Previous choices reduced mean prediction loss for later runs in 108 selected NARPS adults under fixed logistic models.
Previous recorded choices improved later-run prediction in 108 selected NARPS adults under fixed logistic models. The models controlled current and previous money offers and person coefficients. Runs 1 and 2 supplied training choices. Runs 3 and 4 supplied 13,447 eligible test choices. Mean test log loss decreased from 0.248136 to 0.244205 nats per choice. The equal-person gain was 0.003931 nats, or 1.58%. The 95% participant bootstrap interval was 0.002216 to 0.006058 nats. All 2,000 primary samples fitted the group centers and both models again. Five planned checks gave positive overall gain intervals. The equalRange group gain interval included zero. The study reused one selected cohort and tested prediction during an observed choice sequence. It did not test a neural or causal mechanism. The package preserves 437 unchanged source files, fixed source hashes and URLs, all fit coefficients, bootstrap samples, trial links, predictions, source code, and reproduction instructions. Independent fits, source checks, a copied-person check, a future-label guard, and a full repeat run passed.
risky choice · choice history · logistic regression · prediction · NARPS · OpenNeuro · participant bootstrap
References
3- usesPerson slopes reduced test log loss against a person bias model in 108 selected NARPS adults under fixed logistic models sci:sha256:23e1e5766dd32b526e64226ad487117bdf43115ac61fb8114fad718b903082df
This study reused the unchanged NARPS source archive, source manifest, source metadata, loader, group-center solver, and retrieval recipe. It added joint models and independent checks to test previous choice prediction in the same cohort.
- usesdoi:10.18112/openneuro.ds001734.v1.0.5
The analysis used the 1.0.5 NARPS behavioral source files, identified by their recorded fixed-commit URLs and hashes.
- supportsdoi:10.1038/s41597-019-0113-7
The data paper supplies the reported cohort selection facts and the possible reversed response-button flag for sub-056.
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesNext choices reduced retrospective log loss in 108 selected NARPS adults under fixed logistic models
This study used the prior NARPS choice-history result to set the new question. It used the same source manifest and source records. It adapted the reviewed source parser, ridge fit, and joint solver, and retained the retrieval code. It adds matched past, future, and both-choice comparisons and an exact causal reference.