Superficial, rather than true, knowledge interdependence in collaborative learning fosters individual knowledge transfer

Superficial, rather than true, knowledge interdependence in collaborative learning fosters individual knowledge transfer Anne Deiglmayr (anne.deiglmayr@ifv.gess.ethz.ch) ETH Zurich, Research on Learning and Instruction, Universitaetsstrasse 41 8092 Zurich, Switzerland Lennart Schalk (lennart.schalk@ifv.gess.ethz.ch) ETH Zurich, Research on Learning and Instruction, Universitaetsstrasse 41 8092 Zurich, Switzerland collaboration script, each learner becomes an expert for a specific domain before collaborating with other learners who have studied a different domain. To ensure fruitful collaboration, the distribution of expertise within groups typically ensures that “none of the group members has enough information or knowledge to solve the task alone” (Dillenbourg & Jermann, 2007, p. 292), establishing true knowledge interdependence. In fact, differences in prior knowledge and perspectives can lead to fruitful knowledge co-construction, in which ideas are critically evaluated, knowledge is elaborated and restructured, and more abstract representations are derived (Andriessen, Baker, & Suthers, 2003; Schwartz, 1995). When learners integrate and transform their complementary knowledge resources, new knowledge can be created that no individual learner would have been capable of constructing (Deiglmayr & Spada, 2011). On the other hand, research on group information processing shows that much of students` unshared knowledge remains unshared in real group discussions. For example, Buchs, Butera, and Mugny (2004) showed that students studying with a jigsaw collaboration script learned substantially less about their partner’s domain of expertise than about their own, even though they were instructed to teach one another during a face-to-face learning phase. Deiglmayr and Spada (2011) showed that students had severe difficulties integrating interdependent information that was distributed between them. Educators face the challenge of creating knowledge interdependence in a way that ensures that learners’ discussions, and the cognitive activities involved, are focused on the most relevant learning content. Establishing true knowledge interdependence, as in classical jigsaw-type collaboration scripts, may not always be the optimal way to achieve this goal. Rather, we argue that superficial knowledge interdependence is often the better solution. Superficial knowledge interdependence denotes that core structures, such as domain principles and important concepts, remain shared between learners, while only contextual information, such as illustrative examples or application contexts, is distributed between learners. The fact that all relevant structural information is given to all students from the beginning maximizes the chance that each learner becomes familiar with the relevant principles via constructive learning processes, while the distributed Abstract We test the hypothesis that superficial knowledge interdependence is more effective in fostering individual learning from collaboration than the true knowledge interdependence often realized by jigsaw-type collaboration arrangements. Based on research on group information- processing, we argue for the benefits of distributing only contextual information, but not core principles between learners, establishing superficial knowledge interdependence. In a computer-supported collaborative learning environment, 78 university students learned about stochastic urn models. Knowledge interdependence was established by systematically distributing learning materials within student triads, so that students either became experts for an urn model, establishing true knowledge interdependence, or for one of the embedding cover stories, establishing superficial knowledge interdependence. Afterwards, all triads worked on the same collaborative tasks, and were exposed to all models. Results show successful learning across conditions, but superior knowledge transfer in triads collaborating under superficial knowledge interdependence. Benefits were highest for low prior knowledge learners. Keywords: computer-supported collaborative learning; learning through comparison; knowledge interdependence; knowledge transfer Introduction In this paper, we explore different ways of distributing information between collaborative learners, with the goal of promoting the interactive construction of mathematical principles during learning from collaborative comparison of worked examples. In doing so, we address the more fundamental question of what characterizes optimal knowledge interdependence in collaborative learning, as assessed by measures of individual learning and transfer. Collaborative learning has the potential of engaging students in forms of interactive knowledge construction that yield learning outcomes beyond those within the reach of an individual learner (Chi, 2009). However, this requires a certain amount of knowledge interdependence between students, that is, the individual students should hold a certain amount of unshared (unique) knowledge, ideas, and perspectives. The deliberate creation of knowledge interdependence is an important factor in many instructional methods for fostering collaborative learning, with the jigsaw collaboration script as their prototype. In a jigsaw