Using the GUI (3): Create smoothed fixation matrix
Smoothing
This step produces a 3D smoothed fixation map (number_of_item × xSize × ySize). Raw fixation map is smoothed by convolve with a Gaussian Kernel in two ways in iMap4:
- The estimated method - the first dimension coded for each condition in each participant;
- Single-trial method - the first dimension coded for each single trial.
A pop-up will first appear for you to decide the smoothing parameter. iMap4 uses a Gaussian Kernel with the full width at half maximum (FWHM) set to 1° visual angle. iMap4 will automatically convert the visual angle into pixel value, notice that here in this example we will set it to 20 pixel (i.e., the Smoothing parameter as shown below).
You can click Generate on the left to visualize another subject/Trial, or you can change the parameter and click Validate to see the smoothing effect. The smoothing will begin when you click on Done. Finally, a pop-up will show you the directory where the resulting matrixes have been saved.
Alternatively, you can select the Single-trial method. It is the most appropriate if you have a continuous predictor variating at single-trial level. You will need to perform rescaling first to reduce memory usage and computational time. A pop-up will lead you to perform rescaling. (It’s the same one when you click on Rescale in Optional). After that, the procedure will be identical to the estimated method.
There are a few optional preprocessing steps you could apply. Alternatively, you can proceed to visualized the data and perform modeling using Linear Mixed Model