Facilitating Perspective Taking in Groups: An Analysis Using Multiple Conversational Agents

Facilitating Perspective Taking in Groups: An Analysis Using Multiple Conversational Agents

Yugo Hayashi (College of Information Science and Engineering, Ritsumeikan University, Kusatsu, Japan)
DOI: 10.4018/ijssci.2013010101
OnDemand PDF Download:
List Price: $37.50


The present study investigates the nature of egocentric biases in a situation where a speaker is surrounded by social actors with different perspectives. In this context, the author investigated how communication channels function to ease egocentric bias during collaborative activities. To investigate this point, the author used conversational agents as social actors. The present study therefore created a virtual situation where a speaker was surrounded by several speakers. The author hypothesized that the diversity of communication channels available to the audience would increase the awareness of others and facilitate the adoption of an exocentric perspective. The results of the analysis show that participants who engaged in the collaboration task with various communication channels used fewer egocentric perspectives. Studies in egocentrism and communication have not yet investigated the conversational dynamics of multiple speakers. This study therefore provides a new perspective about the kinds of factors that may ease such biases.
Article Preview


Studies in cognitive informatics and cognitive computing have provided many implications on how to develop robots and conversational agents based on human cognition (Wang, 2010). In those studies they guide us how to develop cognitive supporting systems such as problem solving and decision-makings. But there are a few studies that focus on the mechanisms of cognition during ‘collaborative problem solving’. On investigating it there is a need to consider the internal factor such as the cognitive bias that occurs during social interaction.

People use information available to the self when constructing an initial interpretation. This egocentric behavior is characterized as a self-centered behavior or a tendency that all humans naturally have (Keysar, Barr, Balin, & Brauner, 2000). This kind of behavior was previously thought to arise only in communication behaviors with small children (Keysar, Barr, Balin, & Paek, 1998). However, recent studies show that adults also engage in these egocentric behaviors while communicating (Wu & Keysar, 2007). Consider, for example, A and B who are having a conversation on topic X where A is the only person familiar with the topic. Even if A knows that B does not share common knowledge about X, he/she unintentionally converses about the topic based on A’s view point. It is known that egocentric bias gives rise to miscommunication and causes serious errors during interpersonal communication.

Understanding the functions of egocentric bias is not only important for understanding human cognition, but also for designing new computer-supported systems based on human cognitive characteristics. The motivation of the present study relies on the following: (1) no studies investigate the nature of egocentric bias under multiple conversation conditions such as polylogues; and (2) most studies have largely focused on the presence of egocentric bias and few studies have focused on the factors that reduce egocentric biases.

This study tackles the issue of understanding the nature of egocentric bias during collaborative activities in groups. In particular, we focus on the nature of egocentric bias during social interaction with multiple speakers. The present study investigates this throughout using conversational agents as experimental tools. In the next section, we review the literature on egocentric biases.

Egocentric Bias in Collaborative Problem Solving Tasks

In cognitive science, egocentric bias is known as the tendency to consider as potential referents only those objects that serve as potential referents from one’s own perspective, not objects that are in the common ground. Keysar, et al (2000) conducted an experiment in which participants’ eye movements were analyzed and demonstrated that people occasionally use an egocentric heuristic when they communicate. They argued that this egocentric heuristic may sometimes be successful in reducing ambiguities, although it might also cause systematic errors.

Such biases can become a crucial problem in collaborative tasks. Some studies show that difficulties in perspective taking occur during collaborative problem solving in pairs (Hayashi, Miwa, & Morita, 2006; Hayashi, Miwa, 2009; Hayashi, 2012). Hayashi, Miwa, and Morita, (2006), for example, analyzed the characteristics of paired problem solvers with different perspectives and compared them with those having the same perspectives. The protocol analysis showed that pairs holding different perspectives constructed an interactive pattern such as role sharing based on his/her own perspective. An interesting point here is that although participants noticed that it is possible to see the task from a different perspective, they did not try to change their perspective. These results show that an egocentric behavior may occur during collaborative activities as a result of the difficulty in perspective taking. In this study, we regard “egocentric bias” as an attitude of a problem solver in a group who is fixated on his/her own perspective. In contrast, we refer to an “exocentric perspective” as a behavior in which one suppresses his/her perspective and makes an effort to take another’s perspective. However, the kinds of factors that are effective for reducing egocentric biases and fostering exocentric perspectives during social interaction are not yet known. Based on these definitions, we use the terms egocentric and exocentric perspectives to explain our perspective taking behaviors during collaborative problem solving. Therefore this term is different from those terms such as ‘self-assertion’ used in psychology.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 9: 4 Issues (2017): 3 Released, 1 Forthcoming
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing