No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The Knowledge-Learning-Instruction (KLI) Framework: Toward Bridging the Science-Practice Chasm to Enhance Robust Student Learning
Abstract
Despite the accumulation of substantial cognitive science research relevant to education, there remains confusion and controversy in the application of research to educational practice. In support of a more systematic approach, we describe the Knowledge-Learning-Instruction (KLI) framework. KLI promotes the emergence of instructional principles of high potential for generality, while explicitly identifying constraints of and opportunities for detailed analysis of the knowledge students may acquire in courses. Drawing on research across domains of science, math, and language learning, we illustrate the analyses of knowledge, learning, and instructional events that the KLI framework affords. We present a set of three coordinated taxonomies of knowledge, learning, and instruction. For example, we identify three broad classes of learning events (LEs): (a) memory and fluency processes, (b) induction and refinement processes, and (c) understanding and sense-making processes, and we show how these can lead to different knowledge changes and constraints on optimal instructional choices.
... Prior research suggests that instructional principles do not apply universally, but instead seem to depend on the complexity of the knowledge that is the target of instruction (Koedinger et al., 2012). Building on this observation, Koedinger and colleagues (2012) offer the alignment hypothesis, which suggests that the type of instruction needs to match the complexity of the learning goal. ...
... The processes through which students become perceptually fluent are qualitatively different from those through which students learn to make sense of visual representations. Perceptual fluency is acquired via implicit, nonverbal, inductive processes (Gibson, 2000;Goldstone et al., 1997;Kellman et al., 2010;Koedinger et al., 2012). These processes are often unintentional and seem to occur unconsciously (Frensch & Rünger, 2003). ...
... First, identifying surgical anatomy is fundamentally different from content knowledge in undergraduate STEM education in domains like math or chemistry. In undergraduate STEM education, learning is thought to involve the ability to make sense of domain-relevant concepts (Koedinger et al., 2012). Thus, tests of content knowledge that prior studies have used to assess conceptual knowledge of math and chemistry are much closer to sense-making interventions (designed to enhance students' ability to make sense of visuals) than they are to perceptual-fluency interventions (designed to lead to fluent perception of information from visuals). ...
Combinations of perceptual fluency and sense-making competencies contribute synergistically to learning gains in undergraduate science, technology, engineering, and math (STEM) education. However, instructional principles depend on the target of instruction, and in many fields, the targets of instruction are quite different from undergraduate STEM education. Professional learning often involves the application of previously acquired conceptual knowledge in a perceptually complex reality. This paper focuses on the field of surgery, specifically the recognition of surgical anatomy, in which the target of instruction is perceptual ability rather than conceptual knowledge. We conducted two experiments in which 42 and 44 surgical trainees participated in perceptual-fluency and sense-making interventions, followed by tests of their ability to recognize surgical anatomy in real operative images. The results showed that perceptual-fluency interventions contributed to gains in perceptual knowledge relating to surgical anatomy, whereas sense-making interventions did not. We discuss our findings in terms of alignment between instructional design and instructional goals, and the application of advances in learning sciences to adult learning of complex skills.
... It can be contrasted with a claim, which is almost certainly true, that learners in more favorable conditions learn at a more rapid rate than those in less favorable conditions. The educational technologies used in our NSF-funded LearnLab studies arguably provide favorable learning conditions as they implement research-based principles (e.g., varied practice with feedback and explanatory instruction), and many have been improved through iterative data-driven cognitive task analysis and experimental methods (12). A key goal of LearnLab was to identify, in the words of the National Academy of Sciences report, high-ability learners who "learn at a more rapid rate than other students" (13, p. 37). ...
... Second, we found that reaching a reasonable level of mastery (80% correct) requires substantial repeated practice, typically about seven practice opportunities. These results are consistent with learning theories suggesting induction from examples and doing is prominent in human learning (12,41). Third, students' initial performance is highly variable despite students entering the courses in which the data were collected having met prerequisite requirements (for college courses) or age-level requirements (for K-12 courses) and having received verbal instruction. ...
... Books and then recorded lectures have facilitated broader dissemination of knowledge historically, but much emphasis on lecture recording remains today even in online course contexts where interactive practice is feasible and effective (cf., 20). A theoretical postulate consistent with limited accuracy after up-front verbal instruction is that human learning is not simply about the explicit processing, encoding, and retrieval of verbal instruction but as much or more about implicit or nonverbal learning-by-doing in varied practice tasks where interactive feedback is available (12). ...
Leveraging a scientific infrastructure for exploring how students learn, we have developed cognitive and statistical models of skill acquisition and used them to understand fundamental similarities and differences across learners. Our primary question was why do some students learn faster than others? Or, do they? We model data from student performance on groups of tasks that assess the same skill component and that provide follow-up instruction on student errors. Our models estimate, for both students and skills, initial correctness and learning rate, that is, the increase in correctness after each practice opportunity. We applied our models to 1.3 million observations across 27 datasets of student interactions with online practice systems in the context of elementary to college courses in math, science, and language. Despite the availability of up-front verbal instruction, like lectures and readings, students demonstrate modest initial prepractice performance, at about 65% accuracy. Despite being in the same course, students' initial performance varies substantially from about 55% correct for those in the lower half to 75% for those in the upper half. In contrast, and much to our surprise, we found students to be astonishingly similar in estimated learning rate, typically increasing by about 0.1 log odds or 2.5% in accuracy per opportunity. These findings pose a challenge for theories of learning to explain the odd combination of large variation in student initial performance and striking regularity in student learning rate.
... Studies of the ICAP framework have shown that interactive engagement, demonstrated by co-generative collaborative behaviors, is superior to constructive engagement, characterized by generative behaviors such as self-explanation, as well as active or passive learning (Chi et al., 2018). The knowledge, learning, and instruction (KLI) framework that underlies many cognitive tutors highlights the value of designing responses for each student idea (Koedinger et al., 2012). The storylines instructional framework emphasized the collaboration of materials developers, educational researchers, classroom educators, and educational leaders to align materials with disciplinary standards (Edelson et al., 2021;Penuel et al., 2022). ...
... Pedagogical frameworks offer conflicting insights concerning the mechanisms governing CSCL. The KLI framework argues there is insufficient evidence to support the value of collaboration for learning (Koedinger et al., 2012). However, ICAP research suggests that students can deepen their understanding by watching how other students collaborate to address a disciplinary dilemma (Chi et al., 2018). ...
... Designers of guidance face a dilemma between efficiently telling students the answer and giving hints to encourage self-directed learning (Koedinger & Aleven, 2007). Reviews of studies of guidance report conflicting results for design of online adaptive guidance, depending on whether the goal is self-directed inquiry or rapid acquisition of detailed procedures (Koedinger et al., 2012;Gerard, Matuk et al., 2015). Supporting self-directed learning involves enabling students to monitor their own progress, look for opportunities to distinguish ideas, and reflect on their learning. ...
Advances in technology tools and learning sciences research are increasing equitable science outcomes by enabling students to take control of their own investigations and study issues aligned
with their personal and cultural interests. Design teams are using advances in technology to design automated guidance that amplifies the ability of teachers to provide personalized guidance. Further, advances in learning analytics and natural-language processing (NLP) are beginning to scaffold learners and support teachers to guide their students in real time. We report on a confluence of technological advances, including powerful visualizations, collaborative tools, and automated guidance, that, when combined in learning environments, support students to become self-directed learners.
Emerging authorable and customizable environments (ACEs) have the potential to enable teachers to design and customize their instruction using evidence from their students’ work. ACEs can
support dashboards that use learning analytics to synthesize student progress in ways that teachers can use both immediately and when planning future instruction. Further, as students benefit from opportunities to direct their own learning, they gain insights that reinforce their identity as science learners.
... Domains involving human decision-making have long recognized that realistic practice and tailored feedback are key processes for learning. Literature from the learning sciences has demonstrated how practice improves performance across domains, from academic subjects like math and psychology to complex medical procedures and even to professionals' abilities to navigate challenging social interactions [22,29,34]. In prior literature, these simple practice effects have been substantially enhanced when supplemented with explicit feedback [17,33,34]. ...
... Literature from the learning sciences has demonstrated how practice improves performance across domains, from academic subjects like math and psychology to complex medical procedures and even to professionals' abilities to navigate challenging social interactions [22,29,34]. In prior literature, these simple practice effects have been substantially enhanced when supplemented with explicit feedback [17,33,34]. In real-world decision-making settings such as social work, healthcare, and education, where there is typically a long lag (on the order of months or even years) between decisions and corresponding outcomes, such simulation-based training approaches have the potential to greatly accelerate learning. ...
A growing body of research has explored how to support humans in making better use of AI-based decision support, including via training and onboarding. Existing research has focused on decision-making tasks where it is possible to evaluate "appropriate reliance" by comparing each decision against a ground truth label that cleanly maps to both the AI's predictive target and the human decision-maker's goals. However, this assumption does not hold in many real-world settings where AI tools are deployed today (e.g., social work, criminal justice, and healthcare). In this paper, we introduce a process-oriented notion of appropriate reliance called critical use that centers the human's ability to situate AI predictions against knowledge that is uniquely available to them but unavailable to the AI model. To explore how training can support critical use, we conduct a randomized online experiment in a complex social decision-making setting: child maltreatment screening. We find that, by providing participants with accelerated, low-stakes opportunities to practice AI-assisted decision-making in this setting, novices came to exhibit patterns of disagreement with AI that resemble those of experienced workers. A qualitative examination of participants' explanations for their AI-assisted decisions revealed that they drew upon qualitative case narratives, to which the AI model did not have access, to learn when (not) to rely on AI predictions. Our findings open new questions for the study and design of training for real-world AI-assisted decision-making.
... This idea has emerged, for example, from work by Newell and Rosenbloom (1980), who demonstrated a very general power-law decrease in task completion times as a function of trial numbers for a wide variety of tasks. The use of response times has especially been used to measure performance or model student mastery of a given "knowledge component" in intelligent tutoring systems [18,19] and in other learning contexts [20], including problem-solving [21]. One must always keep in mind, though, the possible confound of increasing speed due to retesting effects [22]. ...
... To summarize, let us discuss how our results address our research questions, starting with RQ1 and RQ4. Following expected patterns of accuracy and response time learning curves typically found in studies of learning ( [18]), both algebra and calculus students on average systematically improved their accuracy and decreased their response time per question on a range of physics topics and categories over multiple repeated spaced practices throughout the semester. While calculus students were slightly faster and more accurate than the algebra students, the STEM fluency assignments were still effective and beneficial to both classroom populations in improving student fluency and performance on the assignments. ...
We have investigated the temporal patterns of algebra (N=606) and calculus (N=507) introductory physics students practicing multiple basic physics topics several times throughout the semester using an online mastery homework application called science, technology, engineering, and mathematics (STEM) fluency aimed at improving basic physics skills. For all skill practice categories, we observed an increase in measures of student accuracy, such as a decrease in the number of questions attempted to reach mastery, and a decrease in response time per question, resulting in an overall decrease in the total time spent on the assignments. The findings in this study show that there are several factors that impact a student’s performance and evolution on the mastery assignments throughout the semester. For example, using linear mixed modeling, we report that students with lower math preparation for the physics class start with lower accuracy and slower response times on the mastery assignments than students with higher math preparation. However, by the end of the semester, the less prepared students reach similar performance levels to their more prepared classmates on the mastery assignments. This suggests that STEM fluency is a useful tool for instructors to implement to refresh student’s basic math skills. Additionally, gender and procrastination habits impact the effectiveness and progression of the student’s response time and accuracy on the STEM fluency assignments throughout the semester. We find that women initially answer more questions in the same amount of time as men before reaching mastery. As the semester progresses and students practice the categories more, this performance gap diminishes between males and females. In addition, we find that students who procrastinate (those who wait until the final few hours to complete the assignments) are spending more time on the assignments despite answering a similar number of questions as compared to students who do not procrastinate. We also find that student mindset (growth vs fixed mindset) was not related to a student’s progress on the online mastery assignments. Finally, we find that STEM fluency practice improves performance beyond the effects of other components of instruction, such as lectures, group-work recitations, and homework assignments.
... Nevertheless, the task of learning characters is difficult because there are tens of strokes, hundreds of radicals/components, various structures/positions, and thousands of characters to be memorized Wang, 2011;. Thus, supporting learners of Chinese to develop robust orthographic representations in reading and writing characters is important (Koedinger et al., 2012). ...
... The robustness of each strategy effect merits further investigation. To deal with this issue, we suggest to vary testing times (simultaneous vs. successive) and to provide multiple practice or review opportunities to find an optimal schedule (Koedinger et al., 2012). Second, to secure the internal validity of research, we conducted this study in laboratories with rigorous control, which reduced the external validity of research. ...
... Nevertheless, the task of learning characters is difficult because there are tens of strokes, hundreds of radicals/components, various structures/positions, and thousands of characters to be memorized Wang, 2011;. Thus, supporting learners of Chinese to develop robust orthographic representations in reading and writing characters is important (Koedinger et al., 2012). ...
... The robustness of each strategy effect merits further investigation. To deal with this issue, we suggest to vary testing times (simultaneous vs. successive) and to provide multiple practice or review opportunities to find an optimal schedule (Koedinger et al., 2012). Second, to secure the internal validity of research, we conducted this study in laboratories with rigorous control, which reduced the external validity of research. ...
Written Chinese is unique because of its logographic orthography in nature and the correspondence between Chinese characters, morphemes, and syllables. Therefore, reading acquisition of Chinese is a major challenge for those learning Chinese as a second/foreign language (CSL/CFL). However, studies on reading acquisition in CSL/CFL learners are sparse. It remains unclear how CSL/CFL learners acquire the knowledge of Chinese characters (e.g., the structures including the intricate strokes and square configurations) and establish morphological awareness (e.g., “学” in “学校” and “才学” is the same morpheme, but “面” in “面粉” and “面孔” are two different morphemes). Furthermore, there is a lack of empirical studies on how various linguistic skills that are significantly associated with reading in native Chinese speakers (e.g., orthographic knowledge, phonological awareness, and vocabulary) contribute to sentence/passage reading in CSL/CFL learners with various Chinese proficiency levels.
This Research Topic in Frontiers in Psychology aims to present scientific studies on reading acquisition in CSL/CFL learners that help to reveal the developmental trajectories of reading ability and the contributions of various perceptual, linguistic, and cognitive factors to reading development in CSL/CFL learners. Welcome contributions will focus on reading acquisition in CSL/CFL learners at all levels such as character, word, sentence, and passage. Particular attention will be given to the integration of behavioral, electrophysiological, and neuroimaging techniques to reveal the mechanisms underlying Chinese character recognition and semantic integration during reading.
Topics of interest include, but are not limited to:
• Development of orthographic awareness;
• Development of morphological skills;
• Sentence/passage reading and the contributing linguistic and cognitive factors;
• The relationship between listening comprehension and reading;
• Electrophysiological (e.g., the ERP components N170 and N400) and neuroimaging measures (e.g., activation of the visual word form area in the left fusiform gyrus) of various reading processes.
... Growing efforts in "learning engineering" are beginning to shed light on processes that help recognize what works, why it works, and how to scale what works (Koedinger, Corbett, & Perfetti, 2010;Heffernan & Heffernan, 2014, etc.). Learning engineering was originally introduced by Herbert Simon (1967) and recently formalized as "a process and practice that applies the learning sciences using human-centered engineering design methodologies and data-informed decision making to support learners and their development" (Kessler et al., 2023). ...
... It leverages human-centered design to guide design choices that promote robust student learning, but also emphasizes the use of data to inform iterative design, development, and improvement process. The knowledge-learning-instruction (KLI) framework (Koedinger et al., 2010) and similar efforts such as ASSISTments as an open platform for research (Heffernan & Heffernan, 2014) are excellent examples of learning engineering in practice. The KLI framework advocates for in vivo experimentation, which enables rigorous experimental controls in real learning settings with real students. ...
... 16,17 Computational understandings of learning: Learning theories have for med the basis of computational representations of human learning-learner modeling-that underlie many AI instructional systems. 2 The knowledgelearning-instruction (KLI) framework has supported the representation of human learning in computational models using knowledge components (KCs), a single unit of cognitive function. 19 In the KLI framework, all elements to be learned are organized into related KCs. Competence is estimated within KCs through assessment events such as quizzes. ...
... Competence is estimated within KCs through assessment events such as quizzes. 19 This framework of developing and assessing competence in various KCs has been implemented in many AI instructional systems, where learning material is organized into KCs and learner modeling techniques such as Bayesian knowledge tracing or deep knowledge tracing implement this assessment of competence for individual KCs using learner data. 20 This dynamic estimation of learner mastery supports pedagogical policies such as "mastery learning," where continued instruction or practice in material relevant to the KC is applied until a threshold mastery level. 2 Other strategies in AI instruction have similarly drawn on learning theories. ...
We review some state-of-the-art approaches in the area of human-in-the-loop artificial intelligence (AI) applications, examine some promising human–AI collaboration use cases in adaptive instructional system settings, and also discuss IEEE artificial intelligence and learning technology standards efforts to advance this endeavor.
... From a learning science perspective, individual and collaborative learning each have their own benefits. From the point of view of the Knowledge Learning Instruction (KLI) framework [29], collaborative learning offers opportunities for mutual elaboration and co-construction of knowledge, whereas individual learning may promote induction and refinement as learning mechanisms (cf. [29,41]). ...
... From the point of view of the Knowledge Learning Instruction (KLI) framework [29], collaborative learning offers opportunities for mutual elaboration and co-construction of knowledge, whereas individual learning may promote induction and refinement as learning mechanisms (cf. [29,41]). Previous work by Olsen et al. has shown that combining individual and collaborative learning may yield more effective learning for students than either mode alone [42]. ...
... Kemampuan representasi merupakan salah satu keterampilan tingkat tinggi yang dibutuhkan sebagai bekal untuk menghadapi era abad 21. Representasi digunakan untuk menggambarkan konsep ilmiah (interpretasi), menghasilkan representasi (konstruksi), mengidentifikasi, menjelaskan, menganalisis fitur representasi, menghubungkan berbagai representasi, dan menjelaskan hubungan di antara mereka (Ainsworth, 2018;Koedinger et al., 2012;Ruz & Schunn, 2018). Representasi berisi interpretasi dan penjelasan mengenai ide atau konsep ilmiah dengan menggunakan mode seperti analogi, pernyataan verbal, teks tertulis, diagram, grafik, dan simulasi (Tang et al., 2014). ...
... Supaya mahasiswa lancar dalam menyusun sebuah representasi, maka seharusnya mereka dilatih terlebih dahulu. Kelancaran representasi adalah cara untuk menyatukan berbagai ide tentang bagaimana dan mengapa penggunaan banyak representasi penting bagi siswa, pendidik, dan peneliti pendidikan (Koedinger et al., 2012). Hasil penelitian menunjukkan bahwa penggunaan multi modal representasi dalam kegiatan pembelajaran meningkatkan keterampilan berpikir kritis ilmiah siswa (Sunyono & Meristin, 2018). ...
Concept mastery is a person's basic ability to bring up other abilities in learning. Therefore it is necessary to conduct research on the relationship between mastery of concepts and other content such as the ability to use representations horizontally, which is referred to as cross-mode horizontal translation or HTM. One indicator of someone who has representational abilities is being able to express biological concepts by using representations horizontally. The purpose of this study was to determine the correlation of student mastery of concepts to representational abilities, especially HTM. The research method is a correlational analysis between concept mastery and HTM abilities. The research subjects were 37 students who had taken the Plant Physiology course at the Biology Education Study Program at the Mandalika University of Mataram, in the even semester of the 2019/2020 academic year. Data was collected using essay tests that had been assessed by experts and declared fit for use. Data analysis used a regression test at a significance level of 5%. The results showed that there was a correlation between concept mastery and students' representation abilities in interpreting scientific phenomena horizontally (HTM) in photosynthetic materials. The amount of loan provided by understanding the concept of representation ability is 55.70%.
... This would give each student a more personalized learning experience. This could be accomplished through the use of adaptive learning [18,19]. Also, artificial intelligence could be helpful in the field of social media by making it possible for chatbots to provide efficient and personalized customer service and by making it possible to find and delete unwanted information on social media platforms [20,21]. ...
Artificial intelligence (AI) is a rapidly evolving technology with the potential to revolutionize various sectors, including healthcare, banking, and transportation. AI’s impact on society is substantial, introducing advancements that enhance
numerous aspects of daily life. While AI offers remarkable advancements, it also raises concerns about privacy, accountability, and bias. This paper explores AI’s potential in improving patient care, fraud detection, risk management, and traffic optimization. The research’s implications are vital for policymakers, industry leaders, and researchers, emphasizing the need for responsible and ethical AI frameworks. As AI’s impact grows, careful consideration is essential to harness its benefits while addressing potential risks.
... instruction should produce student outcomes demonstrating improvement, the opportunity for students to expand relative to cognition, and utilize specific teaching strategies appropriately to increase student cognition (Koedinger et al., 2013;Shi et al., 2020). Furthermore, instructional events are expected to generate learning events (Koedinger et al., 2010). therefore, student learning should be outcome specific (Hartikainen et al., 2019;Hatfield, 2015). ...
Assessments play a pivotal role in student performance within higher education courses. In this article the effect of deemphasizing homework assignments and focusing on the course driven project had on undergraduate students’ performance is clearly described. Using student grades as data sets, performance is compared over the Fall 2020, Fall 2021, and Fall 2022 semesters. Grade computation has slightly changed over these three semesters, primarily related to the deemphasizing of homework assignments. During the Covid-19 impacted Fall 2020 course participation was calculated using weekly quizzes instead of through in-class participation. Due to the varied majors of students, a scaffolding approach was used to deliver course content, so all students were allowed the opportunity to build a sophisticated website through project-based learning. While deemphasizing homework assignments did not positively affect student performance, students produced more professional websites for their final project.
... Correspondingly, educational psychologists can also greatly benefit from bringing in new variables of interest from cognitive psychology. Cognitive psychologists have established a variety of different cognitive and metacognitive constructs and processes that support learning (Dunlosky & Metcalfe, 2009, Chapter 3;Koedinger et al., 2012), while educational psychologists have generally lumped them into higher order units or categories to better relate them to other aspects of self-regulation (Entwistle, 1997;Pintrich et al., 1993). An example of educational psychologists who are bridging fields is the theoretical model of self-regulated learning proposed by Efklides (2006Efklides ( , 2011, which describes how metacognitive experiences (e.g., judgments of learning, feelings of difficulty) are closely associated with affect, and how their interactions impact student learning. ...
Understanding how students self-regulate their learning experiences has been at the forefront of many empirical and theoretical advances in both cognitive and educational psychology. Yet, these two fields have traditionally investigated this multifaceted aspect of learning using different approaches, resulting in scientific knowledge that is siloed in separate literatures. The overall aim of this theoretical synthesis and review is to bridge this divide and shed light on potential integrative perspectives and approaches. We compare the theoretical and methodological approaches that have been commonly adopted in these two fields. Next, we identify several factors that contribute to the divide between the two fields. Finally, we discuss three elements that are essential for integrating perspectives and developing a holistic understanding self-regulation of learning: awareness and understanding, innovation and expansion, and collaboration and interaction. Throughout the review, we highlighted how bridging the two divergent, yet complimentary perspectives could facilitate innovative research, extensive practical implications, and improved theory.
... A key theoretical assumption that has emerged from past ITS research is that eventually-correct steps are instrumental in student learning from deliberate practice [12,21,36]. Following past research, we consider a competency to be learned as comprising multiple Knowledge Components (KCs) [22]. We then consider a "practice opportunity" to be attempts to complete a step in a practice problem that "exercises" one or more KCs, including any feedback or instruction toward completing that step. ...
In numerous studies, intelligent tutoring systems (ITSs) have proven effective in helping students learn mathematics. Prior work posits that their effectiveness derives from efficiently providing eventually-correct practice opportunities. Yet, there is little empirical evidence on how learning processes with ITSs compare to other forms of instruction. The current study compares problem-solving with an ITS versus solving the same problems on paper. We analyze the learning process and pre-post gain data from N = 97 middle school students practicing linear graphs in three curricular units. We find that (i) working with the ITS, students had more than twice the number of eventually-correct practice opportunities than when working on paper and (ii) omission errors on paper were associated with lower learning gains. Yet, contrary to our hypothesis, tutor practice did not yield greater learning gains, with tutor and paper comparing differently across curricular units. These findings align with tutoring allowing students to grapple with challenging steps through tutor assistance but not with eventually-correct opportunities driving learning gains. Gaming-the-system, lack of transfer to an unfamiliar test format, potentially ineffective tutor design, and learning affordances of paper can help explain this gap. This study provides first-of-its-kind quantitative evidence that ITSs yield more learning opportunities than equivalent paper-and-pencil practice and reveals that the relation between opportunities and learning gains emerges only when the instruction is effective.
... In order to explain successful mathematics learning with digital tools, the specification of learning activities is crucial for any empirical investigation. We introduce a cognitive process framework for the mechanisms of (mathematics) learning with digital tools that combines core ideas from the psychology of instruction (utilizationof-learning-opportunities models; ULO; Seidel & Shavelson, 2007), cognitive psychology (Knowledge-Learning-Instruction framework, KLI; Koedinger et al., 2012), and mathematics education (content-specific theoretical & empirical analysis). ...
In order to explain successful mathematics learning with digital tools, the specification of learning activities is crucial for any empirical investigation. We introduce a cognitive process framework for the mechanisms of (mathematics) learning with digital tools that combines core ideas from the psychology of instruction (utilization-of-learning-opportunities models; ULO; Seidel & Shavelson, 2007), cognitive psychology (Knowledge-Learning-Instruction framework, KLI; Koedinger et al., 2012), and mathematics education (content-specific theoretical & empirical analysis).
Our framework highlights the mediating role of specific cognitive processes in the cause-and-effect mechanisms of successful learning with digital tools. That is, the digital tool as central predictor includes the design of the environment as well as the inclusion of instructional features. We argue that to find the appropriate to-be-implemented features, knowledge about content-specific learner models (i.e., learning processes and knowledge components) is of use to design instructional events that are well-suited to stimulate the relevant generative cognitive processes.
We demonstrate how this framework can be used to theoretically ground the evaluation of students’ use of such digital tools based on log-files of unobtrusively measured student-tool interactions: We describe how to investigate these processes during learning by linking students’ (external) on-task behavior and their (internal) cognitive processes (Hahnel et al., 2019), so that a causal interpretation on learning effects is possible.
... where the Δ specifies the knowledge domain and the δ refers to the KC or KU (if labeled, else question item). Note that KCs can be identified or refined by the "manual" approaches by the domain expert, some automated methods, or semi-automated methods; there are many examples of research on the identification of the KCs in different domains [37][38][39], but it is beyond the scope of this investigation to make these efforts. ...
The key to the effectiveness of Intelligent Tutoring Systems (ITSs) is to fit the uncertainty of each learner’s performance in performing different learning tasks. Throughout the tutoring and learning process, the uncertainty of learners’ performance can reflect their varying knowledge states, which can arise from individual differences in learning characteristics and capacities. In this investigation, we proposed a multidimensional representation of the evolution of knowledge states of learners to better understand individual differences among them. This assumption about this representation is verified using the Tensor Factorization (TF) based method, a modern state-of-the-art model for knowledge tracing. The accuracy of the Tensor-based method is evaluated by comparing it to other knowledge-tracing methods, to gain a deeper insight into individual differences among learners and their learning of diverse contents. The experimental data under focus in our investigation is derived from the AutoTutor lessons that were developed for the Center for the Study of Adult Literacy (CSAL), which employs a trialogue design comprising of a virtual tutor, a virtual companion and a human learner. A broader merit of our proposed approach lies in its capability to capture individual differences more accurately, without requiring any changes in the real-world implementation of ITSs.KeywordsIntelligent tutoring systemsKnowledge tracingKnowledge states of learnersIndividual differencesTensor-based methodTutoringLearning process
... Each caselet includes problem context, data profile, closeended multiple-choice questions, and feedback with explanations. All caselet questions are categorized into knowledge components linked to DS competencies (Koedinger, Corbett, and Perfetti 2012). Besides, researchers have recently shown increased interest in the metacognition process Copyright © 2023, Association for the Advancement of Artificial Intelligence (www.aaai.org). ...
Data Science (DS) is an interdisciplinary topic that is applicable to many domains. In this preliminary investigation, we use caselet, a mini-version of a case study, as a learning tool to allow students to practice data science problem solving (DSPS). Using a dataset collected from a real-world classroom, we performed correlation analysis to reveal the structure of cognition and metacognition processes. We also explored the similarity of different DS knowledge components based on students’ performance. In addition, we built a predictive model to characterize the relationship between metacognition, cognition, and learning gain.
... Bei den Wissensarten, die die Lehrperson bei der Formulierung der Lernziele aufgreift, wird in der PUT nur noch zwischen deklarativem und prozeduralem Wissen unterschieden. Die Unterscheidung ist weitverbreitet in der Lehr-und Lern-Forschung und der kognitiven Psychologie (siehe Anderson, Bothell, Byrne, Douglass, Lebiere & Qin, 2004;Koedinger, Corbett & Perfetti, 2012). Unter den beiden Kategorien lassen sich die vier Wissensarten der Bloom'schen Taxonomie subsumieren. ...
... However, applying auto-tagging for real-world education is challenging due to data scarcity. This is because auto-tagging has a potentially very large label space, ranging from subject topics to knowledge components (KC) (Zhang et al., 2015;Koedinger et al., 2012;Mohania et al., 2021;Viswanathan et al., 2022). The resulting data * Equal Contribution. ...
Text classification in education, usually called auto-tagging, is the automated process of assigning relevant tags to educational content, such as questions and textbooks. However, auto-tagging suffers from a data scarcity problem, which stems from two major challenges: 1) it possesses a large tag space and 2) it is multi-label. Though a retrieval approach is reportedly good at low-resource scenarios, there have been fewer efforts to directly address the data scarcity problem. To mitigate these issues, here we propose a novel retrieval approach CEAA that provides effective learning in educational text classification. Our main contributions are as follows: 1) we leverage transfer learning from question-answering datasets, and 2) we propose a simple but effective data augmentation method introducing cross-encoder style texts to a bi-encoder architecture for more efficient inference. An extensive set of experiments shows that our proposed method is effective in multi-label scenarios and low-resource tags compared to state-of-the-art models.
... Previous work on novice-AI music co-creation has also found that users systematically tested AI limitations to hone their mental model of the system's behavior [48]. Going further, to better support users in this learning activity, future co-creative systems may ofer designers a set of hands-on miniguided 'experiments' to better understand the system's responses to specifc (extreme) parameter inputs [42]. Systems may also ofer designers opportunities to view sets of examples of input-output pairs to help designers develop useful mental models of an AI tool's generative capabilities and limitations (cf. ...
AI-based design tools are proliferating in professional software to assist engineering and industrial designers in complex manufacturing and design tasks. These tools take on more agentic roles than traditional computer-aided design tools and are often portrayed as “co-creators.” Yet, working effectively with such systems requires different skills than working with complex CAD tools alone. To date, we know little about how engineering designers learn to work with AI-based design tools. In this study, we observed trained designers as they learned to work with two AI-based tools on a realistic design task. We find that designers face many challenges in learning to effectively co-create with current systems, including challenges in understanding and adjusting AI outputs and in communicating their design goals. Based on our findings, we highlight several design opportunities to better support designer-AI co-creation.
... They are also used in libraries, student affairs, school restaurants, and academic programs to give individualized learning, support students, facilitate administrative duties, and foster assessment (Hopcan et al., 2022). As AI technology continues to advance, it is expected to see even more sophisticated learning management systems emerge in the future (Koedinger et al, 2012, Ramesh & Lakshmi, 2018, Saha & Al Amri, 2019. ...
Artificial intelligence (AI) introduces new tools to the educational environment with the potential to transform conventional teaching and learning processes. This study offers a comprehensive overview of AI technologies, their potential applications in education, and the difficulties involved. Chatbots and related algorithms that can simulate human interactions and generate human-like text based on input from natural language are discussed. In addition to the advantages of cutting-edge chatbots like ChatGPT, their use in education raises important ethical and practical challenges. The authors aim to provide insightful information on how AI may be successfully incorporated into the educational setting to benefit teachers and students, while promoting responsible and ethical use.
... Moreover, predictive analytics helps student learning prediction and finds the failure rates performance, and insists that the course can obtain better outcomes in the future [17]. However, data mining techniques resolve the association between attributes and learning [18]. For instance, the cumulative grade point scores of the student are not reflected by ethnicity. ...
... Previous work on novice-AI music co-creation has also found that users systematically tested AI limitations to hone their mental model of the system's behavior [48]. Going further, to better support users in this learning activity, future co-creative systems may offer designers a set of hands-on miniguided 'experiments' to better understand the system's responses to specific (extreme) parameter inputs [42]. Systems may also offer designers opportunities to view sets of examples of input-output pairs to help designers develop useful mental models of an AI tool's generative capabilities and limitations (cf. ...
AI-based design tools are proliferating in professional software to assist engineering and industrial designers in complex manufacturing and design tasks. These tools take on more agentic roles than traditional computer-aided design tools and are often portrayed as "co-creators." Yet, working effectively with such systems requires different skills than working with complex CAD tools alone. To date, we know little about how engineering designers learn to work with AI-based design tools. In this study, we observed trained designers as they learned to work with two AI-based tools on a realistic design task. We find that designers face many challenges in learning to effectively co-create with current systems, including challenges in understanding and adjusting AI outputs and in communicating their design goals. Based on our findings, we highlight several design opportunities to better support designer-AI co-creation.
... Robust learning is the acquisition of new knowledge or skills, which can be applied to new contexts (transfer) or prepare students for future learning (PFL) (Bransford and Schwartz 1999;Koedinger et al. 2012;Schwartz et al. 2005;Richey and Nokes-Malach 2015). Transfer is defined as the ability to use and apply prior knowledge to solve new problems and PFL is defined as the ability use prior knowledge to learn new material (see Figure 3 for a comparison). ...
Metacognition is hypothesized to play a central role in problem solving and self-regulated learning. Various measures have been developed to assess metacognitive regulation, including survey items in questionnaires, verbal protocols, and metacognitive judgments. However, few studies have examined whether these measures assess the same metacognitive skills or are related to the same learning outcomes. To explore these questions, we investigated the relations between three metacognitive regulation measures given at various points during a learning activity and subsequent test. Verbal protocols were collected during the learning activity, questionnaire responses were collected after the learning tasks but before the test, and judgments of knowing (JOKs) were collected during the test. We found that the number of evaluation statements as measured via verbal protocols was positively associated with students’ responses on the control/debugging and evaluation components of the questionnaire. There were also two other positive trends. However, the number of monitoring statements was negatively associated with students’ responses on the monitoring component of the questionnaire and their JOKs on the later test. Each measure was also related to some aspect of performance, but the particular metacognitive skill, the direction of the effect, and the type of learning outcome differed across the measures. These results highlight the heterogeneity of outcomes across the measures, with each having different affordances and constraints for use in research and educational practice.
... In the first example, the learning objective is to improve workers' ability to appropriately rely on ADS outputs in specific cases. The sketch shows a simulated decision-making activity, which provides low-stakes opportunities for workers to practice integrating their own judgments with AI predictions on real historical data while receiving immediate feedback [1,11,14,25]. The second example sketch focuses on honing workers' ability to mentally simulate the model's behavior through repeated practice opportunities on a score guessing exercise, with immediate feedback on the closeness of their guesses. ...
In this short paper, we argue for a refocusing of XAI around human learning goals. Drawing upon approaches and theories from the learning sciences, we propose a framework for the learner-centered design and evaluation of XAI systems. We illustrate our framework through an ongoing case study in the context of AI-augmented social work.
... The studies by Samani and Pan (2021) and by Sana and Yan (2022) followed this principle, using quizzes or homework assignments that referred to previously learned content and combining interleaving with retrieval practice, which is a different desirable difficulty that benefits learning independently of interleaving (see Roelle et al., 2022). Hence, teaching scientific concepts requires the combination of various learning phases (Koedinger et al., 2012(Koedinger et al., , 2013Oser & Baeriswyl, 2001) to master complexity and secure the acquisition of adequate principle-based cognitive skills to achieve lasting learning. ...
Inductive learning, that is, abstracting conceptual knowledge, rules, or principles from exemplars, plays a major role in educational settings, from literacy acquisition to mathematics and science learning. Interleaving exemplars of different categories rather than presenting blocks might be a simple but powerful way to improve inductive learning by supporting discriminative contrast. Although a consistent advantage of interleaving has been demonstrated for visual materials, relatively few studies have examined educationally relevant materials, such as mathematical tasks, science problems, and verbal materials, and their results are mixed. We discuss how interleaving could be made fruitful for school learning of mathematics, science, and literacy acquisition. We conclude that interleaving should be tailored to the specific learning content and combined with supportive instructional measures that assist students in comparing exemplars for discriminating features. Finally, we sketch research gaps that revolve around the use of interleaved learning in the classroom.
Artificial intelligence (AI) can enhance teachers' capabilities by sharing control over different parts of learning activities. This is especially true for complex learning activities, such as dynamic learning transitions where students move between individual and collaborative learning in un‐planned ways, as the need arises. Yet, few initiatives have emerged considering how shared responsibility between teachers and AI can support learning and how teachers' voices might be included to inform design decisions. The goal of our article is twofold. First, we describe a secondary analysis of our co‐design process comprising six design methods to understand how teachers conceptualise sharing control with an AI co‐orchestration tool, called Pair‐Up . We worked with 76 middle school math teachers, each taking part in one to three methods, to create a co‐orchestration tool that supports dynamic combinations of individual and collaborative learning using two AI‐based tutoring systems. We leveraged qualitative content analysis to examine teachers' views about sharing control with Pair‐Up , and we describe high‐level insights about the human‐AI interaction, including control, trust, responsibility, efficiency, and accuracy. Secondly, we use our results as an example showcasing how human‐centred learning analytics can be applied to the design of human‐AI technologies and share reflections for human‐AI technology designers regarding the methods that might be fruitful to elicit teacher feedback and ideas. Our findings illustrate the design of a novel co‐orchestration tool to facilitate the transitions between individual and collaborative learning and highlight considerations and reflections for designers of similar systems.
Practitioner notes
What is already known about this topic: Artificial Intelligence (AI) can help teachers facilitate complex classroom activities, such as having students move between individual and collaborative learning in unplanned ways.
Designers should use human‐centred design approaches to give teachers a voice in deciding what AI might do in the classroom and if or how they want to share control with it.
What this paper adds: Presents teacher views about how they want to share control with AI to support students moving between individual and collaborative learning.
Describes how we adapted six design methods to design AI features.
Illustrates a complete, iterative process to create human‐AI interactions to support teachers as they facilitate students moving from individual to collaborative learning.
Implications for practice: We share five implications for designers that teachers highlighted as necessary when designing AI‐features, including control, trust, responsibility, efficiency and accuracy.
Our work also includes a reflection on our design process and implications for future design processes.
The rapid technological evolution of the last years has motivated students to develop capabilities that will prepare them for an unknown future in the 21st century. In this context, many teachers intend to optimise the learning process, making it more dynamic and exciting through the introduction of gamification. Thus, this paper focuses on a data-driven assessment of geometry competencies, which are essential for developing problem-solving and higher-order thinking skills. Our main goal is to adapt, evaluate and compare Bayesian Knowledge Tracing (BKT), Performance Factor Analysis (PFA), Elo and Deep Knowledge Tracing (DKT) algorithms applied to the data of a geometry game named Shadowspect, in order to predict students’ performance by means of several classifier metrics. We analysed two algorithmic configurations, with and without prioritisation of Knowledge Components (KCs) – the skills needed to complete a puzzle successfully, and we found Elo to be the algorithm with the best prediction power with the ability to model the real knowledge of students. However, the best results are achieved without KCs because it is a challenging task to differentiate between KCs effectively in game environments. Our results prove that the above-mentioned algorithms can be applied in formal education to improve teaching, learning, and organisational efficiency.
The unprecedented global movement of school education to find technological and intelligent solutions to keep the learning ecosystem working was not enough to recover the impacts of Covid-19, not only due to learning-related challenges but also due to the rise of negative emotions, such as frustration, anxiety, boredom, risk of burnout and the so-called “Covid fatigue”. Although this is not a new problem, it was deepened during the pandemic, and we need to face old and new challenges in different ways. Despite focusing only on the learning system’s inefficiencies and on the hegemony of solutions to tackle the learning gap, we also need to shed light on the strengths and the positive aspects of the learning process to promote wellbeing. As highlighted by John Self, an intelligent tutoring system would behave as if it genuinely cared about the student’s success. This note from John Self sheds light on the importance and reflection of what success means and for whom. This study presents a roadmap for positive psychology and artificial intelligence in education. It is intended to identify and understand how the intersection of Positive Psychology and Artificial Intelligence in Education can support the promotion of learning and wellbeing of students, teachers, and other educational stakeholders. As such, a bibliometric analysis of positive psychology and artificial intelligence in education was made as the so-called Positive Artificial Intelligence in Education (P-AIED). The search string was performed in 2021, and the total number of studies gathered was 10,777. After all the PRISMA steps, 256 studies were approved according to the inclusion criteria. The main conclusions were the high number of institutions and researchers with related publications indicate a new trend for the community of AIED; the high number of collaboration from different countries indicates a possible global movement toward P-AIED; Positive Emotion and Engagement were the main Positive Psychology constructs identified in the studies; the lack of well-grounded theories of Positive Psychology indicates an excellent research opportunity; Positive Learning Analytics (P-LA), Positive Educational Data Mining (P-EDM) and Positive Intelligent Tutoring Systems (P-ITS) are three hot topics for P-AIED.
We propose incorporating biophysical data with behavioral data to inform digital learning environments on an individual’s current cognitive state and how it relates to their learning. We used a rule learning paradigm drawn from cognitive psychology to define phases of rule learning across multiple domains. This paradigm can simulate an inductive reasoning framework seen during mathematics education while reducing the number of covariates compared to real-world settings. We combined the time series brain data with behavioral and contextual data in machine learning models for prediction of rule learning phases with the aim of developing approaches to incorporate a mixture of behavioral and neural data into digital learning designs.Keywordsrule learninginductive reasoningfunctional near-infrared spectroscopybrain-computer interfaces
We present a randomized field trial delivered in Carnegie Learning’s MATHia’s intelligent tutoring system to 12,374 learners intended to test whether rewriting content in “word problems” improves student mathematics performance within this content, especially among students who are emerging as English language readers. In addition to describing facets of word problems targeted for rewriting and the design of the experiment, we present an artificial intelligence-driven approach to evaluating the effectiveness of the rewrite intervention for emerging readers. Data about students’ reading ability is generally neither collected nor available to MATHia’s developers. Instead, we rely on a recently developed neural network predictive model that infers whether students will likely be in this target sub-population. We present the results of the intervention on a variety of performance metrics in MATHia and compare performance of the intervention group to the entire user base of MATHia, as well as by comparing likely emerging readers to those who are not inferred to be emerging readers. We conclude with areas for future work using more comprehensive models of learners.Keywordsmachine learningA/B testingintelligent tutoring systemsreading abilitymiddle school mathematics
Students use learning analytics systems to make day-to-day learning decisions, but may not understand their potential flaws. This work delves into student understanding of an example learning analytics algorithm, Bayesian Knowledge Tracing (BKT), using Cognitive Task Analysis (CTA) to identify knowledge components (KCs) comprising expert student understanding. We built an interactive explanation to target these KCs and performed a controlled experiment examining how varying the transparency of limitations of BKT impacts understanding and trust. Our results show that, counterintuitively, providing some information on the algorithm’s limitations is not always better than providing no information. The success of the methods from our BKT study suggests avenues for the use of CTA in systematically building evidence-based explanations to increase end user understanding of other complex AI algorithms in learning analytics as well as other domains.KeywordsExplainable artificial intelligence (XAI)Learning analyticsBayesian Knowledge Tracing
In this paper we argue that artificial intelligence models of learning can contribute precise theory to explain surprising student learning phenomena. In some past studies of student learning, practice produces better learning than studying examples, whereas other studies show the opposite result. We reconcile and explain this apparent contradiction by suggesting that retrieval practice and example study involve different learning cognitive processes, memorization and induction, respectively, and that each process is optimal for learning different types of knowledge. We implement and test this theoretical explanation by extending an AI model of human cognition — the Apprentice Learner Architecture (AL) — to include both memory and induction processes and comparing the behavior of the simulated learners with and without a forgetting mechanism to the behavior of human participants in a laboratory study. We show that, compared to simulated learners without forgetting, the behavior of simulated learners with forgetting matches that of human participants better. Simulated learners with forgetting learn best using retrieval practice in situations that emphasize memorization (such as learning facts or simple associations), whereas studying examples improves learning in situations where there are multiple pieces of information available and induction and generalization are necessary (such as when learning skills or procedures).KeywordsSimulated LearnersRetrieval PracticeWorked Examples
=== Feel free to contact me (Yun) through full-text request in ResearchGate! ===
A central component of many AIED systems is a “domain model,” that is, a representation of knowledge of the domain of instruction. The system uses the model in many ways to provide instruction that adapts to learners. Not all AIED systems have an elaborate domain model, but in those that do, the domain model is central to the system’s functioning. In fact, domain models fulfill so many important functions within AIED systems that entire classes of AIED systems are defined in terms of the types of domain model they use (such as model-tracing tutors, constraint-based tutors, example-tracing tutors, and issue-based approaches to build- ing tutoring systems). Across AIED projects, systems, and paradigms, the types of domain models used span the gamut of AI representations. AIED systems use their domain models for many different purposes, chief among them assessing student work, which is foundational for other functionality.
This chapter reviews major approaches to domain modeling used in AIED systems and briefly touches on the corresponding student models and the way they are used to track an individual student’s knowledge growth. (We do not discuss student models that target other aspects, such as affect, motivation, self-regulation, or metacognition.) We discuss, in turn: rule-based models, constraint-based models, Bayesian networks, machine-learned models, text-based models, generalized examples, and knowledge spaces. These types of models have been studied extensively in AIED research and have been the foundation for many AIED systems that have been proven to be effective in enhancing student learning or other aspects of the student experience. A number of these approaches are now used in AIED systems that are used on a wide scale in educational practice. The chapter discusses how these approaches support key aspects of an AIED system’s behavior and enable the system to adapt aspects of its instruction to individual student variables. We also highlight challenges that occur when applying the different approaches. We look at the use of machine learning and data-driven methods to create or refine domain models, so they better account for learning data and sup- port more effective adaptive instruction. As well, we make note of connections between a system’s domain model and other key components, including the system’s student model. We base this discussion on traditional views of intelligent tutoring systems (ITSs), which divide the system’s architecture into four main components: a domain model, a student model, a pedagogical model, and a problem-solving environment. We focus on systems that support individual learning. Other types of AIED systems are covered in other chapters.
This comparative study aimed to investigate the effects of an Augmented-Reality Game-Based Learning application, “StemUp,” on the English learning performance and motivation of distinct rural and urban groups of young EFL (English as a Foreign Language) learners. Multiple sources of data collection were employed, including pre- and post-tests, Keller’s (1987) ARCS pre- and post-survey, and semi-structured interviews of the fifty participants. Quantitative findings confirmed that the study’s AR game-based intervention resulted in a significant improvement in English performance and motivation for both groups, with the students from the rural area improving more in both English performance and motivation compared to those from the urban area. Motivation and enjoyment, gamification, and learning effectiveness were three important qualitative interview themes. Overall, this mixed-methods study found that the learning design employing “StemUp,” with its unique 3D AR features/affordances, not only improved participant language gains in vocabulary, speaking, and listening, but also enabled participants to be more engaged and motivated in their English learning. The conclusion is that StemUP is a promising tool for assisting EFL learners in both rural and urban areas in their English language development as well as motivation improvement by facilitating an immersive language learning environment for EFL young learners.
The development of deep learning technologies brings opportunities to computer science education (CSEd). Recent CSEd datasets provide actual code submissions that could potentially offer more information on students’ learning for learning analytics, but adding the information needs special considerations. My research focuses on leveraging these code submissions to inform learning analytics. Specifically, it addresses two problems: 1) finding and detecting a meaningful representation of knowledge components (or skills) in students’ programming and 2) using them to provide formative feedback. Previous works have defined or discovered skills in CSEd, but these skills do not follow certain properties of learning. My research proposes to incorporate these properties into data-driven model design, and improves the knowledge components so that they are consistent with theoretical properties, and meanwhile also provide better interpretability than typical deep learning models. Following this, my second project will focus on providing personalized support in student learning.
Students regularly ask, “How can I do well in your course?” They are surprised when I provide a simple answer: Take advantage of the quizzes. Quizzes are not a silver bullet, but they improve students’ recollection of course information and, importantly to students, increase performance on exams. Pre-lecture reading quizzes encourage students to arrive prepared (pre-training), ongoing quizzes promote
regular studying (spacing), and review quizzes help students revisit material from previous topics (interleaving). Central to the present discussion, all of these types of quizzes require students to retrieve information to answer items, which improves performance on later exams (testing effect, retrieval practice). Still, questions remain about how to use quizzes most effectively. In particular, should we use
harder application quizzes or easier factual quizzes to help students do well in the course? That is to say, should we throw students in the deep end early in the learning process or not?
Objectives:
This review aimed to integrate previous research to gain a deeper understanding of which individual factors are associated with reduced accident involvement, and which factors may be linked to success during emergency situations when they do occur. Better understanding how the human will react in these situations, combined with technological enhancements is vital to risk mitigation and ensuring successful performance. This review will also identify gaps in the literature that have yet to be addressed.
Method:
A systematic literature review was undertaken, beginning with 18,319 articles from three multidisciplinary databases. After careful review and exclusion, a final sample of 22 relevant articles were retained. This analysis was spread across various high risk, sociotechnical industries, including aviation, rail, mining, nuclear power, etc.
Findings:
The findings show that previous research has identified cognitive ability, leadership, situation awareness, personality, and risk perception as the most prominently considered factors in reducing accident involvement. Training, skill, situation awareness, and emotional stability were the most commonly associated factors to success through emergencies.
Conclusion:
While the research around individual differences impacting success during emergencies is scarce, this review provides future direction on potential factors influencing reduced accident involvement and/or potential factors that could influence a person's success through a disaster or emergency. This information could be implemented in recruitment and training of front-line workers in high-reliability organisations to reduce risk, increase safety and work towards reducing the number of accidents.
Digital education tools have revolutionized how learning solutions can be personalized for the benefit of all learners. When advances in technology are paired with knowledge of how individuals learn most effectively, more efficacious learning solutions can be developed. In this chapter, the authors contend that not only should learning science principles drive the technologies learners engage with, but combining these principles with the algorithms that personalize a learner's experience will greatly impact learning outcomes at scale. They offer a brief discussion of automation and self-regulation in educational technology, then provide an example of a novel learning solution that pairs personalization—through automated intelligence—with learning science principles to impact outcomes. A practical discussion of how to design and develop learning solutions for maximum impact is shared, as well as best practices for conducting validity and efficacy research to measure the extent to which intelligent self-regulation features are supporting learning outcomes.
Cognitive Load Theory’s (CLT) purpose is to aid in the design of messages, instructional and otherwise, so that learning and message retention are more effective. CLT was introduced in 1998 by John Sweller and his colleagues. They used the constructs of three areas of memory, sensory, working, and long-term memory, to develop a theory to address the limited capacity of working memory. Through these efforts, they created the concepts of intrinsic, extraneous , and germane cognitive load and used these concepts to explain how various loads are placed on working memory. The purpose of this chapter is to describe the three areas of memory, the three concepts of cognitive load, and address various effects created by intrinsic and extraneous cognitive load while guiding instructional designers on best practices to minimize load and maximize performance.
Leveraging a scientific infrastructure for exploring how students learn, we have developed cognitive and statistical models of skill acquisition and used them to understand fundamental similarities and differences across learners. Our primary question was why do some students learn faster than others? Or do they? We model data from student performance on groups of tasks that assess the same skill component and that provide follow-up instruction on student errors. Our models estimate, for both students and skills, initial correctness and learning rate, that is, the increase in correctness after each practice opportunity. We applied our models to 1.3 million observations across 27 datasets of student interactions with online practice systems in the context of elementary to college courses in math, science, and language. Despite the availability of up-front verbal instruction, like lectures and readings, students demonstrate modest initial pre-practice performance, at about 65% accuracy. Despite being in the same course, students’ initial performance varies substantially from about 55% correct for those in the lower half to 75% for those in the upper half. In contrast, and much to our surprise, we found students to be astonishingly similar in estimated learning rate, typically increasing by about 0.1 log odds or 2.5% in accuracy per opportunity. These findings pose a challenge for theories of learning to explain the odd combination of large variation in student initial performance and striking regularity in student learning rate.
Data-driven decision-making and data-intensive research are becoming prevalent in many sectors of modern society, i.e. healthcare, politics, business, and entertainment. During the COVID-19 pandemic, huge amounts of educational data and new types of evidence were generated through various online platforms, digital tools, and communication applications. Meanwhile, it is acknowledged that education lacks computational infrastructure and human capacity to fully exploit the potential of big data. This paper explores the use of Learning Analytics (LA) in higher education for measurement purposes. Four main LA functions in the assessment are outlined: (a) monitoring and analysis, (b) automated feedback, (c) prediction, prevention, and intervention, and (d) new forms of assessment. The paper concludes by discussing the challenges of adopting and upscaling LA as well as the implications for instructors in higher education.
Discipline‐based educational researchers and institutions that provide guidance on Science, Technology, Engineering, and Mathematics instruction have long documented the impact of active learning strategies on student learning and achievement, the lack of incorporation of active learning into undergraduate instruction, and the languishing rates of progression through and completion of undergraduate degree programs. Most research on improving active learning has been found to focus on instructors and the ways they have promoted active learning in their course design. Our aim was to consider the ways that textbook publishers and the online learning platforms they have produced have provided resources to promote active learning. This content analysis has documented the degree to which the digital platform companions to four highly subscribed introductory biology textbooks have provided resources that promote (1) highlighting of informational texts, (2) self‐testing, and (3) generative strategies for learning, including self‐explanation, summarization, and elaboration. Digital platforms largely supported highlighting and self‐testing, but the features of these tools could have been improved to promote more productive engagement in the learning strategies and to provide more explicit instruction about and scaffolding of the practices. Resources that promoted additional generative strategies were more limited. Implications relevant to developers and instructors who wish to leverage digital resources to promote active learning are included and are grounded via empirical support of design and implementation choices.
Reasoning about structure-reactivity and chemical processes is a key competence in chemistry. Especially in organic chemistry, students experience difficulty appropriately interpreting organic representations and reasoning about the underlying causality of organic mechanisms. As organic chemistry is often a bottleneck for students’ success in their career, compiling and distilling the insights from recent research in the field will help inform future instruction and the empowerment of chemistry students worldwide. This book brings together leading research groups to highlight recent advances in chemistry education research with a focus on the characterization of students’ reasoning and their representational competencies, as well as the impact of instructional and assessment practices in organic chemistry. Written by leaders in the field, this title is ideal for chemistry education researchers, instructors and practitioners, and graduate students in chemistry education.
The ability to categorize sketches, ideas, and concepts is an essential, basic skill for designers. Novice- and intermediate-level design students may benefit from making categorization a more explicit step in their project analyses to help them integrate an already developed general skill into the decision making methods that guides their design processes. This step supports the development of more complex and informed approaches to idea generation and the subsequent iterative development of design concepts. While expert-level designers may not often refer to categorization explicitly once they have internalized this ability, articulating basic skills that later become automatic can be an effective step in teaching more complex skills and processes to emerging designers. Additionally, doing this can ensure that design processes become understandable and accessible to wider audiences, including those less inclined to engage in any form of design practice or thinking.
The authors propose that conceptual and procedural knowledge develop in an iterative fashion and that improved problem representation is 1 mechanism underlying the relations between them. Two experiments were conducted with 5th- and 6th-grade students learning about decimal fractions. In Experiment 1, children's initial conceptual knowledge predicted gains in procedural knowledge, and gains in procedural knowledge predicted improvements in conceptual knowledge. Correct problem representations mediated the relation between initial conceptual knowledge and improved procedural knowledge. In Experiment 2, amount of support for correct problem representation was experimentally manipulated, and the manipulations led to gains in procedural knowledge. Thus, conceptual and procedural knowledge develop iteratively, and improved problem representation is 1 mechanism in this process. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
In this Monograph, we present a new theory of children's conceptual development and the empirical research on which that theory is based. The main construct in the theory is the notion of a central conceptual structure. These structures are denned as networks of semantic nodes and relations that represent children's core knowledge in a domain and that can be applied to the full range of tasks that the domain entails. Major transformations are hypothesized to take place in these structures as children enter each new stage of their development. Once formed, the new structures are hypothesized to exert a powerful influence on all subsequent knowledge acquisition. The process by which they exert this effect is believed to be a dynamic one, in which general conceptual insights and more specific task understandings become reciprocally coupled, each exerting a bootstrapping effect on the other. In the first chapter, the general theoretical framework that underlies this conception is spelled out in broad strokes and compared to other contemporary views of conceptual development. In subsequent chapters, more detailed models of children's central conceptual structures are presented for three different domains: number, space, and social interaction. These models are then tested using a mixture of new and previously designed cognitive tasks, which are administered to children from four different age groups (4, 6, 8, and 10 years), three different social classes (high, medium, and low), and four different countries (the United States, Canada, Japan, and China). The results of a 6-year program of instructional research are also summarized and used to clarify and refine the theory. Analytic tools that are employed include computer simulation, item analysis, and linked growth curve analysis. Statistical techniques include latent structure analysis, factor analysis, and Guttman scaling. Problems with previous versions of the theory and implications of the present version are discussed in the final chapter.
The current study examines how holding misconceptions about key problem features affects students' ability to solve algebraic equations correctly and to learn correct procedures for problem solution. Algebra I students learning to solve simple equations using the Cognitive Tutor curriculum (Koedinger, Anderson, Hadley, & Mark, 1997) completed a pretest and posttest designed to evaluate their conceptual understanding of problem features (including the equals sign and negative signs) as well as their equation solving skill. Results indicate that students who begin the lesson with misconceptions about the meaning of the equals sign or negative signs solve fewer equations correctly at pretest, and also have difficulty learning how to solve them. However, improving their knowledge of those features over the course of the lesson increases their learning of correct procedures.
A study was carried out to determine whether knowledge of the internal radical structure of a Chinese character helps a naïve learner to memorize that character. Four groups of Australian subjects who knew nothing about Chinese were asked to learn 24 character/meaning pairs (e.g., Pa-CHEW). Each character was composed of two radicals taken from a set of 16. Every subject was presented with the set of character-meaning pairs three times and then were given each character alone and asked to recall the meaning associated with it. Before seeing any characters, one group (Radicals Before) told about radicals and had 15 minutes to learn the set of 16 radicals thoroughly. Another (Radicals Early) was told about radicals at the first presentation of the stimuli, but were simply asked, as each character was presented, to point out on a chart its component radicals. A third group (Radicals Late) did the same thing, but at the third presentation of the stimuli; while a final group (No Radicals) were told nothing about radicals at all. It was found that memory for the character-meaning pairings was best for the Radicals Early group, suggesting that it is important to highlight the radicals when a character is first presented to the learner.
It is well-documented that most students do not have adequate proficiencies in inquiry and metacognition, particularly at deeper levels of comprehension that require explanatory reason- ing. The proficiencies are not routinely provided by teachers and normal tutors so it is worth- while to turn to computer-based learning environments. This article describes some of our re- cent computer systems that were designed to facilitate explanation-centered learning through strategies of inquiry and metacognition while students learn science and technology content. Point&Query augments hypertext, hypermedia, and other learning environments with ques- tion-answer facilities that are under the learner control. AutoTutor and iSTART use animated conversational agents to scaffold strategies of inquiry, metacognition, and explanation con- struction. AutoTutor coaches students in generating answers to questions that require explana- tions (e.g., why, what-if, how) by holding a mixed-initiative dialogue in natural language. iSTART models and coaches students in constructing self-explanations and in applying other metacomprehension strategies while reading text. These systems have shown promising results in tests of learning gains and learning strategies.
This chapter presents an overview of the current state of Cognitive Task Analysis (CTA) in research and practice. CTA uses a variety of interview and observation strategies to capture a description of the explicit and implicit knowledge that experts use to perform complex tasks. The captured knowledge is most often transferred to training or the development of expert systems. The first section presents descriptions of a variety of CTA techniques, their common characteristics, and the typical strategies used to elicit knowledge from experts and other sources. The second section describes research on the impact of CTA and synthesizes a number of studies and reviews pertinent to issues underlying knowledge elicitation. In the third section we discuss the integration of CTA with training design. In the fourth section, we present a number of recommendations for future research and conclude with general comments.
Cognitive modeling can be used to compare alternative instructional strategies and to guide the design of curriculum materials. We modeled two alternative strategies for fraction division, and the models led to specific empirical predictions of the benefits and drawbacks of each strategy. These insights provided concrete suggestions for developing lessons on fraction division, including a new potential strategy that combines the benefits of the two strategies. This on-going work illustrates the potential of cognitive modeling for informing the design of better mathematics curricula. Background
Two experiments were carried out to examine how adult readers of English learn to acquire the orthographic structure and function of Chinese characters selected from reading material in their first-semester college course in Chinese. The first experi- ment, an online lexical decision task, demonstrated that the learners qtiickly acquired knowledge about the orthographic structure of the characters. With only limited vo- cabulary and without explicit instruction, they were sensitive to the characters' cur- riculum frequency and internal compositional features. Overall, these results were consistent with our previous findings (Wang, Perfetti. & Liu. 2003), The second ex- periment further tested the learners' implicit and explicit learning of the orthographic component function of Chinese characters. In an offline unknown character identifi- cation task, these learners showed difficulty in making use of the functional cue of the known semantic radical without any probing. However, with probing they identi- fied visually the majority of target semantic radicals. In addition, they demonstrated implicit knowledge of the meaning cue of the semantic radical by making more ap- propriate meaning inferences on high-frequenc y radicals than on low-frequency ones. After receiving explicit instruction, the students extracted significantly more meaning information from known semantic radicals, particularly low-frequency ones, compared to their performance prior to such instruction.
This article explores how differences in problem representations change both the per- formance and underlying cognitive processes of beginning algebra students engaged in quantitative reasoning. Contrary to beliefs held by practitioners and researchers in mathematics education, students were more successful solving simple algebra story problems than solving mathematically equivalent equations. Contrary to some views of situated cognition, this result is not simply a consequence of situated world knowl- edge facilitating problem-solving performance, but rather a consequence of student difficulties with comprehending the formal symbolic representation of quantitative relations. We draw on analyses of students' strategies and errors as the basis for a cognitive process explanation of when, why, and how differences in problem repre- sentation affect problem solving. We conclude that differences in external represen- tations can affect performance and learning when one representation is easier to com- prehend than another or when one representation elicits more reliable and meaningful solution strategies than another.
Chunking models offer a parsimonious explanation of how people acquire knowledge and have been validated in domains such as expert behaviour and the acquisition of language. In this paper, we review two computational theories based on chunking mechanisms (the chunking theory and the template theory) and show what insight they offer for instruction and training. The suggested implications include the importance of perception in learning, the cost of acquiring knowledge, the significance of segmenting and ordering instruction material, the role of the variability of the instructional material in acquiring schemata, and the importance of taking individual differences into account.
Participants in 2 experiments interacted with computer simulations designed to foster understanding of scientific principles governing complex adaptive systems. The quality of participants' transportable understanding was measured by the amount of transfer between 2 simulations governed by the same principle. The perceptual con- creteness of the elements within the first simulation was manipulated. The elements either remained concrete throughout the simulation, remained idealized, or switched midway into the simulation from concrete to idealized or vice versa. Transfer was better when the appearance of the elements switched, consistent with theories pre- dicting more general schemas when the schemas are multiply instantiated. The best transfer was observed when originally concrete elements became idealized. These results are interpreted in terms of tradeoffs between grounded, concrete construals of simulations and more abstract, transportable construals. Progressive idealization ("concreteness fading") allows originally grounded and interpretable principles to become less tied to specific contexts and hence more transferable. Cognitive psychologists and educators have often debated the merits of concrete versus idealized materials for fostering scientific understanding. Should chemical molecules be represented by detailed, shaded, and realistically illuminated balls or by simple ball-and-stick figures? Should a medical illustration of a pancreas in- clude a meticulous rendering of the islets of Langerhans or convey in a more styl- ized manner the organ's general form? Our informal interviews with mycologists at the Royal Kew Gardens (personal communication, Brian Spooner and David Pegler, May 1998) indicate a schism between authors of mushroom field guides.
Theoretical linguists have traditionally relied on linguistic intuitions such as gram- maticality judgments for their data. But the massive growth of computer-readable texts and recordings, the availability of cheaper, more powerful computers and soft- ware, and the development of new probabilistic models for language have now made the spontaneous use of language in natural settings a rich and easily accessible alter- native source of data.
Surprisingly, many linguists believe that such ‘usage data’ are irrelevant to the theory of grammar. Four problems are repeatedly brought up in the critiques of usage data—
1. correlated factors seeming to support reductive theories,
2. pooled data invalidating grammatical inference,
3. syntactic choices reducing to lexical biases, and
4. cross-corpus differences undermining corpus studies.
Presenting a case study of work on the English dative alternation, we show first, that linguistic intuitions of grammaticality are deeply flawed and seriously underes- timate the space of grammatical possibility, and second, that the four problems in the critique of usage data are empirical issues that can be resolved by using modern statistical theory and modelling strategies widely used in other fields.
The new models allow linguistic theory to solve more difficult problems than it has in the past, and to build convergent projects with psychology, computer science, and allied fields of cognitive science.
Several investigators have taken protocols of students learning sophisticated skills, such as physics problem solving and LISP coding, by studying examples and solving problems. These investigations uncovered the self-explanation effect: Students who explain examples to themselves learn better, make more accurate self-assessments of their understanding, and use analogies more economically while solving problems. We describe a computer model, Cascade, that accounts for these findings. Explaining an example causes Cascade to acquire both domain knowledge and derivational knowledge. Derivational knowledge is used analogically to control search during problem solving. Domain knowledge is acquired when the current domain knowledge is incomplete and causes an impasse. If the impasse can be resolved by applying an overly general rule, then a specialization of the rule becomes a new domain rule. Computational experiments indicate that Cascade's learning mechanisms are jointly sufficient to reproduce the self-explanation effect, but neither alone can reproduce it.
We argue herein that typical training procedures are far from optimal. The goat of training in real-world settings is, or should be, to support two aspects of posttraining performance: (a) the level of performance in the long term and (b) the capability to transfer that training to related tasks and altered contexts. The implicit or explicit assumption of those persons responsible for training is that the procedures that enhance performance and speed improvement during training will necessarily achieve these two goals. However, a variety of experiments on motor and verbal learning indicate that this assumption is often incorrect. Manipulations that maximize performance during training can be detrimental in the long term; conversely, manipulations that degrade the speed of acquisition can support the long-term goals of training. The fact that there are parallel findings in the motor and verbal domains suggests that principles of considerable generality can be deduced to upgrade training procedures.
Despite a century's worth of research, arguments surrounding the question of whether far transfer occurs have made little progress toward resolution. The authors argue the reason for this confusion is a failure to specify various dimensions along which transfer can occur, resulting in comparisons of "apples and oranges." They provide a framework that describes 9 relevant dimensions and show that the literature can productively be classified along these dimensions, with each study situated at the intersection of Various dimensions. Estimation of a single effect size for far transfer is misguided in view of this complexity. The past 100 years of research shows that evidence for transfer under some conditions is substantial, but critical conditions for many key questions are untested.
Research has shown that it is effective to combine example study and problem solving in the initial acquisition of cognitive skills. Present methods for combining these learning modes are static, however, and do not support a transition from example study in early stages of skill acquisition to later problem solving. Against this background, the authors proposed a successive integration of problem-solving elements into example study until the learners solved problems on their own (i.e., complete example --> increasingly more incomplete examples --> problem to-be-solved). The authors tested the effectiveness of such a fading procedure against the traditional method of using example-problem pairs. In a field experiment and in 2 more controlled laboratory experiments, the authors found that (a) the fading procedure fosters learning, at least when near transfer performance is considered; (b) the number of problem-solving errors during learning plays a role in mediating this effect; and (c) it is more favorable to fade out worked-out solution steps in a backward manner (omitting the last solution steps first) as compared with a forward manner (omitting the first solution steps first).
The situative perspective shifts the focus of analysis from individual behavior and cognition to larger systems that include behaving cognitive agents interacting with each other and with other subsystems in the environment. The first section presents a version of the situative perspective that draws on studies of social interaction, philosophical situation theory, and ecological psychology. Framing assumptions and concepts are proposed for a synthesis of the situative and cognitive theoretical perspectives, and a further situative synthesis is suggested that would draw on dynamic-systems theory. The second section discusses relations between the situative, cognitive, and behaviorist theoretical perspectives and principles of educational practice. The third section discusses an approach to research and social practice called interactive research and design, which fits with the situative perspective and provides a productive, albeit syncretic, combination of theory-oriented and instrumental functions of research. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
In 2 experiments, understanding of historical subject matter was enhanced when students acted as historians and constructed their own models of an historical event. Providing students with information in a web site with multiple sources instead of a textbook chapter, and instructing them to write arguments instead of narratives, summaries, or explanations, produced the most integrated and causal essays with the most transformation from the original sources. Better performance on inference and analogy tasks provided converging evidence that students who wrote arguments from the web sources gained a better understanding than other students. A second experiment replicated the advantage of argument writing even when information was presented as an argument.
In 2 experiments, mechanically naive college students viewed an animation depicting the operation of a bicycle tire pump that included a verbal description given before (words-before-pictures) or during (words-with-pictures) the animation. The words-with-pictures group outperformed the words-before-pictures group on tests of creative problem solving that involved reasoning about how the pump works. In a follow-up experiment, students in the words-with-pictures group performed better on the problem-solving test than students who saw the animation without words (pictures only), heard the words without the animation (words only), or received no training (control). Results support a dual-coding hypothesis (Paivio, 1990) that posits two kinds of connections: representational connections between verbal stimuli and verbal representations and between visual stimuli and visual representations and referential connections between visual and verbal representations.
In the 2 experiments reported here, high school students studied worked examples while learning how to translate English expressions into algebraic equations. In Experiment 1, worked examples were used as part of the regular classroom instruction and as a support for homework. In Experiment 2, students in a remedial mathematics class received individual instruction. Students using worked examples outperformed the control group on posttests after completing fewer practice problems; they also made fewer errors per problem and fewer types of errors during acquisition time, completed the work more rapidly, and required less assistance from the teacher.
Collins, A., Brown, J.S., & Newman, S.E. (1989). Cognitive apprenticeship: Teaching the crafts of reading, writing, and mathematics. In L. B. Resnick (Ed.) Knowing, learning, and instruction: E...
Faded" examples are example problems that provide a solution, but first require students to generate a portion of the solution themselves. Empirical studies have shown that such examples can be more effective teaching aids than completely worked examples that require no work from the student. Cascade is a model of problem-solving skill acquisition that was originally developed to explain other empirical regularities associated with human problem solving and learning, most notably the self-explanation effect. Past research demonstrated that Cascade might also explain the mechanisms underlying the effectiveness of example fading. This paper analyzes new protocol data, and finds that it is consistent with predictions derived from Cascade. Overview Renkl, Atkinson, and Maier (2000) empirically demonstrated the qualitative result that, when learning problem-solving skills, students studying a series of "faded" examples show improved post-test performance over students studying only completely worked examples. Jones and Fleischman (2001) argue that this result can be explained by Cascade (VanLehn, Jones, & Chi, 1991), a computational model of problem-solving skill acquisition. Cascade was originally developed to understand the mechanisms of the self-explanation effect (Chi, Bassok, Lewis, Reimann, & Glaser, 1989; Pirolli & Anderson, 1985). Jones and Fleischman demonstrated that the mechanisms underlying self-explanation might also explain the effectiveness of studying faded examples. Although they showed that Cascade is consistent with the fading result, the explanation involved assumptions that had not yet been tested empirically. Therefore Jones and Fleischman (2001) finished with a small set of predictions and suggestions for new experiments to confirm or dispute Cascade's account. Since that time, Renkl, Atkinson, and their colleagues have run additional experiments, collecting detailed transcripts of subjects studying two types of faded sequences of problems. Although the experiments are not yet complete, we have been able to perform a qualitative analysis of the protocol data for eight of the subjects. Additionally, we have fine-tuned Cascade's knowledge base (but not its underlying mechanisms) to more faithfully model the current data. This paper reports the result of using Cascade to develop a qualitative analysis of the eight subjects. The primary result is that the findings remain consistent with Cascade's account of example fading, as well as the predictions made by Jones and Fleischman (2001).
The goal of computational cognitive neuroscience is to understand how the brain embodies the mind by using biologically based computational models comprising networks of neuronlike units. This text, based on a course taught by Randall O'Reilly and Yuko Munakata over the past several years, provides an in-depth introduction to the main ideas in the field. The neural units in the simulations use equations based directly on the ion channels that govern the behavior of real neurons, and the neural networks incorporate anatomical and physiological properties of the neocortex. Thus the text provides the student with knowledge of the basic biology of the brain as well as the computational skills needed to simulate large-scale cognitive phenomena.The text consists of two parts. The first part covers basic neural computation mechanisms: individual neurons, neural networks, and learning mechanisms. The second part covers large-scale brain area organization and cognitive phenomena: perception and attention, memory, language, and higher-level cognition. The second part is relatively self-contained and can be used separately for mechanistically oriented cognitive neuroscience courses. Integrated throughout the text are more than forty different simulation models, many of them full-scale research-grade models, with friendly interfaces and accompanying exercises. The simulation software (PDP++, available for all major platforms) and simulations can be downloaded free of charge from the Web. Exercise solutions are available, and the text includes full information on the software.
Adults, infants, and other species are able to learn and generalize abstract patterns from sequentially presented stimuli. Rule learning of this type may be involved in children's acquisition of linguistic structure, but the nature of the mechanisms underlying these abilities is unknown. While inferences regarding the capabilities of these mechanisms are commonly made based on the pattern of successes and failures in simple artificial-language rule-learning tasks, failures may be driven by memory limitations rather than intrinsic limitations on the kinds of computations that learners can perform. Here we show that alleviating memory constraints on adult learners through concurrent visual presentation of stimuli allowed them to succeed in learning regularities in three difficult artificial rule-learning experiments where participants had previously failed to learn via sequential auditory presentation. These results suggest that memory constraints, rather than intrinsic limitations on learning, may be a parsimonious explanation for many previously reported failures. We argue that future work should attempt to characterize the role of memory constraints in natural and artificial language learning.
Chi (2005) proposed that students experience difficulty in learning about physics concepts such as light, heat, or electric current because they attribute to these concepts an inappropriate ontological status of material substances rather than the more veridical status of emergent processes. Conceptual change could thus be facilitated by training students in the appropriate ontology prior to physics instruction. We tested this prediction by developing a computer-based module whereby participants learned about emergent processes. Control participants completed a computer-based task that was uninformative with respect to ontology. Both groups then studied a physics text concerned with electricity, including explanations and a posttest. Verbal explanations and qualitative problem solutions revealed that experimental students gained a deeper understanding of electric current.
