siyingxie / VCR_infant

Using multivariate assessments on infant EEG data to investigate visual category representations.

Visual Category Representations in the Infant Brain

This repository provides MATLAB example codes used in the study "Visual category representations in the infant brain ("VCR_infant" in short)".

You can clone this repository to local using:

git clone https://github.com/siyingxie/VCR_infant.git

Example analyses

IMPORTANT: To run the examples, a LIBSVM toolbox is required. Please see the "Dependency" and "Installation" sections for details.

Here are six examples analyses in the study.

• Decode object categories in time (runtime: ~ 2 min)

  example_decodingintime.m

• Decode object categories in time and frequency (runtime: ~ 5 min)

  example_decodingtimefrequency.m

• Time generalization analysis (runtime: ~ 1 min)

  example_decodingtimegeneralization.m 

• Relate category representations in infants and adults (runtime: ~ 1 min)

  example_rsatimegeneralization.m 

• Relate oscillation-based category representations in infants and adults (runtime: ~ 1 min)

  example_rsatimefrequency.m

• Relate category representations to computational models* (runtime: ~ 1 min)

  example_rsatomodels.m

*Gabor filter model and VGG-19 deep neural network trained on object categorization

Dependency

Resource/Software	Source/Identifier
Example data*	[https://osf.io/ruxfg]
LIBSVM toolbox♰	[https://github.com/cjlin1/libsvm; RRID: SCR_010243]

* They will be automatically downloaded while running the examples (via setup.m) if they have not already been downloaded.

♰ The toolbox needs to be downloaded and installed manually. (See "Installation" for details.)

Installation

The examples require LIBSVM (e.g., version-3.25) toolbox, and please add the toolbox as follows:

• To clone the LIBSVM.git repository to the /code directory, please use:

  cd VCR_infant/code
  git clone https://github.com/cjlin1/libsvm.git

• If you are on Unix systems, you will need to compile the toolbox. Please refer to the steps in LIBSVM.readme.

  A quick walk-through:

  • On MATLAB command window, please type:

  >> cd libsvm/matlab
  >> matlabroot % check your $MATLABROOT
  >> edit Makefile 

  • On MATLAB editor, please manually edit the "Makefile":

  a) In line:3, change the "MATLABDIR ?= " to your $MATLABROOT 
  b) save the "Makefile"

  • Back to the MATLAB command window, please type:

  >> make % it will take a few seconds to compile 

  For more details, please see LIBSVM.readme.

Required customized functions

These customized functions are provided in the repository.

• averagetimepoints.m
• averagetrials.m
• correlatevectors.m
• covCor.m
• cvmvnn.m
• setup.m
• timefrexdecomp.m
• vectorizerdm.m

Visualization

The expected outputs from the examples:

• example_decodingintime
• example_decodingtimefrequency
• example_decodingtimegeneralization
• example_rsatimegeneralization
• example_rsatimefrequency
• example_rsatomodels

Results for single participants:

• singlesubject_decodingintime (Related to Figure 1C)
• singlesubject_decodingtimegeneralization (Related to Figure 1G)
• singlesubject_rsatimegeneralization (Related to Figure 2B)
• singlesubject_decodingtimefrequency (Related to Figure 4B)
• singlesubject_rsatimefrequency (Related to Figure 4E)

Model RDMs details:

• visualise_modelRDMs (Related to Figure 3A)