Vapes – a sustainable alternative?! – Promotion of socioscientific decision making through self-regulated learning approaches in sustainability contexts

2.1 Self-regulated learning as a prerequisite basic for dealing with controversial socially relevant evaluation contexts

Self-regulated learning (SRL) is a fundamental objective in the field of educational research, serving to complement traditional content knowledge instruction. ²³ It is imperative for learning across school and extracurricular settings, and it fosters adaptability in lifelong learning. ²⁴ SRL integrates cognitive, metacognitive, and motivational components to assist learners in actively managing and optimizing their learning processes. ²⁵

Given the diversity of theoretical approaches, a universal definition of SRL competence remains elusive. ²⁶ While component models underscore distinct competencies, process models conceptualize SRL as a cyclical process comprising planning, execution, and reflection. ^{25 , 27 , 28} According to Wirth and Leutner (2008), an integrative definition is provided. SRL competence is defined as the ability to autonomously plan, monitor, and evaluate learning, involving ongoing cognitive, motivational, and behavioral decisions. ²⁶ It consists of interconnected sub-competencies that address specific metacognitive demands during learning. ²⁶

Metacognition, a core element of SRL, involves the regulation and reflection of cognitive processes. ²⁹ It includes monitoring progress, applying strategies, and adjusting them to achieve goals. ³⁰ Pintrich et al. (2000) distinguish three types of metacognitive knowledge: declarative (awareness of knowledge), procedural (application of strategies), and conditional (knowing when and why to use strategies). ^{31 , 32 , 33}

Zimmerman (200) ²⁵ defines SRL as a cycle. In the planning phase, learners set goals, choose strategies, ³⁴ and plan time and methods. Motivation supports engagement. ³⁵ In the action phase, learners ³⁴ apply strategies like elaboration or repetition, while monitoring progress and understanding. Motivational regulation helps sustain effort and avoid procrastination. ³⁵ In the reflection phase, learners evaluate performance, assess goal achievement, and make causal attributions. ^{23 , 36} These reflections lead to self-reactions, like adjusting strategies or setting new goals. ^{34 , 35} Through this cycle, learners build the ability to regulate their learning. ^{23 , 36}

SRL is important when learners solve complex problems, as in SSIs, where they connect science facts with social, ethical, and ecological factors, encouraging informed decision-making. ¹⁸ SRL strategies enable learners to control the decision-making process, from problem analysis to selecting appropriate strategies and reflecting on their decisions. ¹⁷

Empirical studies show that metacognitive support and strategy training improve students’ decision-making skills in SSI contexts, especially when activating reflection processes or encouraging independent strategy development. However, SRL elements, such as reflection aids and metacognitive prompts, do not always result in significant overall effects. ^{19 , 37} Other studies emphasize that explicit decision-making models, shifts in perspective, and discovery-based strategy learning stimulate SRL processes, ^{16 , 18} increasing the depth of argumentation and the quality of evaluation. The effect of metacognitively embedded support is evident in the work of Gresch et al. (2013, 2017), who found that combining teaching with systematic decision-making strategies, metacognitive reflection, and SRL techniques improved short-term decision-making skills and enabled long-term application to new contexts. Simple strategy training showed positive effects, but additional SRL phases led to a deeper understanding of the process and more sustainable learning gains. ^{14 , 17}

In chemistry education, SRL must be embedded as an independent educational goal and methodological principle. In chemistry-related SSIs, such as environmental pollution, sustainable energy production, or e-cigarette consumption, SRL provides the basis for conducting experiments, research, and decision-making independently and reflectively. ³⁴

2.2 Socioscientific decision making

Researchers emphasize that engaging with SSI requires multiperspective, informal thinking that involves value decisions. ¹⁸ Students must consider scientific evidence as well as societal, ethical, economic, and environmental factors in order to make informed decisions. Lee and Grace (2012) demonstrate that structured decision-making frameworks enhance students’ ability to integrate diverse perspectives and foster metacognitive reflection in SSI discussions. ¹⁸ Thus, socioscientific decision-making (SSD) is considered a key competence in science education. In the German educational framework of the German conference of ministers of education (KMK), the National Standards define socioscientific decision-making competence as the ability to evaluate information from disciplinary and interdisciplinary perspectives, apply evaluation methods, and critically analyze evidence-based options. ⁸ Students should also be able to make ethically grounded decisions and reflect on the process and its consequences. ^{8 , 38 , 39} Proficient learners can assess scientific data using various criteria, justify their positions, make value-based decisions, and critically examine their reasoning and its implications. This competence requires an understanding of scientific evidence and the integration of economic, environmental, and moral considerations. The KMK emphasizes that effective socioscientific decisions rely on weighing multiple perspectives and criteria. Cultivating this competence enables learners to participate in science-related societal debates and make well-reasoned personal and civic decisions. Different constructs describe the decision-making evaluation processes in science education. The following is a brief description of these constructs, focusing on the PAAWD^R model, the basis for the current study.

One such construct is socio-scientific reasoning, which describes how learners engage with controversies within the framework of SSI. ^{40 , 41} It comprises four central dimensions: complexity, perspective-taking, inquiry, and skepticism. ^{40 , 42} The focus is not on the decision-making process itself but rather on the outcome of thinking, specifically the ability to take an informed position on a controversial issue. In contrast, socioscientific decision-making is a structured, process-oriented framework in which learners formulate a problem, develop possible solutions, and systematically evaluate them in order to make an informed decision. ⁴³ Thus, SSD encompasses advanced skills such as evidence-based argumentation and consciously weighing alternative solutions, which goes beyond the scope of social scientific thinking. ⁴⁰ Although both concepts deal with the examination of science-related social issues, they have different emphases. ⁴⁰

In the context of SSD, Fang et al. (2019) developed a three-phase overarching framework for SSD by analyzing 24 international studies. The first phase, “Formulate the Decision-Making Space”, ⁴⁴ involves identifying the problem, gathering relevant information, and defining evaluation criteria from scientific, social, and ethical perspectives. The second phase, “Posit a Decision-Making Strategy”, ⁴⁴ involves selecting a suitable approach – either compensatory or non-compensatory – to weigh alternatives and make a reasoned decision based on facts and values. ⁴⁵ In the third phase, “Review and Reflect on the Decision-Making Process”, ⁴⁴ learners evaluate their strategies and decisions, which fosters metacognitive growth and the long-term development of SSD abilities. ⁴⁵

Fang et al. (2019) demonstrate through their model that socioscientific decision making encompasses more than just making the “right” choice. It also guides learners through essential cognitive and metacognitive phases, such as understanding the problem, selecting a decision-making strategy, conducting critical follow-up, and reflecting. ⁴⁵ SSD is therefore crucial for analyzing complex, interdisciplinary issues, detecting misinformation, and participating in social discourse. ⁵ As a key competence in science education, SSD enables learners to make scientifically grounded decisions and distinguish facts from opinions. ⁴⁶

The PAAWD^R model, developed by Langlet et al. (2022), supplements the internationally recognized SSD framework established by Fang et al. by translating its three phases into six actionable classroom steps. It is particularly urgent to strengthen SSD, as many young people still lack the knowledge to make informed decisions, especially regarding environmental issues. ^{47 , 48} PAAWD^R was developed to operationalize the National Standards in German science education and offers a subject-specific structure that organizes evaluative reasoning into sequential yet cyclical phases of reflection and decision-making. ⁸ Unlike more abstract models, PAAWD^R provides a clear, time-efficient framework that connects scientific reasoning with value-based decisions, making complex evaluation processes accessible in practical teaching. Although PAAWD^R was originally designed for a national context, it represents a concrete operationalization of Fang’s et al. model, enhancing its applicability across international science education by bridging theory and practice.

The PAAWD^R model divides the science education evaluation process into six steps, based on Fang et al.’s (2019) framework model: Perceiving, Analyzing, Arguing, Weighing, Deciding and Reflecting. ⁴⁷

The PAAWD^R model is distinct from cognitive models in that it incorporates socio-scientific and ethical aspects into the decision-making process. It emphasizes recognizing normative elements, justifying decisions with evidence-based reasoning, and reflecting on implications. The process starts with perceiving conflicts from multiple perspectives and assessing their consequences. ⁴⁷ In the analysis phase, learners derive criteria and gather information to build a solid decision-making foundation. ⁴⁷ These steps align with Fang et al.’s (2019) first phase, “Formulate the Decision-Making Space,” in which learners define the problem and frame it using scientific, ethical, and societal perspectives. Arguing entails distinguishing between factual and evaluative statements, incorporating diverse viewpoints, and critically examining arguments. ⁴⁷ Weighing involves organizing and comparing arguments. ⁴⁷ In the second phase, “Posit a Decision-Making Strategy,” students apply strategies and assess alternatives. The decision step leads to an independent, well-founded judgment. ⁴⁷ Finally, reflection ensures critical evaluation of the process and its consequences. ⁴⁷ This mirrors Fang et al.’s third phase, “Review and Reflect on the Decision-Making Process,” where learners assess their strategies and decisions metacognitively. The cyclical nature of the process highlights the importance of SRL, as learners must plan, monitor, and reflect on their thinking. Metacognitive strategies foster the targeted revision of arguments and conscious reflection on decisions. ⁴⁹

2.3 Artificial intelligence to promote socioscientific decision making and self-regulated learning

Traditional classrooms do not automatically foster informed decision-making in complex scientific contexts. Research on socioscientific issues shows that students need support to evaluate data and make decisions. ⁵⁰ AI-supported learning environments offer new possibilities by providing personalized tools and support for evaluating scientific evidence. Fang et al. (2019) emphasize the importance of learning environments to foster effective decision-making strategies. ⁴⁵ AI systems could fulfill this role through instructional support. Research on AI support for SSD in chemistry is limited, but related studies offer valuable insights. Ariely et al. (2024), for example, showed that automated analysis and feedback on open-ended responses can improve scientific judgment. Their AI-supported system identified patterns in student explanations and offered feedback, enhancing causal-mechanistic reasoning in biology. ⁵¹ Similarly, AI tools could prompt reflection and guide students through chemical decision-making by providing targeted cues and prompts. Early findings suggest that AI can foster SSD by offering support and promoting awareness of alternative courses of action, especially in the context of argumentation – a key step in evaluative processes.

To engage meaningfully in social discourse grounded in scientific evidence, learners must develop scientific argumentation skills. This process entails the construction, support, and critique of explanations of scientific subjects within a scientific community, guided by objective criteria. ^{52 , 53} AI tools can enhance these skills. For instance, Wilson et al. (2023) developed automated evaluation systems that reliably assess student arguments at the competency level, enabling teachers to assign open-ended tasks while maintaining expert-level assessment quality. ^{54 , 55} AI-based feedback systems have also been tested to directly support the writing process. Wambsganss et al. (2025) integrated dynamic learning models into a writing platform, providing automated feedback on logical flaws. Three empirical studies showed this approach significantly improved argumentative writing quality, surpassing traditional instruction. ⁵⁶ Students produced more structured and persuasive arguments over time than with static or generic aids. Another study in higher education found that combining automated feedback with social comparison nudges helped students identify and revise argumentative flaws in practice essays. ⁵⁶ These findings show that AI systems can substantially improve argumentation skills by offering personalized feedback, encouraging revision, and clarifying complex argument structures.

Recent studies show that AI-supported learning environments can effectively foster SRL. For instance, the mobile AI chatbot ‘Study Buddy’ offers targeted support across the SRL phases planning, execution, and reflection in hybrid learning environments. ⁵⁷ Students with initially low SRL levels showed significant improvement when using this tool. Chatbots also serve as structuring aids for goal setting, progress monitoring, and reflection, and are perceived by learners as user-friendly and beneficial for SRL development. ^{58 , 59} Additional research indicates that chatbot-supported goal-setting prompts, especially those based on the SMART framework, can further support learners in online environments. ^{60 , 61}

AI-based learning environments and tutors have proven effective in fostering SRL by adapting task difficulty and providing real-time, personalized feedback that strengthens planning and reflection skills. Studies show that virtual AI coaches significantly increase the use of learning strategies, such as planning tools, compared to control groups. ⁶² Similar approaches are applicable in schools, using intelligent tutoring systems that visualize learning progress or pose metacognitive prompts. Crucially, AI support must explicitly target metacognitive processes by prompting learners to question their understanding, analyze mistakes, and plan next steps. Empirical evidence indicates that AI-supported feedback enhances academic performance, encourages self-reflection, and improves time management. ⁶² Zhai et al. (2023) found that AI-enabled formative assessment motivates students to engage consistently with their learning progress instead of cramming for tests. ⁵⁵ In this sense, AI functions as a metacognitive companion, reminding students of goals, tracking progress, and offering personalized guidance, supporting tasks teachers can rarely perform for every student in large classes. This underscores a reciprocal relationship: while effective AI use requires learners to reflect critically on feedback, ⁶³ AI systems simultaneously promote metacognition by fostering self-monitoring and reflection. ^{57 , 62 , 64} Integrating AI into science education, particularly chemistry, opens promising avenues to strengthen both SSD and SRL. Rather than merely automating processes, AI should prompt continuous learner reflection, making the integration of SSD, SRL, and AI a particularly powerful educational strategy.

2.4 Connection between self-regulated learning, socioscientific decision making and artificial

SRL and SSD empower students to make informed decisions and think critically about scientific and societal issues in our information-saturated and polarized world. Zimmerman’s (2000) ²⁵ phase model of SRL is applicable to evaluating science education, as these processes require systematic analysis, strategic action, and critical reflection. ^{14 , 65} Metacognition, a core component of SRL, supports the regulation and adaptation of cognitive processes, thus supporting reflective decision-making. ^{16 , 18 , 66} This is especially relevant in the context of SSI, where learners confront open-ended, controversial problems that demand the integration of scientific, ethical, and societal perspectives. ^{5 , 18} Studies show that combining metacognitive scaffolding with structured decision-making frameworks enhances learners’ ability to make reasoned, multi-perspective decisions. ^{14 , 16 , 20} Tools such as reflection phases, strategy prompts, and decision-making training help learners assess evidence and deliberate across perspectives. Furthermore, SRL plays an essential role in effective evaluation by fostering goal setting, monitoring, and reflection. ^{17 , 22 , 37}

Argumentation itself can be seen as a metacognitive strategy: learners justify claims, monitor their reasoning, and revise their positions. ⁶⁷ This regulation enables them to scrutinize arguments based on epistemic quality rather than surface plausibility, promoting deeper evaluation and fostering epistemic development. ^{49 , 67 , 68}

Despite this rich body of research, the synergy between SRL and SSD remains under-explored, especially in chemistry education. Empirical studies addressing their intersection primarily stem from biology or environmental science, and often treat SRL only implicitly or marginally. ^{18 , 20 , 37} Chemistry-specific SSI contexts, where students must metacognitively navigate ethical and scientific considerations, are still largely unexamined.

Compounding this gap is the limited use of adaptive metacognitive scaffolding, especially in connection with AI systems. While static metacognitive prompts and decision-making frameworks exist, ¹⁹ they lack personalization and fail to adapt to individual cognitive needs. Systems capable of tailoring feedback are yet to be systematically investigated for their potential to foster evaluation competence. Given the diversity of learners, this omission signals a critical research gap.

AI technologies hold significant promise here. They can support SRL and SSD through personalized feedback, automated reasoning analysis, and reflection prompts. ^{51 , 54} By making thought processes explicit, AI fosters deeper self-reflection, an essential step toward autonomous socioscientific reasoning. ^{62 , 64} Thus, the integration of SRL, SSD, and AI opens new pathways to foster critical, reflective science learning, an imperative for contemporary chemistry education. ²⁴

3.1 Research objective and research question

SRL Although the theory of SSD and SRL has been well researched, there is currently a lack of practice-oriented concepts in chemistry education that promote both competencies in a targeted and integrated manner. An approach combining structured decision-making processes with SRL’s metacognitive strategies appears particularly promising. While SSD focuses on the ability to analyse scientific information systematically, consider different perspectives and make reflective decisions, ¹⁸ SRL could provide the tools to independently control and critically evaluate this process. ^{17 , 36} Several studies demonstrate that metacognitive strategies, such as planning, monitoring and reflecting on decision-making processes, can significantly improve the quality of evaluations and argumentation. ^{16 , 20 , 49} Thus, the literature suggests that promoting both competencies together could generate synergy effects that go beyond training in SSD or SRL separately.

This study addresses a dual gap by examining, for the first time, how SSD and SRL can be promoted together in chemistry lessons through an authentic scenario of SRL intervention (SSI) and adaptive, AI-supported feedback. Not only does this transfer an established approach (linking strategy knowledge, reflection, and SRL; see Gresch et al.; Miller and Byrnes ^{14 , 22} ) to a new subject, it also tests an innovative technology for metacognitive personalization.

Against this background, the present study aims to empirically investigate the mutual promotion of SSD and SRL in chemistry classes. Specifically, the central research question is: ‘To what extent can the implicit promotion of SRL in an evaluation scenario contribute to the simultaneous development of students’ perceived (task-related) socioscientific decision making and self-regulation competence?´

To answer this, a teaching module on the topic of ‘Cigarettes and Vapes’ was developed, integrating implicit SRL elements (e.g. metacognitive guiding questions and reflection prompts) into a structured evaluation process to promote SSD and SRL in learners simultaneously. The following section presents this teaching setting in detail, focusing on how the SRL components are embedded to support the evaluation process.

3.2 Study design

This intervention is based on the finding that metacognitive instruction can improve learning processes in the long term. Kramarski et al. (2002) demonstrated that targeted metacognitive support strengthens problem-solving behaviour and encourages the deliberate use of strategies. While these findings originate from a mathematical context, the principles – particularly the promotion of reflection and planned thinking – can also be applied to science education and the development of SSD. ⁶⁹

Alongside the integration of critical reflection questions that structure the entire evaluation process, this intervention’s central innovative element is the use of an AI-supported adaptive feedback tool. This tool supports students individually in their argumentation process, aiming to improve the content and structural quality of their arguments. It provides adaptive suggestions for improvement and encourages reflection on one’s own thinking through metacognitive stimuli.

The reflection questions go beyond general prompts for reflection and are designed to examine the structure of arguments in depth: They encourage learners to evaluate the quality of their evidence, identify potential counterarguments and critically examine the coherence of their argumentation. ^{14 , 49} In doing so, the concept builds on the work of Yu and Zenker (2020), who demonstrate that critical questions (CQs) are vital for the systematic evaluation of arguments. CQs reveal the weaknesses of arguments and help learners to identify opportunities for improvement. ⁷⁰

By integrating metacognitive support, critical reflection questions and AI-supported feedback in this targeted way, learners can independently control and improve their evaluation processes – a key competence that is often neglected in chemistry classes.

The intervention design follows an experimental pre-post comparison with three groups:

1. Control group (CG): No intervention.

2. Experimental group without AI support (EG-NO): Evaluation unit without adaptive feedback and metacognitive guiding questions.

3. Experimental group with AI support (EG-KM): Evaluation unit with adaptive AI feedback and metacognitive guiding questions.

3.3 Information on the participants

The sample included 50 chemistry pupils, mainly from grammar school (grades 10–12) at German schools. The group composition is shown in Table 1.

The interventions took place in the student laboratory of the RPTU Kaiserslautern-Landau and lasted approximately 6 h. The data was collected by the Technical University of Kaiserslautern-Landau (RPTU) in Rhineland-Palatinate on a completely anonymous basis and with the voluntary consent of the participants. The participants were given comprehensive information and could withdraw their consent at any time. This university’s scientific research requires general approval and notification to the Trier Supervisory and Service Directorate (ADD), which was complied with in the context of this study. The evaluation process was structured based on the PAAWD^R model and its content was adapted. ⁴⁷

3.4 Instruments

To answer the research question, a pre-post comparison of the mean values of the SRL and SSD questionnaires was carried out. These instruments offer insight into students’ perceived awareness of their SRL and SSD processes.

The SSD questionnaire was developed based on established decision-making instruments, such as the Melbourne Decision-Making Questionnaire ⁷¹ and the Decision-Making Questionnaire by Sanz de Acedo, Lizarraga et al. (2009) ⁷² These instruments provide a comprehensive framework for analyzing the key factors that influence decision-making. Additionally, the integration of metacognitive elements follows Miller and Byrnes’s (2001) approach, which emphasizes the role of self-regulated learning in social decision-making situations. ²² The instrument aligns with the six evaluation dimensions of the PAAWD^R model (Perceiving, Analyzing, Arguing, Weighing, Deciding, and Reflecting) ⁴⁷ and has been adapted to specific assessment steps in science education, as described by Langlet et al. (2022). To ensure contextual relevance in a school setting, all literature-based items were modified and translated from English into German. Accordingly, the example items in Table 2 are author-generated translations and adaptations, tailored linguistically and contextually, for use in the present study. The items were rated on a Likert scale (1–4).

The internal consistency is within an acceptable range with Cronbach’s alpha > 0.7. The exact values, as well as a summary of case processing, can be found in Table 2:

The SRL questionnaire is based on eslished instruments ^{35 , 73} and measures cognitive, metacognitive and motivational components of SRL and is structured according to the three phases of SRL according to Zimmerman (2000) ²⁵ (forethought phase, performance phase, self-reflection phase). The questionnaire includes scales to assess cognitive, metacognitive and motivational aspects of SRL.

Items on elaborative and task processing learning strategies assess the cognitive component; items on goal setting, strategy and time planning, self-recording, self-evaluation and adaptive self-regulation assess the metacognitive component; and items on attention focusing, goal orientation, procrastination (reverse coded), intrinsic motivation, self-efficacy and causal attribution assess the motivational component.

These items capture the three phases of SRL and are rated on a Likert scale from 1 (disagree) to 4 (strongly agree). As the questionnaire contains items derived from validated instruments for assessing self-regulation, it has a high level of construct validity.

To determine internal consistency, Cronbach’s alpha was calculated for each subscale. ^{74 , 75}

The reliability values of the scales are within an acceptable range with Cronbach’s alpha > 0.65. ⁷⁵ The exact values and the summary of case processing can be found in Table 3:

The following is a methodological and didactic approach to the six steps of the PAAWD^R model, presented using the teaching example of ‘Sustainability of Vapes’. The study director led the teaching unit, which was structured as a question-based, evolving classroom discussion. During the discussion phases, the experimental group worked individually on tasks supplemented with implicit SRL cues. These cues took the form of metacognitive guiding questions that were highlighted in color and could be used as optional prompts for support. These critical prompt questions aimed to encourage learners to reflect on the task and their thought processes, thereby promoting self-regulation in the learning process. A more detailed description of the methodological implementation and all materials used has been provided in the Supplementary Information (SI).

3.5 Development of the instructional setup

A provocative video on vaping initiates a multi-phase teaching unit fostering scientific reasoning and sustainability education. Students first analyse personal motivations for vaping, deriving subjective and objective analysis criteria (e.g., reusability, health risks, raw materials) which are then structured along the three dimensions of sustainability. Twelve chemistry experiments deepen understanding of core issues such as toxic substances, particulate matter, battery components, and recycling. Based on these findings, students formulate and refine pro/con arguments using a Toulmin-based model. The AI tool, integrated into a custom web interface, supports argument development through adaptive feedback, metacognitive prompts, and confidence evaluations. Students then prioritize criteria, apply decision-making strategies (compensatory and non-compensatory), and make evidence-based decisions. In a simulated parliamentary setting, they present structured pleas and engage in peer evaluation. The process concludes with individual reflection on decision shifts and metacognitive growth, highlighting increased critical awareness and argumentation skills. The learning unit is described in more detail in the Supplementary Information (SI).