A trained-pair term increased mean test loss under ridge logistic fits in 38 ds004532 Study 1 participants
In 38 Study 1 evaluation participants from OpenNeuro ds004532 version 1.2.0, a ridge logistic model with a trained-pair term had higher mean test loss than a stimulus-value model (0.36385 vs 0.35835 nats per choice; 2,271 valid choices). Mean prediction gain was −0.005493 nats per choice (95% selection-aware bootstrap interval −0.011590 to 0.000528). The result does not provide clear evidence that the added term improves prediction under this procedure, and it does not establish a zero population effect. The models, penalty grid, split, and public evaluation choices limit the scope; this is within-session prediction, not an untouched external test.
probabilistic selection task · choice prediction · reinforcement learning · ridge logistic regression · trained pairs · test loss · model comparison · OpenNeuro
References
5- usesThe Choose A minus Avoid B score had opposite mean signs for C/E and D/F in 77 ds004532 Study 1 participants sci:sha256:54f30e78401a97533f0eab68074b546e76b1c4e84d203aef4ab436b29833e136
We used this publication to frame the trained-pair prediction question. We reused its fixed ds004532 source files and source manifest. We inspected its stimulus mapping and parser. The new model analysis uses the same public sample and is not an independent replication of its score result.
- usesWith model values set to 0.5 at the start and a 2% lapse rule, two learning rates reduced new-pair choice loss in 77 ds004532 Study 1 adults sci:sha256:a48a37bc2889fc7ba65003d30165a02cc8464ee71ea8cb1f850364efea8d73ee
We used this related report to distinguish training-to-test learning model prediction from the current test-phase choice model check. We read it after fixing our plan and results. Its scores did not select our sample or model rules. We did not compare losses across the two designs.
- usesdoi:10.18112/openneuro.ds004532.v1.2.0
The analysis used the event files, metadata, and task code from ds004532 version 1.2.0, identified by the saved source manifest and matched external metadata.
- supportsdoi:10.1371/journal.pone.0176205
Background evidence that stimulus discriminability can change learning and transfer measures; it did not supply data for this analysis.
- supportsdoi:10.1038/s41467-018-06781-2
Background evidence that learning context can affect later preferences; it did not supply data for this analysis.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.