Collaborative learning can be described as two or more learners actively pursuing and contributing to a shared goal, or trying to share the effort required to reach that goal. It has become a mainstay of educational practice. During my time as a teacher there was always an expectation that most lessons would involve some element of group work. It was always a bit hit and miss and I was never convinced that it resulted in much learning at all. I’m not alone in this view, as this 2015 opinion piece from Tom Bennett (another ex-teacher and now government advisor and founder of the ResearchEd movement) implies.
Bennett doesn’t dismiss collaborative learning entirely, but does question the extent to which group work can benefit the task at hand, suggesting it can be useful in the following ways:
- In situations where tasks are impossible to achieve without it (e.g. football, orchestra).
- To vary the type of classroom activity.
- To improve students’ ability to cooperate, reason with each other, and listen to each others’ opinions.
The research does find significant inconsistencies in the use of collaborative learning. However, there has been an attempt to link collaborative learning to human cognitive processes and, specifically, the notion of cognitive load. Within a Cognitive Load Theory paradigm, collaborative learning is being reconsidered as a way to overcome human cognitive limitations.
Learning as evolutionary
Cognitive Load Theory is underpinned by the principles of evolutionary psychology, central to which is the proposal that human knowledge structures are themselves evolutionary adaptations. According to evolutionary psychologist David Geary, some types of knowledge are part of this evolutionary journey (what he calls biologically primary knowledge) while other types are learned through formal and cultural education (biologically secondary knowledge). This idea has been incorporated into the Cognitive Load Theory framework, along with other elements drawn from the not-so-long history of psychology. For the sake of our current discussion, I’m going to bypass the many problems that arise with evolutionary psychology, suffice to say that evolutionary psychologists have to make greater leaps than their evolutionary biologist cousins to make their theories work, and often they leap too far and land wide off the mark (a possible future discussion, perhaps).
All knowledge once belonged to someone else
Most, if not all, biological secondary knowledge is acquired from someone else; from teachers, books, newspapers, television, Youtube videos and so on. We then need to be able to keep this information in some way for later access. Generally, we refer to such information as being stored in long-term memory. The human capacity to store such vast amounts of information is a biologically primary skill (we don’t have to learn to remember). This information is, in essence, borrowed from the long-term memories of other people; we imitate others, listen to them and read their words. We then assimilate and reorganise this information in the form of schemas, building vast knowledge structures along the way. We do, of course, still possess the ability to generate ideas and information through problem solving, conditioning and trial and error thinking, but we can’t always get the required information in this way – I can try and learn a musical instrument from scratch and with no assistance whatsoever, but chances are I’ll eventually have to seek help from somewhere (or someone) else. Collaborative learning can help me gain access to the information I don’t hold myself, such as chord progressions or the chords required to play a particular tune.
Learning something novel places pressure (or load) on our mental resources because we don’t have access to previously stored knowledge. These limitations are usually discussed in reference to the capacity and duration of short-term or, more specifically, working memory (and I’ve outlined the differences between them here). If the information we are learning isn’t novel, then we’re going to rely on long-term memory to assist us, which is why learning is often described as cumulative – the more we learn, the easier it is to learn more! Collaborative learning, therefore, extends working memory capacity because it provides multiple people (and, therefore, multiple working memories) to reduce load. But information in long-term memory also has real-world application and operates in conjunction with working memory. If, for example, I was to visit a restaurant I hadn’t eaten at before, I’m going to use my knowledge of restaurants in general (my stored restaurant schema) to orientate myself. Now, some aspects of this new restaurant may well be novel, so my working memory is going to be accessing my restaurant schema while simultaneously updating my long-term memory with new information and linking this to what I already know. These principles are necessary if we are going to appreciate how collaboration operates within a Cognitive Load Theory framework.
Collective Working Memory
The outcomes of collaborative learning are two-fold. In the short-term, group members will hopefully complete some learning task or solve a specific problem. In the longer term, group members will hopefully learn something about cooperation and become more skilled in their ability to work as part of a team. How successful this is, relies on a number of important assumptions.
During collaborative learning, group members are able to access knowledge held by other members, introducing what has been termed collective working memory (see Kirschner, Paas and Kirschner, 2011). This process reduces cognitive load because interacting elements of the task can be distributed amongst multiple working memories (of the group members). According to Kirschner et al. (2018), for complex tasks, collaboration becomes a scaffold for individuals’ knowledge acquisition processes. However, if the scaffolding effect fails to materialise, the task simply adds to extraneous load, a situation we’re attempting to prevent.
Why might collaborative learning fail?
The act of collaboration is evolutionary (it’s a biologically primary skill, in the language of Geary). But, as we know, humans don’t always collaborate effectively. In learning, the aim of collaborative learning is to add to our biologically secondary knowledge by pooling cognitive resources, so collaborative learning is probably best suited to complex tasks. Groups members, therefore, come to the table with differing levels of knowledge and expertise; collaboration allows for the sharing of these knowledge and skill resources. If, however, these differences aren’t recognised prior to the activity, collaboration will most likely fail. Similarly, if group members have no experience of working together, learning is, again, more likely to fail. If group members don’t know what they’re meant to be doing or how they intend to tackle the task, they’re unlikely to learn very much. Individuals within the group may learn incidentally, but the group as a whole is unlikely to be successful.
However, if the group members have previously worked together as a team, the chance that learning will be successful is more likely. This may include an understanding of how to organise information and allocate roles. Similarly, research has discovered that providing groups with scripts on how to collaborate and offering just in time support helps enormously (e.g. Fischer et al. 2013).
What students already know will determine whether or not collaborative learning is necessary. If all group members are already knowledgable about the task then it’s probably not worth it. In some cases, collaborative learning can be detrimental and groups members can experience expertise reversal. For high-complexity tasks, groups members learn in a more efficient way than individual learners, but this isn’t the case with low complexity tasks, so it’s best to opt for individual learning.
To really get to the root of why collaborative learning has produced such inconsistent experimental results, we perhaps need to more closely examine how people learn at the cognitive level. This most certainly includes the influence of prior learning – what knowledge group members are bringing to the table and how this knowledge can be pooled. There is certainly more advantages to collaborative learning than Bennett has suggested, but recognising this perhaps requires a deeper understanding of how the human cognitive systems learns new things and differentiates them from what it already knows. Consequently, it’s perhaps far too early to abandon group work just yet.