Study by the Paderborn University and Bielefeld examines interaction in explanatory situations
When people talk to each other, the conversation partners adapt to one another in various ways. In the context of an explanation, for example, an exchange takes place between what the explainer considers important and what their counterpart wishes to understand. When it comes to technical artefacts – that is, man-made devices, machines or tools – explainers do take their listeners’ interests into account, but nevertheless tend to focus primarily on how the objects work. This is demonstrated by a recent study from the Transregional Collaborative Research Centre(TRR) 318 at Paderborn University and Bielefeld. The findings were published in the “International Journal of Technology and Design Education” and provide important insights for education and explainable artificial intelligence (AI).
Dual perspective – dual nature
The starting point for the study was a theory originating in the philosophy of technology, according to which technical artefacts can be viewed from two perspectives: in terms of their architecture and in terms of their relevance. In other words, on the one hand, the focus is on how they are constructed and how they function; on the other hand, on the purpose they serve and why they are important. This dual perspective is referred to as ‘dual nature’. The researchers, who come from the fields of Computer Science education, educational psychology, and Linguistics, investigated how these perspectives are utilised in explanations. The focus was on how flexibly explainers respond to the interests of their conversation partners.
Explainers adapt their explanations – but not to the full extent
For the study, the researchers conducted a controlled experiment with 72 participants. They used the board game ‘Quarto!’ as an example, which served to highlight typical mechanisms of explanation. The study participants were asked to explain the game in such a way that their conversation partner could subsequently win. Conversation partners who were in on the experiment feigned unfamiliarity with the game whilst expressing an interest in either its structure or its relevance.
The results show that people do indeed respond to the interests of their conversation partner. If, for example, particular interest was shown in the meaning or usefulness of the game, the explainers were more likely to address these relevance-related aspects. In these cases, the proportion of statements relating to relevance rose from 28 to 40 per cent. Nevertheless, explanations of how the game works or is structured – that is, its architecture – dominated overall. Depending on the experimental conditions, architecture-related statements accounted for around 60 to 72 per cent of the explanations. Explainers therefore regard knowledge of the architecture as the foundation for understanding the board game. Statements relating to relevance can further promote understanding.
Implications for education and explainable AI
The study makes it clear that explaining is not a one-sided process. Instead, understanding arises through a shared interaction between the explainer and their counterpart. “We observed that explainers paid close attention to their conversation partner's reactions and adapted their explanations accordingly, provided this was compatible with their own explanation plan,” says lead author Lutz Terfloth from Paderborn University. This so-called ‘monitoring’ enabled co-constructive communication, meaning that both conversation partners actively shaped the explanation process.
The findings provide important insights into how people adapt explanations to their conversation partner and where the limits of this adaptation lie. They highlight that comprehensible explanations should contain information not only about how technical systems work, but also about their meaning and application. This understanding of human explanation processes is relevant to educational research and the development of explainable AI systems. It can help to develop AI systems that are tailored to users’ interests and needs, without omitting information that is important for understanding. “Our aim is to empower users to interact more effectively with technology,” says Terfloth.
The study can be found here: https://link.springer.com/article/10.1007/s10798-026-10084-9
This text was translated automatically.
