Ｅｆｆｅｃｔｓ　ｏｆ　Ｍｅｄｉａ　ａｎｄ　Ｄｉｓｔｒｉｂｕｔｅｄ　Ｉｎｆｏｒｍａｔｉｏｎ　ｏｎ　Ｃｏｌｌａｂｏｒａｔｉｖｅ　Ｃｏｎｃｅｐｔ－Ｌｅａｒｎｉｎｇ:of and on

　　AbstractThe present study explores the effects of media and distributed information on the performance of remotely located pairs of people′s completing a concept-learning task. Sixty pairs performed a concept-learning task using either audio-only or audio-plus-video for communication. The distribution of information includes three levels: with totally same information, with partly same information, and with totally different information. The subjects′ primary psychological functions were also considered in this study. The results showed a significant main effect of the amount of information shared by the subjects on the number of the negative instances selected by the subjects, and a significant main effect of media on the time taken by the subjects to complete the task.
　　Key wordsvideo, audio, distribution of information, collaborative concept-learning.
　　
　　1Introduction
　　
　　Suchman, an anthropologist, argued that actions were always situated in particular social and physical circumstances[1]. In this view actions emerge from moment-by-moment interactions between actors, and between actors and their environments. The social and environmental aspects of cognition have been stressed in Hutchins′ work[2]. He had studied relatively structured decision environments, for example ship navigation and aeroplane piloting. His conclusions are that cognition in such situations is shared both amongst agents in organizationally prescribed roles(pilot, navigator, etc.), and also amongst the artifacts that they use(i.e. information and procedures in the form of charts, maps, plans, routines, etc.). Cognition is not simply in the head of a single individual. Rather aspects of cognition are also in the world, in social groupings and differentiated roles, in the human-designed artifacts and procedures and systems that people use, and also in the legacy of historical and cultural assumptions and ideas[3].
　　Cognitive activities, such as collaborative problem solving, are prime examples of the distribution of cognition across individual minds and their social, physical, and cultural environments. The view of distributed cognition is critical to understanding the processes and structures of collaborative work[4]. The phenomenon of distributed cognition is a wide-ranging one with provocative consequences for theories of the mind, learning, and education[5].
　　Perspectives on learning termed distributed cognition have become increasingly popular in educational research[6]. King has developed a model for peer tutoring that emphasized the importance of understanding the process of learning, as well as the actual learning of particular subject matter[7]. Derry et al. are working toward the development of a model of thinking in interdisciplinary teams in order to better understand the dynamics of group collaboration[8]. Hewitt and Scardamalia consider the implications of the perspective of distributed cognition for classroom practices[9]. They suggest that classroom practices should be redesigned in order to create “knowledge building communities”�groups that are dedicated to advancing the knowledge of the collective. Lebeau characterizes the use of distributed cognition tools, such as medical histories and physical examinations, that are used in medical diagnosis, with the goal of addressing educational questions about learning to use these tools[10].
　　One perspective on the continuum of interpretations of distributed cognition is an elaboration of information-processing psychology, which fits nicely with a standard cognitive science approach[11]. Knowledge is something that individuals can possess, but sometimes that knowledge is distributed around a variety of artifacts and other individuals. Group processes are seen as a way to facilitate individual learning. Information, skills, expertise, and knowledge are distributed(usually according to roles)around the organizational system, sometimes in relatively permanent structures, sometimes in temporary project groups[12]. The situated action perspective, which embraces a diversity of views, challenges several of the fundamental assumptions of the symbolic information-processing framework underlying cognitive science and artificial intelligence[13].
　　We argue that much of the effort to design information technology to support cognition in organization has not addressed its distributed quality. Such systems have tended to focus either on the individual as an isolated decision maker, or on the group as a producer of a decision or policy statement in common[14]. Although some studies have been done to test decision performance of distributed group in group decision-making[15~19], we failed to find any studies that have investigated what would happen if the information shared and used in group concept-learning is distributed differently in the collaborators. In the present study, psychological experimental method was used to explore the effect of distribution of information among collaborators on collaborative concept-learning.
　　Many recent studies have also focused on the effects of video-mediated systems on the communication process, but the classic studies of Chapanis[20] showing no effect of video on quality have not been refuted. Green and Williges found that the quality of the written task products was consistently high and showed no differences among three communication(audio only, audio plus video, and face-to-face)conditions[21]. Their results indicated that face-to-face conditions resulted in significantly less time to complete the task than did other communication conditions, and that high bandwidth video was not necessary in the particular telecommunication task. Video is no better than audio-only for problem solving map tasks[22] and collaborative design tasks[23] either. Overall, thirty years of research has shown very little effect of video on remote communication[24].
　　Most of the video-mediated communication (VMC)research to date has focused on different tasks and settings[25,26]. These studies have compared various forms of VMC with comparable face-to-face interactions, and have found that even the most technically sophisticated VMC systems alter the style of communication, to a more formal style with fewer interruptions and more explicit handovers of turns. Many studies have found that users desire video connectivity, but the evidence thus far on the importance of video for interpersonal communication has been mixed. The question remains whether video has any value in showing facial expressions and gestures. The results showed that non-native SPEAKERs performed significantly better with a video connection than with audio-only[27]. Farmer and Hyatt also found that video connectivity matters when they examined the effects of the fit of task demands for numeric, visual, and conceptual language with the ability of different communication channels to transfer these types of information[28]. But it is still difficult nevertheless to gain an overall picture of the impact of video mediation on communication and collaboration. In the present study, therefore, we looked in particular at the communication media as well as the distribution of information.
　　In our previous study, an important variable was the people′s primary psychological functions[29]. The results showed a significant interaction among communication media, the sound delay time, and the subjects′ psychological functions on the quality and quantity of verbal communication. Those results have important implications for people who have some types of psychological functions to communicate in different communication media conditions. In our present study, we still look the psychology functions as important variable. We used the Behavioral Styles Inventory(BSI)[30], which is used in our previous study, to ascertain the type of the subjects. BSI is based on the Jung′s theory of psychological type[31] that means everyone has different mental functions, Sensing, Intuition, Thinking, and Feeling. Those functions are very important to analyze and interpret individuals′ behavior. In this study, we only focus on two of them, Sensing and Intuition(see[31]). We classified the people based on that function is their more important psychological function, Sensing or Intuition. For example, when the total or dominant attitude of the individual is orientated by the function of Sensing, we speak of a Sensing function.
　　In the present study, Bruner′s concept-learning task[32], an array of 81 cards, was used. The percentage of information shared by collaborators includes three levels, totally same information(81 same cards), 33% same information(27 same cards), and totally different cards(no same card except the positive example). The media variable includes two conditions, video plus audio and audio only. We hypothesized that amount of information shared by the pair would influence the number of the negative instances selected by them, and communication media might influence the time taken by the pair to complete the task.
　　
　　2Method
　　
　　2.1Participants
　　The participants were 120(60 pairs)male undergraduate students. 30 pairs were from Peking University and the other 30 pairs were from Tsinghua University. Two persons in a pair knew each other well prior to participating in the experiments. They were paid for their participation.
　　
　　2.2Task
　　The concept-learning task was used in this study. The cards of this experiment were similar to those that were used in Bruner et al.′s research[32], an array of 81 cards. The cards varied in four attributes: shape(square, circle, or cross), color (red, black, or yellow), number of boarders(one, two, or three), and number of objects(one, two, or three). Thus, the instances comprise the combinations of the four attributes, each with three values. Each instance in the array exhibits one value of each of the four attributes. The subjects′ task was to choose cards for testing, one card at a time, and after each choice the experimenter will tell them whether the card is positive or negative.
　　The concept designated prior to the experiment was “all cards with two black figures, or with circle(s)and two borders.” No.41 card is the positive instance, which was shown to the subjects for illustrating the concept at the beginning of the experiments.
　　During this experiment, the two persons of one pair could talk to each other, collaborated to select card, and then arrived at the concept.
　　
　　2.3Design
　　In this study, there are two variables, media condition and the amount of information shared by the two people. The first variable involved two levels: video-audio and audio-only. The second variable involved three levels. In the first level(Level A), the two persons shared all the information. All the 81 cards were presented in front of each person. In the second level(Level B), the information shared by the two persons was partly same and partly different. One subject had the cards from No.1 to No.54, but the other person had the cards from No. 28 to No. 81. In the last level(Level C), the information shared by the two persons was totally different. One subject had the cards from No.1 to No.41, but the other person had the cards from No. 41 to No. 81. In all three levels of the second variable, two persons in a pair have the same amount of the positive cards and negative cards.
　　In this study there were totally six conditions that are presented in Table 1: audio and video with totally same information, audio and video with partly same information, audio and video with different information, audio only with totally same information, audio only with partly same information, and audio only with different information.
　　
　　In the video-audio conditions, subjects sat in front of the monitors with their cards so that each one was able to see both the partner and himself via video channel and hear the partner via audio channel. In the audio-only conditions, subjects could only hear on another via audio channel because the monitor windows were turned off.
　　Between-subjects design was used in this 2×3 study. The 60 pairs were randomly assigned to one of the six conditions. Each pair participated in only one experimental condition.
　　
　　2.4Experiment environment
　　
　　In the experimental environment, there were two SUNstation 20 workstations with 20-inch monitor, where two windows were displayed. Voice was recorded through microphone and played through internal speaker. Conferencing application was developed in Tsinghua University[33], and audio and video interaction functionality was employed for the experiment.
　　Realtime video was transferred frame by frame. Each frame was encoded with SUN′s CellB format in 320×240 pixels, and higher frame rate would result in heavier CPU load. In the experiment, frame rate was 8~10 fps, which varied with less or much object motion. Considering that delay for each frame was less than 80ms and that the interval between frames was 100~125ms, the upper bound of delay would be 125ms. Audio coding type was ?-law with silence-suppression. Video and audio data was transferred on 10BASE-T LAN with dedicated channel.
　　
　　2.5Procedure
　　
　　The individual subject was seated in separate rooms for the experiment and their dialogues were recorded using an audio recorder.
　　In order to make the subjects understand how to do the experiment, three exercises were conducted before the formal experiment. The nine cards were used in the warm task. The nine cards varied in texture of triangle(plain, striped, or solid), and position of triangle(left, center, or right). After finishing three exercises, the formal experiment began.
　　At the beginning of the formal experiment, the subjects were told the number of the cards that were distributed to each subject. And then the subjects were told that the experimenter had a concept in her mind and that certain cards before them illustrate the concept, others do not, and that it is his task to determine what this concept is. The experimenter began by showing them a card or instance that is illustrative of the concept, a positive instance. The two persons′ task was to choose cards(no time limit)for testing, one card at a time, and after each choice the experimenter would tell them whether the card is positive or negative. When all the 17 positive cards had been selected, the experimenter would ask them to describe the concept that had been arrived at.
　　During the experiment, the subjects were allowed to remove the cards, as they like. And in the video-audio conditions, the subjects could show his partner the cards via the camera.
　　In the experiment, the subjects were asked to select positive instance. If a negative instance was selected, both of the two persons′ pay would be subtracted RMB 0.50. But the subjects′ lowest pay would be above RMB 15.00.
　　
　　3Results
　　
　　Task performance was measured as both the number of the negative instances selected and the time taken by the pair to complete the task.
　　3.1The quality of task performance
　　The quality of task performance was measured as the number of the negative instances selected by the subjects. The more the number of the negative instances selected by a pair, the poorer the performance of that pair.
　　
　　Table 2 shows the number of the negative instances selected in six conditions. A 2×3 between-subjects analysis-of-variance on the number of the negative instances selected by the subjects didn′t show any significant effects or interactions among media and the amount of information shared by the subjects, though the main effect of the latter is very close to a significant level, F(2,54)=3.028, p =0.057. Figure 1 shows that the quality of subjects′ performance is the best when they shared all the information, and it is the worst when they had totally different information.
　　We then grouped the pairs into three types: the two subjects in a pair are of Sensing Function, the two subjects in a pair are of Intuition Function, and the two subjects in a pair are not of the same psychological functions. Table 3 shows the distribution of subjects′ psychological functions in six experimental conditions. A 2×3×3 between-subjects analysis-of-variance on the number of the negative instances selected by the subjects(Figure 2)indicated a significant main effect of the amount of information shared by the subjects, F(2,43)= 5.42, p=0.008.
　　
　　
　　3.2The speed of task performance
　　The speed of task performance was measured as the time taken by the subjects. The longer the time taken by a pair is, the slower the performance of that pair is.
　　Table 4 shows the time taken by the subjects to complete the task in six conditions. A 2×3×3 between-subjects analysis-of-variance on the time taken by the subjects to complete the task(Figure 3)indicated a significant main effect of media, F(1,43)=7.501, p=0.009; the interaction among the subjects′ psychological functions and the media conditions is very close to a significant level, F(2, 43)=3.084, p=０.056. Figure 4 shows that the pairs of Sensing-Sensing(S-S)conducted the task faster in the audio only condition than in the audio-plus-video condition.
　　
　　
　　4Discussion
　　
　　In the present study, Bruner′s concept-learning task[32]was used to explore the effects of the distribution of information and the communication media on the performance of remotely located pairs of people. We also paid attention to the individual differences, i.e. the subjects′ primary psychological functions.
　　
　　4.1The effect of the distribution of information
　　We found that there was a significant main effect of distribution of information on the number of the negative instances selected by the subjects, indicating that information shared and used by subjects influenced the quality of their collaborative work.
　　The results showed that when each subject in a pair has all the information, they performed much better than they have totally different information. This finding has important implications not only for remotely located pairs of students, but also for decision-makers, planners, and researchers who are cooperating in completing a cognitive task. It should be better for each cooperator to have as more information relative to the task as possible.
　　
　　4.2The effect of the communication media
　　The present study did not find any effect of video on the quality of the subjects′ collaborative work, which is consistent with the previous studies[20～23]. But we found that there is a significant main effect of communication media(audio-plus-video and audio-only)on the time taken by the subjects to complete the task, indicating that the media conditions influenced the speed of collaborative work.
　　The results showed that the subjects completed the task faster in the audio-only condition than in the audio-plus-video condition. In other words, it takes more time for people to cooperate when in the audio-plus-video condition than in the audio-only condition. One reason might be that the subjects spent time looking at each other when showing their facial expressions and gestures. Another reason might be that the subjects tried to show their cards to their cooperator in the present study.
　　As to the question whether video has any value in collaborative cognitive task, we think, the answer might vary with tasks, settings, and collaborators. Though video connectivity does not matter in many situations, it helps to the people with language problems[27] or when the task demands the ability of different communication channels to transfer different types of information[28]. In this study, however, video connectivity made remotely located pairs of people work slower when they completed a concept-learning task.
　　
　　4.3The individual differences
　　In the present study, the results showed that the characteristics of the subjects might be fatal variables that influence any effect we try to investigate, which is consistent with the finding of our previous study[29]. The results of our another previous study showed that the phenomenon of slip of tongue may be related to the personality trait of the individual, especially to the degree of the compatibility of the two communicators′ personality traits[34].
　　Although the present findings do not allow us to conclude what kinds of people in a team may perform better, they do suggest that the interactions among media and psychological functions on the performance of collaboration may be more complex than had been assumed initially. Further research is needed to explore this issue. The findings will have important implications for people who have some types of psychological functions to communicate in different communication media conditions.
　　
　　4.4Language communication
　　At an era of computer and Internet use, though language still plays an important role in communication, the dominant face-to-face language communication has changed to distance communication. A new field of language cognition research has been initiated. Do media influence language processing? Compared to face-to-face spoken language processing, what is the difference in distance spoken language processing? What are the new forms of mental representations and strategies? Such issues immediately called the attention of communication scientists. In this paper, however, we did not report and analyze the subjects′ dialogues because we would focus on the effects of media and distributed information on the subjects′ performance.
　　In our previous study, we found that semantic vagueness, disparity of measurement criteria, semantic generality, assuming something as a matter of course(AMOC), and accumulation effect of misunderstanding might be the five important factors leading to misunderstanding in distance verbal communication, and that accumulation effect of misunderstanding often occurs in the condition of audio with longer time delay[34]. Therefore, it is worth analyzing the subjects′ dialogues. In the future, we will further investigate distance spoken language processing.
　　
　　References
　　 1 Suchman L A. Plans and situated actions: The Problem of Human-machine Communication. Cambridge: Cambridge University Press. 1987, 178
　　 2 Hutchins E. In Search of Navigators. New York: Wiley. 1987
　　 3 Norman D A. The Psychology of Everyday Things. New York: Basic Books. 1986
　　 4 Olson G M, Olson J S. User-centered design of collaboration technology. Journal of Organizational Computing, 1991, 1: 61~83
　　 5 Salomon G. Distributed Cognitions: Psychological and Educational Considerations. New York, NY, USA: Cambridge University Press. 1997
　　 6 Moore J L, T R Rocklin. The distribution of distributed cognition: Multiple interpretations and uses. Educational Psychology Review, 1998, 10(1): 97~113
　　 7 King A. Transactive peer tutoring: Distributing cognition and metacognition. Educational Psychology Review, 1998, 10: 57~74
　　 8 Derry S J, DuRussel L A, O′Donnell A M. Individual and distributed cognitions in interdisciplinary teamwork: A developing case study and emerging theory. Educational Psychology Review, 1998, 10(1): 25~56
　　 9 Hewitt J, Scardamalia M. Design principles for distributed knowledge building processes. Educational Psychology Review, 1998, 10: 75~96
　　10 Lebeau R. Cognitive tools in a clinical encounter in medicine: Supporting empathy and expertise in distributed systems. Educational Psychology Review, 1998, 10: 3~24
　　11 Nardi B. Studying context: A comparison of activity theory, situated action models, and distributed cognition. In Nardi, B. (ed.), Context and Consciousness: Acticity Theory and Human-Computer Interaction, MIT Press, Cambridge, MA, 1996, 69~102
　　12 Clegg C. Psychology and information technology: The study of cognition in organizations. British Journal of Psychology, 1994, 85: 449~477
　　13 Patel V L, Kaufman D R, Arocha J F. Steering through the murky waters of a scientific conflict: situated and symbolic models of clinical cognition. Artificial Intelligence in Medicine, 1995, 7(5): 413~438
　　14 Boland R J, Tenkasi R V, Te′eni D. Designing information technology to support distributed cognition. Organization Science, 1994, 5(3): 456~475
　　15 郑全全，朱华燕，胡凌雁等. 群体决策过程中的信息取样偏差. 心理学报，2001，33(1): 68~74
　　16 朱华燕，郑全全. MAU程序和自由讨论的群体决策质量比较. 心理学报，2001，33(6): 552 ~ 557
　　17 郑全全，刘方珍. 任务难度、决策培训诸因素对群体决策的影响. 心理学报，2003，35(5): 669~676
　　18 何贵兵，曾建华. 分布式群体在动态系统控制决策中的内隐学习. 心理学报，2003，35(6): 777~786
　　19 何贵兵，张平. 个人影响力在群体决策整合过程中的作用：对SDS理论的扩展. 心理学报，2004，36(1): 37~43
　　20 Chapanis A. Interactive human communication. Scientific American, 1975. 232: 36~42
　　21 Green C A, Williges R C. Evaluation of alternative media used with a groupware editor in a simulated telecommunication environment. Human Factors, 1995, 37(3): 283~289
　　22 Boyle E, Anderson A, Newlands A. (1994). The effects of visibility on dialogue performance in a cooperative problem solving task. Language and Speech, 1994, 37(1): 1~20
　　23 Olson J, Olson G, Meader D. What mix of video and audio is useful for remote real-time work. The Proceedings of Computer-Human Interaction 1995(CHI95), 1995, 362~368
　　24 O′Conaill B, Whittaker S, Wilbur S. Conversations over video conferences: An evaluation of the spoken aspects of video-mediated communication. Human-Computer Interaction, 1993, 8(4): 389~428
　　25 McDaniel S E, Olson G M, Magee J C. Identifying and analyzing multiple threads in computer-mediated and face-to-face conversations. The Proceedings of Computer-Supported Co-operative Work 1996(CSCW96), 1997, 39~47
　　26 Gale S. Human aspects of interactive multimedia communication. Interacting with Computers, 1990, 2: 175~189
　　27 Veinott E S, Olson J S, Olson G M, Fu X L. Video matters! When communication ability is stressed, video helps. In Proceedings of CHI 97: Human Factors in Computing Systems, 1997, 315~316
　　28 Farmer S M, Hyatt C W. Effects of task language demand and task complexity on computer-mediated work groups. Small Group Research, 1994, 25(3): 331~336
　　29 傅小兰，史美林，伍尚广等. 影响交流效果的媒体条件与心理因素. 计算机研究与发展，1999，36（增刊）：197~201
　　30 Yan G G. Manual: Behavioral Styles Inventory. Beijing, Personnel Testing Authorities. 1997
　　31 Jung C G. Psychological Type. NJ: Princeton University Press. 1971
　　32 Bruner J S, Goodnow J J, Austin G A. Study and Thinking. New York John Wiley & Sons, Inc. 1956
　　33 Wu S G, Shi M L. Support environment for CSCW Research: Design and Implementation of Desktop Computer Conferencing System. The Proceedings of International Conference on Communication Technology 1996(ICCT96), 1996, 187~190
　　34 Zhou G M, Fu X L. Analysis of Misunderstanding in Distance Verbal Communication. Acta Psychologica Sinica, 2000，32(Supplement), 77~81