Data from: Neural mechanisms of resource allocation in working memory
Data files
Apr 10, 2025 version files 411.42 MB
-
README.md
4.60 KB
-
shared.zip
411.42 MB
Abstract
To mitigate capacity limits of working memory, people allocate resources according to an item’s relevance. However, the neural mechanisms supporting such a critical operation remain unknown. Here, we developed computational neuroimaging methods to decode and demix neural responses associated with multiple items in working memory with different priorities. In striate and extrastriate cortex, the gain of neural responses tracked the priority of memoranda. We decoded higher-priority memoranda with smaller error and lower uncertainty. Moreover, these neural differences predicted behavioral differences in memory prioritization between and within participants. Remarkably, trialwise variability in the magnitude of delay activity in frontal cortex predicted differences in decoded precision between low and high-priority items in visual cortex. These results support a model in which feedback signals broadcast from frontal cortex sculpt the gain of memory representations in visual cortex according to behavioral relevance, thus, identifying a neural mechanism for resource allocation.
This repository contains data and code reported in:
Li, H. H., Sprague, T. C., Yoo, A. H., Ma, W. J., & Curtis, C. E. (2025). Neural mechanisms of resource allocation in working memory. Science Advances.
If you use any of the code or data in this repository, include the above reference.
I. Data
Behavioral Data
- Folders:
data/behav_1item
anddata/behav_2item
- Description:
Files in these folders contain trial-by-trial information on condition, target location, and behavioral reports. - Filenames:
Files are named by the subject ID (from S1 to S11) - Variables for the 1-item Experiment:
wm_ang
:- Location of the target (in degree polar angle)
behEst
:- Behavioral reports (in polar angle)
- Variables for the 2-item Experiment:
targ_angs
:- First column: Location of the target (in degree polar angle)
- Second column: Location of the non-target (in degree polar angle)
conditions
:- 1 = valid condition (precue matched the target)
- 2 = invalid condition
behEst
:- Behavioral reports (in polar angle)
sacRT
:- Saccade reaction time (in second)
goodtrial
:- Index for trials that passed exclusion criteria (1=good trials; 0=excluded trials)
sep_cue_angs
:- Angle (degree) of the separator in the precue indicating how the aperture is divided
pri_cue_angs
:- Angle (degree) of the precue indicating the prioritized item
fMRI Data
- Folders:
data/trialData_1item
anddata/trialData_2item
- Description:
Each.mat
file contains voxel activity pattern (#trial X #voxel) for each subject, session, and ROI. Data from the 1-item experiment is only used for voxel selection. - Filenames:
Files are named as subjectID_sessionID_ROI_surf_trialData.mat - Variables for the 1-item Experiment in
data/trialData_1item
:TR
: TR = 0.75 secondc_map
:- Location of the target (in degree polar angle)
dt_mapz
:- Voxel activity patterns (#trial X #voxel)
- Variables for the 2-item Experiment in
data/trialData_2item
:TR
: TR = 0.75 secondc_all
:- First column: Location of the target (in degree polar angle)
- Second column: Location of the non-target (in degree polar angle)
- Third column: conditions (1 = valid condition; 2 = invalid condition)
dt_allz
:- Voxel activity patterns (#trial X #voxel)
II. Outputs of Analysis
- Folder:
mdata/decoded
- Description:
Contains decoding results for each subject and ROI. Each.mat
file includes - Filenames:
Files are named as subjectID_sessionID_ROI_decoded_750vox.mat - Variables:
est
: Decoded locationsunc
: Decoded uncertainty
In both variables, the first column represents the target and the second column represents the distractor (non-target).p
: Hyperparameters used in decodingvox_mask
: index for selected voxelsnvox
: number of voxels for decodingsess_2item
: sessions decoded
- Folder:
mdata/VPmodel
- Description:
Contains fitting results of a priority-dependent variable precision model to the behavioral data (used for Supplementary Fig.2). - Variables:
berr_mean
: mean absolute behavioral error (valude, invalid)berr_std
: Sdv of behavioral error (valude, invalid)bestPar
: best-fit parametersnLLVec
: negative log-likehoodparMat
: parameters under different intial points
III. Code
-
Main Script:
RunwmPriority_wrapper.m
to decode two items for each trial from the voxel activity of a particular subject and ROI.wmPriority_genModelDecode_2item.m
is called by the wrapper to read and sort the data for decoding with leave-one-run-out cross validation procedure. - Decoding Algorithms:
The algorithms and code for conducting Bayesian decoding of 2 items from neural (fMRI) activity are located in the folderDecode_2item
. - Summary:
Runplot_summary.m
to plot summary of the results shown in the paper. The summary results are saved insummary.mat
- Others:
cat_struct.m, cpsFigure.m, plot_errorbar.m, plot_label.m are miscellaneous functions for plotting or sorting the data
Human subjects data
Explicit written consents were obtained from all the participants, for data collection and for publishing de-identified data in the public domain. The data published did not contain any identifiable information. Subject’s names are replaced by anonymous id numbers.