Tutorial widgets
This page includes all widgets used throughout the tutorials. In the event the widgets cannot be access, please send us a message at jculham<at> uwo.ca. Video demonstrations of the widgets are also available here.
Tutorial 2
Widget 1: Fitting a sine wave
A predictor function can be scaled by a beta weight to fit simulated data for two voxels and a Pearson correlation coefficient (r) can be determined.
Widget 2: Fitting one predictor to the localizer
A model with one predictor (visual stimulation vs. baseline) can be used to fit different voxels from one run of one participant for the course localizer data.
Widget 3: Fitting one predictor with a square wave
Comparison of correlations with and without convolution of the predictor
Tutorial 3
Widget 4: Fitting 4 predictors to the localizer data
A model with four predictors (faces, hands, bodies, scrambled) is used to fit activation for 3 voxels.
Widget 5: Fitting a baseline predictor
The effect of erroneously including a redundant baseline predictor. To make the two predictors completely redundant, we used the unconvolved (box car) versions of the predictors. The redundancy would be slightly less with convolved predictors; nevertheless, redundancy would be suboptimal.
Tutorial 4
Serial correlation
Allows you to calculate the correlation between the original residuals function and a shifted version.
Tutorial 6
Linear drift
How linear trend removal affects the GLM
Temporal filtering (Two conditions + Baseline)
How temporal filtering affects the GLM. To perform high-pass filtering, set the minimum frequency to remove frequencies below that value. To perform low-pass filtering, set the maximum frequency to remove frequencies above that value. To do band-pass filtering, set minimum and maximum frequencies to remove frequencies at either extreme.
Temporal filtering (Localizer data)
Same as previous widget, but on Localizer timecourse
Tutorial 7
Deconvolution
Adjust each beta weight to minimize the error.