4TU.CEE is co-organisor of the SEFI2020 conference. This blog is part of the SEFI2020 blog series.

Blog by Marta Gavioli, PhD student at TU Delft

Prof. Pierre Dillenbourg was the keynote speaker on the second day of the SEFI2020 Conference. He started his presentation with the statement: “every speech should try to convey just one idea” and his was:

“The added value of Augmented Reality (AR) for educational purposes lies in the difference between AR environments with reality, not in their resemblance.”

To support this idea, he presented several AR projects developed by the CHILI (Computer-Human Interaction Lab for learning and Instruction) of EPFL (Lausanne, Switzerland), together with the lessons learnt from those projects. These examples were beautiful and really entertaining, so it would be a pity to go too much into details in this blog post and spoil the surprise and the WOW effect. For this reason, I would like to redirect you to the website of prof. Dillenbourg’s lab (, where you can find a collection of videos of every project, or to the recorded SEFI presentation in the conference app. In this blog post, I would like to focus only on the first example presented, because it initiated a series of very interesting reflections among the attendees.

A wrong paradigm

At the beginning of his presentation, prof. Dillenbourg explained how, in education, we often assume the following (wrong) implications regarding the learning gains stemming from an AR activity:

visual fidelity–> immersion –>engagement –>learning

These are not true for a variety of reasons. First of all, it is not true that a successful AR activity requires high visual fidelity. In this regard, the protagonists of the first presented AR project are a group of Swiss vocational students, who normally spend one week working in a company and one week in school. In these cases, there is often a mismatch of what students are allowed to do as apprentices and what they are taught in school. In the company, students’ tasks are often very basic, such as moving boxes in a warehouse. On the contrary, the school wants students to learn more abstract concepts to be prepared for their future job, for example the basic principles of logistics and how to manage a warehouse. Because this learning experience was not possible in the real world, Dillenbourg’s team developed a very sophisticated learning activity which implements a mix of projectors, sensors, small-scale models, paper, and augmented reality. In this case, AR allowed students to manipulate the environment, trying out different arrangement of the warehouse shelfs. Students were also able to see things that are normally invisible, such as the potential trajectories of forklifts. Moreover, students could experience variety, being able to play around with different factors while deciding on how to load the trucks. The activity was immersive even if it had low visual fidelity to the real environment. The winning characteristic was its high cognitive fidelity, so that students could perform real tasks in an augmented environment.

The relationship between engagement and learning is not linear

Screenshot from the keynote presentation


Even more interestingly, Dillenbourg also discussed some challenges his team faced during the activity development. In fact, during the first trial, the researchers noticed that no learning gains were measured even if students were very engaged. Surprisingly, the reason was that students were so engaged in the manipulation of the AR environment that they did not pay attention to the concepts they were supposed to learn. The researchers found out that the relationship between engagement and learning is not linear, and the only way to achieve learning was to disengage the students a bit and find a way to force them to reflect. They did so by adding a task where the teacher could guide them in the reflection and reification process.

The role of the teacher

Screenshot from the keynote presentation

An interesting reflection stemmed from one question from the audience. All the examples presented during the keynote speech showcased very advanced and sophisticated technological tools which made many people wonder if teachers will ever be able to learn how to develop and use them. However, prof. Dillenbourg pointed out how, in this case, the best teacher is not the most technologically fluent, but the one who can push student to reflect, building upon what students experienced during the AR activity. From this perspective, a debriefing stage after the activity is essential to help students construct knowledge.

A final remark

Prof. Dillenbourg clarified that we cannot say that AR is good for learning, in the same way we cannot say books are good. There are good books and bad books as there are good or bad AR activities. Like everything in education, how good an activity is depends on many variables, such as our goal and how we implement it.




Be Sociable, Share!