Differences between poker | 您所在的位置:网站首页 › poker-faced › Differences between poker |
Differences between poker-faced and affective, ‘real’ and ‘virtual’ news readers
Zs. Ruttkay $,& E.A. Konijn % H. Hondorp $
$ HMI, Dept. of Computer Science % Dept. of Communication Science, & Faculty of Information Technology
University of Twente Free University, Amsterdam Pázmány Péter Catholic University The Netherlands The Netherlands Hungary
Abstract
Much attention has been paid to the simulation and display of emotional states in applications in which an Embodied Conversational Agent (ECA) takes the role of a real human who is expected to have and show emotions. However, what to do if the ECA replaces a human where the protocol prescribes a poker-face, likein the case of TV news reading? The question is whether in a noticeably non-real scenario, the user profits from the benefits of an emotional layer, even if this contradicts to the ‘genre-practices’? To answer this question, we are conducting an experiment where real and virtual news readers are evaluated, with different amounts of emotional display both in facial expression and in voice. In this paper the goals, design, and preliminary results of an ongoing experiment is discussed] Introduction In the ECA community, much attention has been paid to the simulation and display of emotional states in applications such as emergency training [Gracth 2000] or tutoring [Kim 2004], where it is a natural and common practice to show one’s emotions. However, there are cases, such as TV news readers, when the real professional does not show emotions. What to do if one designs an application in which a virtual character replaces a real, non-emotional person? In order to answer this question, we designed a study in which differences between “real” and “virtual” news readers can be investigated in their effects of varying degrees of emotional display on liking, trustworthiness, and information transfer to the recipient or user. The experiment is still going on, so in this short paper we report about the goals and design of the experiment, together with dilemmas we came across. By the time of presentation, the first results will be available.
’Real‘ news-readers, as we know them from TV, usually do not display emotions however dramatic or joyful the events they are reporting on, although positive events of the lighter category may sometimes be accompanied with a smile. The overall impression is, however, that the real news readers (RNR) are instructed not to display emotions in any way, presumably to provide an objective and steady, person-independent reporting of events (Leathers, 1997). Traditionally, information media, and thus news items had to be factual, representing an objective reality. By contrast, emotions were reserved to fictional characters in fantasy worlds (e.g., feature films and games). However, today, hybrid program types flourish and will undoubtedly affect how we interact with virtual characters. Actually, ECAs are hybrid creatures:
usually taking the role of real people’s cast, but for the user they are clearly non-realistic, existing on the computer screen. When the news reader is a virtual character (like the famous Ananova was 1 , or the less known Reana 2
is), one may wonder whether the virtual news reader (VNR) will be regarded as belonging to a ‘reality registration genre’ (e.g., News) or to a ‘fictional fantasy genre’ (e.g., Lara Croft or conversational agent Sybille) (see Konijn & Hoorn, 2004). ‘The poker face’ may be seen as a critical feature of the genre ‘news-report’ as contrasted with emotional displays of actors in fictional or fantasy genres. Therefore, the question is whether a VNR should adopt the poker-faced protocol of RNRs or a ‘fictional news reader’ protocol with emotional reactions? On the one hand, viewers are used to the poker-faced news readers. On the other hand, such news readers look unnatural from a certain perspective: announcing dramatic events in a disciplined, neutral way is not what human beings usually do and will not help to establish an affective bond. Arguments for a positive effect of emotions are given by the CASA paradigm of Nass as well as from several ECA evaluation studies (e.g., Dehn & Van Mulken, 2000; McBreen, Anderson & Jack, 2001;
Prendinger & Ishizuka, 2003;
Reeves & Nass, 1996; Ruttkay & Pelachaud, 2004, Brave, Nass & Hutchinson 2005): People generally expect and award a similar protocol for virtual and real life characters. However, other studies report that virtual characters deviating from the real-life protocol were liked by users (e.g., dog embodiment in Koda & Maes, 1996; split cast of role in André & Rist, 2000). In persuasion studies, emotions are generally seen as effective means to influence attitudes |
CopyRight 2018-2019 实验室设备网 版权所有 |