Adaptive Feedback Triggers Students Interest in Challenging Tasks: The Mediating Role of Positive Surprise

• In Event: The Interplay of Feedback and Motivation: Perspectives on How Context, Student, and Feedback Characteristics Matter

Abstract

Instructions should motivate students to perform challenging tasks, like writing and revising argumentative texts. Feedback can trigger students' situational interest in challenging tasks (Harks et al., 2014). The use of artificial intelligence makes it possible to adapt feedback to student texts. The further development of the systems follows the assumption that stronger adaptivity makes feedback more interesting. However, it is unclear if data can support this assumption. Further, research is needed on the motivational processes associated with adaptive feedback. Our work investigates effects of feedback adaptivity on students' interests and considers the role of emotions in the process.

The control-value theory (Pekrun, 2006) describes how feedback relates to motivation depending on students' emotional responses. Feedback can trigger strong emotions, moderated by feedback characteristics (Goetz, 2018). Feedback that aligns with student performance can trigger interest because it connects to student performance and allows them to excel. When instruction exceeds learning expectations, it can cause positive surprise (Valdesolo et al., 2017), and being surprised is often positively related to students' interest in the task (Dohn, 2010). In the present study, we investigate whether adaptive feedback triggers more interest than non-adaptive feedback and whether positive surprise can explain this effect.

We conducted an experimental study with a between-subject design and three measurement points (see Figure 1 for the procedure). Students from grades 9 to 12 (56% female, mean age 16.99 years [SD = 1.28]) wrote an argumentative text on climate change and received automated either non-adaptive feedback (N = 247) or adaptive feedback (N = 100) to revise the text. Both groups saw one of the same four feedback messages. In the adaptive group, an algorithm created from a preliminary study selected the best matching feedback message. In the non-adaptive group, the messages were randomly assigned. We asked students about their situational interests and their emotions (i.e., feeling amazed, astonished, surprised). Analyzing the data considering the dependency between three measurement points and distinguishing between state and trait interest, we specified a latent state-trait-occasional model (Castro-Alvarez et al., 2022).

Results show that students receiving adaptive feedback showed higher interest than students receiving non-adaptive feedback (β = 0.22). Exploring the role of emotions, we found that students receiving adaptive feedback were more surprised by the feedback (β = 0.35), and students' surprise was positively related to interest (β = 0.18). Results indicate that surprise explains the effect between feedback adaptivity and interests, indicated by a significant indirect effect (β = 0.05).

The results highlight the importance of adaptive learning opportunities to trigger students' interests in challenging tasks. The results are in line with the control-value theory, showing that feedback adaptiveness predicted students' affective responses, which are related to students' interest in revising their texts. In the presentation we will discuss why adaptive feedback might facilitate surprise and thus interest more strongly than non-adaptive feedback and hypothesize that students might be amazed by the possibility of an algorithm that adapts feedback to their performance.

Authors

• Thorben Jansen, Leibniz Institute for Science and Mathematics Education
  Presenting Author

• Lars Höft, DIPF | Leibniz Institute for Research and Information in Education
  Non-Presenting Author

• Jan Luca Bahr, Leibniz Institute for Science and Mathematics Education
  Non-Presenting Author

• Jennifer Meyer, Leibniz Institute for Science and Mathematics Education
  Non-Presenting Author