AI & SOCIETY

, Volume 23, Issue 2, pp 139–145

Social intelligence design: a junction between engineering and social sciences

Authors

    • Department of PsychologyKobe Gakuin University
  • Naohiro Matsumura
    • Graduate School of EconomicsOsaka University
Editorial

DOI: 10.1007/s00146-007-0139-9

Cite this article as:
Miura, A. & Matsumura, N. AI & Soc (2009) 23: 139. doi:10.1007/s00146-007-0139-9
  • 130 Views

1 Overview: the meaning and achievements of SID2006

This special issue of AI and Society contains a selection of papers presented at the 5th International Workshop of Social Intelligence Design held at Osaka University Nakanoshima Center, Osaka, Japan, in March 2006.

Nishida (2001) defined social intelligence design (SID) as a research field attempts to integrate understanding and designing social intelligence. On the one hand, it involves engineering approaches concerning design and implementation of systems and environments, ranging from group/team oriented collaboration support systems that facilitate common ground building, goal-oriented interaction among participants, to community support systems that support large-scale online-discussion. On the other hand, it involves scientific approaches addressing cognitive and social psychological understanding of social intelligence, and provides a means for predicting and evaluating the effect of a given communication medium on the nature of discussions, interaction dynamics, and conclusions. In addition, it encompasses pragmatic considerations from economy, sociology, ethics and many other disciplines, for social intelligence design has a direct relation with the society. The engineering and analytical approaches are complementary to each other and should be integrated intimately, for good systems cannot be built without good understanding and vice versa.1

The 5th workshop focused on the impact and significance of information technology in our lives, work, home, and on the move. In this workshop we consider Social Intelligence as the ability of people to relate to, understand, and effectively interact with others. The central question was how social intelligence was mediated through the use of emerging technologies. We distinguished the following three themes:
  • Development, operation, and evaluation of support systems or tools for SID

  • Observation and modeling of psychological and behavioral processes of e-community

  • Social intelligence design by pilot program and computer-aided simulation.

Within a social scientific context, especially psychology, research on “intelligence” has a long history, over one hundred years. In fact, the foundational work of A. Binet, L’ Étude Experiméntale de I’Intelligence, was published in 1903. However, the concept of “social” intelligence is a comparatively recent development, and even the meaning of the term “intelligence” itself has changed over the years. In a classical definition, “intelligence” did not have a “social” meaning and was defined as a personal and often innate property of the mind that encompassed many related mental abilities, such as the capacities to reason, plan, solve problems, think abstractly, comprehend ideas and language, and learn. On the other hand, “intelligence” in SID bases the definition on social factors. In other words, it is grounded in acquisition of abilities through interaction with the environment and the development/evolution within this process. These abilities are not only learned from public teachings, but also from the everyday environment. Furthermore, the word “social” tends to be used in equivocal definitions by the context. SID uses the word “social” as personal interaction, that is, not at the individual level, but at the group or community level. It is similar to the social psychological idea of “interaction” and “relationships” as being social, and is based on group and community exchanges.

In these times of societal decline and the impoverishment of regional security, the bulk of real life human interaction or relationships is attenuating and people are getting fewer opportunities for social interaction. It may be further necessary to use technical support to acquire of various social skills. For such a purpose, in addition to rational techniques or systems that can provide effective support for human behavior, we will have to develop “emotional” ones that enable us to control human mentality. That is why people who use these techniques are closely related to each other and form a society. In other words, such a society is fundamentally based on their emotional connections and their control of them. Collaboration between engineering researchers and social scientists is needed to make an active contribution towards the development of social intelligence design.

With this view in mind, we emphasized the role of SID2006 as the junction between engineering researchers and social scientists and for both productive research activities and significant social contributions under a common goal for exploring SID in the real world. If engineering researchers and social scientists know the needs and seeds of each other’s research, a new idea might emerge, which leads to novel “social intelligence” as unique research findings. So far, engineering researchers have at least tended to look at mainly systems or product designs, whereas social scientists have tended to mainly look at mental or behavioral modeling for individuals or interpersonal behaviors mostly based on multiple laboratory experiments or one-shot questionnaire surveys. Based on the robust achievements of our previous workshops, the 5th workshop was the opportunity to apply them to the real world and by measuring and analyzing actual social phenomenon, explore a new research development.

2 Brief introduction of the papers in the special issue

For this purpose, three top researchers from different fields were invited to make keynote speeches at the workshop; Dr. Thomas Erickson from the field of informational science, Dr. Hisao Nojima from the field of cognitive science, and Dr. Tomio Kinoshita from the field of social psychology. All three speeches clearly demonstrated the importance of SID research in each field and effectively stirred up the interests of the participants. Twenty-four research papers were subsequently presented and they generated a lively discussion from various standpoints. The SID2006 workshop, advocating the role of a meeting point for both engineering researchers and social scientists, produced sufficient amounts of a variety of papers from both fields and generated an active academic interaction between them. At the present, it can be said that the workshop carried out the mission successfully. This special issue includes the paper that was the basis for Dr. Erickson’s keynote speech and ten presented papers selected by the editors.

Dr. Erickson focuses on large groups of people that exhibit social intelligence from the standpoint of a designer. He examines the ways in which coherent behavior directed towards individuals or collective goals is produced in face to face situations and discusses how it can be supported in online systems used by geographically distributed groups. The various design examples for the digital interactions presented in his speech (also in his paper in this issue) represent the optimal path in which we should follow.

The selected papers here were newly peer-reviewed after the workshop and categorized into either engineering or social scientific. We will briefly introduce the selected papers in each category in order to explain their contribution to the workshop.

3 Engineering perspective

Six papers are included in this category, which are briefly introduced below. The “engineering” perspective in this issue represents the direction of these papers. Their final goal is to develop some support systems or tools suitable for SID. Four of them are regarding the functions of humanoid social robots or human-like agents. To make robots or agents more adaptable to our daily life, it is important for them to interact within the social rules attached to their role. It is essential for these robots or agents to not only have a centralized and sophisticated artificial brain, but they must also have “social” intelligence, because their roles and the rules that guide them will be defined by society.

Since humans tend to understand other humans’ actions in a goal-directed manner and they will expect the social robot to behave similarly, one of the most important skills that a social robot needs to have in order to be accepted is a detection function for human’s intended behaviors. Mohammad and Nishida focused on detecting intended behavior signals (IBSs) in a perception module. IBSs are defined as the component(s) of sensor signals that originate from actual human body movements as intended by a human. They proposed an interactive adaptive perception scheme that uses four important behavioral features of humans to amplify the signals originating from an intended behavior with respect to the signals originating from an unintended behavior or other noise sources, such as instrumental noise. A simple experiment was conducted using a simulated robot in a real environment and the results showed that the proposed scheme could effectively attenuate the unintended signal components. Their system could be useful in many application areas in a social environment, for example, intelligent robotic guides in museums that can detect interested visitors and intelligent carts in big supermarkets that can detect users interested in help.

In natural human–human communication, we always use both verbal and nonverbal information to speculate about our speaking partners’ intentions. Although it is true that nonverbal information (i.e. facial expressions, gestures, and gazes) is important for smooth communication, we often unconsciously express such information that robots or agents would have a difficulty recognizing and reacting to it. Ohmoto et al. focused on the unconscious expressions used when people are lying. Telling a lie is a typical behavior in which we may unconsciously express our intentions. They made a real-time system that simultaneously measures the gaze direction and facial features without making a model of a face and then conducted two laboratory experiments to investigate the usefulness of their method for discriminating between lies and truths in situations similar to actual communication. The results showed the robustness of their method for discriminating between lies and truths and suggest that it is necessary for to make these discriminations to pay attention to multimodal nonverbal information. Based on their results, it is expected that in the near future an advanced system will be developed that can enable us to carry out the full automatic detection of lies.

Xu et al.’s final goal is to develop an interactive robot that can gradually adapt to human instructions using alignment-based learning through a nonverbal communication channel, and that is capable of mutual adaptation. As the first step in this direction, they designed an experimental environment using a Wizard of Oz (WOZ) method to observe how and whether or not mutual adaptation occurs in human–human communication. They obtained three findings that were observed frequently: alignment-based action, symbol-emergent learning, and environmental learning. Although not all of the experimental findings in the human-human communication tests apply to other (human–robot and human–learning robot) experiments as they are, it will undoubtedly be helpful in designing human–robot or human–learning robot interfaces.

Rienks et al.’s interest is in the development of technology for supporting meetings. Technology in the field of meeting support ranges from completely passive objects like microphones, to pro-active autonomous actors, such as virtual meeting participants. They focus on the latter, which are able to act autonomously. They elaborate on the concept of pro-active meeting assistants, in particular software agents that aim at assisting the meeting process, thereby facilitating more effective and efficient meetings. Furthermore, they conducted an experiment using a WOZ method, where they simulated several forms of pro-active meeting assistants designed to streamline the meeting process. The results showed that meeting efficiency can be improved by using pro-active assistants with respect to situations in which no meeting assistants are employed.

The other two papers classified in this category have characteristics different from the above-mentioned four papers. One takes a more practical approach and the other a more fundamental and theory-based approach. Shoji et al. designed and developed a blog-based system called “PLASIU”, which supported a creative decision-making process, and applied it in a real-world job-hunting process for university students. Their results confirm the importance of “conception” (they defined this word as awakening of a new viewpoint) through interaction in decision making and suggested that their PLASIU system could efficiently support such interactive processes. Conception induced through the interaction with others (professional coaches, in this case), which is based on an activity log stored in their system, can be regarded as a kind of social intelligence.

Notsu et al. have an engineering view that proposes a novel medium for interactions and also places importance on a psychological perspective when they examine their basic framework, the visual assessment of clustering tendency (VAT) algorithm. Their framework depends on the theory of “Naïve psychology” which is one of the most classical and influential theories in modern psychology. They adapt the VAT algorithm to the idea of cognitive balance for improving the mutual understanding between people. In addition, they derive two important notions of “imbalance” that characterize individual personality: crisp and fuzzy. They apply this idea to actual data including complicated relationships. Various fundamental psychological theories, such as naïve psychology, would be helpful when modeling human interaction processes related to social intelligence and Notsu et al. have one such example.

4 Social scientific perspective

Four papers on SID from a social scientific perspective are also included in this special issue. Social scientific studies, which do not necessarily apply directly to the development of support systems or tools suitable for SID, can offer significant insight to engineering researchers through the careful observation and exact modeling of the social, psychological, and behavioral processes of an e-community.

Although, it is empirically true that interpersonal communication sometimes has a creative power for developing social intelligence, the developing process has not yet been fully investigated. Suzuki et al. focused their attention on group discussion processes to establish a measurement for evaluating the conversational impressions from group discussions, and to make an exploratory investigation on their interactional processes that may affect the formation of those impressions. Their research can be characterized by using both quantitative and qualitative methods, which are indispensable for understanding the complex nature of interpersonal communication. Their analysis revealed that each communication scene had a unique pattern in the configuration of the questions presented by the moderator and the interviewees’ responses. Based on the results, they discussed how the distributional patterns of the discussions might have contributed to their positive and negative values in terms of two evaluation factors, conversational activeness and conversational sequencing.

Morio and Buchholz focused on “anonymity” in computer-mediated communication (CMC), an old and new problem that has been discussed for a long time. Anonymity is one of the most salient characteristics of CMC, and many previous studies have already dealt with it and evaluated its impact on interpersonal communication. The novel point of their work is that they focused on the cultural differences in regards to anonymity being used as a motivational factor for interpersonal communication. They empirically showed that autonomy was often valued in Western societies, whereas Eastern societies tended to emphasize affiliation, They also suggested that individuals in Western societies would gravitate toward online communities that allowed for lower levels of anonymity, while individuals in Eastern societies would be more likely to seek out online communities that promoted higher levels of anonymity. We need to consider such cultural differences when designing online communication systems to maximize social intelligence.

Furutani et al. and Moriyama et al. both focused on “self-efficacy” in participating in Internet communities. Self-efficacy can be defined as the cognition of being able to do certain things. Social intelligence would be developed under a communication environment in which Internet users can actively join and contribute to them. The users would gain higher self-efficacy through active participation and contribution. Both studies intended to investigate what factor of the Internet communication experience would have a significant effect on the self-efficacy of users. They took different approaches and focused on different factors. Furutani et al. conducted a questionnaire survey on randomly sampled Japanese Internet users and suggested that the use of the Internet did not affect self-efficacy, but cognition of the network-changing possibility did affect it. In particular, the cognition that users could connect with people or groups with heterogeneous social backgrounds by using the Internet positively affected their self-efficacy. On the other hand, Moriyama et al. conducted a practical field survey for junior high school students and suggested that Internet use could promote a feeling of effectiveness in a student’s daily life and their self-efficacy was affected by both the processing and creation abilities of information use. Based on these suggestions, they proposed some suggestions for improving classroom activities in information education.

Footnotes
1

See http://www.ii.ist.i.kyoto-u.ac.jp/sid/ for the series of the Social Intelligence Design workshops.

 

Copyright information

© Springer-Verlag London Limited 2007