold cross-validation.

Jargon: False alarm rate (FAR) = frequency of incorrectly identifying novel images as previously seen; false memories = mistaken recollections of events that didn't occur.

Finding: Treatment-seeking individuals and mental health professionals view Ecological Momentary Assessment (EMA) and Ecological Momentary Interventions (EMIs) as valuable for providing real-time, tailored support for non-suicidal self-injury, leading to a proposed traffic-light model that adjusts interventions based on changing risk levels.

Why it matters: This demonstrates how mobile technology can enable dynamic, personalized behavioral interventions that adapt to user states in real-time, relevant for designing responsive HCI systems and understanding human-AI collaboration in mental health contexts.

Method: Semi-structured interviews with 30 individuals and 14 professionals after a four-week EMA period, analyzed using reflexive thematic analysis.

Jargon: EMA = real-time data collection via mobile devices; EMI = just-in-time digital interventions delivered through mobile technology; NSSI = deliberate self-harm without suicidal intent.