About

Bruce Sherin is a Professor in the Learning Sciences at Northwestern University, with a focus on conceptual change in science. His work primarily explores how our everyday understanding of the natural world evolves over time and through instruction. Sherin's early research involved designing and studying innovative interventions for physics education, emphasizing the role of symbolic representations. He is known for explicating the theory of Symbolic Forms, which captures the conceptual structures that successful physics experts learn to recognize in equations. More recently, his research has addressed methodological issues in studying conceptual change, applying natural language processing techniques to interview protocols. Additionally, Sherin is the designer and developer of Tactic Text, a web-based tool related to his research interests.

Research topics

• Computer Science
• Pedagogy
• Mathematics education
• Psychology
• Epistemology
• Sociology
• Linguistics
• Knowledge management
• Engineering
• Multimedia
• Geography
• Engineering ethics

Selected publications

• From Sound to Code: Creative Workflows and Problem-Solving in Music and Coding Tasks

  Proceedings. · 2025-06-10

  articleOpen access

  This paper explores the creative workflows and problem-solving strategies individuals use when composing music with code.By analyzing the experiences of seven participants with varying levels of musical and coding expertise, we identified two primary creative workflows for developing and encoding musical ideas: direct translation, where participants developed musical ideas using non-code representations and iteratively decomposed them into code; and tinkering, where participants developed musical ideas through experimentation with code.Additionally, we describe two methods used to resolve discrepancies between their mental conception of music and the coded output, which we call trial-and-error and analytical sound debugging.Our findings highlight the complementary roles of musical intuition and music theory in shaping participants' reasoning and problemsolving processes for developing and encoding their musical ideas.We conclude by discussing design implications for these findings, both regarding music+coding platforms and curricula, and extending to broader hybrid STEAM contexts.

  PublisherOA PDFDOI

• Learning Trajectories in Large-Scale Communities

  Proceedings. · 2025-06-10

  articleOpen access1st authorCorresponding

  Over two decades ago, Lave and Wenger introduced an image of learning that remains popular to this day.In that image, learners are viewed as participating in a community of practice, and learning is conceptualized as a movement from the periphery of that community toward the center.In this paper, we describe attempts to produce plots of learning trajectories for online communities that capture this movement.Our goal in this paper is to show it is possible to produce plots that capture qualitative differences in trajectories.

  PublisherOA PDFDOI

• Processes Matter: How ML/GAI Approaches Could Support Open Qualitative Coding of Online Discourse Datasets

  Computer-supported collaborative learning/˜The œComputer-Supported Collaborative Learning Conference · 2025-06-10 · 1 citations

  articleOpen access

  Open coding, a key inductive step in qualitative research, discovers and constructs concepts from human datasets.However, capturing extensive and nuanced aspects or "coding moments" can be challenging, especially with large discourse datasets.While some studies explore machine learning (ML)/Generative AI (GAI)'s potential for open coding, few evaluation studies exist.We compare open coding results by five recently published ML/GAI approaches and four human coders, using a dataset of online chat messages around a mobile learning software.Our systematic analysis reveals ML/GAI approaches' strengths and weaknesses, uncovering the complementary potential between humans and AI.Line-by-line AI approaches effectively identify content-based codes, while humans excel in interpreting conversational dynamics.We discussed how embedded analytical processes could shape the results of ML/GAI approaches.Instead of replacing humans in open coding, researchers should integrate AI with and according to their analytical processes, e.g., as parallel co-coders.

  PublisherOA PDFDOI

• Processes Matter: How ML/GAI Approaches Could Support Open Qualitative Coding of Online Discourse Datasets

  ArXiv.org · 2025-04-02

  preprintOpen access

  Open coding, a key inductive step in qualitative research, discovers and constructs concepts from human datasets. However, capturing extensive and nuanced aspects or "coding moments" can be challenging, especially with large discourse datasets. While some studies explore machine learning (ML)/Generative AI (GAI)'s potential for open coding, few evaluation studies exist. We compare open coding results by five recently published ML/GAI approaches and four human coders, using a dataset of online chat messages around a mobile learning software. Our systematic analysis reveals ML/GAI approaches' strengths and weaknesses, uncovering the complementary potential between humans and AI. Line-by-line AI approaches effectively identify content-based codes, while humans excel in interpreting conversational dynamics. We discussed how embedded analytical processes could shape the results of ML/GAI approaches. Instead of replacing humans in open coding, researchers should integrate AI with and according to their analytical processes, e.g., as parallel co-coders.

  PublisherOA PDFDOI

• Computational models as tools for supporting responsive teaching

  Behaviour and Information Technology · 2024-07-30 · 6 citations

  article

  It is widely agreed that science instruction should help students build new knowledge on the foundation of their prior knowledge. Responsive teaching refers to a family of teaching strategies that pursue and build on student ideas. We introduce a particular approach to responsive teaching and examine how it can be supported by the use of computational models. We analyse an 8th grade science teacher’s facilitation of a class discussion near the end of a lesson on sound. We present a moment-by-moment characterisation of her responsive teaching moves, highlighting the ways she used a computational model to help students articulate and examine their thinking. Our findings make empirical contributions to literature concerned with responsive teaching and literature concerned with the role of computational models in constructivist approaches to instruction.

  PublisherDOI

• Novel Technologies and Epistemic Considerations in Studying Knowledge-in-Use and in-Transition

  Proceedings. · 2024-06-10

  articleOpen accessSenior author

  This symposium brings together researchers interested in moment-by-moment reasoning and learning processes drawing from data across multiple disciplines and populations.A commonality of the research programs represented in the symposium is that they all contribute to elaborating Knowledge Analysis as a methodology, and draw upon the Knowledge in Pieces heuristic epistemological framework.This symposium engages the participants in methodological reflections on new directions for advancing how knowledge analyses are conducted.In particular, the symposium is framed around the dual issues of (1) how to approach and coordinate the analysis of multiple kinds and levels of data to shed light on learning processes and (2) the advantages and tradeoffs of human collaboration with computational methods. Symposium overviewReflections on methodological insights and considerations are important for learning sciences researchers to understand new possibilities for analysis, especially as new technologies for analysis of verbal and interactional data continue to be developed.The infusion of new methods, such as textual analysis using big data sets and Epistemic Network Analysis, brings new possibilities for researchers who typically do video-based analyses of knowledge in use and in transition.This symposium proposes conversation across a unified set of papers by researchers who broadly are informed by the Knowledge in Pieces theoretical framework to frame their study of moment-by-moment reasoning and learning processes.Knowledge in Pieces (KiP; diSessa, 1993) is a heuristic epistemological framework.KiP models knowledge as a complex system of interconnected elements and learning as changes in the composition of networks activated in response to the sense-making demands of a given context.A primary concern of the KiP research agenda is the development of theoretical characterizations of knowledge and learning.Over the years, a number of theoretical constructs and models have been generated to describe novice and expert sense-making across domains (see diSessa, 2018; diSessa & Levin, 2021 for overviews).Methodological descriptions are typically embedded in individual papers, but elevating a discussion of methods across research programs and studies can spark innovation and new insights.Methodological papers on Knowledge Analysis provide a needed foundation for articulating the common assumptions and practice of Knowledge Analysis from a Knowledge in Pieces perspective (Barth-Cohen, Swanson, & Arnell, 2023;diSessa, Sherin & Levin, 2016;Parnafes & diSessa, 2013).However, more recently-developed technologies and approaches, such as Eye-Tracking, Epistemic Network Analysis, and Thematic Analysis, afford new questions and issues for how to pursue Knowledge Analysis.For example, recently, Orrill and colleagues (this session) have used KiP-inspired Epistemic Network Analysis to study mathematics teacher learning and design PD.The results of such work are consequential as Orrill's work demonstrates that teachers have access to a number of different knowledge resources and that a crucial issue for teachers is coordinating the activation of their knowledge resources.This has strong implications for the design of PD.Question that orient this session are: How do we approach and coordinate analyses of data involving multiple different sources and levels to shed insight into learning processes? What are the advantages and tradeoffs of human collaboration with advanced technologies in conducting analyses of learning data?The symposium will begin with co-chair, Mariana Levin, setting the stage for the focus on methodological reflection across research programs in the symposium.Hillary Swanson will then provide a framework for reflecting on processes of theory building, informed by Collins ' and Ferguson's (1993) Epistemic Forms and Games and applied to the case of theorizing within the epistemological assumptions of the Knowledge in Pieces perspective.The framework she proposes will orient participants to the aims of the work presented in

  PublisherOA PDFDOI

• Prompts Matter: Comparing ML/GAI Approaches for Generating Inductive Qualitative Coding Results

  arXiv (Cornell University) · 2024-11-10

  preprintOpen access

  Inductive qualitative methods have been a mainstay of education research for decades, yet it takes much time and effort to conduct rigorously. Recent advances in artificial intelligence, particularly with generative AI (GAI), have led to initial success in generating inductive coding results. Like human coders, GAI tools rely on instructions to work, and how to instruct it may matter. To understand how ML/GAI approaches could contribute to qualitative coding processes, this study applied two known and two theory-informed novel approaches to an online community dataset and evaluated the resulting coding results. Our findings show significant discrepancies between ML/GAI approaches and demonstrate the advantage of our approaches, which introduce human coding processes into GAI prompts.

  PublisherOA PDFDOI

• Engaging Stakeholders in Deliberation for Organizational Design

  Proceedings. · 2024-06-10

  articleOpen accessSenior author

  Deliberation involves applying knowledge, synthesizing ideas, weighing different options and reflection by stakeholders who negotiate a decision that leads to implementation.While we know about many of the cognitive processes involved in deliberation, we know little about how stakeholders become involved in deliberation in real-world, non-binding contexts.This study investigates the processes of building up a Human-Computer Interaction (HCI) certificate program for undergraduate students at a private university.We analyzed 334 messages from 57 stakeholders' email exchanges, meetings, and interview transcripts to reveal the stakeholder-seeking process and outreach strategies.Results showed that deliberation started from a small group of active stakeholders who accounted for most of the work and outreach.Our findings reveal that an effective stakeholder involvement strategy is the key for the outcome of deliberation to factor in implementation in the observed case.We also consider the types of learning required by individuals and the group in deliberation. Deliberation in the classroomDeliberation entails exploring diverse perspectives, challenging assumptions, and reaching consensus through open discussion.This process aims to facilitate idea exchange and alignment among participants toward common goals (Locke, 2019;Paul, 2017).Civic education scholars have argued that students should have chances to bolster their civic engagement experience (Levine, 2008) including practicing democratic deliberation where learners utilize their experiences collaboratively to find solutions to the challenging issues impacting their lives and communities (Levine, 2000;Munoz & Wrigley, 2012).People need to learn how to do deliberation to be effective civic actors (Levine, 2022).Practicing deliberation in the classroom prepares students to make collaborative decisions and adopt broader perspectives.However, the transition from the classroom to real-world settings reveals the absence of prearranged environments conducive to deliberation.In the real-world scenarios, citizens often confront situations where they must initiate deliberative processes from scratch, and the essence of implementation emerges in those unscripted scenarios.Thus, it is essential for civic education to provide students with the necessary knowledge and abilities to engage as activists, extending beyond mere participation in discussion to adopt the competence of initiating and sustaining deliberation at both individual and group levels. Deliberation in the real worldPolitical scientist Robert Dahn (1989) proposed mini-publics, such as citizens' assemblies, as a process for engaging citizens in dealing with public issues.In mini-publics, people are randomly selected as the representatives of a bigger population.They work together to deliberate and learn from one another to make collective decisions for public good.Mini-public participants hold equal rights.Every participant has an equal chance of being selected to represent the broader population.Mini-publics thus promote deliberation and representation better than existing political decision-making processes.Unfortunately, mini-publics often fail to have political impact in practice.For instance, in 2019 the Scottish government initiated a mini-public, with 104 participants to deliberate the nation's future amidst Brexit challenges.Despite positive feedback on its execution, concerns linger about the feasibility of recommendations and their impact on government policies.The outcome of mini-publics failed to result in policy changes.Studying how mini-publics work looks promising from the normative perspective, but it needs a clear set of rules and structures for the process to result in policy changes, especially when mini-publics are frequently used for giving advice (Setala, 2021). Binding and non-binding contextsIn a binding context, the decisions made during deliberation must be carried out, in the way a jury's decision in a court of law is enforced after the trial, their verdict settles the case.The jury's decision is accepted as binding ahead of the deliberation, which with certainty leads to implementation.Other examples of binding decisions include company staff executing the deliberative decisions of a board meeting, or city staff performing the resolutions from city council members' deliberation.The outcome of deliberation in each case is accepted as binding before the processes of deliberation begins.Conversely, in a non-binding context, decision makers are

  PublisherOA PDFDOI

• Knowledge in New Pieces (KiNP): Exploring Diverse Areas of Contemporary Youths’ Intuitive Technosocial Knowledge

  Proceedings. · 2024-06-10 · 1 citations

  articleOpen accessSenior author

  Since Knowledge in Pieces, diSessa's influential conceptual change framework, was introduced 35 years ago, the worlds that youth inhabit, navigate, and author have changed dramatically.New computing technologies have gained widespread use in daily personal, professional, and academic life.Social norms have broadened and shifted, making previously taboo topics a part of daily headlines and conversations.Advances in science have changed our understanding of global phenomena.Given the centrality of intuitive knowledge to the Knowledge in Piece framework, we propose that these changes to youth's daily lives could result in similar changes to their thinking and reasoning that merit documentation.Moreover, should new intuitive knowledge pieces prove to be widespread, formal education and curricula could benefit from reflecting them.This symposium brings together researchers studying youth's intuitive knowledge across a range of fields to share findings and organize toward a future research agenda. Symposium overviewRoughly 35 years ago, diSessa published a foundational article, Knowledge in Pieces ( 1988), creating a new paradigm in the field of conceptual change, especially in science education.In contrast to traditions that paralleled children's development of scientific ideas with the history of science (Wiser & Carey, 1983) or that described learners as progressing through series of coherent theories as trained scientists do (McCloskey, 1983b(McCloskey, , 1983a)), diSessa drew on empirical studies of physics education to posit that learners have varied intuitive knowledge pieces (phenomenological primitives or p-prims) about how the world works that they pick up in the course of everyday life (e.g., "more force implies more effect," "forces act in balance").The Knowledge in Pieces (KiP) framework suggests that these p-prims are highly contextual and persist through efforts to change or replace them (Smith III et al., 1994).Since its introduction, KiP has informed many subsequent studies on conceptual change (diSessa & Sherin, 1998), including within mathematics ("Transfer in Pieces," Wagner, 2006) and social justice ("Ideology in Pieces," Philip, 2011).More recently, researchers have argued that KiP paired with interaction analysis reframes competence and counters "conventional characterizations of learners as systematically deficit" (diSessa et al., 2016, p. 18) Given the centrality of everyday knowledge in forming the foundational elements of reasoning, and the radical changes in the contexts and technologies children are exposed to daily, scholars (Blikstein & Rosenbaum, 2023) have called for considerations of how the intuitive knowledge that youth develop in the course of their contemporary techno-social lives may differ from that of the youth with whom diSessa worked 30+ years ago.How might youth's everyday interactions with new media and sources of information outside of school shape their thinking and learning?How might such thinking and learning be augmented by the fact that today, many youth spend more time on digital devices than interacting with peers "in real life"?In other words, what intuitive knowledge pieces are generated by youth in their contemporary techno-social lives, in the multiple contexts in which they live, and how do these pieces guide conceptual development?Understanding the intuitive knowledge youth draw on as they learn new conceptual domains and exploring the everyday experiences from which such knowledge pieces emerge would build on complementarities in recent KiP and interaction analysis approaches (diSessa, et al., 2016).Should such knowledge in new pieces (KiNP) be identified and validated through coordinated research programs, they could have implications for at least three audiences: educators and curriculum developers in better understanding and building on students' prior knowledge, researchers in examining educational contexts in more precise and culturally relevant ways, and the field as a whole in conceptualizing dynamic learning spaces and practices that are attuned to students' lived experiences.

  PublisherOA PDFDOI

• How Co-Designing Computational Modeling Activities Helped Teachers Implement Responsive Teaching Strategies

  2023-06-21 · 2 citations

  articleOpen access

  In recent years, science education has shifted focus, from content to practice. This is reflected in the NGSS, which advocate learning science concepts through engagement in science and engineering practices. Theory building is a central activity of science and computational modeling is a key practice through which contemporary scientists construct theory. In this paper, we discuss an 8th grade science teacher's implementation of a computational modeling lesson. The teacher had co-designed the computational modeling microworld and lesson with the research team over the preceding summers. We investigate the teacher's activity during a whole-class discussion near the end of the lesson, to understand her responsive teaching strategies and how the co-designed technology supported her in eliciting and responding to student ideas. We examine the transcript from a follow-up interview to understand her experience implementing the co-designed technology and responsive teaching strategies, and to identify foci of future co-design iterations.

  PublisherOA PDFDOI

Frequent coauthors

• Miriam Gamoran Sherin

  11 shared

• Victor R. Lee

  Stanford University

  9 shared

• Daniel C. Edelson

  8 shared

• Andrea A. diSessa

  8 shared

• Rosemary S. Russ

  University of Wisconsin–Madison

  8 shared

• Uri Wilensky

  Northwestern University

  5 shared

• Kenneth D. Forbus

  5 shared

• Hillary Swanson

  5 shared

Education

• Ph.D., Education

  University of California, Los Angeles

  1990

• M.A., Education

  University of California, Los Angeles

  1985

• B.A., Education

  University of California, Los Angeles

  1983