How to display PLS results of ERP data All the PLS results are saved in '*_ERPresult.mat' files. The result windows will provide you a convenient way to display: 1. Significant LVs: Plot Singular Values 2. Electrode Saliences 3. Stable effects (bootstrap results) 4. Average Amplitudes 5. Detail View for Electrodes 6. TaskPLS Scores (Scalp Scores, Design Scores) 7. BehaviourPLS Scores and Correlations When you open the ERP results window, the default plot is the electrode saliences for the first LV. If the file contains bootstrap results, stable differences are also displayed. Clicking on the wave will display the channel name and LV number. - File menu From the File menu, you can export the plot to a file or print out the plot. - Edit Menu Electrodes can be selected by clicking on the electrode name or choosing 'Select Electrodes with Rubberband' from Edit menu. Clicking on the selected electrode again will deselect the electrode. Choose 'Select All Electrodes' or 'De-Select All Electrodes' to deal with a large number of electrodes. - Tool Menu 'Display Options': select which ERPLVs to display. Note that for TaskPLS, the last LV is the GrandAverage, and will be essentially zero. 'Bootstrap Options': The default threshold for the bootstrap results is 95% of the range in the bootstrap ratios for the LV. As the bootstrap ratio is similar to a Zscore, the threshold should typically be >~2.5. - View Menu 'Average Amplitude': this will open the window to plot ERP grand averages. The default display is the average for the first condition, and the bootstrap ratios for the first LV. Clicking on a wave will display the channel and condition names. - View Menu (p.2) If you click 'Display Condition Difference' under 'Tool menu, you can also monitor the difference between conditions. 'Plot Detail': If any electrodes have been selected, a detailed view, including bootstrap results, will be plotted. (currently no legend is plotted - colours are the same as in the results window) 'Plot Singular Values': This opens a bar graph window, which, through its 'View' menu, you can display: 1. Observed Singular Values for the LVs. 2. Proportion Crossblock Covariance accounted for on each LV. 3. Permutation test results (if the number of permutations is greater than 0). This plot indicates which LVs are significant. - View Menu (p.3) 'Plot Design Scores': This window displays the contrasts distinguishing tasks for each LV. The default is to plot the results for the first LV, and first group. Use the 'View' Menu to switch between the following plot types. 1) Scatterplot of ScalpScores (BrainScores) vs Design Scores: This plot is ideal for determining if there are any outlier subjects, or groups of subjects. Clicking on a datapoint will identify the subject, and highlight their ScalpScore in each condition. Selected subjects can be removed from the datamat in the 'Edit' menu. You will be prompted to modify the datamat when the window is closed. 2) Design Scores: This bar graph plots the design scores in the order of the conditions. This plot is useful for more clearly visualizing which conditions differ. 3) Design LV: is proportional to the Design scores. - View menu (p.4) 'Plot Canonical Scores': This plots scalpscore across the time periods. 'Contrasts Information': This displays the contrasts data in bar graph. 'Plot Scalp Scores': This window displays the correlations of behaviour with Scalp scores (brain scores) for each LV. The default is to plot the behaviour-scalp score scatterplot for the first LV (first condition, first behaviour). Use the 'Plot' Menu to switch between the following plot types: - View menu (p.5) 1) Scatterplot of ScalpScores (BrainScores) vs Behaviour: This plot is ideal for determining if there are any outlier subjects, or groups of subjects. Currently, you must use the legend to identify subjects. The scroll bar on the right will move through the conditions, and the scroll bar on the bottom will move through the behaviours. Under the 'Option' Menu, clicking on 'Change Plot Dimension' will redisplay the scatterplots so that more than one plot is on the screen at a time. Currently, each behaviour will only be plotted in a single column. 2) Correlation bar plot: This is available only if bootstraps were run, and is activated by 'Hide BrainScore Plot' or 'Show Correlation Plot' in the 'Plot' Menu. These are barplots of the brain-behaviour correlations, with the 95% confidence interval for the correlation indicated. Correlations where the confidence interval includes zero are considered unreliable.