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)