The 95 percent interval included zero for a fixed older history forecast gain in one ds005364 split
In one fixed OpenNeuro ds005364 version 1.0.0 split, adding three summaries of choices and feedback older than five trials to a model with five recent trials and the current displayed reward ratio changed mean test log loss from 0.523497 to 0.522923 nats per choice. The estimated gain was 0.000574 nats per choice (0.110% of reference loss), with a paired 95% bootstrap interval from -0.001248 to 0.002596 across 2,000 draws. The interval includes zero, so this test did not establish a forecast benefit; it also does not establish equivalence or that people use only five trials of history. The result applies to this source version, fixed split, target period, and model pair.
choice forecasting · reinforcement learning · choice history · logistic regression · bootstrap · reversal learning
References
3- usesCurrent reward ratios did not lower test log loss with five prior trials in a fixed ds005364 split sci:sha256:de3a6375338d8513cf8a92ea53a65c5bd976fe367984dee9e0c87eb226efcef7
We used the unchanged event parser and fitting code, source bytes and manifest, fixed 48-person fit and 47-person test split, and prior reference coefficients. These supplied the reference model and enabled a checked older history comparison.
- usesdoi:10.18112/openneuro.ds005364.v1.0.0
This study analyzed task event files from OpenNeuro ds005364 version 1.0.0; the pinned source files and hashes are recorded in the package.
- doi:10.1038/s41598-024-69772-y
This paper describes the source study task and recruitment methods.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.