Finger Sensory Integration

This repository contains matlab code and pre-processed data to reprodce key analyses and figures in the paper:

Arbuckle, Pruszynski, & Diedrichsen (2022). Mapping the integration of sensory information across fingers in human sensorimotor cortex. Journal of Neuroscience. [link to paper]

Data

The data folder contains the following files:

• beha_all.mat : behavioural data for all trials and participants
• fmri_BAxx_betas.mat : activity patterns from brodmann area xx
• fmri_selectivity.mat : single-finger selectivity results
• fmri_modelFits.mat : representational model results
• fmri_regionG.mat : group-average second-moment matrices for each region (used in model analyses)

Code

The analysis code is in the following file:

• fsi_ana.m (finger sensory integration analysis)

This code can reproduce the key analyses reportd in the paper, namely the single-finger selectivity and representational model analyses and plots from the first-level GLM activity patterns. See the Usage section for more information.

Dependencies

The code in this repo uses functions from these (freely) available toolboxes. Be sure to add them to your path.

• RSA toolbox
• PCM toolbox - develop branch
• Plotlib toolbox - dev branch, for plotting
• Dataframe toolbox - for pandas-like functionality

Usage

Please download the entire repository to use. The variable dataDir in fsi_ana.m must point to the data folder. If you pull this repo, the default path for dataDir should be correct (by default, it assumes there is a subfolder called data wherever fsi_ana.m is saved).

Please note that, by default, verbose updates about analysis progress will be displayed to the user. To turn this off, please set verbose = 0; in fsi_ana.m.

Behaviour: To calculate behavioural performance on the detection mismatch task, execute the following matlab commands:

fsi_ana('beha:do_analysis');

Single-finger selectivity: To produce the single-finger selectivity plot, execute the following matlab commands:

D = load('data/fmri_selectivity.mat');
fsi_ana('plot:selectivity',D);

You can also reproduce the selectivity analysis yourself:

D = fsi_ana('selectivity:do_analysis');

To speed up the analysis, you can change how many simulated datasets are generated by adjusting the numSim variable under the selectivity:do_analysis case (default is 1000). These simulated datasets are used to estimate the influence of measurement noise (see paper methods).

Representational model analysis: To produce the representational model fit plot, execute the following matlab commands:

D = load('data/fmri_modelFits.mat');
fsi_ana('plot:modelFits',D);

You can also reproduce the representational model analysis:

D = fsi_ana('model:do_analysis');

The default is to re-do the analysis for all six regions. To adjust this, edit the roi variable in the model:do_analysis case.