Fitting utility power to immediate reward choices increased test loss for two delayed rewards in 178 AgeRisk people under bounded hyperbolic discount with 2% lapse
We tested whether a fitted money utility power improved choice prediction in AgeRisk, OpenNeuro ds004711 version 1.0.0. We fitted immediate reward choices and tested choices between two delayed rewards in all 178 eligible people. Under bounded hyperbolic discount with a fixed 2% uniform lapse, fitted power increased test loss by 0.05053 nats per choice. The 95% person bootstrap interval was 0.03276 to 0.06985. It reduced training loss by 0.02607 nats per choice. Power reached a bound in 92 people. The package contains source data, a fixed local plan, code, all test probabilities, six planned secondary checks, fixed power profiles, and independent numerical checks. A complete fresh source repeat gave the same fits and profiles. The result concerns this offer split and model. It does not measure a brain signal or recover true utility.
decision making · delay discounting · utility power · choice prediction · AgeRisk
References
3- usesImmediate reward fits did not establish lower hyperbolic loss: 178 AgeRisk people, two delayed rewards, linear utility, 2% lapse sci:sha256:373f71baa39a7484ea9b4cce794a3dc4c9bb45e809a249d6bb4156446b0153fd
The prior work supplied the first source files, source hashes, and the row selection rule. Its fixed utility results informed the question. The present code fits free utility power and tests a different comparison.
- usesdoi:10.18112/openneuro.ds004711.v1.0.0
The analysis uses delay discounting event and participant files from AgeRisk, OpenNeuro ds004711 version 1.0.0.
- doi:10.3758/s13415-022-01060-6
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesAll reward BART choices lowered run 2 log loss versus cashout balloon choices in 187 AgeRisk people under fixed logistic models
Used its preserved source tag record to identify AgeRisk version 1.0.0 and its fixed commit. We retrieved BART event files directly. We reused no delay discounting task data, code, or results.