What is Galvanic Skin Response (GSR)?
The human body is comprised of various systems that respond to stimuli. Just like the body’s “fight or flight” response may cause our heart rate to increase when we hear a frightening noise, our skin has electrical properties that change based on our arousal levels. These electrical properties are commonly referred to as our Galvanic Skin Response (GSR). Fortunately, for scientific purposes, our GSR is fairly easy to measure with a few pieces of technology. Insights drawn from GSR data can reveal an incredibly accurate and fine-tuned look at what we experience on a moment to moment basis. Among the many aspects of the skin’s electrical properties, skin conductance (SC) is perhaps the most widely studied and applied measurement in research.
Like any substance, our skin contains certain conductive and resistive properties. That is, it permits (i.e., conducts) a portion of an electrical current to pass from one point to another, while not permitting (i.e., resisting) a portion of that same electric current. Typically, when we think about conducting electricity, we think of metal objects, such as lightning rods. The reason why we use metals for conducting electricity is that they are comparatively good conductors of electricity. More specifically, they do not resist the flow of electricity very much. By contrast, our skin is a relatively poor conductor of electricity. Meaning, it resists more of an electric current than it actually conducts over a given distance. So, the question is why do we rely on skin conductivity in science, if our skin is a poor conductor of electricity?
The answer is in our sweat.
Unlike the epidermis and dermis layers of our skin — which are poor conductors of electricity — our sweat is rich with electrolytes such as sodium and chloride, which makes our sweat a surprisingly decent conductor of electricity (though not quite as good as metal).
What’s more, our bodies don’t just sweat when we are hot. Every square inch of our skin is releasing traces amounts of sweat, regardless of temperature.
The essential piece of the puzzle is the fact that we sweat in response to the stimuli we perceive. This could be in the form of observable stimuli (e.g., a sudden loud noise) or non-observable stimuli such as our thoughts (e.g., did I leave my stove on this morning?!). More often than not, however, the data collected through EDA relates to stimuli in our environments. This is likely due to the fact that our environment is rapidly changing, and often, our thoughts change with them. Unless you’re daydreaming during a company meeting, odds are good that fluctuations in your EDA response is due to observable, external stimuli.
When it comes to EDA, more sweat is not necessarily better from a data collection standpoint. Indeed, contrary to what we experience, areas like our armpits don’t contain the right type of sweat gland for EDA use. Areas like our armpits have high concentrations of apocrine sweat glands, which do not respond as quickly in response to changes in environmental stimuli.
Instead, the best sweat glands for EDA are eccrine sweat glands. Unlike apocrine glands, eccrine sweat glands are distributed across every inch of our skin’s surface in varying degrees of concentration. As a general guideline, areas along our fingers, head, palms of our hands, and soles of our feet have the highest concentration of eccrine sweat glands. Somewhere in the neighborhood of 400 eccrine sweat glands per square centimeter (400/㎠), depending on the body area in question.
Other areas tangential to these regions, such as the proximal areas of our fingers and toes, wrists, ankles, and the instep of our feet are also fairly rich with eccrine sweat glands too. By contrast, regions of our torso — along our abdomen, chest, and back — have relatively low concentrations of apocrine glands. Interestingly, males and females average the same total number of eccrine sweat glands across the entire skin’s surface.
How does GSR technology work?
The idea behind GSR technology is deceptively simple. The majority of GSR sensors on the market today fall into the category of “exosomatic measurement with direct current”. While this category is a mouthful, it means all GSR measurements take place outside the body (i.e., exo- = external; somatic = body), through the aid of a direct electric current.
During GSR data collection, a small direct current is applied by one electrical lead attached to the participant’s skin surface. The current is far below a perceivable threshold and poses no possible health risk to participants. Typically, this is somewhere in the ballpark of 0.5 V, about three times less than a hearing aid battery. Due to this low voltage, the electrical current is not large enough to penetrate our epidermis layer.
A second electrical lead is placed close by the first one. Usually, within 2 inches of each other. The purpose of this second lead is to “read” the level of direct current that passes through the surface of the skin that spans between the two leads.
This works because there may be several hundred eccrine sweat glands between these two leads. As the person unconsciously “sweats” in response to changes in environmental stimuli, the level of skin conductivity between these leads changes from one moment to the next. It is important to note, however, there is a short lag period between when a stimulus occurs and when our sweat glands respond. Often, this latency period is between 1 – 3 seconds, depending on the magnitude or intensity of the stimulus response as well as the participant’s individual skin characteristics. As a result, when reviewing GSR data, don’t be surprised if any “peaks” in GSR don’t line up perfectly with the presentation of a stimulus.
Modern GSR units are designed to capture several dozen or hundred measurements each second. Devices labeled as 120Hz sample at a maximum rate of 120 times per second, while devices set to 2,000Hz sample at a maximum of rate of 2,000 times per second. While the latter can seem like overkill, certain reaction time tasks do warrant extreme precision. A good rule of thumb to follow is that you should sample at a rate of at least twice your potential stimuli presentation rate and/or expected between subject differences to ensure that you’re accurately capturing a meaningful sample, rather than noise or other study artifacts. However, depending on the task, some recommend sampling at least three or four times your expected differences (Wescott, 2010).
If, after your data have been collected, you decide that your sampling rate was higher than necessary, you can always downsample to a lower level (e.g., 120 Hz to 30 Hz). It’s a better strategy to downsample afterwards, rather than wish you had used a higher sampling rate from the start. Just like a haircut, you can always take more away afterwards, but you can’t add more back on after its finished.
Where should GSR leads be placed for the most accurate results?
In a recent study, van Dooren et al. (2012) conducted a comparative study looking at 16 different GSR lead placements to determine which areas afforded the most consistent and large results. Each participant was equipped with all 16 GSR lead placements simultaneously while they watched a series of brief videos intended to elicit different types of emotional responses. Lead placement included the proximal region of the fingers (index and ring fingers), multiple regions and orientations along the wrist, shoulder, back, buttock, inner thigh, abdomen, chest, ankle, calf, forehead, armpit, upper arm, and neck.
van Dooren and colleagues hypothesized that the fingers, feet, and forehead would provide the most consistent and robust data among all lead placement areas, because these areas have the highest concentration of eccrine sweat glands.
Results from their study suggested that while each area provided meaningful and useable data, the fingers and feet — not the forehead — were among the most responsive regions in regards to skin conductivity amplitude (i.e., largest peaks in response to stimuli). In the next tier down, the various wrist locations, shoulders, calf, and chest revealed moderate skin conductance responsiveness. In the bottom tier, areas such as the abdomen, back, armpit, and upper arm provided the lowest skin conductance responsiveness overall.
What are some trade-offs that affect GSR lead placement?
Although van Dooren et al. (2012)’s study provides an initial recommendation for GSR lead placement, sometimes contextual factors require you to make tradeoffs. A common situation is when your participants will have to (or may choose to) use both hands to complete a task, such as driving an automobile, or typing on a keyboard. In these situations, the ideal proximal finger location for GSR lead placement (or the wires connecting it) is susceptible to being bumped, resulting in noisy data collection. Other circumstances, such as the need to wear personal protective equipment (PPE), may restrict the location of GSR leads even further. For example, riders on a motorcycle may need to wear thick, long riding gloves. In these situations, finger and wrist locations may not be possible.
Another factor to consider is how much ambulatory movement the participant will experience while completing your study. Large muscle movements, such as walking and lifting objects can quickly introduce noise to your data. In an ideal setting, GSR data collection should take place while participants are comfortably seated, experience minimal vibration, and do not need to exert large forces to interact with devices.
Keeping these tradeoffs in mind can make the difference between successful and unsuccessful data collection efforts, when considering the use of GSR in your next study.
Why use biometric research?
As an example, one of the limitations of survey-based studies is that participants are typically administered questions following an experience or interaction with a product or system. As a result, what you are actually testing with these surveys is the participants’ “aggregate” experience. That is, their initial reactions, discoveries, mistakes, and final thoughts all summed up into one survey. Post-hoc study interviews help remedy this issue, but cannot resolve it completely. Especially when your testing session is longer than 30 minutes, this aggregate experience is suspect to biases, such as described by the “peak end rule”, as well as primacy and recency effects.
By contrast, biometric technology give us an incredibly accurate, reliable, and fine-tuned look at what people experience on a moment to moment basis, in real time. That is, it samples participants’ physiological responses several hundreds of times each a second. These data afford a rich, reliable, and honest look at what people physically experience when they interact with a product or system.
How can biometric research help?
Through the use of biometric technology, we can go beyond survey measures to better answer the following types of questions:
- Where do users expect information to be?
- Which parts of a design do users attend to?
- Which parts of a design to users NOT attend to?
- How long do users fixate on certain elements?
- At what level of arousal are users when X occurs?
- Which of my two system versions elicit the largest emotional/arousal response?
The additional insights obtained with biometric research data allows researchers to make more informed recommendations, improving a product or system’s usability, functionality, and desirability.
Braithwaite, J. J., Watson, D. G., Jones, R., & Rowe, M. (2015). A guide for analysing Electrodermal Activity (EDA) & Skin Conductance Responses (SCRs) for psychological experiments (Revised version 2.0). Retrieved (6 April, 2018) from http://www. biopac. com/wp-content/uploads/EDA-SCR-Analysis. Pdf.
van Dooren, M., & Janssen, J. H. (2012). Emotional sweating across the body: Comparing 16 different skin conductance measurement locations. Physiology & behavior, 106(2), 298-304.
Wescott, T. (2010). Sampling: what nyquist didn’t say, and what to do about it. Wescott Design Services, Oregon City, OR.
About the Author
Joe O’Brian | Senior Human Factors Scientist | Research Collective
Joe O’Brian is a Senior Human Factors Scientist at Research Collective. He has co-authored articles on topics ranging from judgment and decision making to education and healthcare technologies. At Research Collective, his contributions include project planning, observational and biometric research, and advanced statistical analysis for major automotive and healthcare organizations. Joe can be found on LinkedIn here.