Before someone in your organisation goes blowing the budget on a set of virtual reality headsets, it might be worth having a look at the findings of a new study by Makransky, Terkildsen and Mayer. The research enthusiasts among you may recognise Mayer for his Cognitive Theory of Multimedia Learning (CTML) and this theory does play a part in the study.
The paper is a fascinating read and I can’t capture all of it but I will attempt to explain some of the main findings.
The test subjects were 52 university students who were learning about protein expression through a computer simulation of a laboratory task. In the low-immersion condition, the students viewed the simulation on a regular computer and manipulated the apparatus with a mouse. In the high-immersion condition, the students donned Virtual Reality (VR) headsets that made use of a mobile phone. These headsets provided a 3D version of the lab that the students could explore. They could select objects by clicking a button on the side of the headset.
In addition, the researchers manipulated the ‘redundancy effect’ by providing text in some simulations and text with narration in other simulations. This is an effect that is well known in Cognitive Load Theory (CLT) and CTML and it arises when subjects are asked to read information while listening to verbal narration at the same time. This causes interference because, as Mayer would argue, both sets of words need to be processed in the auditory channel of working memory. It is therefore better to have either the written words or the narration rather than both. This is the reason why you should never read-out or talk over text that you present on a PowerPoint slide.
The authors point out that there are two competing hypotheses about the effect of low immersion versus high immersion (actually, they comprehensively cover a range of ideas but I will stick to the two main ones). Firstly, arguments based on those of John Dewey would predict the high-immersion environment is more authentic, giving more of a practical experience and that this should therefore lead to more learning than low-immersion. Conversely, arguments based upon CTML or CLT would suggest that high-immersion VR would introduce more extraneous details to process, increasing overall cognitive load and degrading learning.
Students were randomly assigned to conditions, with every student attempting one low-immersion and one high-immersion activity. Interestingly, the simulations were in English but the surveys and post-tests were in Danish – I’m not sure how that would effect things although it may increase the cognitive load of the simulation, depending on how fluent the students were in English.
The researchers measured cognitive load directly using an Electroencephalograph (an EEG), a device that records electrical activity in the brain via electrodes attached to the head. The EEG was initially trained by making measures of people performing typical working memory tasks such as accurately reciting lists of digits (digit span tasks) or performing mental arithmetic, as well as looking for correlates with subjects’ own reports of cognitive load. A company has developed proprietary software to process all of this, as well as subtract effects due to people blinking or other muscle movements. I find this quite impressive.
After the simulation, students answered questions directly assessing their knowledge, as well as transfer questions that required them to apply their new knowledge in novel situations. The low-immersion condition led to higher scores on both tests but this was only statistically significant for the knowledge tests. The transfer test items had quite a low reliability and so, with a sample this size, it would have been hard to see an effect.
The main finding was that the low-immersion, computer version of the simulation was more effective than the high-immersion, VR headset version. However, the subjects did report a significantly greater ‘sense of presence’ when using the headsets, which is hardly surprising.
The EEG allowed the experimenters to calculate the proportion of time that students were cognitively overloaded. For the first simulation task, there was no significant difference but in the second task, the VR headset students spent a significantly greater amount of time overloaded. This is consistent with the predictions of CLT and CTML.
Interestingly, the redundancy effect was entirely absent. This is odd but it may be due to a combination of how it was tested and how subjects responded. Most evidence for the redundancy principle comes from comparisons of narration versus text and narration rather than text versus text and narration as used here. In addition, some subjects appeared to ignore the text and just listen to the narration. According to CLT and CTML, this should be effective because it makes use of the modality principle of mixing narration with visual images. I think this area is worth exploring further because it might hint at something we haven’t fully worked out.
So should your school go out and buy those VR headsets? The best we can say from this research is that they may be motivational. For instance, they could be used as a whole class or individual reward for effort or they might help generate excitement about a topic (although this will not necessarily persist once the headsets are taken off). However, if you are buying them because you think your students will learn more then this study suggests you are wrong.