The DOI matches its Crossref record; 104 of the 104 references it cites were found in a registry or an index; 25 of the 110 sources its claims cite were checked for support.
The Cognitive Architecture of Digital Externalization
Open accessAbstract
This review is aimed at synthesizing current findings concerning technology-based cognitive offloading and the associated effects on learning and memory. While cognitive externalization (i.e., using the environment to outsource mental computation) is a highly useful technique in various problem-solving tasks, a growing body of research suggests that the offloading of information into the environment (and digital storage in particular) can have negative effects on learning... As a result, in many instances, only the gist of information (or its location) is retained, while learners often gain the illusion of having memorized that information.
Overview
CoChat Summary
TL;DR
This review argues that digital cognitive offloading is best understood through cognitive load theory: high intrinsic load, high extraneous load, low perceived/actual working memory capacity, and high perceived value of information shape when learners externalize information. It warns that digital externalization can help performance but also reduce memorization, create illusions of learning, and increase vulnerability to memory manipulation, with generative AI potentially amplifying these risks.
Key findings
- The paper’s model treats intrinsic cognitive load and extraneous cognitive load as central to digital externalization, with extraneous load acting as a “triple barrier”: it can prevent information from entering working memory, inhibit storage in long-term memory, and prevent externalization through technology.
- Offloading is more likely when learners have lower actual or perceived working memory capacity, when extraneous load is higher, when intrinsic load is higher, and when the perceived value of information is higher.
- The review reports that offloading often improves task performance but can lower memory performance; for example, in the reviewed studies, 4- to 11-year-old children and 11-year-olds using laptops offloaded more difficult items, and offloading could improve recall for some tasks while harming memorization in others.
- The paper states that offloading can produce an illusion of having learned and can increase susceptibility to memory manipulation; it also notes that current generative AI may risk weakening acquisition of biologically secondary knowledge and essential skills.
Methods
- This is a review article synthesizing current research on technology-based cognitive offloading, learning, and memory, and it develops a conceptual model from the reviewed literature.
- The paper draws on studies of non-digital and digital offloading, including children’s offloading behavior, Internet/search use, smartphone-based retrieval, and AI-related externalization.
- The main analysis is theoretical synthesis: it integrates findings with cognitive load theory, biologically primary vs. secondary knowledge, metacognition, and value-based memory models to propose a “cognitive architecture of digital externalization.”
Limitations
- The paper does not report a new empirical dataset or original experiment; it is a review and model paper, so its conclusions depend on the studies it summarizes.
- The author explicitly notes that more research is necessary to determine the positive and negative effects of the Internet and AI on cognition and learning, and to test the proposed model and hypotheses.
- Evident from the text: several claims about generative AI, long-term developmental effects, and “biological pointers” are presented as hypotheses or future research directions, not established findings.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
25 Valid · 1 Not valid · 70 Not checked · 39 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“educational computer games ('serious games') often require learners to internalize the rules of such a game, thereby causing extraneous cognitive load”
What the sources say
- Partly supportedcites Skulmowski & Xu, 2022 sourceExperimental — accuracy not yet measured
In “Challenge 1: Interactive Learning Media” (page unavailable), the paper says that learning a serious game's rules can induce unnecessary cognitive load. It describes this as a potential drawback, however, rather than establishing that games often cause extraneous load in this way. (LLM, conf=0.84)
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
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“with several studies on issues besides learning and memory often showing considerable advantages accrued using this strategy”
What the sources say
- Not matchedcites Armitage etal., 2020
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Partly supportedcites Dunn & Risko, 2016 sourceExperimental — accuracy not yet measured
Section 4.3 reports faster and more accurate performance when participants were forced to physically rotate toward the text, supporting an advantage of non-digital offloading outside a learning-and-memory task. But Section 3.3 says spontaneous rotation showed no observable performance benefit, and the other favorable example in Section 1.2 uses an on-screen calculator. The paper therefore does not establish that several studies show considerable advantages from non-digital externalization. (LLM, conf=0.84)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
Every source the paper cites for this claim was checked and supports it.
In the paper
“not receiving temporal costs for offloading also reduced memory performance.”
What the sources say
- Supportedcites Grinschgl, Papenmeier, and Meyerhoff (2021a) source
In Experiment 1 > Results (page unavailable), participants in the no-lockout condition offloaded more and performed less accurately on the subsequent memory test than those facing a lockout. The paper interprets this as a negative effect of greater offloading on memory accuracy. (LLM, conf=0.93)
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.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“research on non-digital forms of externalization is used to demonstrate how deeply embedded the tendency of offloading is in humans”
What the sources say
- Not matchedcites Armitage etal., 2020
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Supportedcites Dunn & Risko, 2016 source
The Introduction describes cognitive offloading through physical actions and manipulation of external artifacts as prevalent. Section 1.2 says people readily engage spontaneously in external normalization, a non-digital form of offloading. Together, these support the citing sentence’s broad characterization. (LLM, conf=0.88)
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
Every source the paper cites for this claim was checked and supports it.
In the paper
“humans have a set of certain innate capabilities, such as being able to speak or to recognize faces, that do not require explicit instruction”
What the sources say
- Supportedcites Paas & Sweller, 2012 source
The Abstract identifies speaking and face recognition as abilities learned without explicit instruction. It also describes biologically primary knowledge as something humans have evolved to acquire. (LLM, conf=0.95)
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
89 more claims · 13 with their cited sources checked
Create a free account to see all 135 claims and to verify the 1 reference still open.
- It may state two separate claims as one. 12
- Probably not a substantial claim. 32
Citations
Cites
Advances in Neural Information Processing Systems 19
(2007) · The MIT Press eBooks · doi:10.52202/085713 · Cited by 15,539
Denoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712
ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education
Kasneci, E., Seßler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., et al. (2023) · doi:10.35542/osf.io/5er8f · Cited by 7,552
ChatGPT for good? On opportunities and challenges of large language models for education
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., et al. (2023) · Learning and Individual Differences · doi:10.1016/j.lindif.2023.102274 · Cited by 7,326 · cited 1×
“While there are many potentials for this technology in learning and instruction (for an overview, see Kasneci et al., 2023) and productivity in general (Noy & Zhang, 2023), the drastic change in everyday life these tools are likely to introduce can hardly be underestimated.”
BackgroundBarsalou, L. W. (1999) · Behavioral and Brain Sciences · doi:10.1017/s0140525x99002149 · Cited by 6,887
Clark, A., & Chalmers, D. (1998) · Analysis · doi:10.1093/analys/58.1.7 · Cited by 6,333 · cited 3×
“Instead, the cognitive system is thought to include bodily resources as well as environmental artifacts (Clark & Chalmers, 1998; Hollnagel, 2001).”
BackgroundCognitive Architecture and Instructional Design
Sweller, J., van Merrienboer, J. J. G., & Paas, F. G. W. C. (1998) · Educational Psychology Review · doi:10.1023/a:1022193728205 · Cited by 6,023 · cited 6×
“One of the most influential theories of instruction is cognitive load theory (CLT; Sweller et al., 1998, 2019).”
InfluentialPhotorealistic Text-to-Image Diffusion Models with Deep Language Understanding
Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E., Ghasemipour, S. K. S., Ayan, B. K., Mahdavi, S. S., Lopes, R. G., et al. (2022) · arXiv · doi:10.48550/arxiv.2205.11487 · Cited by 2,528 · cited 1×
“These two principles share striking similarities with generative AI starting with random spots that are refined using a model derived from training with large image datasets (e.g., Saharia et al., 2022).”
Cognitive Architecture and Instructional Design: 20 Years Later
Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019) · Educational Psychology Review · doi:10.1007/s10648-019-09465-5 · Cited by 2,337 · cited 6×
“One of the most influential theories of instruction is cognitive load theory (CLT; Sweller et al., 1998, 2019).”
BackgroundMethodInfluentialCarey, S. (2009) · doi:10.1093/acprof:oso/9780195367638.001.0001 · Cited by 2,322
Cited by
Gkintoni, E., Antonopoulou, H., Sortwell, A., & Halkiopoulos, C. (2025) · Brain Sciences · doi:10.3390/brainsci15020203 · Cited by 309
Chen, A., Xiang, M., Zhou, J., Jia, J., Shang, J., Li, X., Gašević, D., & Fan, Y. (2025) · Computers & Education · doi:10.1016/j.compedu.2024.105198 · Cited by 115
Cognitive offloading or cognitive overload? How AI alters the mental architecture of coping
Chirayath, G., Premamalini, K., & Joseph, J. (2025) · Frontiers in Psychology · doi:10.3389/fpsyg.2025.1699320 · Cited by 58
“On the other hand, the same tools may create cognitive overload: an erosion of introspection, over-reliance on algorithmic feedback, and anxiety induced by hyper-monitoring and optimization (Grinschgl et al., 2021; Skulmowski, 2023).”
Placebo or Assistant? Generative AI Between Externalization and Anthropomorphization
Skulmowski, A. (2024) · Educational Psychology Review · doi:10.1007/s10648-024-09894-x · Cited by 65
“While the aspect of whether users perceive AIs as human-like partners will be discussed in a later section, a cognitive explanation for not giving credit to AIs could be that humans perceive generative AIs merely as a tool to offload parts of their work (Skulmowski, 2023).”
InfluentialSelf-citationSkulmowski, A. (2024) · Educational Psychology Review · doi:10.1007/s10648-024-09869-y · Cited by 110
“…focusing on touchscreen activities and gestures) and using artificial intelligence to externalize demanding aspects of complex activities (e.g., Wahn et al., 2023; for an overview of digital externalization, see Skulmowski, 2023) to leave more cognitive resources for the actual learning process.”
Self-citation
Metadata
- Authors
- Skulmowski, A.
- Affiliations
- Karlsruhe University of Education (DE)
- Corresponding author
- Skulmowski, Alexander
- Venue
- Educational Psychology Review
- Published
- 2023-10-10
- Volume
- 35
- Issue
- 4
- ISSN
- 1040-726X (ISSN-L)
- 1573-336X
- OpenAlex
- W4387477515
- Cited by
- 76
- References
- 94
- Citations by year
- 2026: 46 · 2025: 18 · 2024: 12
- FWCI
- 19.59
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Educational Psychology Review (publishedVersion, cc-by)
- Funders
- Pädagogische Hochschule Karlsruhe
- Topics
- Social Sciences › Psychology › Experimental and Cognitive Psychology › Visual and Cognitive Learning Processes
- Social Sciences › Psychology › Developmental and Educational Psychology › Innovative Teaching and Learning Methods
- Life Sciences › Neuroscience › Cognitive Neuroscience › Memory Processes and Influences
- Keywords
- cognitive offloading, cognitive load, intrinsic cognitive load, extraneous cognitive load, working memory capacity, long-term memory, learning, memory manipulation, generative AI, technology-enhanced learning, value of information
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 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (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: The Cognitive Architecture of Digital Externalization. https://app.cochat.ai/papers/10.1007/s10648-023-09818-1
- The paper
- Skulmowski, A. (2023). The Cognitive Architecture of Digital Externalization. Educational Psychology Review. https://doi.org/10.1007/s10648-023-09818-1
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.
