---
request_id: data_request.khalvati-kiani-rao-2021-raw-behavior-matlab
status: blocked
priority: high
request_type: author_request
generated_at: 2026-07-26T14:22:05.080429+00:00
---

# Khalvati-Kiani-Rao raw behavioral MATLAB files

## Email

Contact path:
Use the article's data-availability contact path. If files are shared under non-redistribution terms, keep the raw MATLAB files outside git and commit only schema/provenance metadata plus aggregate checks that the license allows.

Subject: Request for Khalvati/Kiani/Rao 2021 macaque behavioral MATLAB files

Dear Dr. Kiani / Khalvati / Rao team,

I am curating behavtaskatlas, an open-science atlas of sensory-guided decision-making tasks and reusable behavioral-model fits.

I am working with the public source-data ZIP and POMDP-Confidence code for:

Khalvati K, Kiani R, Rao RPN. Bayesian inference with incomplete knowledge explains perceptual confidence and its deviations from accuracy. Nature Communications, 2021.

The public code appears to reference behavioral MATLAB files named `beh_data.monkey1.mat` and `beh_data.monkey2.mat`, but I could not find those files in the Nature source-data ZIP or the public POMDP-Confidence code archive. Would you be willing to share those analyzed/raw behavioral MATLAB files, or point me to their current public location?

The immediate use is to build a provenance-preserving trial-level harmonizer for the macaque RDM confidence task, recover signed choice and response-time fields where available, and fit psychometric / DDM models with explicit citation and caveats. I would also appreciate any redistribution terms you want observed for derived canonical trial tables or aggregate model-fit outputs.

Thank you,
behavtaskatlas curator

## Requested Files

- `beh_data.monkey1.mat` (required): MATLAB behavioral data struct for monkey 1, referenced by the public POMDP-Confidence fitting code. Evidence: fit.py references beh_data.monkey1.mat in the public code archive.
- `beh_data.monkey2.mat` (required): MATLAB behavioral data struct for monkey 2, referenced by the public POMDP-Confidence fitting code. Evidence: fit.py references beh_data.monkey2.mat in the public code archive.

## Purpose

Replace figure-source and direction-choice proxy fits with harmonized trial-level records for the Kiani-Shadlen macaque RDM confidence task, so the atlas can fit psychometric and DDM models against real choice/RT outcomes rather than inferred accuracy summaries.

## Blocker

The Nature source-data ZIP and public POMDP-Confidence code archive provide figure source tables and model code, but not the behavioral MATLAB files referenced by the code. Without those files, signed direction choice, raw session identifiers, and response-time fields cannot be recovered from the public package.

## Affected Atlas Records

### Datasets

- `dataset.khalvati-kiani-rao-rdm-confidence-source-data` - Khalvati-Kiani-Rao macaque RDM confidence source data

### Papers

- `paper.khalvati-kiani-rao-2021` - Khalvati K, Kiani R, Rao RPN

### Protocols

- `protocol.macaque-rdm-confidence-wager`

### Slices

- `slice.macaque-rdm-confidence-wager`

### Findings

- `finding.khalvati-kiani-rao-2021.accuracy.accuracy-no-sure-target.m1`
- `finding.khalvati-kiani-rao-2021.accuracy.accuracy-no-sure-target.m2`
- `finding.khalvati-kiani-rao-2021.accuracy.accuracy-sure-available-direction-chosen.m1`
- `finding.khalvati-kiani-rao-2021.accuracy.accuracy-sure-available-direction-chosen.m2`
- `finding.khalvati-kiani-rao-2021.chronometric.direction-choice-proxy.accuracy-no-sure-target.kiani-shadlen-m1`
- `finding.khalvati-kiani-rao-2021.chronometric.direction-choice-proxy.accuracy-no-sure-target.kiani-shadlen-m2`
- `finding.khalvati-kiani-rao-2021.psychometric.direction-choice-proxy.accuracy-no-sure-target.kiani-shadlen-m1`
- `finding.khalvati-kiani-rao-2021.psychometric.direction-choice-proxy.accuracy-no-sure-target.kiani-shadlen-m2`

## Provenance Checks

- article data availability: The article data-availability statement says analyzed data are available from R.K. upon reasonable request. https://www.nature.com/articles/s41467-021-25419-4
- code reference: The POMDP-Confidence public code references behavioral MATLAB files named beh_data.monkey1.mat and beh_data.monkey2.mat. https://github.com/koosha66/POMDP-Confidence `fit.py`
- public archive check: Local inspection of the public code archive found no .mat, .csv, .npy, .p, .pkl, .npz, or .zip behavioral data files. https://github.com/koosha66/POMDP-Confidence
- source data check: The Nature Communications source-data ZIP contains figure source tables, not the raw behavioral MATLAB files referenced by the code. https://static-content.springer.com/esm/art%3A10.1038%2Fs41467-021-25419-4/MediaObjects/41467_2021_25419_MOESM3_ESM.zip

## License And Redistribution Question

Please confirm whether behavtaskatlas may redistribute derived canonical trial tables and aggregate model-fit outputs produced from the shared files. If raw files cannot be redistributed, the atlas can keep them outside git and publish only provenance, field mapping, validation checks, and allowed aggregate summaries.

## Status Handoff

After sending this request, update the source `data_request` record from `ready_to_send` to `requested` and add an event with `event_type: sent`, the send date, actor, and evidence URL/path. Repository validation requires that event before accepting the `requested` status.

