mTRF-Toolbox is a system identification toolbox that permits the
fast computation of the linear stimulus-response mapping of a
sensory system in the forward or backward direction. Forward
modelling can be used to estimate multivariate temporal response
functions (mTRFs) that describe the mapping between user-defined
stimulus features and multi-channel neural recordings, or to predict
the spectrotemporal dynamics of neural responses to novel stimuli.
Backward modelling can be used to perform stimulus reconstruction by
mapping the multivariate population responses back to the features
of the causal stimulus. mTRF Toolbox facilitates the study of
ecologically-relevant stimuli such as speech, music or motion, and
is suitable for analysing different types of electrophysiological
signals including EEG, MEG, ECoG and EMG.
RaceModel is a psychophysics toolbox for testing the race model on
reaction times (RTs) generated using a redundant signals paradigm.
Bi- or tri-sensory RT distributions can be compared the statistical
facilitation predicted by the unisensory RT distributions. The race
model can be generated under the assumption that RTs on separate
sensory channels are statistically independent (Raab's model),
perfectly negatively dependent (Miller's bound) or perfectly
positively dependent (Grice's bound). RaceModel provides rountines
for computing geometric measures of multisensory gain, empirical and
predicted benefits as well as modality switch effects.
PERMUTOOLS is a statistical software package for multivariate
permutation testing. By comparing the magnitude of the test
statistic of interest with those obtained using permutations of the
data, it provides powerful, distribution-free hypothesis testing.
Family-wise error rate is controlled using the maximum statistic
method, making it suitable for multivariate or multiple permutation
tests. PERMUTOOLS offers a range of test statistics including the
t-statistic (paired, unpaired), the F-statistic (unpaired), the
correlation coefficient (Pearson, Spearman), as well as measures of
effect size with bootstrapped confidence intervals (Cohen's d,
Hedges' g, Glass' delta).