[Posted on 14 August, 2018 by Martyna Galazka-Carney]
Recently I came across an article which I could not stop thinking about long after I read its last sentence. The article, from the Journal of Personality and Social Psychology (JPSP), was written by Dariusz Dolinski from the SWPS University of Social Sciences and Humanities in Wroclaw, who analysed 45 empirical articles published in the last six issues of the JPSP. The article, although only considering this particular journal, reported an interesting finding: only 4 of the 45 articles comprising 290 individual studies went beyond reporting how the participants answered questionnaires and surveys and only 18 individual studies (6%) measured actual behaviour. Further reading revealed that among those studies the term ‘behaviour’ was a widely used concept which included not only how individuals acted in – or reacted to – certain experimental situations, but also how many computerised cognitive tasks the participants were able to solve. Actual behaviour (something other than a sitting posture and finger movements required to solve computerised tasks) was only really studied in one developmental study which reported how pre-schoolers acted in various social interactions – presumably because these participants were too young to fill out questionnaires or solve computerised tasks.
Acknowledging the limitations of the Dolinski’s study, the findings were intriguing for bringing to light what researchers have always realised – studying the often variable and subtle human behaviour is simply quite difficult. First, it is difficult to define and validate behaviour, then it is difficult to exactly replicate it and finally to record it. With recording – a necessary evil in this type of research – it becomes a challenge to overcome rigorous ethical committees rightly guarding participants’ personal information. Finally, statistically speaking, behaviour in sociological research is often binary, meaning that an experimenter may only have one chance to record a participant in a certain situation (they either acted in an expected way or not) which requires extremely large sample size. Limited sample size and the dichotomous character of the dependent variable in turn posits limits on how the data can be analysed, impeding the construction of viable models. Noting these challenges, it becomes obvious why the reliance on questionnaires and surveys is an easier avenue.
My personal interest in Dolinski’s findings focus on the challenges he lists in studying behaviour. Specifically, I ask: are these challenges really challenges or could we begin to expand the definition of ‘behaviour’? In millennial terms, should we ‘install updates’ in the way we measure behaviour? One way of doing so would be to focus on the use of eye-tracking and sensor technologies to measure the physiological changes in the human body such as skin conductance, heart rate and pupil dilation. The last decade especially has seen an increased use of these technologies in psychological and sociological research, and in recent years, they have been reaching their operational and analytical potential with more sophisticated and detailed measurements of the time-locked changes in the autonomic nervous system. Could the use of this type of ‘behaviour 2.0’ be the answer in saving the status of psychology as a study of human behaviour?
On some level, yes. Physiological reactions seem to address many of the limitations in studying human behaviour noted by Dolinski. First, because these reactions can be measured in a number of participants and/or in the same participants (to a certain extent) allows for validation and replication. Second, because the data, rather than a video or audio recording of the participant, is a series of data points, it is naturally de-identified allowing for the protection of the participant’s personal information. Next, because the generated data is not binary but continual, the analysis can be done using standard hypothesis testing. Finally, one participant’s recording can generate an extremely large number of data points making collecting data a feasible task.
I would like to note however, that relying on physiological data is not an answer to all problems and we are still in some ways away from ideal. The primary limitation still lies in the need to standardise these measurements. In order to address this, I believe that more research from different labs and with different equipment is necessary to establish valid standards, including issues concerning data filtering and how to deal with insufficient or missing data. There is also a need to correlate different measures of arousal, for instance skin conductance with pupil dilation, in order to really establish their relationship. Finally, although measurements of physiological arousal allow us to form hypotheses, it should remain a supplement rather than end-all-be-all solution. Researchers should continue to note and rely on the physical behaviour of the participant, how engaged or motivated is the participant during the testing session as well as how the participant answered question-based assessments. For instance, an individual with an autism diagnosis may show clear pupil dilation when presented with angry faces, but to get a comprehensive picture of the individual the experimenter should note his outward behaviour as well as the scores on standardized questionnaires designed to measure emotional processing (such as Toronto Alexithymia Scale 20 among others). All these concerns can and hopefully will be addressed in the next few years. Some of the current exciting research focuses on physiological reaction arousal in individuals with ESSENCE – an area that the GNC is at the forefront of. I encourage you all to stay tuned. For now, having the option to also examine ‘behaviour 2.0’ sheds some light on Dolinski’s concerns regarding the status of behaviour in psychological research.