Collaborative Interactions: The Process of Joint Production and Individual Reuse of Novel Ideas

Abstract

Collaborative Interactions: The Process of Joint Production and Individual Reuse of Novel Ideas Mark U. McGregor (markmc@pitt.edu) Michelene T.H. Chi (chi@pitt.edu) Department of Psychology and the Learning Research and Development Center, University of Pittsburgh, 3939 O’Hara Street, Pittsburgh, PA 15260 Abstract Collaborative problem solving involves the active exchange and interaction of ideas between two or more people and such interac- tive exchanges can result in the joint production of co-constructed ideas, some of which may be novel. We analyzed verbal data of pairs of students collaboratively solving problems posed by a com- puter workplace simulation (a banking business), and then indi- vidually solving two transfer problems, in order to examine the frequency of occurrence of co-constructed novel ideas, and the subsequent individual reuse of these co-constructed ideas. The results show that in collaborative interactions, about 20% of the task-relevant ideas were produced jointly, whereas about 80% of the utterances were produced individually (i.e., they were self- explanations). However, about half of these jointly produced ideas (or 10%) were novel. Moreover, individual collaborators were able to reuse these jointly constructed ideas to solve transfer problems. Finally, more interactive collaborative pairs produced a higher proportion of jointly constructed ideas than less interactive pairs, and individual members of more interactive pairs reused jointly constructed ideas more than low interactive pairs. Introduction In classrooms and workplaces, individuals frequently learn by collaborating with others, in tasks such as solving phys- ics problem (Kneser & Plotzner, 2001), planning (Barron, 2000), and learning electricity (van Boxtel, van der Linden, & Kanselaar, 2000). Although operational definitions of collaboration vary widely both within and across various fields (e.g., Psychology, Education, Artificial Intelligence, CSCL), for the purposes of this paper, we define collabora- tion as the active exchange and interaction of ideas between two or more individuals attempting to discover solutions or create knowledge together (Damon, 1984). While some of the results of previous collaboration research are inconsis- tent, the majority support the conclusion that compared to solving a problem alone, collaborative problem solving is often more efficient, and in some conditions, more effica- cious than individual learning (SCANS, 1991; Webb & Palinscar, 1996). Most of the initial research on collaborative learning focused on the environmental conditions under which collaborative learning was more effective than individual learning. Some examples of such environmental factors are group composition, task features, context, and communica- tive medium (Dillenbourg et al., 1996). However, these mediating factors also interact in a highly complex manner, and this complexity has made the resulting examination of this complexity has made the resulting examination of how these multiple interactions produce collaborative learning effects a very difficult pursuit. In part due to this difficulty, an alternative approach to the study of collaboration focuses on the interactive proc- esses that are thought to underlie successful collaborative learning. Examples of such processes are observing peers’ strategies, engaging in productive argumentation, explaining one’s own thinking, sharing knowledge, and providing cri- tique (Azmitia, 1988; Bos, 1937; Coleman, 1998; Hatano & Iganaki, 1991; King, 1990; Phelps & Damon, 1989; Webb, Troper, & Fall, 1995). Many of these processes are captured more-or-less in Webb & Palinscar’s (1996) ‘Input-Process- Output’ model of group (collaborative) processes. The process component of the model contains four common collaborative learning processes: (a) resolving con- flict and controversy, (b) giving and receiving explanations, (c) providing emotional and motivational support, and (d) co-constructing new ideas. The first two processes result in the generation of explanations, either to resolve a conflict, or to explain a problem or solution, and such explanation generation is known to produce learning gains (Chi, Bassok, Lewis, Reimann, & Glaser, 1989; Palinscar & Brown, 1984). The third process is generally comprised of personal- ity factors (e.g., emotion regulation, motivation, social skills, and attitudes), each of which effects how collabora- tors interact with one another. The fourth process, co- construction, can potentially result in novel ideas, or ideas that no collaborator previously possessed explicitly. While Webb & Palinscar’s (1996) four collaborative processes do drive collaborative learning effects, the first three processes are not unique to collaboration -- that is, each can also be observed and implemented in individual learning environments. For example, conflict and contro- versy can arise within oneself when one thinks more deeply about, or attempts to integrate new information with, one’s prior beliefs. The resolution of such conflict (or self-repairs) produces learning gains (Chi, 2000). Similarly, giving help or explanation is akin to self-explaining one’s own thinking without interacting with another individual, and such self- explanations foster learning (Chi et al., 1989). Similarly, receiving explanations has always been shown to be helpful as well, but less so than giving explanations. Providing emotional and motivational support is also not unique to collaboration – individuals are capable of supporting their