At current phase as a PhD candidate, I have two broad research interest. One is computer-supported collaborative learning (CSCL) and the other is cognitive science. The two research fields are often considered as distinct and sometimes even contradictory by many researchers, due to their differences in theory-orientation, unit of analysis, and so on. But I always tell myself that there should be a way to combine the two, and we as researchers need to find it. That is because in my faith, a phenomenon as complex as collaborative learning requires a comprehensive way to explore, both at individual and group level. In a collaborative group, individual never exist isolated from other group members, and the group is not merely an empty entity, but comprises individuals. There should be a dialectical relation between individual and group. It means, to help understand what has happened in a collaborative learning episode, we need not only examine the individual, but also the group level, and more importantly, how these two levels interact and affect each other. I think the paper written by Janssen et al. (2010) made my lingering ideas clearly and explicitly.
The paper began with reviewing effect-oriented CSCL research, which primarily examined the effects of collaborative learning with a range of dependent variables such as student achievement and motivation. Yet increasing studies have discovered that not all collaborative learning has positive effects (e.g. Kirschner et al., 2009), and this “black box” approach failed to gain a deeper understanding of the differential learning effects. Eventually, the field realized the need for studying the interaction processes occurring between group members. Throughout years, this line of research has tried to address interaction processes such as giving detailed explanations, negotiating meaning, co-constructing solutions, and reasoning. However, according to Janssen et al. (2010), this type of research also has drawbacks. For example, it could probably result in a tendency to zoom in on the interaction processes without systematically considering how they emerge or affect group performance and individual learning.
To help solve these potential problems and thus make the black box more transparent, the authors advocated to additionally adopt Cognitive load theory (CLT, Sweller, 1994; Sweller & Chandler, 1994) to reveal the complex processes that happen within individual’s head, together with human cognitive architecture. CLT has shown that learning environments can only be effective when they are designed in such a way that they facilitate schema construction and schema automation. In a situation which is too demanding on cognition, schema construction and schema automation are hard to occur. If apply CLT to CSCL, all active sharing, discussing, remembering, and reorganizing of task-relevant information are considered to be information processing, and impose cognitive load on individuals. As long as the communication and coordination between the group members are effective, the intrinsic cognitive load (i.e. the difficulty of the task) could be divided, and each individual would invest less cognitive efforts, comparing to carry out the task alone. On the other hand, the communication and coordination also generate extraneous cognitive load for individuals, and consequently ask for additional cognitive efforts. Collectively, the question of whether collaborative learning produces better learning results than individual learning depends on whether the decrease in intrinsic load overweighs the increase in extraneous load. In my opinion, the trade-off caused by communication and coordination matters more in online collaborative contexts, where communication and coordination are more of asynchronous nature and would gender more extraneous load correspondingly. Except for promising uses of the CLT approach, it also has limitations, mainly in ignoring particular processes involved in inter-individual communication and coordination. To be specific, it still remains unclear which processes are effective in lessening intrinsic load and which processes are ineffective in imposing extraneous load.
The paper ends in proposing to combine the process-oriented and CLT approach together so as to cancel out their disadvantages respectively, and thereafter allows to address new research questions like 1) the effectiveness of communication and coordination processes for learning, 2) the relationship between antecedents, collaborative processes, and consequences, and 3) alternative measurement of cognitive load in collaborative processes.
Personally, this combined research method pushed me to rethink how I conceptualize collaborative learning and what researchers could do to delve into the phenomenon. I guess my whole idea starts with the following two sentences. On page 143, it says “To gain a more fundamental understanding about collaborative learning, it is important to study both the antecedents and consequences of the collaborative process”. I totally agree with it. When you see two group members working on the task separately, instead of doing it together collaboratively (i.e. the process), you may want to ask: Why they didn’t work together? Was it because of an unhappy interaction occurred before (i.e. the antecedence)? What are the results of them not working together? Would they still be able to carry out the task with satisfactory quality (i.e. the consequence)? I consider this antecedence-process-consequence analysis as a horizontal dimension of collaborative learning. Moreover, the authors also mentioned that “both the activities that occur within the heads of the learners and the activities that occur between learners should be studied at the same time” (p.146). I name this simultaneous analysis of intra-individual and inter-individual as a vertical dimension of collaborative learning. It suggests to investigate individual and group level learning at the same time point, and determine how they interact with each other. Based on my knowledge of collaborative learning, only when all factors discussed above (namely antecedence, process, consequence, intra-individual, and inter-individual) have been taken into account could we possibly explain variant learning results, at both group and individual level. For example, why this group failed while that group succeeded? Why within a successful group, some individuals learned while others did not? Why individuals in a failed group also learned something? From my perspective, a systematical framework of multi-level analysis is likely to offer convincing answers to these questions.
However, there are still several places I hope the paper could clarify with more details. Although the paper did distinguish cooperative and collaborative learning, defining the former as associating “with division of labor among group members” and the latter as “a continuous mutual effort of group members to learn by solving problems together” (p.140), it used the term collaborative learning throughout the paper, from my interpretation, to represent both cooperative and collaborative learning since “there are several important similarities” between the two. Yet, regarding the aforementioned trade-off between types of cognitive load, cooperative learning normally sees relatively less communication and coordination with comparison to a collaborative one, so the extraneous load should logically be lower. Therefore, I would recommend to differentiate these two learning situations for the sake of analyzing the phenomenon with more specifics and accuracy.
Another concern I have is that although cognitive load is a key component of this paper, it did not explore enough schemas and how they change as the task progresses. It could partially because that schemas are hard to be visualized and presented. Yet physical artifacts such as concept maps or essays might be an alternative solution. But again, the uneasy part is how to use these tools to reflect schemas in a timely manner. This deserves more careful consideration.
Of course, there are additional issues need to be resolved to pursue this new combined approach. One of them is how to measure cognitive load in collaborative situations. In fact, cognitive load measurement for individuals still remains a challenge as well (Brunken, Seufert, & Paas, 2010). That is one of the areas our research group attempting to contribute to. Besides, the complex interplay between task characteristics, learner characteristics, and group characteristics also increases difficulty for this new line of research. Nevertheless, if we are willing to try out the new approach, I believe new findings would be reported to further unpack the complicated phenomenon of collaborative learning.
References
Brünken, R., Seufert, T., & Paas, F. (2010). Measuring cognitive load. In J. L. Plass, R. Moreno, & R. Brünken (Eds.), Cognitive Load Theory, (pp.181-202). New York: Cambridge University Press.
Janssen, J., Kirschner, F., Erkens, G., Kirschner, P. A., & Paas, F. (2010). Making the black box of collaborative learning transparent: Combining process-oriented and cognitive load approaches. Educational Psychology Review, 22(2), 139-154.
Kirschner, F., Kirschner, P. A., & Paas, F. (2009). Individual and group-based learning from complex cognitive tasks: Effects on retention and transfer efficiency. Computers in Human Behavior, 25(2), 306-314.
Sweller, J. (1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4, 295-312.
Sweller, J. & Chandler, P. (1994). Why some material is difficult to learn. Cognition and Instruction, 12(3), 185-233.