Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review
Open accessThe DOI matches its Crossref record; 41 of the 41 references it cites were found in a registry or an index; 12 of the 25 sources its claims cite were checked for support.
Abstract
Generative artificial intelligence (GenAI) is increasingly integrated into higher education, where it supports writing, feedback, problem solving, and research-related tasks while also raising concerns about cognitive offloading and learner dependence. This scoping review maps the literature on the relationships among GenAI, cognitive offloading, and learner agency in higher education. Following PRISMA-ScR guidance, searches were conducted on 24 June 2026 across six database sources: Web of Science Core Collection, Scopus, EBSCOhost ERIC, EBSCOhost Education Source, APA PsycInfo via ProQuest, and ProQuest Education Collection via ProQuest Social Science Premium Collection. English-language peer-reviewed or scholarly journal literature published from 2022 to 24 June 2026 was eligible for inclusion. Data were charted using a structured extraction matrix and synthesized through descriptive mapping and theory-informed configurative synthesis. After duplicate removal, 5327 records were screened by title and abstract, 691 full-text reports were assessed, and 123 studies were included in the final charting matrix and synthesis. The review shows that learner agency is conceptualized as a multidimensional construct involving self-regulation, reflective judgement, intentionality, and responsible action. Across the literature, GenAI is best interpreted through an interpretive dual-pattern account: agency-supportive patterns are associated with self-regulated learning, self-efficacy, feedback literacy, and reflective engagement, whereas agency-eroding patterns are associated with cognitive offloading, overreliance, dependence, uncritical uptake, and weakened judgement. These patterns should be interpreted as a configurative synthesis of heterogeneous evidence rather than as causal effect estimates. Overall, the findings suggest that the educational value of GenAI depends less on the technology itself than on how it is pedagogically embedded, with augmentation-oriented and scaffolded use being more supportive of learner agency than replacement-oriented use.
Overview
CoChat Summary
TL;DR
This PRISMA-ScR scoping review maps 123 studies on generative AI, cognitive offloading, and learner agency in higher education, arguing that GenAI’s educational value depends less on the technology than on pedagogical embedding. It proposes an interpretive dual-pattern account: agency-supportive uses linked to self-regulation, self-efficacy, feedback literacy, and reflection, versus agency-eroding uses linked to offloading, overreliance, dependence, and weakened judgement.
Key findings
- Across 123 included studies, learner agency is conceptualized as multidimensional—self-regulation, reflective judgement, intentionality, and responsible action—with L. Xia et al. (2025) proposing key abilities, active actions, and essential mental characteristics, and Yang et al. (2024) distinguishing receptive, resistive, resourceful, and reflective engagement.
- Theme-family counts show 118 studies contributed to agency-supportive patterns, 108 to agency-eroding/offloading patterns, 116 to regulatory/evaluative mechanism-related constructs, 86 to learner agency/autonomy conceptualization, and 121 recorded agency/autonomy construct signals.
- Recurring mechanism-related constructs include self-regulation, metacognition, critical thinking, self-efficacy, trust, feedback literacy, AI literacy, and cognitive offloading/reliance/dependence; self-efficacy is described as ambivalent because it can accompany dependence or “false self-efficacy,” and trust is framed as needing calibration rather than being uniformly desirable.
- Boundary conditions reported as supporting augmentation over replacement include scaffolded or guidance-based GenAI use, metacognitive prompts, reflection journals, AI/feedback literacy development, and continued teacher/human mediation.
Methods
- Scoping review following PRISMA-ScR guidance, registered in INPLASY (INPLASY202650152); searches conducted 24 June 2026 across Web of Science Core Collection, Scopus, EBSCOhost ERIC, EBSCOhost Education Source, APA PsycInfo via ProQuest, and ProQuest Education Collection via ProQuest Social Science Premium Collection.
- Eligible literature was English-language peer-reviewed or scholarly journal literature published from 2022 to 24 June 2026; screening identified 8020 records, removed 2693 duplicates, screened 5327 titles/abstracts, sought 717 reports, could not retrieve 26, assessed 691 full texts, and included 123 studies.
- Three authors screened with fourth-author adjudication; agreement statistics included Fleiss’ kappa = 0.7103 at title/abstract and 0.4884 at full text. Data were charted with a structured extraction matrix and synthesized through descriptive mapping and theory-informed configurative synthesis, not meta-analysis or causal modeling.
Limitations
- The paper states that it focused on English-language peer-reviewed or scholarly journal literature from 2022 to 24 June 2026, so other languages, books, theses, conference proceedings, institutional reports, and grey literature may be missed; it also excluded K-12, teacher/faculty adoption, institutional implementation, technical system development, and clinical populations unless directly relevant to student learning processes.
- The paper states that the included literature is methodologically heterogeneous, no formal critical appraisal or risk-of-bias assessment was conducted, and many included studies used cross-sectional self-report data; it also notes concentration in language-learning, writing, and Chinese/EFL higher education settings, limiting generalizability.
- The paper states that 26 of 717 sought reports could not be retrieved, and that the GenAI search block emphasized ChatGPT, generative AI, GenAI, large language model, LLM, GPT, GPT-3, and GPT-4 terms, so studies indexed only under terms such as Gemini, Bard, Claude, Copilot, AI chatbot, conversational agent, AI assistant, or foundation model may have been missed.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures


Verification
Claims the paper makes
8 Valid · 1 Not valid · 26 Not checked · 177 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“These studies reinforce the idea that overreliance is better understood as a patterned behavioral and psychological response than as mere frequency of use”
§ Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review > 2. Conceptual Background > 2.2. Cognitive Offloading, Overreliance, and Dependence
What the sources say
- Supportedcites Hou et al., 2025 source
The Introduction (p. 3) describes reliance as patterns of engagement and argues that counting acceptance of AI recommendations does not capture the complexity of problem-solving. The questionnaire description (p. 5) distinguishes reflective, cautious, collaborative and thoughtless use, while the Discussion (p. 15) emphasizes the reasoning processes behind those behaviours. This supports the claim that overreliance is more than mere frequency of use. (LLM, conf=0.88)
- Not checked — no full text of the sourcecites Zheng & Wang, 2026 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Studies on pedagogical partnerships, research competence, and writing support indicate that the educational value of GenAI depends on whether students remain the active agents of meaning making rather than outsourcing that role to the system (Wang & Zhang, 2026; Zhu & Yang, 2026).”
§ Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review > 6. Implications > 6.1. Pedagogical Implications
What the sources say
- Supportedcites Wang & Zhang, 2026 source
The discussion of mediating pathways (p. 7) distinguishes strategic delegation that maintains student agency and engagement with complex synthesis from intellectual outsourcing. The Discussion (p. 23) reports that such delegation can free cognitive resources for higher-order reflection, supporting the claim that students must remain engaged in meaning-making even when they offload tasks. (LLM, conf=0.90)
- Supportedcites Zhu & Yang, 2026 source
The Discussion’s Practical Implications section (§5.2.1) recommends critical assessment of AI output rather than using it as a substitute for independent thinking. It also warns that over-reliance can undermine student autonomy, supporting the citing sentence’s paraphrase about students retaining an active role. (LLM, conf=0.90)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Conversely, augmentation-oriented pedagogy can be understood as a way of keeping technological cognition active by requiring students to inspect prompts, compare outputs, justify revisions, and explain why AI suggestions are accepted or rejected (Bluet et al., 2025; Federico et al., 2026).”
§ Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review > 5. Discussion
What the sources say
- Partly supportedcites Bluet et al., 2025 sourceExperimental — accuracy not yet measured
The Discussion (page not provided) reports that watching tool-making recruits the technical-reasoning network and proposes that teaching cues focus learners’ attention on important steps, potentially strengthening their reasoning. That supports the general idea of keeping technical reasoning engaged, but the paper does not study augmentation-oriented pedagogy, students’ use of AI, or the listed practices of inspecting prompts, comparing outputs, and justifying AI-related decisions. (LLM, conf=0.84)
- Not checked — no full text of the sourcecites Federico et al., 2026 source
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Self-regulation recurs across the reviewed literature as a key construct linking GenAI use with agency-related outcomes.”
§ Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review > 2. Conceptual Background > 2.3. Self-Regulation as a Recurring Linking Construct
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The updated evidence base consisted of 123 included studies that met the full eligibility criteria and had a substantive connection to GenAI, higher education learners, and learner agency, cognitive offloading, reliance/dependence, or related mechanisms.”
§ Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review > 4. Results
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
119 more claims
Create a free account to see all 212 claims and every source verdict.
- It may state two separate claims as one. 51
- Probably not a substantial claim. 55
Citations
Cites
PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation
Tricco, A. C., Lillie, E., Zarin, W., O'Brien, K. K., Colquhoun, H., Levac, D., Moher, D., Peters, M. D. J., Horsley, T., Weeks, L., et al. (2018) · Annals of Internal Medicine · doi:10.7326/m18-0850 · Cited by 43,565 · cited 1×
“The review process was further informed by the PRISMA-ScR reporting framework to ensure transparency in searching, screening, data charting, and synthesis (Tricco et al., 2018); the completed PRISMA-ScR checklist is provided in Appendix A.”
Scoping studies: towards a methodological framework
Arksey, H., & O'Malley, L. (2005) · International Journal of Social Research Methodology · doi:10.1080/1364557032000119616 · Cited by 36,742 · cited 1×
“This approach is consistent with the foundational rationale of scoping reviews as a method for mapping key concepts, research trends, and knowledge gaps in developing fields (Arksey & O’Malley, 2005; Levac et al., 2010; Munn et al., 2018).”
Scoping studies: advancing the methodology
Levac, D., Colquhoun, H., & O'Brien, K. K. (2010) · Implementation Science · doi:10.1186/1748-5908-5-69 · Cited by 15,268 · cited 1×
“This approach is consistent with the foundational rationale of scoping reviews as a method for mapping key concepts, research trends, and knowledge gaps in developing fields (Arksey & O’Malley, 2005; Levac et al., 2010; Munn et al., 2018).”
Munn, Z., Peters, M. D. J., Stern, C., Tufanaru, C., McArthur, A., & Aromataris, E. (2018) · BMC Medical Research Methodology · doi:10.1186/s12874-018-0611-x · Cited by 11,385 · cited 1×
“This approach is consistent with the foundational rationale of scoping reviews as a method for mapping key concepts, research trends, and knowledge gaps in developing fields (Arksey & O’Malley, 2005; Levac et al., 2010; Munn et al., 2018).”
Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., Shen, Y., Li, X., & Gašević, D. (2025) · British Journal of Educational Technology · doi:10.1111/bjet.13544 · Cited by 805 · cited 1×
“Together, these studies indicate that metacognition shapes whether AI outputs are treated as raw material for thinking or as authoritative content to be adopted with minimal scrutiny (Teng, 2025; Urban et al., 2025; Yao et al., 2025; Fan et al., 2025).”
Xu, X., Qiao, L., Cheng, N., Liu, H., & Zhao, W. (2025) · British Journal of Educational Technology · doi:10.1111/bjet.13599 · Cited by 359 · cited 12×
“These findings indicate that self-regulation is not merely an additional benefit of AI use; it is one of the conditions under which GenAI becomes educationally valuable in the first place (Xu et al., 2025; Teng, 2025).”
InfluentialLee, H.-Y., Chen, P.-H., Wang, W.-S., Huang, Y.-M., & Wu, T.-T. (2024) · International Journal of Educational Technology in Higher Education · doi:10.1186/s41239-024-00447-4 · Cited by 273 · cited 1×
“These approaches keep students cognitively engaged and make it more difficult for them to drift into passive acceptance or superficial completion (Lee et al., 2024; Joo et al., 2026; Lukešová & Jennings, 2025; Sun et al., 2026; R. Zhou et al., 2025).”
Self-citationZhang, L., & Xu, J. (2025) · The Internet and Higher Education · doi:10.1016/j.iheduc.2024.100978 · Cited by 168 · cited 5×
“L. Zhang and Xu (2025) found that more frequent GenAI use could simultaneously enhance students’ confidence and efficiency while intensifying technological dependence.”
InfluentialZhan, Y., & Yan, Z. (2026) · Assessment & Evaluation in Higher Education · doi:10.1080/02602938.2025.2471821 · Cited by 162 · cited 7×
“From an agency perspective, these findings are important because they suggest that GenAI can support not only task performance but also students’ capacity to engage with feedback more deliberately and productively (C. Zhou & Wang, 2026; Zhan & Yan, 2025).”
InfluentialYang, Y., Luo, J., Yang, M., Yang, R., & Chen, J. (2024) · Studies in Higher Education · doi:10.1080/03075079.2024.2327003 · Cited by 129 · cited 5×
“In this respect, learner agency is best seen as a continuum of increasingly active and reflective participation rather than as a binary attribute that learners either possess or lack (Jiang et al., 2026; Lai et al., 2026; Yang et al., 2024).”
Influential
Cited by
Wei, J., & Shang, Y. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1965371 · Cited by 1
“…thinking, cognitive offloading, learner agency, psychological outcomes, and epistemic autonomy, generally reporting conditional effects (Alubthane, 2026; Balart et al., 2026; Li et al., 2026; Ni and Li, 2026; 10.3389/fpsyg.2026.1965371 Suazo et al., 2026; Wang et al., 2026; Zhai et al., 2024).”
InfluentialAI-powered student support: a conceptual paper on ethical AI in education
Νικολοπούλου, Κ. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1801972 · Cited by 1
GenAI-Assisted Design Education: Task Appraisals, Course-Value Reaffirmation, and Responsible UsePreprint
Song, Y. (2026) · Research Square · doi:10.21203/rs.3.rs-10611624/v1
Fan, M., & Chang, P. (2026) · Behavioral Sciences · doi:10.3390/bs16091671 · Cited by 6
“The same move has been made independently on engagement outcomes: Y. Wang et al. (2026) separated reflective from thoughtless AI use and argued that research has attended to how much students use AI at the expense of how they use it.”
Psychological Distress and Artificial Intelligence Dependency: The Moderating Role of Extraversion
Yıldırım, Z. E. (2026) · Current Perspectives in Social Sciences · doi:10.53487/atasobed.1920593
Metadata
- Authors
- Wang, G., Wang, W., Yang, D., & Ren, J.
- Affiliations
- Tsinghua University (CN)
- Harbin Institute of Technology (CN)
- Hubei University Of Economics (CN)
- Venue
- Behavioral Sciences
- Publisher
- MDPI AG
- Published
- 2026-07-08
- Volume
- 16
- Issue
- 7
- Pages
- 1150
- ISSN
- 2076-328X (ISSN-L)
- PMC
- PMC13405335
- OpenAlex
- W7167731096
- Semantic Scholar
- e5dc0be82e088dc7c5d627a05c57669df16f9755
- Cited by
- 20
- References
- 40
- Citations by year
- 2026: 20
- FWCI
- 34.59
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Behavioral Sciences (publishedVersion, cc-by)
- Preprints.org (acceptedVersion, cc-by)
- PubMed (publishedVersion)
- PubMed Central (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (acceptedVersion)
- Topics
- Social Sciences › Education › Artificial Intelligence in Education
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Social Sciences › Psychology › Developmental and Educational Psychology › Innovative Teaching and Learning Methods
- Keywords
- generative AI, cognitive offloading, learner agency, higher education, student self-regulation, self-efficacy, feedback literacy, overreliance, reflective judgment, AI-assisted writing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review. https://app.cochat.ai/papers/10.3390/bs16071150
- The paper
- Wang, G., Wang, W., Yang, D., & Ren, J. (2026). Generative AI, Cognitive Offloading, and Learner Agency in Higher Education: A Scoping Review. Behavioral Sciences. https://doi.org/10.3390/bs16071150
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.