In the mid-1950s, Egon Brunswik distinguished between two modes of human cognition: intuitive judgment based on perception, and analytic thinking based on explicit reasoning. This distinction has recently been formalized by means of a computational model called the “Precise/Not Precise” (PNP) model (Sundh et al., 2021). PNP has been shown to differentiate intuitive from analytic processing across tasks such as multiple-cue learning, monetary lotteries, and perceptual judgments (Sundh et al., 2021; Collsiöö et al., 2023a; 2023b; Sundh et al., 2025). A key limitation of the PNP model is that it only considers a single analytic rule at a time. I therefore propose an extended “Piecewise Precise/Not Precise” (P-PNP) model that allows for rule switches within a task, capturing what we colloquially call a “change-of-mind”. Testing this model requires model and parameter recovery studies. Because the P-PNP model is prone to local minima, I must rely on large-scale grid search which makes it computationally heavy to fit. Fitting P-PNP to 1,000 participants already required several days of runtime on my desktop PC. My planned model recovery involves approximately 250,000 simulated participants (but can be scaled down, if need be), each requiring optimization over multiple parameter settings. The workload is well beyond the capacity of any resources that are currently available to me. I therefore request access to NAISS compute resources to perform large-scale model recovery of the P-PNP model. Hopefully, this project will contribute both to computational/cognitive modeling methodology and to the field of cognitive psychology.