Science Approach – Barnard et al. (2000)

8.2 Example of a Science Approach to HCI Research

As an example of a science approach to HCI research, Barnard, May, Duke and Duce (2000) argue that early HCI research was guided by a vision. The latter was of using theory from psychology and cognitive science to develop engineering tools for HCI design. Given the extensive advances in computing technology, they present the case for developing new forms of deep theory, based on generic systems of interactors. Such an overlapping, layered structure of macro- and micro-theories could then serve an explanatory function. The latter would be of human behaviour generally and human-computer interaction behaviour specifically. The resultant psychology theory-based understanding would serve as the basis for developing engineering tools to support the design of interactive behaviours for using interactive computer systems.

On what grounds might Barnard et al’s work be classified as a science approach to HCI research? First, a science approach to HCI research is a way of addressing the topic or problem of designing human-computer interactions by understanding such interactions as explanation and prediction. Barnard et al. argue that the new psychology theories, such as they propose, are candidate contributors to understanding the phenomena of humans interacting with computers. The theories can serve as the basis for developing engineering tools to support the design of such interactions. Second, a science approach to HCI research requires the performing of actions, to progress that approach to the topic or problem of human-computer interaction. Barnard et al. argue that the understanding, offered by new macro-theories in psychology, such as interacting cognitive subsystems, constitute the products of HCI research. The latter can inform the design of human-computer interactions indirectly by means of engineering tools, derived from such theories.

Third, a science approach to HCI research requires the evaluating of the success of the actions performed to progress that approach. Barnard et al’s science approach to HCI research, for example, can be evaluated in terms of their psychology theories’ success in understanding, that is, explaining and predicting, the phenomena of human-computer interactions. Their science approach can also be evaluated with respect to its support for HCI design by means of associated engineering tools. Fourth, a science approach to HCI research requires the cumulating of the successes as a way of establishing whether the topic or problem of designing human-computer interactions has been addressed or not. Barnard et al. report no assessment and so no such cumulating of successes. This is not surprising, given the early stage of the research into macro-theories, as demonstrated by their work.

Conclusion: On balance, Barnard et al’s (2000) research can be classified as a science approach to HCI research. It satisfies most of the criteria. It assumes the ability of new theories of psychology, such as they propose, to contribute albeit indirectly to the design of human-computer interactions. The contribution takes the form of the understanding of the latter and by the engineering tools, derived from that understanding. This classification suggests that Barnard et al. could decide to support any of their future HCI macro-theory research, either on the basis of the science approach presented earlier (see 8.1) or on the basis of the science framework, which follows.

[1]

***PB you can see it (Barnard et al, 2000) best as a case study in what is possible for the application of theory. I am a bit sad that you missed the methodological significance of syndetic modelling in what it provides for systematic reasoning from principles to application avoiding the influence of ad hocery…..

+++JL You are right. I did, and still do, miss the methodological significance of syndetic modelling. Part of the reason may be my failure to understand whether the syndetic model is a feature of ICS, of the expert system or of both. I would very much welcome clarification on this point. The latter would certainly be included in any second edition of the book.

8.3.3.

The Barnard et al. deep theory research (see 8.2) is re-used as an illustration of a possible science-based design research cycle. The latter might comprise specifying and implementing a system of additional interactors informed video-communications channels to an interactive teleconferencing system. The addition of the latter might be in response to user requirements for improved video-communication. Re-using the Barnard et al. research example for the science design research exemplar is intended better to support researchers attempting to apply the specific science framework.

The four third-level boxes represent the science design research cycle, starting on the left with the Specific Science Problem and terminating on the right with the Specific Science Solution, which solves the specific science problem. In terms of the deep theory research illustration, the specific science problem might be understanding the attentional integration of an additional video-communications channel for interactive teleconferencing systems. The specific science solution might be a deep, interactors-based theory, embodying the understanding of that attentional integration of an additional video-communications channel. The limitation of the specific science problem, relative to the possible and plausible specific applied problem, might be justified by the early stage of deep theory and the associated research, as exemplified by that of Barnard et al.

8.3.5 Example Application of a Science Framework to a Science Approach to HCI Research

The research of Barnard et al. (2000) can be considered to constitute a science approach to HCI (see 8.2). They note that early HCI research sought to apply psychology theory to HCI design. To accommodate the increasing range of advances in computing technology, however, Barnard et al. propose new forms of deep theory, based on generic systems of interactors. Such an overlapping, layered structure of macro- and micro-theories would then serve an explanatory function for human behaviour, including human-computer interactions. The latter would inform HCI design by means of the development of engineering design tools. What potential does the Barnard et al. science approach to HCI research offer the science framework proposed here?

First, the specific science framework is for a discipline, as an academic field of study.

Potential: Barnard et al’s science approach considers cognitive theory to belong to the scientific discipline of psychology. Their deep theory of systems interactors constitutes an instance of cognitive theory.

Second, the specific science framework is for science, as understanding.

Potential: Barnard et al’s science approach is concerned primarily with understanding as explanation. No reference is made to prediction, which is not surprising, given the early stage of their research.

[2]

***PB There are two things (1) we did look at generalisation as a means to prediction in the expert system; (2) in syndetic modelling, we also looked at the idea of testing conjectures about how an interactive system of both user and device would perform – part of the significance of syndetic modelling you did not mention and makes me a bit sad

+++ JL My reference here is to ICS, as an example of cognitive theory, supporting understanding, as explanation (of psychological phenomena within its scope). The absence of any reference to prediction applies to the latter. This would appear consistent with (1) – in which prediction appears to be a feature of the expert system. The point about syndetic modelling is addressed above under [1].

Third, the specific science framework is for the general problem as the design of human-computer interactions, based on science understanding.

Potential: Barnard et al’s science approach addresses the general problem of understanding in the form of the explanation of human-computer interactions, although not so far of their prediction or their design.

[3]

***PB See previous comment it is not just understanding it is forward looking reasoning. This is potentially much more powerful than the traditional psychology model “prediction”

+++JL This is a very interesting claim and well worth unpicking. A brief summary of how ‘forward looking reasoning’ differs from ‘prediction’ would be very welcome. It would also help to clarify the earlier points raised in [1] and [2]. Again, an obvious candidate for a second edition update.

Fourth, the specific science framework is for the particular scope of human-computer interaction design on the basis of science understanding to do something as desired.

[4]

***PB Don’t understand that last point

+++JL If I have understood your position, ICS is intended as a basis for the creation of engineering tools to support the design of interactive human-computer systems. This claim implies 2 types of scope and the relationship between them. First, the scope of ICS and second, the scope of the engineering tools. In its absence, I am offering an expression of the latter as – ‘human-computer interaction to do something as desired’. What is your expression of the scope of ICS – the psychological phenomena it is intended to explain, the engineering tools, the creation of which it is intended to support or both or indeed some other?

Potential: Barnard et al’s science approach is intended to address both the general problem of understanding and the particular scope of engineering tool derivation to support the design of human-computer interactions as desired.

Fifth, the specific science framework is for science research as the diagnosis of science design problems and the prescription of science design solutions, as they relate to performance for the acquisition and for the validation of knowledge to support practices.

Potential: Barnard et al’s science approach proposes a new type of deep theory, based on systems of interactors.

Validation is premature at this stage of the research as is prediction and its indirect application to the design of human-computer interactions by the development of engineering design tools.

Sixth, the specific science framework is for science knowledge, as theories, models, laws, data, hypotheses, analytical and empirical methods and tools and supports practices.

Potential: Barnard et al’s science approach proposes a new type of deep theory, including models, based on systems of interactors. Support for practices has yet to be addressed.

Seventh, the specific science framework is for science practices, as science-based trial and error/implement and test.

Potential: Barnard et al’s science approach is concerned primarily with explanation and the associated theory.

Given the early stage of the research, unsurprisingly no reference is made to prediction or to any details, concerning a possible contribution to engineering design tools.

Conclusion: Barnard et al’s approach to HCI as science could be further developed with respect to scientific practice of prediction and validation of their proposed theory. Also, with respect to its application to the practices of HCI design by way of engineering tools. The science framework proposed here is considered to have potential for contributing to such developments.

Conclusion

This concludes the presentation of the specific science approach and the specific science framework for HCI research and their exemplification. Also presented are the science design research exemplar and the lower-level science framework.

12.3.1 Morton, Barnard, Hammond and Long (1979) Interacting with the Computer: a Framework.

Summary

The Morton et al. framework was prompted by the introduction in the 70s of the interactive computer. It is, then, perhaps the earliest framework of its kind. The main aim of the framework is to describe the behavioural phenomena of people interacting with the newly introduced computers. Also, to apply the models and methods of cognitive psychology, including linguistics, to the better understanding of these phenomena.

On the basis of observational studies of computer users at work, Morton et al. conclude that the information technology industry’s model of human-computer interaction is essentially computer-centric. On occasion, also designer-centric. The resulting interactive systems are difficult to understand and to use. This is clearly demonstrated by a number of observational studies (Long, Hammond, Barnard, Morton, and Clark, 1980 and 1982). Morton et al. propose an alternative to the information technology industry’s model. The alternative is one inwhich ‘systems match people’ rather than ‘people match systems’. That is, the systems are user-centric. They argue that cognitive psychology research is well placed to characterise and to understand such a mismatch between people and computers. Understanding is to be achieved by the use of empirical tools. The latter include – field and laboratory studies and conceptual tools, such as models and representations. For example, block interaction, information structures and state transition models. The resultant output to designers is intended to support more user-centric design. Consistent with the paper’s title, the research required to acquire and to validate such an output, is neither included nor specified.

Key Concepts

The key concepts of the Morton et al. framework are presented in Italics and the related main General Frameworkvconcepts (see 3.4 and 10.2.2) are presented in Bold. Scope (as People, Computers, Systems, Interactions, Behavioural Phenomena and Performance) – Particular Scope.

Cognitive Science (as Psychology and Linguistics) – Discipline, and Output to Designers (as User-Centric) –

General Problem and Knowledge.

Tools (as Empirical Tools/Methods – Field Studies, Observational Studies, Laboratory Studies)- and (as

Conceptual Tools – Models as Block Interaction, Information Structures, State Transition) – Research,

Knowledge and Practices. Representations (as Domain – Computer Representation) and (as Problem – User Knowledge, Computer, Interface, General Systems, Machines, Procedures; as Natural Language – Lexical, Syntactic and Semantic; as Mismatch People/Computers) – Research and Knowledge.

Assessment

Since the Morton et al. framework comprises a science framework and the concept of its application (Output to Designers), the related General Framework concepts can also be found in the specific framework instantiations (for applied see 7.3 and for science see 8.3). In addition, the related General Framework concepts may also be referenced in the associated design research exemplars (for applied see Figure 7.1 and for science see Figure 8.1).

The General Framework does not reference specific lower-level concepts, such as Models and Representations, as proposed by Morton et al. These concepts, such as the Block Interaction, the Information Structures and the State Transition Models are all different concepts with different descriptors, compared with those of the General Framework. The superordinate concepts of Research and Knowledge (see 3.4 and 10.2.2), however, are coherent as concerns the Models and Representations.

With respect to the framework of Morton et al., the General Framework is considered to be generally complete. However, the completeness is much dependent on the coherence between the two sets of concepts. It is not considered as complete, to allow for the possibility of disagreement, as to which concepts form part of the Morton et al. framework. Also, of those concepts, which are key. Unlike the General Framework, all the concepts of Morton et al. are not explicitly identified. [2]

Dissemination

Concerning the dissemination of the Morton et al. framework, the HCI research literature suggests the disseminators to be limited to the framework’s creators, for example Barnard (1991) and Long (1987). However, not Morton, who subsequently changed research field. The dissemination content consists of the application of a few highly selective concepts and their illustration, plus the general concept of applied cognitive psychology. Note, the latter is not exclusive to the framework. The dissemination media comprise research papers. The dissemination means consist essentially of using the framework for illustrative purposes and as the basis for other frameworks. There is no evidence of any case-studies attempting to validate the original framework

12.3.6 Barnard (1991) Bridging between Basic Theories and the Artefacts of Human-Computer Interaction.

Summary

Barnard begins by reviewing the bridging required to relate cognitive theory to the behaviours observed in human-computer interactions. Such bridging, he argues, is one of the many needed, if science is to support the design of interactive systems. Barnard proposes a framework to bridge this gap, based on Long (1987 and 1989 – see also 12.3.4). He develops further the idea that tasks are analysed to produce an acquisition (he prefers discovery) representation. The latter is generalised (he prefers assimilated) by the sciences and in particular cognitive psychology. The tasks are then particularised (he prefers contextualised) to produce an application representation. The latter is synthesised for application to tasks. He argues that such a framework, comprising bridging, discovery and application representations, is potentially valuable. It facilitates specification of comparison between, and evaluation of, the many different paradigms and practices, operating in the field of HCI. Barnard then expands the discovery representation to include assumptions leading to the design of observational, experimental and scenario studies. The application representation is expanded, in turn, to include a family of cognitive task models The latter comprise goal formation and action, expressed in terms of process configuration, procedural knowledge, record contents and dynamic co-ordination and control. Finally, the cognitive science representation is expanded in terms of a model – interactive cognitive subsystems. The latter includes its principles of operation and the HCI phenomena, which it characterises. The characterisation comprises the behaviours, observed in human-computer interactions.

According to Barnard, the ultimate impact of basic theory, such as the interactive cognitive subsystems model, on design can only be indirect. That is through explicit application representations. Alternative and additional forms of such representations, he argues, need to be created, developed, and evaluated. On this view, the direct theory-based product of an applied science HCI is not an interface design. It is an application representation capable of providing principled support for reasoning about designs.

Key Concepts

The key concepts of the Barnard framework are presented in Italics and the related main General Framework concepts (see 3.4 and 10.2.2) are presented in Bold. Scope (as People, Behaviour, Computers, Systems, Tasks and Interactions) – Particular Scope. Cognitive Psychology, (Applied Science) Paradigm, Basic Theory, Comparison, Evaluation – Discipline. Principled Design Support, Design (as Direct/Indirect, Reasoning) – General Problem and Research, Representations (as Bridging – Discovery, Assimilation, Contextualisation and Synthesis). Research and Knowledge.

Reviews by Date

12th Mar 2021 by Nichel

I am happy with the book This book is written in a very formal scientific style I am happy with the book and the breadth of material it covers