The Effectiveness of Intelligent Tutoring Systems on Self-Regulated Learning: A Synthesis of the Literature

• In Event: Advances in Self-Regulated Learning Research Using Learning Technologies: A Synthesis of the Empirical Evidence

Abstract

Objectives. Intelligent tutoring systems (ITSs) have the potential to enhance SRL and learning outcomes. AIEd systems model, support, and augment learners’ ability to dynamically, accurately, and efficiently monitor and regulate their cognitive, affective, and metacognitive knowledge and skills during learning, reasoning, and problem-solving across tasks, topics, and contexts. Various affordances have been isolated that have tremendous impact on learners‘ SRL (e.g., intelligent scaffolding delivered by pedagogical agents), however, other affordances of ITSs have had different effects on SRL (e.g., type, timing, and sequence of SRL interventions) and various learning outcomes.

Theoretical Framework. In our synthesis of the literature, we used Winne’s (2018) theory of SRL which states that learning occurs through a series of four cyclical phases, where metacognitive monitoring and regulation are the hubs of SRL. These processes are captured as events that unfold over time and across several phases. Specifically, the four phases involve: (1) understanding the task, (2) setting goals and making plans to accomplish goals, (3) using learning strategies planned out in (2), and (4) making adaptations to (2-3). This model has been used extensively in the ITS and SRL literature (see Aleven et al., 2016; Azevedo et al., 2018; Biwas et al., 2018; Johnson & Lester, 2016; Lajoie et al., in press). We used it to specificy exactly what, when, how, and why different theoretically-based ITS affordances have impacted SRL and outcome measures differentially.

Findings. Our synthesis will focus on major issues found in the literature including the (1) mapping findings and different pedagogical methods or interventions (e.g., timing, type, and/or delivery of scaffolding by artificial and human agents) and their impact on learning; (2) foci of SRL (e.g., metacognitive knowledge, skills, or both) compared to less-researchersd areas (e.g., emotion regualtion skills) of ITSs; (3) differential impact of metacognitive pedagogical approaches based on the individual differences of learners (e.g., age, prior knowledge, task familiarity, etc.) and ITSs (e.g., tasks and domains, etc.); (4) delivery system of the SRL pedagogical interventions (e.g., teacher, peer, expert, or artificial agent such as a virtual human, fading, scaffolding techniques, etc.); (5) optimal sequencing (e.g., declarative followed by procedural and then conditional knowledge and skills?) and timing (e.g., period, extensive, random) of SRL modeling or training; (6) effectiveness of different SRL pedagogical strategies on performance; and, (7) types of quantitative and qualitative measurements (e.g., self-report, trace data) used to assess (changes in) in instrcutional outcomes and SRL behaviors.

Significance. We propose new methods for detecting, tracking, measuring, and analyzing multimodal SRL data that have specific non-static attributes (i.e., frequency of use, time-dependent patterns of use), offer novel ways to examine and understand the role of these processes across ITSs that can advanced instructional science (see Figure 1). Our presentation will also focus on embedding affordances typically used in non-ITSs that can potentially induce, support or foster SRL with ITSs such as open-learner models (from MCOOs to allow learners to metacognitive monitor their metacognitive skills), agency from game-based learning environments (to foster self-efficacy while dealing with emotion regulation), etc.

Authors

• Roger Azevedo, University of Central Florida

• Elizabeth Brooke Cloude, University of Central Florida

• Daryn Dever, University of Central Florida

• Megan Wiedbusch, University of Central Florida