Skip to main content

Outline

This week, as the title suggests, we’re talking about robots, avatars, and mice. A bit more specifically, the following questions are addressed:

  1. Are there gender differences in learning with educational robots?
  2. Do we perform better with avatars similar to ourselves? If so, does it matter whether we make them ourselves?
  3. How does distributed practice improve memory?

Let’s get to it!

Educational Robots and Cognitive Load

The first article this week looks at educational robots and learning. Specifically, the authors were interested in whether gender differences exist when learning from educational robots (Chen, Hwang, & Wang, 2021). Artificial intelligence (AI) has been widely used in education settings, particularly since the development of natural language processing (NLP). When looking at AI, NLP allows the computer/app/robot to communicate with people in their natural languages. The current study used NLP to interact with learners, helping them to learn Chinese idioms. The idea with integrating educational robots into this environment is that when we create “joyful” learning, students are more focused and attentive (Chen et al., 2021). However, adding more components may come at a cost with information processing. Past research has indicated that gender differences exist in cognitive load, but few have assessed this in relation to educational robots. If you need an update on Sweller’s cognitive load theory, be sure to check out LSW Issue #35!

In the study, robots would read the idiom that students selected on a screen. After the idiom, students would then play a game for review. The design utilized both intelligent robots and game-based learning. The robot reading aloud idioms while text and images were on the screen followed Mayer’s (2009) findings that a combination of words and pictures help to improve learning outcomes. Results showed that cognitive load was indeed significantly higher for girls than boys. While there was no significant difference in learning performance, the authors point out that boys trended higher than girls (Chen et al., 2021). It was also noted that some girls had negative scores, suggesting that an educational robot may actually cause learning disturbances for girls. These findings echo past work illustrating that boys are more apt to explore new environments and have higher self-confidence when doing so (Ring, 1991; Webley, 1981). It is important to note the study presented was conducted with a 4th grade classroom. Thus, the results can only truly be generalized to this population. However, this finding does suggest that research should be conducted with adults to assess whether this difference continues to persist over the lifespan.

Key Takeaway: When considering incorporating educational robots into a learning environment, it's important to be aware that it may not prove to be beneficial for all learners. As we’ve mentioned before, bells and whistles might be enticing, but may also lead to distractions.

Read More ($): Chen, B., Hwang, G.H., & Wang, S.H. (2021). Gender Differences in Cognitive Load when Applying Game-Based Learning with Intelligent Robots. Educational Technology & Society, 24 (3), 102–115.

“The source of avatar design and the similarity of the avatar to a person can influence performance and subjective experience.”Rahill & Sebrechts (2021)

Avatar Similarity and Performance

The second study in our issue continues with game-based learning, with a focus on avatars. The Proteus Effect is a phenomenon observed in virtual worlds - a person’s behaviors change depending on the characteristics of their avatar. For example, think of the difference between making an avatar that looks like you and using the random generator while playing The Sims. We would be less likely to steal, cheat on our partner, or lie when playing an avatar that more closely resembles us. Researchers Rahill and Sebrechts (2021) were interested in avatar similarity, as well as customization source. Past research suggests that avatar customization leads to increased motivation, involvement, and enjoyment - but does it matter whether the player customizes the avatar themselves? The current study evaluates the relative importance of similarity and construction source.

The authors utilized a within-subjects design, in which all 40 participants engaged in creating similar and dissimilar avatars, as well as having the experimenter create similar and dissimilar avatars for them. Participants engaged in a game of tennis on the Nintendo Wii for each of the conditions. Results showed significant main effects of avatar similarity and construction source on performance. Performance was higher with similar avatars, as well as self-constructed avatars. The authors also found that presence and perception of performance were higher for similar, self-constructed avatars. Overall, the similar avatars showed the best results throughout the study. These findings have important implications for game-based learning, especially those utilizing customizable avatars.

Key Takeaway: For game-based learning environments, prompting learners to create avatars similar to themselves may improve presence, control, and overall performance.

Read More ($): Rahill, K. M. & Sebrechts, M. M. (2021). Effects of Avatar player-similarity and player-construction on gaming performance. Computers in Human Behavior Reports.

Spaced Training and Memory

Our last article for this week reiterates one of the best replicated findings in psychology. A recent article in Current Biology covers a new way to look at the spacing effect (aka distributed practice). The spacing effect suggests that memory is improved when we distribute learning over time, rather than in quick succession. While research on distributed practice has been replicated time and time again, the question of how increased spacing affects the neural pathways of individual memories remains. Thus, Glas et al. (2021) assessed this using an everyday memory task with mice. The mice were split into two groups, one with breaks between maze trials and one that completed trials in succession.

As expected, the researchers found that the prefrontal cortex (an area of the brain important for planning and decision making) was necessary for task performance. Specifically, the crucial region was the dorsomedial prefrontal cortex. While the detailed neural information in this paper excited me, I’ll spare you the nerdy details and leave you with the knowledge that the brain area activated as expected. The neural information that was of particular interest was whether the specific neural ensemble that activated was affected by the trial spacing. Indeed, they found that mice with increased trial spacing (more time between trials) showed a more precise reactivation. This means that mice with more spaced trials are more likely to have memory retention and retrieval. When looking at mice with massed training, memory retrieval was worse than in any spaced training mice. Within massed training, retrieval was better after 24- and 48- hours over 2.5 hours. So, if massed training must be used, it should be at least 24 hours prior to testing.

The big finding from this study is that spaced training, or distributed practice, is going to be more beneficial for learning!

Key Takeaway: Remember to use distributed practice/spaced training to promote better memory retention and retrieval! See LSW Issue #43 for more information on the benefits of distributed practice.

Read More (Open Access): Glas, A., Hübener, M., Bonhoeffer, T., & Goltstein, P. M. (2021). Spaced training enhances memory and prefrontal ensemble stability in mice. Current Biology, 31, 1–10.

Pets of Learning Science Weekly

This week we're featuring a photo of Scout, from human Emily G. Scout enjoys chasing after small rodents, burrowing under blankets, and going on long walks in Central Park! He would probably make the cutest avatar imaginable. Thanks for sharing, Emily!

Send us your pet pics at editor@learningscienceweekly.com.

Wondering why we’re including animal photos in a learning science newsletter? It may seem weird, we admit. But we’re banking on the baby schema effect and the “power of Kawaii.” So, send us your cute pet pics -- you’re helping us all learn better!

The LSW Crew

Learning Science Weekly is written by Kaitlyn Erhardt, Ph.D. and edited by Julia Huprich, Ph.D. Our head of growth and community is Julieta Cygiel.

Have something to share? Want to see something in next week's issue? Send your suggestions: editor@learningscienceweekly.com