Finding: This meta-analysis of 17 studies found that behavioral interventions can reduce ultra-processed food intake, with the most effective interventions combining health information with self-regulatory techniques like goal setting, self-monitoring, feedback, and action planning.
Why it matters: The identification of effective behavior change techniques (particularly goal-setting and self-monitoring) provides evidence for behavioral science principles that could apply to other domains of behavior change and goal achievement.
Method: Systematic review using the Nova food classification system, analyzing behavior change techniques and delivery characteristics across trials and pre-post studies from 2009-2025.
Jargon: Ultra-processed foods (UPFs) - industrially manufactured foods with multiple additives; Nova classification - system for categorizing foods by degree of processing; BCTs - behavior change techniques.
Ariana Modirrousta-Galian, Tina Seabrooke, Yaniv Hanoch et al. (5 authors) β’ PsyArXiv
Finding: An inductive learning intervention that trains people to classify news headlines as true or false improved their ability to discern news veracity, but adding performance-contingent badges backfired when combined with easy-to-hard training due to declining performance awareness undermining motivation.
Why it matters: This demonstrates how gamification elements can interact with learning difficulty progression in counterintuitive ways, providing crucial insights for designing educational interventions and understanding motivation in learning contexts.
Method: Three preregistered experiments tested inductive learning with immediate feedback, comparing different gamification approaches (badges) and training progressions (easy-to-hard vs. hard-to-easy).
Jargon: Inductive learning (IL) = learning categories by observing and classifying examples; Bayesian evidence = statistical approach measuring strength of evidence for hypotheses.
β Pre-registered | β Includes replication β οΈ No control group mentioned
Jamie Kiltie, Liam Satchell, Sophie Cawley et al. (5 authors) β’ PsyArXiv
Finding: In mock investigative interviews using a round-robin design (10 interviewers Γ 23 witnesses), researchers found weak agreement between interviewers and witnesses on rapport ratings, and no significant relationships between coded behaviors and perceived rapport or between rapport and interview performance measures.
Why it matters: This challenges assumptions about behavioral indicators of rapport and highlights the subjective nature of interpersonal connection assessment, which is relevant for HCI systems that attempt to detect or facilitate rapport in human-computer or computer-mediated interactions.
Method: Round-robin methodology where each interviewer conducted interviews with each witness about different mock crime videos, followed by behavioral coding of recorded interviews and rapport ratings from both parties.
Jargon: Round-robin design = experimental setup where every participant in one group interacts with every participant in another group.
β οΈ Very small sample (n=10) | No control group mentioned | Single study (not replicated)