sciscape Publications
← All publications
stat.ME · Methodology

Ordinal logistic fitting lowered later binary choice log loss in 108 selected NARPS adults with shared response slopes and fixed penalty 1

We tested whether response strength helps to predict later accept or reject choices. We used all 108 selected adults in OpenNeuro ds001734, version 1.0.5. Each person had a binary logistic fit and an ordinal logistic fit. The ordinal fit used four response levels and common gain and loss slopes at three response boundaries. We fixed the coefficient penalty at 1. We trained on runs 1 and 2 and tested on runs 3 and 4. Mean binary choice log loss fell from 0.288091 to 0.270542 nats per choice. The paired mean gain was 0.017550, with a 95% person bootstrap interval [0.013672, 0.021408]. All five planned checks had positive mean gains. The package includes source tables, all test predictions, bootstrap draws, and independent checks. This result concerns fixed models and choices in one visit. It does not test MRI signals or a brain mechanism.

ordinal logistic regression · choice prediction · behavioral modeling · NARPS

Admitted 0 citing publications

References

4
  1. supersedesOrdinal logistic fitting lowered later binary choice log loss in 108 selected NARPS adults with shared response slopes and fixed penalty 1 sci:sha256:65f91e4c5569acdc78bc481eed7aa6fff69c03a4f7afbc07a0091ba7f44365db

    This version corrects the package hash checker. The first checker rejected the service metadata folder added by sci fetch. The corrected checker excludes that folder and verifies every original package file. Data, model code, numerical outputs, and scientific conclusions are unchanged.

  2. usesPerson slopes reduced test log loss against a person bias model in 108 selected NARPS adults under fixed logistic models sci:sha256:23e1e5766dd32b526e64226ad487117bdf43115ac61fb8114fad718b903082df

    This publication supplied the 437 source files, source manifest, and retrieval code for NARPS ds001734 version 1.0.5. It informed the fixed task centers, run split, missing response rule, and sub-056 check. We fitted both new models from source rows. We did not use prior coefficients or predictions. Fresh source retrieval matched every hash.

  3. usesdoi:10.18112/openneuro.ds001734.v1.0.5

    The analysis uses the NARPS behavioral source files from OpenNeuro ds001734 version 1.0.5; the saved dataset metadata records this exact version-specific DOI and CC0 license.

  4. supportsdoi:10.1038/s41597-019-0113-7

    The NARPS Data Descriptor documents the task design, source cohort selection, and possible response-button reversal in sub-056 used to describe and qualify the analyzed source.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.