Abstract

An active research goal for human-computer interaction is to allow humans to communicate with computers in an intuitive and natural fashion, especially in real-life interaction scenarios. One approach that has been advocated to achieve this has been to build computer systems with human-like qualities and capabilities. In this paper, we present insight on how human-computer interaction can be enriched by employing the computers with behavioral patterns that naturally appear in human-human negotiation scenarios. For this purpose, we introduce a two-party negotiation game specifically built for studying the effectiveness of haptic and audio-visual cues in conveying negotiation related behaviors. The game is centered around a real-time continuous two-party negotiation scenario based on the existing game-theory and negotiation literature. During the game, humans are confronted with a computer opponent, which can display different behaviors, such as concession, competition, and negotiation. Through a user study, we show that the behaviors that are associated with human negotiation can be incorporated into human-computer interaction, and the addition of haptic cues provides a statistically significant increase in the human-recognition accuracy of machine-displayed behaviors. In addition to aspects of conveying these negotiation-related behaviors, we also focus on and report game-theoretical aspects of the overall interaction experience. In particular, we show that, as reported in the game-theory literature, certain negotiation strategies such as tit-for-tat may generate maximum combined utility for the negotiating parties, providing an excellent balance between the energy spent by the user and the combined utility of the negotiating parties.