EDA
Frazer Findlay, CEO
Acquisition
· Specific vs non-specific skin conductance response
· Precise time-locking to stimulus important
· Store electrodes cool and sealed, so the gel doesn’t dry out. And use extra gel, especially if electrodes get old
· If extra gel, the electrode needs to sit on the subject for a while in order to make a good connection
· In the scanner: needs filtered patch panel eliminate high-frequency noise; if not set up correctly, data is noisy also needs physical amplifier (STP100C) if you’re not exclusively using a wireless tool
· Radio-translucent electrodes (look the same as the carbon fiber ones but are different): arrive dry, use extra gel (Gel101, NOT a regular gel!)
· If subjects get cold hands (or feet if measuring there) is problematic; ensure good room temperature.
· Use camera systems to mark artifacts (sneezing, moving aso.), monitoring room temperature; you can use additions of focus areas to highlight such noise online during acquisition how to exclude such periods, you can use connect end points
Processing
· The little ‘I’ (info) button on the right hand bar of a channel is not there right after data collection, it only appears after something has been done to the data. You can check what has been done. This one shows that the data has already been resampled from 2,000Hz to 100Hz (Biopac samples everything at 2,000Hz by default)
· Missed the exact data selection and filtering option: remove high-frequency noise, happens if contact between electrode and skin is poor
· FIR: finite-impulse response à use this one, nice and sharp cutoff without much data distortion
· IIR: … impulse response
· Usually creates a new channel with filtered data to preserve original data but this can be turned off if desired
· Now convert peaks of stimulus presentation into stimulus event markers
· If you only have a handful of stimulus events, you don’t need to use the latency option (use zero); you can use it if you have hundreds of different ones, pick this option
· Now we have light bulbs à in event palette, you see stimulus types 1 and 2 (1=happy, 2=sad)
Preferences
· The first time you use the EDA option, you are asked to set your preferences; set those and then use these presets for ALL participants!
· Tonic: absolute signal coming from the electrode
· Phasic: transformation of tonic waveform applied to events à proper filtering of waveform important to get a proper phasic signal
o Recommended option: “Smoothing baseline removal”
o Second option: high-pass filter à
· Stick with smoothing baseline, high-pass filter can sometimes take a while to settle down
· Threshold as low as you can to identify stimulus responses; recommended .02 micro siemens but can be lower
o Baseline: baseline = the moment of stimulus onset and then compared to peak à baseline can shift over time
o Opening parentheses “(“: beginning of SCR crossing our .02 threshold; as the signal on the phasic waveform drops down, it adds a closing parenthesis “)”
o If you click on one, it places the cursor at the exact position of threshold crossing
o Check lightbulbs: timing when stimuli were presented and investigating signal there
o Channel 1 tonic; after SCR analysis, the phasic waveform was created (here channel 5)
o A response needs to happen AFTER the stimulus presentation; at least 1 second (signal transmission) and less than 4 seconds. If you see an immediate response, that is likely due to something that happened previously; i.e. exclude responses too close (minimum) or too far away (max) from your stimulus presentation
o Either report within a time window (e.g. 10 or 20 seconds), or report per event
o Using the time period option, we get a summary of all different types of analyses; by event would be better for block design
o Event markers now show “1” and “2” for type of stimulus (happy and sad, respectively)
Summary measures
o How many responses occur within an area?
o Select area; use channel info bar and select Event count à number shown is number of events (droplets)
o Can also do amplitude (mean amplitude)
o Can also use find cycle or even better: Analysis à Epoch analysis