PaperlayerView exact revision on arXiv

Source-linked AI summary

CircuTutor: Transforming Static Circuit Problems into Intelligent and Dynamic Tutoring

Ziyu Luo, Xiaorui Ma, Lin Chen, Xiaoming Chen

arXiv:2609.08254v1cs.AI

TL;DR

Learners often find it difficult to connect invisible DC-circuit quantities with observable behavior, while conventional materials and simulators provide limited reasoning-focused diagnosis. CircuTutor parses textbook problems into validated circuit tasks, uses SPICE-compatible simulation to generate state-based explanations and adaptive practice, and reports improved conceptual learning and learner experience in a controlled evaluation. The framework also suggests a possible extension of simulation-grounded tutoring beyond circuit education.

  • Problem

    DC-circuit learners must connect invisible current, voltage, resistance, and power with observable outcomes, but conventional materials and simulators provide limited support for explaining behavior changes and diagnosing incorrect reasoning.

  • Method

    CircuTutor parses textbook circuit problems into validated structured tasks, uses SPICE-compatible simulation to compute circuit states, and converts those states into animations, causal explanations, misconception diagnoses, and adaptive follow-up exercises.

  • Results

    CircuTutor improved conceptual understanding, transfer performance, learning efficiency, and learner experience in a mixed-methods controlled study.

  • Takeaways & Limitations

    Simulation-derived circuit states can support intelligent and dynamic tutoring by grounding visual explanations, causal reasoning, misconception diagnosis, and follow-up exercise generation.

  • Takeaways & Limitations

    Future work must evaluate larger and more diverse learner populations, additional circuit structures and misconception categories, longer-term knowledge retention, and component-level contributions.

Abstract

from arXiv · show

Learning direct current circuit concepts requires learners to connect invisible physical quantities, such as current, voltage, resistance, and power, with observable outcomes such as bulb brightness. Conventional textbook materials and general-purpose circuit simulators provide opportunities for problem solving and exploration but offer limited support for explaining why circuit behavior changes or diagnosing the reasoning behind incorrect answers. We present CircuTutor, a circuit-state-driven intelligent tutoring system that transforms static textbook circuit problems into an interactive tutoring workflow. CircuTutor first uses multimodal problem parsing to extract the textbook question, circuit topology, component parameters, switch states, and answer options, which are converted into a structured task and validated through circuit simulation. Learners can then interactively explore the circuit (by changing parameters) and submit an answer while a SPICE-compatible solver computes physically consistent circuit states. After the learner submits an answer, CircuTutor presents a before-and-after circuit state animation corresponding to the selected operation, organizes the simulated state changes into a causal reasoning chain that explains the underlying circuit behavior, maps answer discrepancies to likely misconceptions, and generates adaptive follow-up exercises targeted at the diagnosed misconception. Our experimental results demonstrate that CircuTutor effectively improves conceptual learning and the overall learning experience. The proposed framework demonstrates how simulated circuit states can be transformed into intelligent and interactive tutoring for circuit education, with the potential to generalize to other STEM domains.

1 Introduction

CircuTutor addresses persistent difficulties in connecting invisible circuit quantities to observable behavior by turning static circuit problems into an interactive, simulation-grounded tutoring workflow. It combines structured problem parsing, circuit-state explanations, misconception diagnosis, and adaptive follow-up exercises.

  • Motivation: Learners often struggle to connect current, voltage, resistance, and power with observable outcomes such as bulb brightness.Misconceptions include treating batteries as constant-current sources, confusing current with voltage or resistance, and reasoning locally rather than about the whole circuit.
  • Motivation: Existing interactive simulations support parameter manipulation and dynamic exploration, but effective circuit learning also requires specific, timely, explanatory feedback.Simply marking an answer correct or incorrect may not distinguish different misconceptions behind the same response.
  • Approach: CircuTutor transforms static textbook circuit problems into an interactive tutoring workflow through multimodal problem parsing and circuit simulation.The system extracts questions, answer options, parameters, circuit elements, and topology, then validates the resulting structured task.
  • Approach: The system converts simulated circuit states into before-and-after animations, causal reasoning chains, misconception diagnoses, and adaptive follow-up exercises.These components connect learner operations and answer discrepancies to physically consistent circuit behavior and targeted practice.
  • Evaluation: A mixed-methods controlled study reported improvements in conceptual understanding, transfer performance, learning efficiency, and learner experience.The introduction summarizes validation of the complete CircuTutor workflow across these learning and experience outcomes.
  • Broader relevance: The framework is proposed as a potential paradigm for translating simulation-derived state representations into intelligent tutoring support beyond circuit education.The paper identifies other simulation-based STEM domains involving links between invisible system states and observable phenomena as possible extensions.

2 Related Work

Related work establishes that simulations can connect abstract models with observable behavior, but effective learning depends on instructional guidance and explanatory feedback. CircuTutor builds on this work by grounding feedback and diagnosis in reliable circuit simulation and evidence-constrained rules.

  • Interactive Simulations for Physics Education: Interactive simulations let learners manipulate parameters, observe system responses, and connect abstract models with observable phenomena.PhET is presented as a representative research-informed simulation platform for conceptual exploration across physics topics.
  • Interactive Simulations for Physics Education: Simulations are more effective when integrated with instructional guidance and appropriate learning activities rather than used as isolated exploratory tools.This issue is especially important for DC circuits because key quantities such as current, voltage, resistance, and power are not directly visible.
  • Visual Explanations and Representations: Visual explanations help learners coordinate diagrams, numerical values, verbal descriptions, and conceptual models across multiple representations.Related work emphasizes integrating text and graphics and helping learners relate information distributed across representational forms.
  • Feedback and Intelligent Tutoring: Effective feedback is timely, specific, and directed toward task-related reasoning rather than merely reporting correctness.In physics, elaborated feedback should explain what was incorrect, why it was incorrect, and how reasoning can improve.
  • Feedback and Intelligent Tutoring: CircuTutor grounds circuit outcomes in reliable simulation and maps answer discrepancies to likely misconceptions through an evidence-constrained rule set.A constrained language model varies adaptive follow-up exercises, while post-submission explanations and diagnoses remain tied to simulation-derived circuit states.

3 Method

CircuTutor converts textbook circuit problems into validated, interactive simulations and uses shared simulated states to provide visual, causal, diagnostic, and adaptive tutoring support.

  • 3.2 Multimodal Problem Parsing: CircuTutor parses textbook problem images into validated structured tasks containing circuit topology, parameters, answer options, and visual mappings.Schema and simulation validation ensure only complete, internally consistent, solvable tasks reach learners.
  • 3 Method: The architecture links parsing, interactive simulation, post-submission explanation, misconception diagnosis, and adaptive exercises through a feedback loop.The shared circuit state grounds visual comparison, reasoning, diagnosis, and follow-up generation in the same physical solution.
  • 3.3 Learner Interaction and Circuit Simulation: Learners adjust component parameters or switch states while a PySpice-based solver converts each interaction into a SPICE-compatible circuit state.The system records interaction traces and models open and closed switches as high- and low-resistance paths.
  • 3.3 Learner Interaction and Circuit Simulation: After answer submission, CircuTutor compares before-and-after states and maps voltage, current, power, and brightness changes onto the rendered circuit.Brightness changes are derived from simulated power differences using a tolerance that filters negligible solver-level differences.
  • 3.4 Misconception Diagnosis and Follow-up Exercise: The system organizes state changes into a dependency chain from topology and resistance through current, component power, and observable brightness.The active topology transition determines which causal steps are included.
  • 3.4 Misconception Diagnosis and Follow-up Exercise: A rule-based matcher compares predicted and simulated bulb-change patterns to select an evidence-consistent misconception and generate a targeted follow-up exercise.Follow-up formats include contrastive circuit variants, parameter-exploration tasks, and multiple-choice explanation questions; generated tasks undergo physics validation.

4.1 Overview

CircuTutor was evaluated through complementary studies addressing both learning outcomes and learner experience and the quality of its tutoring support.

  • 4.1 Overview: A between-subjects controlled evaluation compared CircuTutor with a control condition on learning outcomes and learner experience.The evaluation examined overall learning effectiveness and experience.
  • 4.1 Overview: A second evaluation used expert review and learner feedback to examine the quality of CircuTutor’s tutoring support.Together, the evaluations addressed both outcomes and support quality.

4.2 Evaluation of Learning Outcomes and Experience

A randomized study of 30 students compared CircuTutor with a control condition across DC-circuit learning tasks, measuring learning outcomes, interaction efficiency, and learner experience.

  • Participants: 30 students were randomly assigned to a control group or CircuTutor group, with 15 participants in each condition.All participants had prior introductory DC-circuit instruction but had not used CircuTutor.
  • Conditions and learning tasks: Both groups completed four DC-circuit reasoning tasks covering topology, resistance, current redistribution, bulb power, and brightness variation.Tasks used different configurations and parameter values while requiring coordination of the same underlying physical quantities.
  • Measures for Learning outcomes: Post-test and transfer-test scores were primary learning outcomes, with raw learning gain calculated as post-test minus pre-test performance.Transfer tasks used unseen circuit configurations and parameter combinations.
  • Analysis: The analysis used ANCOVA for post-test and transfer scores, mixed-effects logistic regression for first-attempt correctness, and log-transformed linear modeling for completion time.Condition was the between-subjects factor, with pre-test score as the ANCOVA covariate and participant and task as random intercepts for logistic regression.
  • Learning outcomes: After controlling for pre-test performance, CircuTutor produced significantly higher post-test scores than the control group, F(1, 27) = 5.89, p= .022.The control group had numerically higher pre-test scores, but that difference was not statistically significant, p= .170.
  • Experience: CircuTutor participants showed higher first-attempt accuracy (OR= 3.64, p= .012), shorter task-completion time (ratio= 0.60, p= .004), and higher interest/enjoyment and value/usefulness.Retry count did not differ significantly, IRR= 0.71, p= .126; mean SUS was 73.33 (SD= 14.57).

4.3 Evaluation of Tutoring Support

CircuTutor’s tutoring support was evaluated for physical reliability, diagnostic appropriateness, and understandability through expert review and learner feedback. Physical correctness received the strongest expert assessment, while causal reasoning and remedial relevance were comparatively weaker.

  • Evaluation design: Expert review covered 24 representative cases spanning four misconception categories and included simulations, animations, reasoning chains, diagnoses, and follow-up exercises.Three physics-education experts independently evaluated the cases.
  • Expert evaluation: Clarity and causal completeness received mean expert ratings of 3.54 and 3.53, while remedial relevance received the lowest expert rating at M= 3.05.The remedial relevance rating had SD= 1.12.
  • Learner feedback: Learners rated adaptive follow-up exercises highest (M= 3.87), followed by misconception diagnosis (M= 3.80) and circuit state animation (M= 3.73).Causal reasoning received the lowest learner rating, at M= 3.13.
  • Learner feedback: Learners reported that current and power changes supported brightness explanations, while diagnosis clarified errors and follow-up exercises supported application to related circuits.The main difficulty concerned the presentation and organization of causal reasoning.

5 Conclusion

CircuTutor integrates parsing, simulation, interactive exploration, and adaptive feedback to help learners understand circuit changes and incorrect answers. The authors report improved learning outcomes and identify broader evaluation and ablation studies as future work.

  • Conclusion: CircuTutor unifies multimodal problem parsing, task validation, SPICE-compatible simulation, state animation, causal reasoning, misconception diagnosis, and adaptive follow-up exercises.The system separates learner exploration and answer submission from post-submission feedback.
  • Conclusion: Participants using CircuTutor achieved significantly higher post-test and transfer performance, lower mental effort, and higher learning motivation than the control group.Expert and learner evaluations also rated physical correctness and diagnostic accuracy highly.
  • Future work: Future work will examine larger and more diverse learner populations, additional circuit structures and misconception categories, and longer-term knowledge retention.The authors also plan component-level ablation studies to assess individual feedback components.
Loading 2609.08254v1…