The Believable Stranger or: Do We Need Culturally Portable Embodied Agents?
Vienna Workshop "Agent Culture", 24-25 August 2001
Sabine Payr
Austrian Research Institute for Artificial Intelligence
Where do problems in „cultural portability" arise in thedevelopment of embodied agents? In order to outline the issues thatshould be addressed and to separate them from the issues that do notenter into this discussion, I will try first to draw the border ofwhat I call the „cultural background". Doing this with theproblems of cross-cultural communication in mind, we can leave outany traits that can be considered universal: on the one hand, theseare properties and behaviors that all humans share as a species, andon the other hand, the undeniable (but probably over-rated) effectsof the ongoing globalization brought about by mass media, mobility,and the Internet. It is also necessary to draw the line between thecultural background and rational discourse, where, ideally at least,any communicative act can be questioned, justified and negotiated.With this separating line in mind, a different look at e.g. nonverbalcommunication becomes possible, which is then not only classified bythe „channels" it uses (kinesics, proxemics, etc..) but also bythe communicative rationality of the signs. Not surprisingly,breakdowns in cross-cultural communication are particularly severewhen it is impossible to „rationalize" them, i.e. make them thesubject of rational discourse.
We can then ask how the cultural background, now implicitlypresent in any agent architecture, can be extracted as something likea set of „cultural parameters" in agent design. I will discussOrtony’s approach to underpin his model of emotions with atheory of personality to ask where culture-specific characteristicswould come into this picture - on the level of the expression ofemotions, of response tendencies, of emotions or of personality?Agent researchers currently tend towards a „universalist" viewof personality and emotions. However, there are studies showing thatcomplex emotions like the attitude towards children and parents,differ significantly across cultures.
However, another question has to be asked first: Is it necessaryor even desirable for agents to be culturally adapted or adaptive tobe believable? Foreigners, aliens, or phantastic characters are,after all, as believable as the woman next door. Do we conceive,then, the encounter between the agent and the user as intra- orintercultural?