Publications
Peer-reviewed research on attention, decision-making, and gaze.
Six papers, published or under review. Each is peer-reviewed and, where possible, backed by open data and code - because work that is both reviewed and reproducible is a stronger signal than either alone.
Confidence tracked performance when users knew they could be seen, and barely tracked it when they thought they were acting privately.
Gerlofs, D. J., & Kingstone, A. (2026). Metacognition and the dual function of social gaze. Journal of Experimental Psychology: General, 155(8), 1967.
So what for product: Self-reported confidence is a usable signal in some contexts and a misleading one in others. Knowing which context you are in is what makes it safe to build a metric on.
Belief about a source's intent shifted decisions by about 30 percentage points - more than the information the source actually provided.
Gerlofs, D. J., Roberts, K. H., & Kingstone, A. (2025). Perceived intent drives gaze interpretation. Attention, Perception, & Psychophysics, 87, 2323–2331.
So what for product: Credibility signals, source badges, and ranking positions can move decisions more than the content they label - an effect products rarely measure, and one that differs across user segments.
Gaze, gesture, and speech work as a single coordination system - remove one channel and the others compensate.
Haraped, L., Gerlofs, D. J., Huang, O. C. H., Hickling, C., Bischof, W. F., Sachse, P., & Kingstone, A. (2025). Coordinating attention in face-to-face collaboration: The dynamics of gaze, pointing, and verbal reference. Cognitive Science, 49(10), e70123.
So what for product: Real-time collaborative interfaces that reduce coordination to a single signal (e.g., cursor position) miss the multimodal system people actually rely on - especially on shared screens and in XR.
Awareness that your gaze might be seen shifts your decisions as much as interface salience does.
Gerlofs, D. J., Roberts, K. H., Anderson, N. C., & Kingstone, A. (2022). Eye spy: Gaze communication and deception during hide-and-seek. Cognition, 227, 105209.
So what for product: In tools where activity is visible - collaborative platforms, social commerce, presence-aware apps - users respond not just to what they see, but to what they imagine others see about them.
Confidence ratings can track how confident someone felt a moment ago rather than the evidence in front of them.
Zheng, Y., Wang, L., Gerlofs, D. J., Duan, W., Wang, X., Yin, J., … & Kwok, S. C. (2022). Atypical meta-memory evaluation strategy in schizophrenia patients. Schizophrenia Research: Cognition, 27, 100220.
So what for product: A confidence rating is not always a read of the evidence - it can carry over from the judgment before it. Anything that treats self-reported certainty as a signal should account for that drift.
Search performance collapses past a visual-complexity threshold rather than degrading gradually - from 82% to 20% accuracy in the condition shown, against a 1-in-9 chance floor.
Gerlofs, D. J., et al. Visual complexity and search performance. Under review.
So what for product: Knowing where the complexity threshold sits for your product gives design teams an empirical line to defend - the difference between a heuristic ("avoid clutter") and a criterion.