An open dataset of cognitive test results.
Neuropsych is compiling an open dataset of results from its tests, in English and Spanish. It contains de-identified records from adults who consented to research use and no direct identifiers. Each release is published under CC-BY 4.0 with a DOI.
Counted on . A usable first attempt is the first time a person takes a test, as an adult with age and education on file, with the validity checks passed.
Usable first attempts by test and language
Counts are assessed against three thresholds. At 250 a task supports preliminary reference values; at 600, flagging of unusually low scores; at 1,000, norms stratified by age, sex and education. Each bar has one segment per threshold.
The first release is issued when ten of the tests each reach 250 usable first attempts in one language. Later releases are annual, or earlier when the counts change materially.
| Test | English | Spanish |
|---|---|---|
| Loading the counts. | ||
The 12 largest samples are shown. Counts refresh weekly.
Contents of each release, and exclusions
Files follow the Psych-DS layout. The dashed rule marks content excluded by design.
- Description
- A
dataset_description.jsonnaming the release, its version and its licence. - Codebook
- Per task: every variable, its unit and direction, the scoring rule and the version history.
- Attempts
- One record per first attempt: scores, age band, education, language, device type and task version, with both languages in one file.
- Trials
- Each response, its latency and the item presented, for tasks that log trials.
- Participants
- Participants under 18, and participants who did not consent to research use.
- Small cells
- Cells with fewer than 5 participants are suppressed.
- Identifiers
- Names, email addresses and any other direct identifier.
Administer the battery to your own participants
The procedure is identical in English and Spanish. One step depends on participant consent.
Direct participants to the tasks
Participants complete the battery in English or Spanish. Each task has a stable URL, so a protocol can administer any subset.
Obtain the data export
Participant-level results are returned trial by trial for tasks that log trials, and as one record per administration otherwise. Exports are prepared manually.
Verify the scoring
Every scoring rule and the source of each normative table are documented on the scoring engine page and can be reviewed before data collection.
Administer a subset
A protocol may use the full battery or selected tasks. No task is required.
Ethical review remains with the investigator's institution. We provide task descriptions and data storage and retention details for the application. Contact us with the study design, the tasks required and the expected sample size.
Citing a release
Each release carries its own version number and DOI.
Dominguez, C. (2026). Neuropsych Open Cognitive Dataset (Version 1) [Data set]. Neuropsych. DOI assigned at first release.
Licensed under CC-BY 4.0. Reuse requires attribution.