Waveform Quantification & Statistics

ERP Component Measures

1. We can measure the physical properties of the ERP components:
   1. Amplitude: how large is the deflection?
   2. Latency: when does the deflection begin or peak?

Amplitude

1. What should I quantify?
   1. Amplitude: How strong was the response in voltage (µV)?
      1. Peak Amplitude = Baseline-to-peak Voltage, amplitude of only the peak value
      2. Mean amplitude: average voltage of all these points or points within the time window
2. Peak amplitude vs. mean amplitude
   1. Peak amplitude: can be appropriate for quantifying sharp peaks
   2. Mean amplitude: can better capture the strength of the response when the component is more broad and shallow

Latency

1. What should I quantify?
   1. Latency: When did the process occur in time (ms)?
      1. Peak latency: The time from the stimulus when the component peaks
      2. Onset latency: Time when component deflection begins
      3. Response: What is the timing of the response?
2. Fractional Area Latency: alternative to peak latency
   1. Peak latencies are also affected by noise in the data
   2. When a component achieves some threshold of its total voltage in the window
   3. 50% fractional-area latency
      1. Measure time point at which 50% of area under the voltage curve defined by the time window has accrued
   4. Less prone to outlier voltages and noise

Measurement Parameters

1. Time window of measurement
   1. Whether you are measuring latency or amplitude, you need to tell your software how to identify the component or peak
      1. Example: Peak amplitude or latency-- find the most negative value between 140 and 190 ms
2. How do I choose the right time window?
   1. Examine prior research on that component or experimental design
      1. Timing of window
      2. Size/width of that window (ms)
         1. E.g., How wide is the typical N170 component?
   2. Examine your grand averages
      1. Combine conditions for timing
      2. Your time window should include most of your participants’ data
3. How do I choose the right electrodes?
   1. Which sites typically show the effect you are interested in?
      1. Select based on the literature
      2. Check the scalp distribution

Difference Waves

Difference waves: Quantify condition difference

1. Difference waves
   1. Show how ERP components differ between conditions
   2. Similarities in source waveforms (e.g., perceptual inputs) are eliminated but differences remain
2. Created by subtraction
   1. Condition B waveform - Condition A waveform
   2. Quantification carried out on difference waves
3. Trade-offs:
   1. Benefit: simplicity
   2. Risk: doesn’t allow isolation of effect to a particular condition
   3. Always look at source waveforms

Data Spreadsheets

From waveforms to spreadsheets

1. Quantified data measures from the waveform get pulled into a spreadsheet
   1. The numbers going into the spreadsheet are values for amplitudes/latencies
      1. Each participant has their own row
      2. Each condition is in a different column
      3. Organized by electrode/hemisphere
         1. Example: Columns: PO7(electrode) Faces (condition), PO7 Cars, PO8 Faces, and PO8 Cars, and Rows: Participants

Quantification Walkthrough: Time Window of Interest

1. Open Matlab, Set Path to ERPlab plugins, open EEGlab
2. Load a Grand Average (GA) to determine the time window of interest
   1. ERPlab > Load existing ERPset > Select your GA file
      1. You can see loaded ERPset in the ERPsets menu
3. ERPlab > Measurement tool > Make sure your ERPset is selected in the left pane, and modify the settings
   1. Below are the settings used in this video
      1. Measurement: "Mean amplitude between two fixed latencies"
      2. Bins: 1 & 2
      3. Channels: 9 (CPz)
      4. Window: 400 to 600
   2. Click Viewer to see the GA file's plot of Bins 1 &2 at CPz, with the time period from  400 to 600 ms highlighted
4. Viewer Navigation Tools
   1. Under the plot is information about the File, Bin, Channel, Measurement, Window, Latency, and measurement Value.
   2. To the right of the plot, from top to bottom, displays:
      1. Measurement Type
      2. X and Y scales, positive can be plotted up or down
      3. Plotted values from selected Bins, Channels, and Files
         1. Clicking "All" next to these options displays all of the previously selected Bins, Channels, or Files.
      4. Window highlight color, and value line color can be changed in "Color for Measurement"
5. Try changing the measurement window in the Measurement Tool (here used 240-440ms) to shift the highlighted area on the Viewer
   1. The values exported from the Measurement Tool come from the Time Window that you set in the Measurement Tool. Make sure you are highlighting the areas you are interested in.
6. Once you decide on your time window, load the individual participant averages, rather than the GA

Quantification Walkthrough: Extract Individual Data

1. Load individual participant averages using ERPlab > Load Existing Datasets
2. Return to the Measurement Tool (ERPlab > ERP Measurement Tool)
3. Adjust the .erp files selected
   1. From ERPset Menu: select the appropriate erpsets
      1. Here used 2:11
4. Check the settings are what you want to examine and click Viewer
5. View selected bins, at the selected channel, for each file you selected
   1. Scroll through each selected file and check if the window you defined works for each of them
   2. If you have more than 1 channel selected, you can overlay the channels using "All"
   3. Return to the Measurement Tool to either change settings, or save an output file.
6. Under "Save output file as", select Browse, name the output file, and navigate to where you will save the file
   1. Specify the format of the .txt file ("One ERPset per line (wide format)" used here)
7. Click Run
   1. This will save a .txt file with the data you specified in the Measurement Tool
   2. Click Yes on the yellow pop-up which says exporting an excel file is no longer supported by ERPLAB.
8. Open your .txt file (you can open it with excel)
   1. The data for each bin at each electrode that you specified will be in the file, which you can then use for statistical analysis