Learning from complex cognitive tasks: comparing groups to individuals

The effects of individual versus group learning (in triads) on efficiency of retention and transfer test performance in the domain of biology (heredity) among 70 high-school students were investigated. Applying cognitive load theory, the limitations of the working memory capacity at the individual level were considered an important reason to assign complex learning tasks to groups rather than to individuals. It was hypothesized that groups will have more processing capacity available for relating the information elements to each other and by doing so for constructing higher quality cognitive schemata than individuals if the high cognitive load imposed by complex learning tasks could be shared among group members. In contrast, it was expected that individuals who learn from carrying out the same complex tasks would need all available processing capacity for remembering the interrelated information elements, and, consequently, would not be able to allocate resources to working with them. This interaction hypothesis was confirmed by the data on efficiency of retention and transfer test performance; there was a favorable relationship between mental effort and retention test performance for the individual learners as opposed to a favorable relationship between transfer test performance and mental effort for the students who learned in groups. Learning from Complex Cognitive Tasks: Comparing Groups to Individuals Collaborative learning models are based on the premise that certain types of learning are best achieved interactively rather than through a one-way transmission process (Johnston, James, Lye, & McDonald, 2000; Littleton & Häkkinen, 1999; Slavin, 1983, 1995; Veerman, 2000; Veerman & Veldhuis-Diermanse, 2001; Weigel, 2002). Although collaborative learning is emerging as a promising educational approach, research on its effects on learning has been highly inconclusive (Kester & Paas, 2005). We believe that these inconclusive results have, among other things, been caused by a lack of attention to the structures constituting human cognitive architecture (Sweller, Van Merriënboer, & Paas, 1998) when designing collaborative learning environments. Research stressing the potential of collaborative learning shows that collaborative learning environments can stimulate and/or enable learners to engage in activities that are valuable for learning. It is important to note that these positive results were found in studies that implemented ‘extra’ measures to ensure that participants engage in effective collaboration, primarily using highly constrained and scripted collaborative learning environments (Dillenbourg, 2002). It has become clear that placing learners in a group and assigning them a task does not guarantee that they will work together, engage in effective collaborative learning processes, and show positive learning outcomes (Soller, 2001). However a controlled collaborative learning environment is also not a guarantee for success (Beers, 2005; De Westelinck, De Craene, & Kirschner, 2005). Results of these recall studies suggest that the collaboration process is detrimental for group-member performance even though the environment was constrained and communication and coordination were bound to rules and kept to a minimum (Weldon & Bellinger, 1997). Different results were found when the tasks used were problem-solving tasks instead of recall tasks. When learners had to work with the information elements, relate them to each other, and by doing so find the solution to a problem (Laughlin, Hatch, Silver, & Boh, 2006). This time, participating in a group facilitated the performance of the individual group member. The type of the task seems to be an important factor in determining whether collaboration is beneficial or not. The research in which individuals and groups are compared regarding their recall or problem-solving performance implies that individual learning is superior to group learning for relatively simple recall tasks, and that group learning is superior to individual learning for relatively complex problem-solving tasks. A possible explanation for this can be found in cognitive load theory (CLT: Paas, Renkl, & Sweller, 2003, 2004). CLT is mainly concerned with individual learning from complex cognitive tasks. It assumes that individuals cannot process an unlimited number of information elements in their working memory (WM). Applying the principles of CLT, this study considers groups as information processing systems consisting of multiple (limited) WMs which can create a collective working space. From this theoretical point of view, multiple collaborating WMs always provide more processing capacity, but whether this capacity can be used effectively depends on the type of task. It can be argued that a group has more effectively available processing capacity than an individual information processing system for tasks in which the relevant information needs to be shared among working memories for learning to commence. In a group, the cognitive load imposed by a task can be shared among group members, and by doing so free-up WM capacity at the individual level that can be used to deal with more complex problems and construct higher quality cognitive schemata compared to an individual working alone. Therefore, the limitations of the WM-capacity at the individual level can be argued to be an important reason to assign complex learning tasks to groups rather than to individuals. However, creating a collective working space is only possible if the relevant knowledge held by each individual group member is communicated and coordinated within the group (Salas, Simms, & Burke, 2005). Structure and control of knowledge communication and coordination are very important for collaborative learning environments to be effective. The beneficial effect of being able to share the cognitive load within a group could be annulled by the costs of communication and coordination between the group members, the so called cognitive and social transaction costs. This concept of transaction costs is more and more used in the field of learning and especially collaborative or cooperative learning (i.e., learning in groups) (Ciborra & Olson, 1988; Yamane, 1996). Within a collaborative or cooperative learning environment these transaction costs are “the costs of setting up, enforcing, and maintaining the reciprocal obligations, or contracts, that keep the members of a team together [and]...represent the “overhead” of the team...linked to the resources (time, skills, etc.) employed to allow a work team to produce more than the sum of its parts” (Ciborra & Olson, p. 95). In our situation, they refer to the specific cognitive load that has to be taken into account when learners are communicating with other learners and coordinating both the carrying out of the task and the communication between each other. When communicating and exchanging information learners are forced to come up with and agree upon a common solution by combining and integrating their individual ideas into a shared and collective one. Because CLT has exclusively focused on individual learners performing an individual task, the cognitive load associated with initiating and maintaining communication and coordination the transaction costs have not received specific attention. However, collaborative learning environments can only be effectively designed if those costs are taken into account. The transaction costs can be argued as imposing intrinsic, germane, or extraneous cognitive load on learners. Intrinsic load is imposed when communication and coordination are inherent to a collaborative learning situation and/or environment, one cannot exist without the other. Germane load is imposed when the transaction costs are effective for learning because they foster shared understanding, trust, mutual performance monitoring, common ground, argumentation, coordination, and positive cognitive conflicts (Leitão, 2000) which have been shown to facilitate the learning process. Extraneous load is imposed when the transaction costs are ineffective for learning because it fosters errors, conflicts, unnecessary duplication, etc. (Bernard & Lundgren-Cayrol, 2001). Especially the extraneous or ineffective cognitive load should be minimized for collaborative learning to be effective. If these costs are not controlled and minimized, the freed-up WM-capacity at the individual level could be used for non-essential or non-learning related communication instead of constructing high quality cognitive schemata. The advantage of being able to share the cognitive load that a complex task causes could be annulled by too high transaction costs. Taking both the complexity of the task and the transaction costs into account, a prerequisite for groupbased learning being more effective than individual learning would be that the demands involved in carrying out the task alone exceed the sum of the cognitive resources that a single individual can supply and the resources needed to deal with the ineffective social transaction costs of communication and coordination of the knowledge between the group members. In this study, it is hypothesized that when performing complex tasks group members will be able to collaborate with one another in a fashion that reduces the high intrinsic cognitive load and therefore will be able to develop higher quality schemata than learners working individually. Higher quality schemata would allow those working in groups to attain higher performance on transfer tasks with less investment of mental effort than individual learners. By contrast, it was expected that those learning from carrying out the same complex tasks individually would need all of their processing capacity for remembering the interrelated information elements, and consequently, would not be able to allocate resources to working with and applying them. This would allow those working individually to attain higher performance on retention tasks with less investment of mental effort than group members. Group members will be able to solve a problem by collaboratively combining the information elements that are distributed across the multiple working memories in the group. Consequently, there will be no need for group members to remember all information elements.