Changing money utility exponents from 0.5 to 2 increased fitted hyperbolic rates by a geometric mean factor of 9.57 in 118 ds002843 v1.0.1 adults with a fixed 2 percent lapse

We tested fixed money utility assumptions in 118 selected adults from OpenNeuro ds002843 version 1.0.1. We fitted visit 1 choices and predicted all 14,140 valid visit 2 choices. The primary model used hyperbolic discounting, a logistic choice rule, and a fixed 2 percent uniform lapse. Changing the utility exponent from 0.5 to 2 increased fitted rates by a geometric mean factor of 9.57. The 95 percent paired person bootstrap interval was 8.69 to 10.47. Rate rank correlation was 0.99901. The equal-person later log loss difference was 0.01458 nats per choice, with interval -0.00078 to 0.02804. This interval does not establish equal prediction performance. The secondary accuracy difference was -0.957 percentage points. The rate shift persisted with wider bounds. Prediction loss depended more on the lapse assumption. This is a sensitivity test of a previously studied cohort. It does not estimate true utility or test new people, new offers, neural activity, or treatment effects. The package preserves source data, provenance, code, fits, every prediction, bootstrap draws, figures, and repeat instructions. All 1,175 source files matched a fresh public clone. Independent scalar checks passed. A complete numerical repeat matched all twelve result files by SHA-256.

intertemporal choice · discounting · utility curvature · hyperbolic discounting · model sensitivity · behavioral neuroscience · OpenNeuro ds002843