Challenging the mechanism for the implicit association test
Shows that response caution, rather than automatic association alone, explains much of the IAT's signature effect across a large set of topics.
Managing control over
individual and
collective decisions.
I study how much control people have over the expectations, emotions, and beliefs that shape social judgment, decision-making, and organizational behavior, using computational tools to separate what can be changed from what quietly drives behavior outside awareness.
Kyle J. LaFollette, PhDUniversity of Chicago Booth School of Business
Roman Family Center for Decision Research
Roman Family Center for Decision Research
April 2026Nature Human Behaviour paper on the mechanism behind the IAT.
February 2026Communications Psychology paper on learning transfer across task, person, and experience.
September 2025Scientific Reports paper on crowd emotion judgment and racial composition.
July 2025Joined Chicago Booth as Principal Researcher in the RF-CDR.
July 2025PNAS paper on nonlinear reinforcement-learning dynamics.
April 2026Nature Human Behaviour paper on the mechanism behind the IAT.
February 2026Communications Psychology paper on learning transfer across task, person, and experience.
September 2025Scientific Reports paper on crowd emotion judgment and racial composition.
July 2025Joined Chicago Booth as Principal Researcher in the RF-CDR.
July 2025PNAS paper on nonlinear reinforcement-learning dynamics.
A computational social psychology of control.
Methodologically, I combine behavioral experiments, diffusion decision models, reinforcement learning models, social network analysis, experience sampling, dynamical systems approaches, and theory-constrained equation discovery algorithms, decomposing behavior into the parts that are controllable and the parts that are not. I am especially interested in using computational tools to build interpretable models of how emotional and cognitive processes unfold together over time, and to identify the control variables - inputs in system dynamics and control theory - that shape these processes for achieving desired performance.
Implicit expectations
Emotion dynamics
Decision diffusion models
Reinforcement learning
Social networks
Equation discovery algorithms
Selected publications.
Full list on Google Scholar ↗Task, person, and experiential characteristics drive the transfer of learning
Examines when learning carries forward across contexts and when environments hide the skills people have actually acquired.
Judgment of crowds as emotional increases with the proportion of Black faces
Uses process models to show how racial composition shapes emotion inference in crowds through evidence accumulation.
Data-driven equation discovery reveals nonlinear reinforcement learning in humans
Uses equation discovery algorithms to uncover interpretable, nonlinear rules for how people update expectations after feedback.
Current projects.
These interactive windows give a small, playful glimpse into the problems I study: how leaders structure teams, how people share emotion with AI and one another, how social judgments unfold through decision processes, and how expectations update over time.
Expectations about organizational dynamics
Rearrange a network of workers
Drag workers closer together or farther apart. Click a worker to change their affective tendency and see how leaders can control information flow, emotional health, and productivity.
positive
steady
concerned
explorer
negative
Teams0
Emotional Health0%
Productivity0%
Technology and social-emotional health
Share an emotion with AI
Share a feeling with an AI friend and watch a monitored emotional trajectory respond, asking how technology can return control over emotional dynamics as they unfold.
Want to share what you're feeling today?
High Happiness
Low Happiness
live
Implicit expectations about emotion
Tune an emotion-inference process
Adjust a diffusion decision model for judging whether a face is emotional, separating controllable response caution from evidence and noise in fast social inference.
Is this person emotional?
DecisionEmotional
RT0 ms
How expectations update
Inspect an assembly line
Report the chance that the next phone will work, then inspect it and observe how control variables shape the way expectations update as the system reveals itself.
0%50%100%
Report your expectation to begin building a personalized updating trajectory.
reported expectation
fitted updating rule
working
defective
Teaching and service.
Graduate
Course instruction for researchers learning computational tools.
Python for Psychological ScienceA graduate methods course focused on Python programming, data workflows, and computational tools for psychological science.
Undergraduate
Courses and seminars in quantitative and computational psychology.
Quantitative Methods in PsychologyAn undergraduate statistics course covering research design, probability, statistical inference, data visualization, and psychological data analysis.
Computational Psychology & NeuroeconomicsA special topics seminar on computational approaches to decision-making, value, learning, and affect.
Statistical consulting
Roman Family Center for Decision ResearchStatistical Consultant supporting behavioral science and marketing faculty and students at Chicago Booth.
Army Research Institute consultingConsulting on cognitive and affective learning and decision-making systems for Army Research Institute Grant #W911NF-22-1-0238.
Service
Journal serviceReviewer for outlets including PNAS, Scientific Reports, Emotion, Journal of Statistical Software, and Frontiers in Psychology.
Equation discovery in behavioral scienceEducating researchers on data-driven equation discovery methods, including talks at the Automated Scientific Discovery of Mind and Brain Workshop at Princeton and the Knowledge Lab at the University of Chicago.
FlexDDM, flexible decision diffusion modeling for all behavioral scientists.
FlexDDM is for students and career academics alike who want to build, validate, fit, and learn about a large variety of decision diffusion models. The package is powerful, using C++ machine code for fitting, but accessible through minimal front-end Python, helpful tutorials, and a point-and-click interface. It is flexible enough to run a wide range of models, including custom models.
Get it from Microsoft
Contact Me.
I am especially interested in collaborations on emotion dynamics, implicit expectations, behavioral process models, digital and AI-mediated sociality, and team or organizational systems.