Data from: A mosaic of whole-body representations on the human precentral gyrus

Deo, Darrel 1 ; Okorokova, Elizaveta2; Pritchard, Anna3; Hahn, Nick1; Card, Nicholas2; Nason-Tomaszewski, Samuel3; Jude, Justin4; Hosman, Thomas5; Choi, Eun Young1; Qiu, Deqiang3; Meng, Yuguang3; Wairagkar, Maitreyee2; Nicolas, Claire4; Kamdar, Foram1; Iacobacci, Carrina2; Acosta, Alexander4; Hochberg, Leigh5; Cash, Sydney4; Williams, Ziv4; Rubin, Daniel4; Brandman, David2; Stavisky, Sergey2; AuYong, Nicholas3; Pandarinath, Chethan3; Downey, John6; Bensmaia, Sliman6; Henderson, Jaimie1; Willett, Francis1

Published Jun 08, 2026 on Dryad. https://doi.org/10.5061/dryad.mpg4f4rg5

Data files

Jun 08, 2026 version files 384.23 MB

README.md

8.93 KB
T11.mat

14.15 MB
T12_control.mat

11.89 MB
T12.mat

18.12 MB
T15.mat

26.97 MB
T16.mat

40.25 MB
T17.mat

259.79 MB
T5.mat

13.04 MB

Abstract

This dataset contains intracortical neural recordings from human motor cortex collected during a delayed movement task designed to probe the neural representation of attempted or executed movements across multiple body parts. Data were acquired from 20 microelectrode arrays implanted in precentral gyrus across 8 participants with paralysis due to spinal cord injury, amyotrophic lateral sclerosis, or brainstem stroke, all enrolled in brain-computer interface clinical trials. Participants were visually cued by text on a computer monitor to attempt or perform isolated movements of specific body parts. This dataset includes the neural activity recorded during these experiments, together with associated task and trial information used to quantify movement-related tuning across the body. These data support the analyses and figures reported in the accompanying paper: "A mosaic of whole-body representations on the human precentral gyrus".

Overview

Neural data were collected from eight participants while each performed a text-cued instructed delay task. This repository contains the data for participants T5, T11, T12, T15, T16, and T17. Data for participants C1 and C2 are hosted separately on DABI and are available at https://dabi.loni.usc.edu/projects/K1KK0H4SRS11.

Code associated with the dataset is available at: https://github.com/d-r-deo/pcg-mosaic

During the instructed delay period, a red square and text appeared at the center of the computer monitor, indicating to the participant that they should prepare to make the specified movement. The instructed delay period lasted a random amount of time. After the delay period, the square turned green, indicating the movement period, at which point the participant moved as immediately as possible.

Each participant was asked to:

Continue attempting the movement if they could not overtly move the body part; or
Hold the posture of the completed movement, if they were able to overtly move the body part, until the text changed to "Return," at which point the participant relaxed and returned to a neutral posture.

All movement cues appeared in randomized order. The ranges of delay and movement periods may differ for each participant, as they were chosen based on each participant's comfort level.

Data Details

For each experiment with a given participant, data were collected in a series of blocks. Neural data between blocks was not recorded. Data can be loaded with MATLAB or Python (scipy.io.loadmat).

Dataset Fields

Each participant's dataset contains the following fields:

Field	Dimensions	Description
`tx`	T × F	Binned threshold crossing counts (20 ms bins). T = number of 20 ms time steps in the experiment; F = number of features (multi-unit channels concatenated with sorted units from spike-sorted data).
`tx_blkMeanSub`	T × F	Block-wise mean-removed threshold crossing counts (20 ms bins). Same dimensions as `tx`.
`blockNum`	T × 1	Block number at each time step. Gaps in recordings occur at block transitions (neural activity between blocks was not recorded).
`trialCue`	R × 1	Movement condition for each trial (see Movement Condition Numbering below). R = number of trials.
`goCue`	R × 1	Row indices into the time dimension (e.g., rows of `tx`) marking the transition from 'delay' to 'move'.
`state`	T × 1	Trial state at each time step: `0` = Delay, `1` = Go, `2` = Return.
`stateNames`	1 × 3 cell	String names for the trial states (`0` = Delay, `1` = Go, `2` = Return).
`cueList`	46 × 1 cell	Movement condition names (strings). The index of each name corresponds to a `trialCue` number (e.g., if`trialCue(1) = 3`, then trial 1 was HEAD – Turn Down).
`binWidth`	scalar	Bin size in seconds (`0.02` = 20 ms).
`chanSets`	1 × C cell	Each cell contains a vector of electrode-channel or sorted-unit indices for a particular array/feature set. C = number of array/feature sets.
`chanSetNames`	1 × C cell	Corresponding names for each array/feature set in `chanSets`.
`hemisphere`	string	The hemisphere of the brain where the electrodes were located.
`details`	string	Brief details about the dataset (participant ID, date of experiment, block numbers included).
`sorted`	T × S	Binned threshold crossing counts (20 ms bins) of sorted units only. S = number of sorted units.
`sortedChanIDs`	1 × S	Mapping from each sorted unit (column of `sorted`) to the electrode channel on which it was detected (see `chanSets` for the array a sorted unit belongs to.

Dataset Names and Repetitions

Dataset	Repetitions per movement condition
T5	22
T11	12
T12	20
T15	7–8
T16	13–14
T17	15–16
C1	14
C2	12–13

T12_control — Supplementary Control Dataset

T12_control It's a special dataset from a control experiment performed with participant T12. During this control experiment, T12 performed a limited set of movements restricted to all four limbs. See T12_control.neural.cueList the full list of movements used in this control experiment.

T12_control contains two sub-structures:

T12_control.neural — A data structure with the same fields as those described in Data Details.
T12_control.opticalFlow — An accompanying data structure quantifying movement of the hands and feet using optical-flow analysis.

The opticalFlow Sub-structure has the same fields as the core data, with the following additions:

Field	Dimensions	Description
`opticalFlow`	M × 4	Optical-flow movement quantification. M = number of movie frames; 4 = number of effectors (right hand, left hand, right foot, left foot, in that order).
`columnOrder`	4 × 1 cell	String names of the effectors corresponding to each column of `opticalFlow`.

Note: The goCue within the opticalFlow Sub-structure is indexed relative to the opticalFlow matrix (not the neural data).

Movement Condition Numbering Scheme

#	Condition	#	Condition
1	Do Nothing	24	MOUTH – Open
2	EYEBROWS – Raise	25	NOSE – Wrinkle
3	HEAD – Turn Down	26	RIGHT ANKLE – Down
4	HEAD – Turn Left	27	RIGHT ANKLE – Left
5	HEAD – Turn Right	28	RIGHT ANKLE – Right
6	HEAD – Turn Up	29	RIGHT ANKLE – Up
7	LEFT ANKLE – Down	30	RIGHT ARM – Raise Left
8	LEFT ANKLE – Left	31	RIGHT ARM – Raise Right
9	LEFT ANKLE – Right	32	RIGHT HAND – Close
10	LEFT ANKLE – Up	33	RIGHT HAND – Open
11	LEFT ARM – Raise Left	34	RIGHT LEG – Raise Left
12	LEFT ARM – Raise Right	35	RIGHT LEG – Raise Right
13	LEFT HAND – Close	36	RIGHT TOES – Curl
14	LEFT HAND – Open	37	RIGHT TOES – Open
15	LEFT LEG – Raise Left	38	RIGHT WRIST – Down
16	LEFT LEG – Raise Right	39	RIGHT WRIST – Left
17	LEFT TOES – Curl	40	RIGHT WRIST – Right
18	LEFT TOES – Open	41	RIGHT WRIST – Up
19	LEFT WRIST – Down	42	SPEAK – "Ban"
20	LEFT WRIST – Left	43	SPEAK – "Bat"
21	LEFT WRIST – Right	44	SPEAK – "Can"
22	LEFT WRIST – Up	45	SPEAK – "Cat"
23	LIPS – Pucker	46	TONGUE – Out

Dataset-Specific Additional ConditionsThe

T16 dataset includes two additional movement conditions:

#	Condition
47	EYES – Up
48	EYES – DownThe

T17 dataset includes two additional movement conditions:

#	Condition
47	HUM – Hi
48	HUM – Low

Human subjects data

We confirm that the informed consent form signed by participants permits sharing of coded data that does not contain other identifying information. Neural data contain no personally identifiable information (PII).