Separate trial completions with a 12-pump cap allowed reverse orders for 91.0% of strict cash-out score pairs in 261 ds000030 v1.0.0 records
We tested how explosion trials limit person orders in the public scanner balloon task. We used all 263 event records from OpenNeuro ds000030 version 1.0.0. One record had an incomplete trial. One valid record had no cash-out score. The primary comparison used 261 people and 4,817 reward trials. The target was each person's mean count over separate trial completions without a burst, with each count capped at 12. The recorded history before each trial was held fixed. Hidden counts could vary separately within their allowed integer ranges. Correct action prefixes and terminal labels were required. We imposed no common policy or probability model. Of 33,764 strict cash-out score pairs, 30,718 (91.0%) permitted a strict reverse order. The score order was fixed in 3,024 pairs. Endpoints touched in 22 pairs. Two pairs had a strict reverse order in every allowed completion. Possible reverse-order shares were 81.7% at a cap of 9 and 63.3% at a cap of 6. Each pair was tested separately. The result is not an observed error rate or a measure of a stable risk trait. A separate source parser, exact fractions, exhaustive small examples, fresh source downloads, and a second analysis run passed. The package preserves all source event files, hashes, code, a local plan, a bound proof, trial links, person and pair tables, figures, and reproduction instructions.
Balloon Analogue Risk Task · censored outcomes · partial identification · rank stability · cash-out score · OpenNeuro ds000030 · behavioral neuroscience
References
3- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The analysis uses the scanner Balloon Analogue Risk Task event files from OpenNeuro ds000030 version 1.0.0.
- doi:10.3758/s13428-021-01634-1
- doi:10.1038/sdata.2016.110
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesUnder hidden independent uniform burst points, the 95% interval for early first-pump burst effects included zero for next-trial intentions capped at six in 263 ds000030 records
Used its event parser and retrieval code through publication 333a58db-1e0c-4863-9c96-8239990c9d53. The package preserves this code source and the MIT notice.
- usesA burst adjustment reduced score variance in 183 ds000030 BART records at a cap of 12 and kept the expected mean under hidden independent uniform burst points from 1 to 12
The copied retrieval scripts and adapted trial parser derive from this MIT code through publication 333a58db-1e0c-4863-9c96-8239990c9d53. We kept the license and checked the trial table against the later source package.
- usesInverse probability weights estimate pump intentions capped at 12 without bias in complete BART trials under hidden independent uniform burst points from 1 to 12
Used its fixed ds000030 source event archive, source manifests, retrieval code, and event parser. The parser was adapted. The new analysis estimates a capped mean under a known burst probability model. Both works use the same records.