Science or No Science

One of the most frustrating things about the HCI community is being under-recognised as a science by the other disciplines. The main argument consists of two parts; the first is the lack of motor themes (central themes) in mainstream research, and the second is the absence of frameworks or models. Many community members have urged to end the “unruly fashion” studies and suggested establishing order by expanding interdisciplinary research.

Incoherence and inadequacy

In Reeve’s opinion, two factors have caused anxiety to HCI research members; incoherence and inadequacy[4]. He argues that the incoherence stems from the nature of HCI research, which is interdisciplinary. It has accumulated great diversity over time with an absence of uniformity with regard to theories, methods, and instruments as compared to other well-established disciplines. Based on my understanding, HCI is quite different from other traditional disciplines and only emerged since the innovation of the computer. HCI research is still exploring and extending its boundaries. Especially when incorporated with other well-established disciplines (e.g. psychology), even using their frameworks and models as our research tools or methods, similar to how computer science borrows mathematical tools to solve their problems. In the HCI research community, different research branches focus distinctly on multimodality rather than unimodality. These factors have resulted in incoherence in HCI research.

 

Reeve’s argues that the inadequacy of HCI is due to the absence of rigorous quantitative laws and formulae when compared to mature “hard science” (e.g. physics, chemistry, or biology). In comparison to “soft science“ (e.g. sociology, psychology), HCI lacks the qualitative frameworks or models [2], [4]. Based on the case study of the process of computer mouse design, the author argued the HCI is a scientific design process. Applying cognitive scientific concepts for designing a mouse is an example of conceptualising design space as applied science (using cognitive science to achieve the best HCI Performance) [1], [4]. The scientific design space should contain more dimensions, such as economic, organisational, and social system rather than just artefacts [4].

Reeves also describes a layered model that was initially presented by Newell and Card in 1985, where “temporal bands map to different action units, associated memory and cognitive capacities” [1]. These rationality bands are similar to cognitive science with regard to time and ordering set of psychology and bounded rationality, as shown in Figure 1.

 

Figure1 HCI in scientific order discipline (original from [4])

The argument implicitly presented evidence that HCI can satisfy both the reductionism theory and positivistic philosophy of science. When judged against the traditional definition of science, HCI does not have appropriate rigour. However, HCI research is undoubtedly built atop other fundamental science disciplines, such as psychology, anthropology, computing science, etc.

Existing research practices were used to adopt cognitive science (methodology) to evaluate the task-oriented interface through a hypothesis-driven experimental format. Although the downside is that the research practice has been treated as an instance of applied cognitive science research, it satisfied cumulative, replicable, and generalised findings that are usually considered the features of scientific research. However, those one-off experiments/designs were the result of “implications for design”, which demotivated incremental and repeating studies [3]. To overcome the adverse effects, we should think from the perspective of rationalised scientific design space and systematically take each separate research practice as a block to build our research frameworks with [7].

Interdisciplinary

From discussion above, we can see the importance of HCI being interdisciplinary research, as Roger gave it the characterisation as an “eclectic interdiscipline”. We could start constructing a rationalised disciplinary order that can eliminate the unruly fashion. Furthermore, if HCI needs to become a rigorous interdisciplinary branch, then we need to identify the boundaries and clarify what kind of contributions should be classified under HCI. Otherwise, HCI may not get due “credit” for research [3].

The HCI community should not be overly worried about the problem of not being recognised as a scientific discipline. Interdisciplinary research sometimes finds it difficult to draw a clear line of splitting from other research. If we need to draw a sharp boundary, then the kinds of research that cannot take place without HCI’s contribution can be analysed. The framework can build on top of the other disciplines, such as how data science is built on top of statistics, computer science, social science, etc.

 

[1]         CARD, S.K., MACKINLAY, J.D., and ROBERTSON, G.G., 1990. The design space of input devices. In Proceedings of the SIGCHI conference on Human factors in computing systems ACM, 117-124.

[2]         COLE, S., 1983. The hierarchy of the sciences? American Journal of Sociology 89, 1, 111-139.

[3]         KOSTAKOS, V., 2015. The big hole in HCI research. interactions 22, 2, 48-51.

[4]         REEVES, S., 2015. Human-computer interaction as science. In Proceedings of The Fifth Decennial Aarhus Conference on Critical Alternatives Aarhus University Press, 73-84.


Author: Bing Zhai

One response to “Science or No Science”

  1. Matt Wood says:

    Interesting stuff Bing, you’ve clearly got a good understanding of the issues in the paper, and you raise some nice points. You outline the ‘problem’ nicely, along with the difficult (tiresome!?) argument of HCI as ‘science’. You make a convincing conclusion around HCI shouldn’t be too concerned with being a science or not, whilst highlighting we perhaps need clarity in what constitutes a ‘contribution’ in HCI. On this point, it is sometimes amusing to me that papers in HCI are required to list their ‘contributions’ to the field like a shopping list! Also nice the point around HCI building ontop of other disciplines

    There were a few points I felt needed a little more clarity – I didn’t quite catch your argument about thinking from a rationalised design space to take on different device designs, and note traditional psychology and sociology don’t ‘normally’ have qualitative frameworks or models – mainstream psychology is still alarmingly quantitative!

    Overall though, a good understanding of the issues, and am interested to see where your line of argument continues.

Leave a Reply

Your email address will not be published. Required fields are marked *