2 Games User Research & Physiological Evaluation
2.3 Physiological Evaluation
2.3.1 Physiological Measurements
The human nervous system (that has an essential role in the control of behaviour) is split into two parts: one is the central nervous system (CNS), which contains the majority of the nervous system and consists of the brain and the spinal cord. The other is the peripheral nervous system (PNS or PeNS), which contains the nerves and nerve cells outside of the brain and spinal cord. The main task of the PNS is to connect the CNS to the limbs and organs. As the PNS transmits our physical sensations, and unlike the CNS is not protected by the skull and bone of the spine, it makes it easier to access measurements via our skin. The PNS itself is also divided into the somatic nervous system (SoNS), which regulates voluntary bodily activity, and the autonomic nervous system (ANS), which takes care of our unconscious responses and controls internal organ functions. Controlling unconscious responses make the ANS more suitable for
physiological evaluation. The ANS is also divided into two subsystems: sympathetic nervous system (SNS), which is to mobilise the body’s nervous system fight-or-flight response, and the parasympathetic nervous system (PSNS), which is responsible for the stimulation of rest-and- digest or feed-and-breed activities.
Sensor technologies enable researchers to use physiological measurements for testing or quantifying a user’s feelings. Depending on the research question (dimensions of feeling to explore) there are various measures that can be taken using different physiological sensors. Common physiological measures in game research include skin conductance (SC),
electromyography muscle measures (EMG), Electroencephalography (EEG), skin temperature, respiration rate and electrocardiography (ECG), where several measurements can be computed from ECG such as interbeat intervals (IBIs), heart rate (HR), heart rate variability (HRV) (Kivikangas et al., 2011a).
This thesis utilised galvanic skin response (GSR) computed from SC and facial EMG, with the aim to cover both dimensions of users feelings (see Figure 2-3), hence the next two sections (2.3.1.1 and 2.3.1.2) explore these two measures in more detail and the reasons they were chosen (despite other possible measurements) to answer the thesis’s research questions.
Common approaches distinguish physiological analysis on a temporal dimension: Studying phasic psychophysiological and behavioural responses at game events (points in time) (e.g.(Ravaja, Turpeinen, Saari, Puttonen, & Keltikangas-Järvinen, 2008) ) and studying tonic responses to variations of in-game variables (time span) (e.g.(Mandryk & Atkins, 2007)).
Figure 2-3 Russell, Weiss, & Mendelsohn, (1989) model of arousal and valence. The relationship between arousal/valance and physiological signals has since been applied to measure player experience. Explanations of other mentioned physiological measures are not covered here, as this was not the focus of this thesis. However there are plenty of resources available; for example (Cacioppo et al., 2007) explored these measures in detail. In terms of the applications of these measures, section 2.4 provides some examples of how studies used them.
2.3.1.1 Galvanic Skin Response (GSR)
Arousal is commonly measured using SC (Lang, 1995), also known as galvanic skin response (GSR which is used in studies conducted for this thesis - when SC is measured as a direct response to a stimulus) or electro dermal activity (EDA - when SC is measured over time) and depends on how it is computed (Boucsein, 1992).The conductance of the skin is directly related to the production of sweat in the eccrine sweat glands. In fact, subjects do not even have to be sweating to see a difference in GSR because the eccrine sweat glands act as variable resistors on the surface. As sweat rises in a particular gland, the resistance of that gland decreases even though the sweat may not reach the surface of the skin (Stern, Ray, & Quigley, 2001). GSR has a linear correlate to arousal (Lang, 1995) and reflects both emotional responses as well as cognitive activity (Boucsein, 1992).
It is recommended to place GSR sensors to the fingers, palms or toes as there are more sweat glands in those areas, which make them more likely to react to changes in PNS. Although electrodermal activity can be measured from any of these sites, the values obtained are not necessarily comparable.
Because of relative ease of measurement and quantification combined with its sensitivity to psychological states and processes, GSR measures have been applied to a wide variety of questions ranging across examining attention, information processing and emotion (Cacioppo et al., 2007).
GSR has been closely linked with the psychological concepts of arousal and attention.
Woodworth & Schlosberg (1954) supported this indexing relationship by noting that tonic GSR is generally low during sleep and high in activated states, such as rage or mental work. They also related phasic GSR to attention, noting that such responses are sensitive to stimulus novelty, intensity and significance. This thesis utilises phasic GSR since it focuses on the analysis and visualisation of player’s arousal state and mapping this to in-game micro events.
There are two methods for measuring GSR: One, exosomatic: which relies on the passage of an external current across the skin. Second, endosomatic: which is recording the skins potential response without an external current. Most commercial physiological measurement kits use exosomatic methods for recording GSR, and this is the method of choice among researchers (Fowles, 1986).
The principle in the measurement of skin resistance or conductance is that of Ohm's law, which says skin resistance (R) is equal to the voltage (V) applied between two electrodes placed on the skin surface, divided by the current (I) being passed through the skin (R=V/I). Therefore, if the current is held constant then it is possible to measure the voltage between the electrodes (which will vary directly with skin resistance). Alternatively, if the voltage held constant, then the measure of the current flow would be skin conductance (which will vary directly with the reciprocal of skin resistance). Conductance is expressed in units of Siemens and measures of skin conductance are expressed in units of microSiemens (!S). Chapter 5 provides details on computing GSR measurements as utilised in the BioSt tool and study reported in Chapter 6. For studies reported in Chapters 3 and 4, GSR data was gathered using the BIOPAC hardware system, sensors and software from BIOPAC Systems Inc. This was measured by using two passive SS3LA BIOPAC electrodes. The electrode pellets were filled with TD- 246 skin
conductance electrode gel and attached to the ring and little fingers of the participant’s left hand. Electrodes sites should be in a natural condition and not be cleaned by alcohol or abrasion. However, it is recommended for participants to wash their hand with a nonabrasive soap before
having the electrodes attached on to clean and dry skin. Room temperature, humidity and time of day are two environmental factors that should be controlled (Hot, Naveteur, Leconte, & Sequeira, 1999). Boucsein (1992) recommended a room temperature of 23 C.
2.3.1.2 Facial Electromyography (EMG)
As discussed earlier emotions can be interpreted in a two dimensional model: arousal and valence (Russell et al., 1989). While GSR can help us on uncovering feelings related to arousal, the final study of this thesis (Chapter 6) utilised measurement of facial muscles as a way to interpret valence. Hence this section looks into electromyography (measurement of muscle activity) and more specifically facial muscles.
To measure whether a muscle is active or not, an EMG electrode needs to be attached to the surface above a muscle to be able to sense the slightest activation (Lang, 1995). Therefore, depending on the placement of EMG sensors, most muscle activities can be measured. However, measurement of facial muscles is the most established for evaluating valence (Fridlund &
Cacioppo, 1986). Especially, measurements of brow muscles (corrugator supercilli) and cheek muscles (zygomaticus major) to indicate positive or negative reactions to game events (Hazlett, 2008).
As EMG sensors measure signals released due to a muscle activity they need a reference for comparison. This reference sensor should be placed on an area without any muscle; for facial measurement it is common to attach the reference sensor to an ear lobe. Placing sensors on a participant’s head can be intrusive and introduce movement artefacts. Also as facial muscles will be easily activated (for example by talking) participants need to be informed not to talk or move their head while signals are recorded, thus blocking any form of talk aloud protocol. Although these artefacts mean careful data interpretation, analysis of EMG signals is not complicated.
Chapter 5 explains EMG analysis for BioSt tool in detail, where NeXuS-10 MKII device and sensors were used to record GSR and facial EMG for the BioSt tool. For Zygomaticus major (smiling) and corrugator supercilii (frowning) facial muscle activity were measured using passive EMG sensors on a player’s cheek, brow, and ear lobule (for ground sensor). Recording software was a custom C++ application using the NeXuS SDK to collect raw data from the device and display the recording timestamp on the computer screen.
As discussed earlier, physiological measures are often applied in conjunction with other user research methods to provide context to these sensitive measurements. So far this chapter provides the introduction to physiological measurements and discussions on their contributions
and limitations. The next section looks at how these measures are used in combination with other user research tools.