Documentation

Overview

• Demonstrations - full listings
• Get Started
• Download instructions
• NMSM 2019, Noesselt’s lab 3rd Modelling Symposium, University of Magdeburg
• Indices and tables
• Frequently Asked/Anticipated Questions
• Contact information
• Information for contributors
• Information for contributing programmers
• Authors
• Contributors
• Acknowledgements
• Citations
• Copyright information
• Changelog
• References

Detailed

• Demonstrations - full listings
  • demo_fmri_correlation_searchlight
  • demo_fmri_distatis
  • demo_fmri_rois
  • demo_fmri_searchlight_lda
  • demo_fmri_searchlight_naive_bayes
  • demo_fmri_searchlight_rsm
  • demo_meeg_dataset_operations
  • demo_meeg_timefreq_searchlight
  • demo_meeg_timelock_searchlight
  • demo_meeg_timeseries_classification
  • demo_meeg_timeseries_generalization
  • demo_surface_searchlight_lda
  • demo_surface_tfce
• Get Started
  • Download instructions
  • Matlab experience
  • Cognitive neuroscience experience
  • Multivariate pattern analysis
  • CoSMoMVPA concepts
  • Prerequisites
  • Get your environment ready
  • Next steps
• Download instructions
  • Get the code and example data
  • Keep your code up to date
• NMSM 2019, Noesselt’s lab 3rd Modelling Symposium, University of Magdeburg
  • Contents
  • Miscellaneous
• Indices and tables
• Frequently Asked/Anticipated Questions
  • General
  • How do I …
• Contact information
  • General
  • Bugs and other issues
  • Contributions
• Information for contributors
  • Directory locations and naming conventions
• Information for contributing programmers
  • Code development
  • Build system
  • Matlab code guidelines
  • CoSMoMVPA-specific guidelines
  • Test suite
  • Good tests
• Authors
• Contributors
• Acknowledgements
• Citations

Full documentation

• Demonstrations - full listings
  • demo_fmri_correlation_searchlight
  • demo_fmri_distatis
  • demo_fmri_rois
  • demo_fmri_searchlight_lda
  • demo_fmri_searchlight_naive_bayes
  • demo_fmri_searchlight_rsm
  • demo_meeg_dataset_operations
  • demo_meeg_timefreq_searchlight
  • demo_meeg_timelock_searchlight
  • demo_meeg_timeseries_classification
  • demo_meeg_timeseries_generalization
  • demo_surface_searchlight_lda
  • demo_surface_tfce
• Get Started
  • Download instructions
    • Get the code and example data
      • For the impatient
      • For the patient
    • Keep your code up to date
      • External dependencies
      • Developers
  • Matlab experience
    • Advanced beginner
    • Competent user
    • Proficient user
  • Cognitive neuroscience experience
  • Multivariate pattern analysis
    • Multivariate pattern analysis (MVPA)
      • Starting simple: the univariate case
      • More measurements: the multivariate case
      • Patterns
      • Further reading
  • CoSMoMVPA concepts
    • Dataset
      • Attributes
        • Sample attributes
        • Feature attributes
        • Dataset attributes
      • Targets
      • Chunks
      • Samples + sample attributes + feature attributes + dataset attributes = dataset
    • Dataset operations
      • Slicing
      • Slicing-related functions
    • Classifier
    • Measure
    • Neighborhood
  • Prerequisites
  • Get your environment ready
  • Next steps
    • Analysis gallery
• Download instructions
  • Get the code and example data
    • For the impatient
    • For the patient
  • Keep your code up to date
    • External dependencies
    • Developers
• CoSMo 2013 Multivariate Pattern Analysis Workshop
  • Contents
    • Introduction
      • Goals of the workshop
      • MVPA Dataset
      • Code for the workshop
      • Sample Dataset
      • Plan for the day
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification with Naive Bayes classifier
      • Write your own Nearest-neighbor classifier
      • Build wrapper for Matlab’s SVM classifier
        • Wrapper for two classes
        • Wrapper for multiple classes
      • Cross-validation part 1: N-Fold Partitioner
      • Cross-validation part 2: using multiple classifiers
      • Cross-validation part 3: using a dataset measure
      • Visualization of DSMs
        • Visualizing dissimilarity matrices
        • Centering the data
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • Define voxel selection for a searchlight
        • Compute voxel offsets in a sphere
        • Using the sphere offsets for voxel selection
      • Reading material
      • Write a searchlight function that computes a generic dataset measure
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• CIMeC 2014 Advanced Neural Decoding
  • Contents
    • Introduction
      • Prerequisites
      • Goals of this course
        • Not covered in this course
        • Code and data needed for this workshop
      • Sample Dataset
      • To cover in this course
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Classification analysis with cross-validation
        • Background
        • Single subject, n-fold cross-validation classification
          • Part 1: implement n-fold crossvalidation
          • Part 2: use a partitioner
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• Royal Holloway University of London CoSMoMVPA workshop
  • Contents
    • Introduction
      • Prerequisites
      • Goals of this course
        • Not covered in this course
      • Sample Dataset
      • To cover in this workshop
      • Tentative schedule
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Classification analysis with cross-validation
        • Background
        • Single subject, n-fold cross-validation classification
          • Part 1: implement n-fold crossvalidation
          • Part 2: use a partitioner
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • Using CoSMoMVPA multiple-comparison correction
        • Rationale
        • Exercise with fMRI data
        • Exercise with time-generalization data
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• CIMeC hands-on methods course, part 1 (6 April-2 May 2016)
  • Contents
    • Introduction
      • Parts
      • Prerequisites
      • Goals of the fMRI / MVPA part of the course
        • Not covered in this course
      • Sample Datasets
        • AK6 dataset
        • Haxby2001 dataset
        • MEG obj6 dataset
        • MEG TAC dataset with anatomical scan
      • Links
      • Assignments
        • Preparation
        • Format
        • Assignments must be your own work
        • Exercise 1 - deadline 23:59, 19 April 2016
        • Exercise 2 - deadline 23:59, 26 April 2016
        • Exercise 3 - deadline 23:59, 03 May 2016
        • Exercise 4 - deadline 23:59, 17 May 2016
        • MEG Exercises - deadline 23:59, 15 June 2016
      • FAQ
      • Schedule
      • Contact
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Classification analysis with cross-validation
        • Background
        • Single subject, n-fold cross-validation classification
          • Part 1: implement n-fold crossvalidation
          • Part 2: use a partitioner
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • General MEEG analysis toolboxes
      • MEEG Searchlights
        • Part 1: MEEG univariate contrast
        • Part 2: MEEG time-course searchlight with classifier and cross-validation
        • Part 3: MEEG time-course searchlight with split-half correlation measure
        • Part 4: MEEG channel-time-frequency searchlight with Naive Bayes classifier
      • MEEG time generalization
        • Requirements
        • Reading material
        • The time generalization method
      • Using CoSMoMVPA multiple-comparison correction
        • Rationale
        • Exercise with fMRI data
        • Exercise with time-generalization data
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• PRNI 2016 workshop
  • Contents
    • Introduction
      • Overview of the workshop
      • Format
      • Prerequisites
      • Goals of this course
        • Not covered in this course
      • Datasets
        • AK6 dataset
        • MEG obj6 dataset
      • Tentative schedule
      • Contact
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • General MEEG analysis toolboxes
      • MEEG Searchlights
        • Part 1: MEEG univariate contrast
        • Part 2: MEEG time-course searchlight with classifier and cross-validation
        • Part 3: MEEG time-course searchlight with split-half correlation measure
        • Part 4: MEEG channel-time-frequency searchlight with Naive Bayes classifier
      • MEEG time generalization
        • Requirements
        • Reading material
        • The time generalization method
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • Using CoSMoMVPA multiple-comparison correction
        • Rationale
        • Exercise with fMRI data
        • Exercise with time-generalization data
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• LABMAN 2017 course
  • Contents
    • Introduction
      • Overview of the workshop
      • Format
      • Prerequisites
      • Goals of this course
        • Not covered in this course
      • Datasets
        • AK6 dataset
        • MEG obj6 dataset
      • Tentative schedule
      • Contact
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Double dipping
        • Background
        • Double dipping illustration
      • Classification analysis with cross-validation
        • Background
        • Single subject, n-fold cross-validation classification
          • Part 1: implement n-fold crossvalidation
          • Part 2: use a partitioner
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • General MEEG analysis toolboxes
      • MEEG Searchlights
        • Part 1: MEEG univariate contrast
        • Part 2: MEEG time-course searchlight with classifier and cross-validation
        • Part 3: MEEG time-course searchlight with split-half correlation measure
        • Part 4: MEEG channel-time-frequency searchlight with Naive Bayes classifier
      • MEEG time generalization
        • Requirements
        • Reading material
        • The time generalization method
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • Using CoSMoMVPA multiple-comparison correction
        • Rationale
        • Exercise with fMRI data
        • Exercise with time-generalization data
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• NMSM 2019, Noesselt’s lab 3rd Modelling Symposium, University of Magdeburg
  • Contents
    • Introduction
      • Overview of the workshop
      • Format
      • Prerequisites
      • Goals of this course
        • Not covered in this course
      • Datasets
        • AK6 dataset
        • MEG obj6 dataset
      • Tentative schedule
      • Acknowledgements
      • Contact
    • Exercises
      • Dataset basics
        • Reading material
          • fMRI dataset basics
        • Make sure your setup works
        • Very first exercise
        • Load and view anatomical dataset
          • Loading datasets with a mask
          • Operations on datasets
      • Split-half correlation-based MVPA with group analysis
        • Reading material
          • Background
          • Exercise
        • Part 1 (single subject analysis)
        • Part 2 (group study)
      • Classification analysis
        • Reading material
        • Background
        • Single subject, single fold split-half classification
      • Classification analysis with cross-validation
        • Background
        • Single subject, n-fold cross-validation classification
          • Part 1: implement n-fold crossvalidation
          • Part 2: use a partitioner
      • Double dipping
        • Background
        • Double dipping illustration
      • Using CoSMoMVPA measures
        • Rationale
        • The measure concept
        • Split-half correlations using a measure
        • Classifier with cross-validation using a measure
      • Using CoSMoMVPA neighborhoods for regions of interest
        • Rationale
        • Exercises
          • Part 1
          • Part 2
          • Part 3
          • Part 4 (advanced)
      • Use the searchlight with a neighborhood and a measure
        • ‘I love it when a plan comes together’
        • Part 1
        • Part 2
      • General MEEG analysis toolboxes
      • MEEG Searchlights
        • Part 1: MEEG univariate contrast
        • Part 2: MEEG time-course searchlight with classifier and cross-validation
        • Part 3: MEEG time-course searchlight with split-half correlation measure
        • Part 4: MEEG channel-time-frequency searchlight with Naive Bayes classifier
      • MEEG time generalization
        • Requirements
        • Reading material
        • The time generalization method
      • Representational similarity analysis
        • Reading material
        • Visualizing dissimilarity matrices
        • Comparing dissimilarity matrices
        • Target dissimilarity matrix searchlight
      • Surface-based fMRI searchlight
        • Reading material
        • Required toolboxes
          • Background
          • Exercise
        • Part 1 (cortical thickness)
        • Part 2 (Classification analysis)
      • Using CoSMoMVPA multiple-comparison correction
        • Rationale
        • Exercise with fMRI data
        • Exercise with time-generalization data
    • CoSMoMVPA functions
    • CoSMoMVPA functions - header signature files
    • CoSMoMVPA functions - skeleton files
    • Runnable examples
    • Runnable examples - skeleton files
  • Miscellaneous
    • CoSMoMVPA unit tests
    • Demonstrations
• Indices and tables
• Frequently Asked/Anticipated Questions
  • General
    • How should I cite CoSMoMVPA?
    • What is the history of CoSMoMVPA?
    • What are the main features?
    • What does CoSMoMVPA not provide?
    • Does it integrate with PyMVPA?
    • Does it run on GNU Octave?
    • How fast does it run?
    • What should I use as input for MVPA?
    • Who are the developers of CoSMoMVPA?
    • Which classifiers are available?
    • Which platforms does it support?
    • What future features can be expected?
    • How can I contact the developers directly?
    • Is there a mailinglist?
    • Why do you encourage balanced partitions?
    • Does the LDA (linear discriminant analysis) classifier use shrinkage / normalization?
  • How do I …
    • Find the correspondence between voxel indices in AFNI and feature indices in CoSMoMVPA
    • Get ECoG data in a CoSMoMVPA struct
    • Get temporal data in a CoSMoMVPA struct
    • Run group analysis
    • Make an intersection mask across participants
    • Compute for a group of participants who were scanned with MRI the overlap of their masks
    • Run group analysis on time-by-time generalization measures
    • Use LIBSVM
    • Use surface-based mapping with a low-resolution output surface
    • Correct for multiple comparisons
    • Do cross-modal decoding across three modalities
    • Compute classification accuracies manually
    • Make a merged hemisphere from a left and right hemisphere
    • Merge surface data from two hemispheres
    • Visualize and store multiple fMRI volumes
    • Average along features in a neighborhood
    • Select a time interval in an MEEG dataset
    • Select a particular channel type in an MEEG dataset
    • Use only a subset of channels for my analysis?
    • Should I Fisher-transform correlation values?
    • Average samples in a deterministic manner?
    • Select only a subset of features in a neighborhood?
    • Use multiple-comparison correction for a time course?
    • Classify different groups of participants (such as patients versus controls)?
    • When running an MEEG searchlight, have the same channels in the output dataset as in the input dataset?
    • Save MEEG data when I get the error “value for fdim channel label is not supported”?
    • Run a 2 x 2 within-subject ANOVA?
    • Use a FieldTrip source dataset that uses a ‘fake’ channel structure
    • Run cosmo_montecarlo_cluster_stat on a cluster with multiple nodes
    • Why should I consider re-meaning when doing representational similarity analysis (RSA)?
    • Import BrainStorm data
    • Compute the correlation between two dissimilarity matrices
    • Use the generalization measure with different durations for training and test set?
    • Get the classifier weights after training a classifier?
    • Get coordinates of voxels in a CoSMoMVPA fMRI dataset?
    • Compute representational similarity across chunks?
    • Run group analysis on time generalization results?
• Contact information
  • General
  • Bugs and other issues
  • Contributions
• Information for contributors
  • Directory locations and naming conventions
    • Meta - naming conventions in the documentation
    • Setting the Matlab path
    • Organization of files and directories
    • Setting up the documentation build system
• Information for contributing programmers
  • Code development
    • Contributing using git
      • Initial git / github setup
      • Proposing a change in the code (a.k.a. submitting a Pull Request (PR)
      • Notes on committing
  • Build system
  • Matlab code guidelines
    • Maximum line length is 75 characters
    • Indentation is 4 spaces (no tabs)
    • Use lower-case letters for variable names
    • Throw an (informative) error early
    • Do not repeat yourself
    • Write in normal, understandable english
    • Document functions
    • Pre-allocate space for data
    • Use vectorization
    • Use clear variable names
    • Avoid side effects
    • Use onCleanup if an earlier state needs to be reset
    • Tests should not require user interaction
    • Do not use global variables
    • Avoid long and complicated expressions
    • Use sprintf or fprint when formatting strings
    • Avoid using eval
    • Minimize using try and catch
    • No private functions
    • No file duplication
    • Check input arguments
  • CoSMoMVPA-specific guidelines
    • Writing exercises
    • Documentation tests
    • Compatibility notes
  • Test suite
  • Good tests
• Authors
• Contributors
• Acknowledgements
• Citations