Brief Presentation and Key Project Results

Abstract

poliTE—Social Appropriateness for Artificial Assistants examined the phenomenon of social appropriateness through the lens of ‘cultural techniques’ to understand socially shared contexts of action and their potential transferability to human–machine interactions.

1.1 poliTE—Social Appropriateness for Artificial Assistants

poliTE—Social Appropriateness for Artificial Assistants examined the phenomenon of social appropriateness through the lens of ‘cultural techniques’ to understand socially shared contexts of action and their potential transferability to human–machine interactions. poliTE was a joint project by CITEC (Center for Cognitive Interaction Technology) at Bielefeld University and FoKoS (Institute for Advanced Studies “Shaping a humane future”) at the University of Siegen, funded by the Federal Ministry of Education and Research, Department 616 “Interactive Technologies for Health and Quality of Life” under the motto “Bringing technology to the people” (grant number 16SV7880K).

The project’s starting point was the prediction that interactive systems capable of learning will continue to gain importance in more and more areas of life. For example, the professional and private everyday lives of many people will be affected by the use of digital voice assistants, social robots, and virtual agents. For interpersonal interactions, every culture has developed a set of more or less complex rules and techniques governing behaviour, interactions, and (social) appropriateness that for example determine when apologies, greetings, congratulations, reprimands, or other social practices and rituals are appropriate, as well as how these interactions should unfold. Such cultural techniques of behaviour, and the customs, behavioural norms, and conventions that regulate them are passed on—often in unwritten form—and acquired by individuals over the course of their socialisation. Thus, modes of behaviour are judged as more or less socially appropriate depending on the person, role, situation, context, culture, time, etc. Especially when our everyday lives are increasingly permeated by assistance technology, we face questions about the relationship between cultural techniques of behaviour and ‘intelligent’ technological systems: can and should technical systems be equipped with the ability to interact in a socially appropriate manner?—this was the key focus of the poliTE research project.

As part of poliTE, an interdisciplinary literature search in natural sciences and humanities was performed to investigate philosophical, empirical, and technological aspects of social appropriateness. The results were then systematized. To gain an overview of the interdisciplinary research and theory landscape, a comprehensive topic and actor analysis (TAA) was performed on the genesis, change, and the conditions of possibility and perception of different forms of socially appropriate behaviour and appropriateness judgements associated with them, allowing a set of recurring topics, research interests, and people to be identified. The topic and actor analysis included extensive term searches in literature databases,¹ journal analyses,² and conference analyses,³ as well as a survey (mixed sample of specialized and non-specialized audience) and in-depth free research on selected topics. In this process we accumulated a total of around 5,000 scientific publications dealing with aspects of social appropriateness or judgement thereof, which are publicly available and browsable in the form of a database at http://www.polite-data.netzweber.de. In the second project phase, these results were sorted and systematized, and theories from scientific branches such as psychology, philosophy, sociology, linguistics, anthropology, and theories of culture and media were examined and adapted accordingly to identify any potential of epistemologically fruitful transferability to human–machine interactions in general and to technical systems design in particular.

1.2 Summary of research results

A central result of the analysis of these interdisciplinary approaches to the phenomenon of social appropriateness is the FASA model, a model of the factors (FA) of social (S) appropriateness (A), cf. Chap. 4. The FASA model was established with a special focus on potential transferability to human–machine interactions and to technical systems design. It consists of the following five factors:

• a) «Type of Action, Conduct, Behaviour, or Task» (Sect. 4.1): Actions or behaviour (or sequences thereof) are always viewed as socially appropriate as part of (as a concrete realization of) a certain type of action, conduct, behaviour, or task. The terms ‘behaviour’ and ‘action’ are typically used as synonyms in this book. A more precise distinction can be made by stipulating that behaviour is not deliberate but represents an organic or motor response to stimuli, while action is more cognitively targeted.

• b) «Situational Context» (Sect. 4.2): Actions and behaviours are embedded within situations—whether an action/behaviour is socially appropriate depends on the situation in which it unfolds, among other things. «Situational Context» refers in particular to the specific time and place of a concrete interaction, which has a specific spatial structure, and can for example be shaped by the degree of formality (intimate, familial, private, semi-private, public, etc.) of the interaction, the typical roles, behavioural/action and status requirements placed on the interacting parties, their typical hierarchies of preference, the participants’ definition of the situation, and so on.

• c) «Individual Specifics»⁴ (Sect. 4.3): «Individual Specifics» addresses individual influences on the construction and/or perception of social appropriateness. Among other things, what is considered appropriate depends strongly on the individual characteristics of the interacting parties, such as their physical, psychological, and cognitive constitution and situational state, or their age, gender, etc.

• d) «Relations between Interacting Agents»⁵ (Sect. 4.4): In an interaction, the interacting agents do not meet as solitary agents but as actors who are engaged in relating to one another. Social relations between the participants of an interaction can, for example, take the form of cultural or societal relations such as social closeness/distance, status, respect, degree of familiarity, power constellations, etc.

• e) «Standards of Customary Practice» (Sect. 4.5): In social practice, there are implicit norms of action and behaviour. In many cases, there are standards of customary practice about what is considered socially appropriate. These can be understood as a set of customs on a spectrum ranging from ‘specific’ (possibly divergent) group customs (e.g., customs within different families, sports teams, or companies) on the one hand, which in extreme cases may be negotiated once and for all or alternatively may be constantly situationally renegotiated, to ethically justifiable, and hence ‘general’, regulative norms on the other hand.

Within these factors, there may be sub-factors, which are described as factor criteria in our terminology. As discussed above, a situation is for example constituted by the time of the interaction, its environment, the participants’ roles, etc.—in this sense, the factor criteria are what ‘make’ something a situation. The factors and factor criteria themselves are abstract—factors and factor criteria may not be sensually perceptible. By contrast, so-called ‘observables’ can be perceived and observed with senses or sensors: a person’s concrete clothing, smell, hairstyle, etc. are sensually perceptible characteristics from which people and potentially also technical systems can infer parameters of factor criteria and factors.

Observables require interpretation. For instance, what does a display reading ‘19 °C’ actually mean in a specific situation? On a thermostat display, most people would consider this a comfortable room temperature, but on a kettle, it would indicate that the water is still far from boiling (and you need to wait a little longer for your tea, for example). The same applies to observables that function as symbols of meaning in the context of social appropriateness. When observables are interpreted by an observer and provide meaningful clues about particular facts or contexts to this observer, we call them ‘indicators’.

The five factors of social appropriateness as presented above, together with the many factor criteria and observables/indicators associated with them, determine what is considered socially appropriate in the presence of complex conditions and interdependencies. They also influence the construction and performance of socially appropriate behaviour, as well as judgements of appropriateness in interpersonal interactions. Regarding human–machine interactions and the design of technical systems, some of the conditions that apply to the construction and evaluation of socially appropriate types of action, conduct, behaviour, or task are the same, while some others are very different. The FASA model allows the aspects of a human–machine interaction to be inspected to determine which factors are being considered with what realisations, which factors could or should be considered, and which factors cannot or should not be considered.

These interdependencies between the factors of social appropriateness, their criteria, and any observables and indicators can also be structured into other heuristic approaches to the phenomenon of social appropriateness: for instance, social appropriateness tree structures are presented in Chap. 5. These tree structures classify the factors, their criteria, and some associated observables/indicators, offering another way to approach social appropriateness besides the FASA model and the explanations of individual factors of the model in Chap. 4. Finally, the complex interdependencies mean that the explanations about theoretical approaches to specific factors in this book are closely linked. We refer to other potentially applicable factors and factor criteria when describing the respective theories.