Teachers can easily spot the potential benefits of group work:
- students have lots to learn from each other
- they might benefit from teaching each other
- children who work well together might grow up to be adults who work well together
And so forth.
At the same time, group work creates additional working-memory burdens. Students must use limited WM resources to manage the cognitive work before them — and they must simultaneously use WM to manage the social demands of the task:
- who is leading the discussion?
- is it my turn yet?
- if I disagree with my popular group mate, will I add grief to my lunchroom dynamics?
- why hasn’t that kid stopped talking yet?
For these reasons, I’ve historically been skeptical about group-work enthusiasm. For instance, all the excitement about the jigsaw method is — as far as I can tell — not supported by a strong research base.
Step 1…
Perhaps we can find a good balance here: a group work strategy that reaps the benefits without ramping up WM problems. Even better: we might find research to help us in this question.
Three years ago, I wrote about a study that tried to find that balance. The authors’ hypothesis: if students practice group work, that practice will lower the subsequent WM demands of group tasks. Sure enough, the study’s authors found that students performed better at group work tasks once they’d had a chance to practice at them.
This conclusion came with an important caveat:
- The benefit shows up only if the groups face cognitively challenging problems. Collaboration doesn’t help with basic tasks.
To help our students succeed at group work, step 1 is: have them practice group work.
…Step 2?
Step one gives us a good start. What else can we do?
A more recent study (Wang, Li, & Liu 2026) considers this question:
How much task-specific prior knowledge do students need to learn from group work?
To answer this question, researchers worked with 100+ 5th graders at an urban school in China.
- Students in Group A read a packet explaining how to derive formulas for the area of triangles, trapezoids, and parallelograms.
- Students in Group B read the same packet, and also studied worked-examples of those mathematical processes.
They then studied worked examples:
- Half of students in both groups worked solo.
- The other half worked in groups of three.
So, did the worked examples in the initial packet matter? Did that “task-specific prior knowledge” help students learn more — either solo or in groups?
Helpful Answers…
We’ve got LOTS of data to consider, so let’s focus on the headlines.
- Headline #1: researchers compared “low task-specific prior knowledge” results to “high task-specific prior knowledge” results eight times. In all eight cases, 5th graders who studied worked examples before they attempted them on their own scored higher than those who didn’t. In some cases, students with high levels of task-specific prior knowledge scored 30% higher than their counterparts.
In brief: students who saw those worked examples in the initial packet consistently scored higher.
- Headline #2: on near-transfer problems, students who both previewed the worked examples and worked in groups learned even more when they faced challenging problems.
In this study, group work benefits show up most strongly with a specific combination of factors:
If students preview worked examples before they solve problems, and
if they work in groups, and
if the problems they face create high cognitive demands,
then students get the most benefit from group work.
That’s a lot of “ifs.” But in some classes, the potential benefits group work might make all that careful preparation worthwhile.
…and Important Hesitations
Like all research studies, this one requires some important cautions. We have one study with 5th grade math classes in China. We’re happy to have that study, and grateful the researchers did all that work. But we shouldn’t immediately assume that these results apply everywhere.
If you teach 9th grade, or teach reading, or teach in a culture quite different from China’s, you might not get the same results that these researchers got.
I also want to sound an unusual note of caution. The researchers summarize this study — in part — by saying that students with “high prior knowledge” learn more from group work. While I understand that shorthand, I think it might mislead readers.
The phrase “high prior knowledge” typically describes students who know something because they learned it well in the past. For instance: students in the US typically take Algebra I before taking Geometry. Students who have high prior knowledge of algebra — because they did well last year — benefit from that prior knowledge in their geometry classes. Those who struggled with algebra — and thus have low prior knowledge — may find geometry a painful lift.
This study, however, specifically excludes students who have “high prior knowledge” (according to this definition). That is: they pre-tested students to find out how much they knew about calculating the area of various shapes — and excluded students who already knew about the formula for triangles, etc.
The participants’ “prior knowledge” came not from the past — as the phrase is typically used — but from the worked examples in packet they read. These students had “task-specific prior knowledge” given to them at the beginning of the lesson; they did NOT have “high prior knowledge” in the way that cognitive scientists understand that phrase.
As is so often the case, we need to be very careful about terminology to understand this research correctly.
The Big Picture
If group work overwhelms students’ working memory, it will almost certainly not help them learn. However, we have increasingly helpful research-informed guidance on getting the benefits of this pedagogy. If we
- Give students the chance to practice group work,
- Prepare them for the specific work that they will be doing,
- Ask them to do complex cognitive work…
they are likeliest to learn the most from collaboration with their classmates.
Wang, Y., Li, Q., & Liu, X. (2026). Boundary conditions of collaborative learning using worked-examples: joint moderating effects of task-specific prior knowledge and task complexity on transfer performance and cognitive load. Frontiers in Psychology, 17, 1855051.
