Learning disabilities (LDs) encompass disorders of those who have difficulty learning and using academic skills, exhibiting performance below expectations for their chronological age in the areas of reading, writing, and/or mathematics. Each of the disorders making up the LDs involve different deficits; however, some commonalities can be found within that heterogeneity, such in terms of learning self-regulation and metacognition. Unlike in early ages and later educational levels, there are hardly any evidence-based evaluation protocols for adults with LDs. LDs influence academic performance but also have serious consequences in professional, social, and family contexts. In response to this, the current work proposes a multimodal evaluation protocol focused on metacognitive, self-regulation of learning, and emotional processes, which make up the basis of the difficulties in adults with LDs. The assessment is carried out through analysis of the on-line learning process using a variety methods, techniques, and sensors (e.g., eye tracking, facial expressions of emotion, physiological responses, concurrent verbalizations, log files, screen recordings of human-machine interactions) and off-line methods (e.g., questionnaires, interviews, and self-report measures). This theoretically-driven and empirically-based guideline aims to provide an accurate assessment of LDs in adulthood in order to design effective prevention and intervention proposals.

This chapter focuses on the challenges of measuring self-regulatory processes and learning outcomes while learning from multiple representations with advanced learning technologies (ALTs). More specifically, we present challenges associated with measuring learners' processes and outcomes while using multiple representations with ALTs. ALTs are technology-based learning and training systems designed to teach learners of all ages to learn, reason, problem solve, and understand different topics and domains such as physics, math, history, biology, medicine and can include different systems such as multimedia, hypermedia, serious games, simulations, intelligent tutoring systems, immersive virtual environments and so forth. The structure of our chapter is as follows: (1) a brief overview of the major issues related to using multiple representations for learning with ALTs; (2) a description of the analytical techniques for understanding multiple representations and their goals for successful learning with ALTs; (3) a discussion of the issues associated with using multiple representations to measure and assess learning outcomes and how we can use ALTs as research tools to capture self-regulated learning; and (4) a presentation of challenges and future directions that need to be addressed by researchers.

Game-based learning environments enable students to engage in authentic, inquiry-based learning. Reflective thinking serves a critical role in inquiry-based learning by encouraging students to think critically about their knowledge and experiences in order to foster deeper learning processes. Freeresponse reflection prompts can be embedded in game-based learning environments to encourage students to engage in reflection and externalize their reflection processes, but automatically assessing student reflection presents significant challenges. In this paper, we present a framework for automatically assessing students' written reflection responses during inquiry-based learning in Crystal Island, a game-based learning environment for middle school microbiology. Using data from a classroom study involving 153 middle school students, we compare the effectiveness of several computational representations of students' natural language responses to reflection prompts-GloVe, ELMo, tf-idf, unigrams-across severalmachine learning-based regression techniques (i.e., random forest, support vector machine, multi-layer perceptron) to assess the depth of student reflection responses. Results demonstrate that assessment models based on ELMo deep contextualized word representations yield more accurate predictions of students' written reflection depth than competing techniques. These findings point toward the potential of leveraging automated assessment of student reflection to inform real-time adaptive support for inquiry-based learning in game-based learning environments.

Game-based learning environments are designed to foster high levels of student engagement and motivation during learning of complex topics. Game-based learning environments allow students freedom to navigate a space to interact with game elements that foster learning, i.e., agency. Agency has been studied in learning, and it has been demonstrated that increased student agency results in greater learning outcomes. However, it is unclear what is the level of agency that is required to demonstrate this effect, and whether this effect applies only to learning or to problem solving and affect during game-based learning as well. To investigate how the level of student agency impacts learning, problem solving, and affect, a study was conducted with 138 college students interacting with a game-based learning environment for microbiology, Crystal Island. This study is an extension of a previous study that examined the impact of agency on learning and problem-solving behaviors during game-based learning with Crystal Island. Students were randomly assigned to either a High Agency condition, a Low Agency condition, or a No Agency condition. It was found that students in the Low Agency condition achieved significantly higher normalized learning gain scores than students in the No Agency condition, and marginally higher normalized learning gains than the High Agency condition. Post-surveys of interest and presence indicated that students in the No Agency condition were less interested, and perceived themselves as less present in the virtual environment, than students in the other conditions. Students in the No Agency condition also experienced less frustration, confusion, and joy than the other agency conditions, indicating a less cognitively stimulating experience. Overall the results indicate that a moderate degree of agency provided to students in game-based learning environments leads to better learning outcomes without sacrificing interest and without yielding a negative emotional experience, demonstrating how even low levels of agency can positively impact learning, problem solving, and affect during game-based learning. © 2019 Elsevier Ltd

[No abstract available]

Modeling student knowledge is critical in adaptive learning environments. Predictive student modeling enables formative assessment of student knowledge and skills, and it drives personalized support to create learning experiences that are both effective and engaging. Traditional approaches to predictive student modeling utilize features extracted from students' interaction trace data to predict student test performance, aggregating student test performance as a single output label. We reformulate predictive student modeling as a multi-task learning problem, modeling questions from student test data as distinct “tasks.” We demonstrate the effectiveness of this approach by utilizing student data from a series of laboratory-based and classroom-based studies conducted with a game-based learning environment for microbiology education, Crystal Island. Using sequential representations of student gameplay, results show that multi-task stacked LSTMs with residual connections significantly outperform baseline models that do not use the multi-task formulation. Additionally, the accuracy of predictive student models is improved as the number of tasks increases. These findings have significant implications for the design and development of predictive student models in adaptive learning environments.

Emotions are a core factor of learning. Studies have shown that multiple emotions are co-experienced during learning and have a significant impact on learning outcomes. The present study investigated the importance of multiple, co-occurring emotions during learning about human biology with MetaTutor, a hypermedia-based tutoring system. Person-centered as well as variable-centered approaches of cluster analyses were used to identify emotion clusters. The person-centered clustering analyses indicated three emotion profiles: a positive, negative and neutral profile. Students with a negative profile learned less than those with other profiles and also reported less usage of emotion regulation strategies. Emotion patterns identified through spectral co-clustering confirmed these results. Throughout the learning activity, emotions built a stable correlational structure of a positive, a negative, a neutral and a boredom emotion pattern. Positive emotion pattern scores before the learning activity and negative emotion pattern scores during the learning activity predicted learning, but not consistently. These results reveal the importance of negative emotions during learning with MetaTutor. Potential moderating factors and implications for the design and development of educational interventions that target emotions and emotion regulation with digital learning environments are discussed.

Analyzing multimodal multichannel data about self-regulated learning (SRL) obtained during the use of advanced learning technologies such as intelligent tutoring systems, serious games, hypermedia, and immersive virtual learning environments is key to understanding the interplay among cognitive, affective, metacognitive, and social processes and their impact on learning, problem solving, reasoning, and conceptual understanding in learners of all ages and contexts. In this special issue of Computers in Human Behavior, we report six studies conducted by interdisciplinary teams' use of various trace methodologies such as eye tracking, log-files, physiological data, facial expressions of emotions, screen recordings, concurrent think-alouds, and linguistic analyses of discourse. The research studies focus on how these data were analyzed using a combination of traditional statistical techniques as well as educational data-mining procedures to detect, measure, and infer cognitive, metacognitive, and social processes related to regulating the self and others across several tasks, domains, ages, and contexts. The results of these studies point to future work necessitating interdisciplinary researchers' collaboration to use theoretically based and empirically derived approaches to collecting, measuring, and modeling multimodal multichannel SRL data to extend our current models, frameworks, and theories by making them more predictive by elucidating the nature, complexity, and temporality of underlying processes. Lastly, analyses of multimodal multichannel SRL process data can significantly augment advanced learning technologies by providing real-time, intelligent, adaptive, individualized scaffolding and feedback to address learners' self-regulatory needs.

The goal of this study was to use eye-tracking and log-file data to investigate the impact of prior knowledge on college students' (N = 194, with a subset of n = 30 for eye tracking and sequence mining analyses) fixations on (i.e., looking at) self-regulated learning-related areas of interest (i.e., specific locations on the interface) and on the sequences of engaging in cognitive and metacognitive self-regulated learning processes during learning with MetaTutor, an Intelligent Tutoring System that teaches students about the human circulatory system. Results revealed that there were no significant differences in fixations on single areas of interest by the prior knowledge group students were assigned to; however there were significant differences in fixations on pairs of areas of interest, as evidenced by eye-tracking data. Furthermore, there were significant differences in sequential patterns of engaging in cognitive and metacognitive self-regulated learning processes by students' prior knowledge group, as evidenced from log-file data. Specifically, students with high prior knowledge engaged in processes containing cognitive strategies and metacognitive strategies whereas students with low prior knowledge did not. These results have implications for designing adaptive intelligent tutoring systems that provide individualized scaffolding and feedback based on individual differences, such as levels of prior knowledge.

65 undergraduate students from a North American university interacted with MetaTutorIVH, an agent-based multimedia learning environment that fosters self-regulated learning (SRL) strategy use (e.g., metacognition) while presenting information on several human body systems. Participants completed a self-paced task to study the influence of relevant content and an agents' expressed emotions on metacognitive judgments, gaze behaviors, and science learning. The goal of this study was to examine eye gaze behavior and metacognitive process use relative to their perceived relevancy of content based on discrepancies between multimedia materials and artificial agent's facial expressions. Results indicate an increase in overall fixation duration on multimedia content (e.g., text, diagram) when the text was perceived as less relevant, specifically revealing longer time spent fixating on text when the text was perceived as less relevant compared text and diagrams that were perceived as relevant. Further analyses reveal an increase in diagram fixation duration when the text was judged as being less relevant, but when compared to instances where the text and diagrams were both seen as only somewhat relevant. Across trials, there was no indication of the actual relevancy of content influencing the gaze behaviors of participants.