Often the question arises, if galvanic skin response (GSR) and facial expressions are both indications of emotional response, what is the difference between the two? When should you measure one or the other? We can take a look at an example with real data. 104 participants watched a documentary featuring several magic tricks, while their biometric responses (including GSR and facial expressions) were recorded.
One magic trick featured was the “Beuchet Chair Illusion”. This is a visual perception illusion using forced perspective to make things look abnormally small. Let’s take a look at the response pattern.
We can see the GSR pattern in the top track. Commonly, we measure GSR in terms of aggregated ‘peaks’, which is shown here. Peaks occur when there is a sudden change in the number of sweat glands that are activated – an indication of emotional arousal. As we can see here, there isn’t much change in the number of GSR peaks as our audience observes the illusion. This means the illusion isn’t causing a large amount of emotional stimulation.
If we look at the bottom track though, we can clearly see a large rise in the response pattern. This purple line is an indication of overall emotional expressivity displayed on the face, and is the average of all our respondents. We can break this response down into specific emotions (e.g., joy) or expressions (e.g., smile), but the overall average of emotional expressivity tends to be a good indication of how engaged the audience is with what they are viewing. In this case, the strong rise demonstrates that our audience was visibly reacting to this illusion, indicating that they are emotionally engaged with the illusion.
So the Beuchet Chair Illusion was engaging, but not overly stimulating. We could interpret this illusion to be relaxing – it’s engaging enough to hold attention, but it’s not going to cause you to break a sweat (literally!).
Let’s take a look at another example, the Rubber Hand Illusion (watch the actual clip here). This is a sensory illusion where the participant’s brain is tricked into thinking a fake rubber hand is their own through synchronous touch (a paintbrush is brushed against both the fake hand and the real hand simultaneously). In this particular scene, after the participant has become assimilated to the rubber hand, the experimenter suddenly pulls out a hammer and strikes the rubber hand, causing a startling sensation.
This illusion is much more stimulating: the GSR peaks (top track) more than double during the course of this scene. Facial expressivity (bottom track) is also high, experiencing a sharp rise and indicating the audience is thoroughly engaged with the magic trick. Thus, this trick is both stimulating and engaging. We could interpret this stimulus as being exciting – it prompts both a visible facial reaction and causes some increase in the number of sweat glands being activated.
The diagram below demonstrates a simple method to interpreting this mix of biometric signals.
While they do provide different measures, GSR and facial expressions are both measuring broader emotion. Thus, the response pattern between the two signals does tend to correlate (not perfectly, as in the examples above). People may also have a range of responses to the same stimulus. In these cases, we must use all the information together, including expressions of specific emotions (e.g., joy , surprise, anger) or the frequency of GSR peaks to provide further context. Nonetheless, this diagram is a handy way of quickly interpreting a mix of signals, and demonstrates how our biosensors provide subtly different information.
In the end, which biosensor is more relevant to you will depend on the goals of your stimulus. Although the only way to get the complete picture is to use them all together, your goals may be more suited to particular outcomes. Get in contact with us, and we can work on a tailored solution that best fits the scope and goals of your project.