The road to quality data can be rocky at times. Overwhelmed by the dos and don’ts and better-nots, it is almost impossible to not feel swamped by the complex demands of multi-modal experimental setups and the constant pressure for outstanding data that sets you apart from the crowd.
We’re here to help you travel on the bright side of the road to best data quality. As a first stop along the way, let’s check on your experimental environment as great data begins right there.
We recently kicked off a dedicated series of blog posts to hook you up with everything you need to know to create the optimal physical framework for your biometric research. In the course of the last three weeks we have been feeding you with our hottest tips on eye tracking, facial expression analysis, and GSR. In case you missed them, no worries – just follow the links and click yourself smart.
Let’s wrap up our series this week with electroencephalography (EEG). We have a hunch you’ll be quite thrilled at how surprisingly simple it is to arrange for a proper study environment with just a few small steps (so much for mystery).
Let’s get you on the right track now.
There is no substitute for clean data
If you follow our blog on a regular basis, these words might not come as a novelty to you. We have brought it up numerous times, and yet we can’t emphasize it enough – there is no substitute for clean data. It’s the one dogma you simply can’t get around if you’re shooting for strikingly good EEG data (which you surely are). So embrace it, dance around it, make friends with it. Only then you’re off to the high road in EEG research.
In its essence, the message is evident: Reduce data contamination to a minimum. That, of course, means before and while you acquire data. Don’t try “cleaning up” heavily contaminated or very noisy data during post-processing. Seriously, you don’t want to opt for that solution.
Sounds easy to achieve in theory? It is in practice, too. Here’s what you can do.
1. Electrode placement:
EEG records electrical activity generated by the brain. As the electrodes read the signal from the head surface, it is important that they are in proper contact with the skin throughout the recording for an accurate EEG reading. And that’s exactly what you’re striving for. Now how do you get at it?
- Instruct respondents to wash their hair before the session and not use any styling products. Why? Greasy residues from hair styling products (gel, wax, spray etc.) and shampoos can impact the signal transmission to the electrodes inducing unwanted artifacts in the recording or, at worst, rendering data acquisition even impossible.
- Always clean the recording sites with an alcohol swab to remove sweat, tallow and old skin particles which could deteriorate data quality.
- Assess the physical characteristics of your prospect respondents prior to running the experiment: EEG readings from (entirely) bald participants can pose challenges as the skin of their scalp is harder, resulting in higher resistance and reduced data quality. Recording EEG from women generally yields better data quality as female skulls are thinner. On the other hand, respondents with extremely thick hair might turn out to be tricky, too, as the electrodes might not adhere firmly to the head surface – a definite no go for clear data reading.
2. Body Movement:
Keep in mind that any sort of body and head movements – even minor ones – can impact your recording. Movement-related muscle activity generates much stronger electrical activity compared to brainwaves, resulting in a complete masking of the cognitive effects. In both stationary and mobile EEG setups you definitely do well to instruct respondents to keep movements to a minimum. Whenever bodily movements cannot be avoided, develop a sharp eye for movement-induced artifacts, place annotations in the data, inform respondents and make appropriate adjustments if need be. Be on the lookout for any of the following:
- Respondents should not be chewing gum.
- Facial activity such as laughing, smiling, chatting, excessive yawning or blinking will induce artifacts.
Don’t think overly complicated when it comes to clean data. One key to accurate measurements is keeping noise and distractions from the surroundings at a minimum. Noise from adjacent rooms, corridors (people chatting, for example) or even the street in front of the building (noise from traffic) can constitute possible sources of disturbance that might interfere with the EEG signal. If possible, have a dedicated, sound-proof room for your study without people frequently dropping by. Ideally, choose a remote area without the daily hustle and buzzle throwing your respondents for a loop.
As trivial as it may appear, unfavorable lighting conditions may in fact mess with data. Flickering light, fluorescent lighting or rooms lit up too brightly may cause your respondents to squint at the screen. Needless to say this will induce severe artifacts that should be avoided under all circumstances. If possible, try to arrange for stable lighting conditions that make your respondents feel at ease rather than to force them into eye acrobatics to accomplish the task (which is guaranteed to make you chuckle, however does not lead to the desired result).
5. Pretest and Impedance Check:
Before kicking off your experiment, check if everything (sensors, respondents etc.) is set up properly and working smoothly. Ensure low impedances at all electrodes to be sure that the recorded signal is picking up brain waves instead of electrical noise from the surroundings. Only then you can be really sure of receiving a stable, reliable EEG signal. Don’t let the slogan “hindsight is easier than foresight” become your mantra – be smarter than that and pretest!
We’re here for you to help smarten up your data. Contact our experts at iMotions to get even more tips and tricks that will make your data shine!