Researchers Jeroen Olieslagers and Wei Ji Ma have introduced ROSETTE, a new cognitive framework designed to model human planning through mental simulations. In human cognition, one proposed planning mechanism is the mental simulation, or rollout, of potential action sequences to evaluate future outcomes. Although rollout-based models have achieved great success in artificial intelligence, cognitive models that describe how people use rollouts remain scarce because they are difficult to fit to behavioral data.
Existing methods rely on computationally intensive simulations that result in approximate, noisy estimates of latent planning parameters. To address these limitations, the authors developed ROSETTE, which stands for Rollouts with Stochastic Early Termination and Tractable Estimation. The framework provides analytically tractable, simulation-free model fitting specifically tailored for finite goal-directed tasks with absorbing goals.
Under this formulation, ROSETTE treats planning as a sequence of mental simulations formalized as an absorbing Markov chain. In this process, the model deterministically accumulates analytically computed expected rollout evidence and performs a variable number of policy-update cycles.
The researchers fit ROSETTE to experimental data from human participants across two multi-step planning tasks, finding that it well accounted for a rich set of summary statistics of behavior. By making this specific class of absorbing-goal rollout models tractable, the framework provides a testable account of how people may simulate and evaluate possible futures.
Subject of Research: Computational modeling of human planning and mental simulations
Article Title: Rollout models of human planning with exact action probabilities
Article References: Olieslagers, J., & Ma, W. J. (2026). Rollout models of human planning with exact action probabilities. PLOS Computational Biology, 22(11), e1014945. https://doi.org/10.1371/journal.pcbi.1014945
Image Credits: AI Generated
DOI: 10.1371/journal.pcbi.1014945
Keywords: ROSETTE, human planning, mental simulation, rollout models, cognitive modeling, absorbing Markov chain, simulation-free fitting, goal-directed tasks, absorbing goals, policy-update cycles, multi-step planning, computational biology, behavioral data, latent planning parameters, action sequences, expected rollout evidence, Jeroen Olieslagers, Wei Ji Ma
Cite Scienmag News
Drew Townsend. (November 20, 2026). ROSETTE Framework Models Human Planning Simulations with Analytically Tractable Rollouts. Scienmag. https://scienmag.com/rosette-framework-models-human-planning-simulations-with-analytically-tractable-rollouts/
Drew Townsend. "ROSETTE Framework Models Human Planning Simulations with Analytically Tractable Rollouts." Scienmag, 20 November 2026, https://scienmag.com/rosette-framework-models-human-planning-simulations-with-analytically-tractable-rollouts/. Accessed 20 November 2026.
Drew Townsend. "ROSETTE Framework Models Human Planning Simulations with Analytically Tractable Rollouts." Scienmag. November 20, 2026. https://scienmag.com/rosette-framework-models-human-planning-simulations-with-analytically-tractable-rollouts/








