CITATION — REFERENCE ENTRY
Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task — arXiv
- Key
- kosmyna-et-al-2025-brain-chatgpt
- Authors
- Kosmyna, Nataliya; Hauptmann, Eugene; Yuan, Ye Tong; Situ, Jessica; Liao, Xian-Hao; Beresnitzky, Ashly Vivian; Braunstein, Iris; Maes, Pattie
- Issued
- 2025-6
- Type
- article
- Container
- arXiv
Raw CSL JSON
{
"DOI": "10.48550/arXiv.2506.08872",
"URL": "https://arxiv.org/abs/2506.08872",
"type": "article",
"title": "Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task",
"author": [
{
"given": "Nataliya",
"family": "Kosmyna"
},
{
"given": "Eugene",
"family": "Hauptmann"
},
{
"given": "Ye Tong",
"family": "Yuan"
},
{
"given": "Jessica",
"family": "Situ"
},
{
"given": "Xian-Hao",
"family": "Liao"
},
{
"given": "Ashly Vivian",
"family": "Beresnitzky"
},
{
"given": "Iris",
"family": "Braunstein"
},
{
"given": "Pattie",
"family": "Maes"
}
],
"issued": {
"date-parts": [
[
2025,
6
]
]
},
"number": "2506.08872",
"container-title": "arXiv"
}
Claims
-
Kosmyna and colleagues describe cognitive debt as a condition in which repeated reliance on external systems such as LLMs replaces the effortful cognitive processes needed for independent thinking, deferring mental effort in the short term at the price of long-term costs such as diminished critical inquiry, increased vulnerability to manipulation, and decreased creativity.
"This pattern reflects the accumulation of cognitive debt, a condition in which repeated reliance on external systems like LLMs replaces the effortful cognitive processes required for independent thinking. Cognitive debt defers mental effort in the short term but results in long-term costs, such as diminished critical inquiry, increased vulnerability to manipulation, decreased creativity. When participants reproduce suggestions without evaluating their accuracy or relevance, they not only forfeit ownership of the ideas but also risk internalizing shallow or biased perspectives."
-
EEG connectivity was strongest in the brain-only group, intermediate in the search engine group, and weakest in the LLM group, which showed up to 55% lower total connectivity (dDTF magnitude) than the brain-only group in some frequency bands.
"EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. [...] The LLM group showed the least extensive connectivity, with up to 55% reduced total dDTF magnitude compared to the Brain-Only group in low-frequency semantic and monitoring networks."
-
The authors conclude that their findings support an educational model that delays AI integration until learners have engaged in sufficient self-driven cognitive effort.
"Taken together, these findings support an educational model that delays AI integration until learners have engaged in sufficient self-driven cognitive effort. Such an approach may promote both immediate tool efficacy and lasting cognitive autonomy."
-
In an EEG study of 54 participants across three sessions writing essays with a large language model, with a search engine, or with no tools, the LLM group showed the weakest brain connectivity; an additional fourth session with role reversal involved 18 participants.
"Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed three sessions under the same condition. In a fourth session, LLM users were reassigned to Brain-only group (LLM-to-Brain), and Brain-only users were reassigned to LLM condition (Brain-to-LLM). A total of 54 participants took part in Sessions 1-3, with 18 completing session 4... EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity."
-
Participants chose one of several SAT essay prompts and had 20 minutes to write. The LLM group could use only ChatGPT; the search engine group could use any website except LLMs; the brain-only group could use no websites or tools.
"The participants were instructed to pick a topic among the proposed prompts, and then to produce an essay based on the topic's assignment within a 20 minutes time limit. Depending on the participant's group assignment, the participants received additional instructions to follow: those in the LLM group (Group 1) were restricted to using only ChatGPT, and explicitly prohibited from visiting any websites or other LLM bots. [...] Search Engine group (Group 2) was allowed to use ANY website, except LLMs. The Brain-only group (Group 3) was not allowed to use any websites, online/offline tools or LLM bots"
-
The authors list as limitations a small participant pool from a few nearby institutions, use of ChatGPT only, no division of writing into subtasks, EEG analysis limited to connectivity, and findings specific to essay writing in an educational setting; they call for longitudinal studies. The paper is marked as a preprint under review.
"In this study we had a limited number of participants recruited from a specific geographical area, several large academic institutions, located very close to each other. [...] we cannot directly generalize the obtained results to other LLM models. [...] Our findings are context-dependent and are focused on writing an essay in an educational setting and may not generalize across tasks. Future studies should also consider exploring longitudinal impacts of tool usage on memory retention, creativity, and writing fluency."
-
Self-reported ownership of essays was lowest in the LLM group and highest in the brain-only group. In session 4, participants who switched from the LLM to writing unaided showed reduced alpha and beta connectivity, which the authors interpret as under-engagement.
"In session 4, LLM-to-Brain participants showed reduced alpha and beta connectivity, indicating under-engagement. Brain-to-LLM users exhibited higher memory recall and activation of occipito-parietal and prefrontal areas, similar to Search Engine users. Self-reported ownership of essays was the lowest in the LLM group and the highest in the Brain-only group. LLM users also struggled to accurately quote their own work."
-
The 54 analysed participants were aged 18 to 39 (mean 22.9) and were all recruited from five universities in the greater Boston area: MIT, Wellesley, Harvard, Tufts, and Northeastern.
"These 54 participants were between the ages of 18 to 39 years old (age M = 22.9, SD = 1.69) and all recruited from the following 5 universities in greater Boston area: MIT (14F, 5M), Wellesley (18F), Harvard (1N/A, 7M, 2 Non-Binary), Tufts (5M), and Northeastern (2M)"
-
In session 1, 15 of 18 LLM-group participants (83.3%) failed to provide a correct quotation from the essay they had just written, compared with 2 of 18 (11.1%) in each of the search engine and brain-only groups.
"In the LLM‑assisted group, 83.3 % of participants (15/18) failed to provide a correct quotation, whereas only 11.1 % (2/18) in both the Search‑Engine and Brain‑Only groups encountered the same difficulty."
-
Participants were divided into LLM, search engine, and brain-only groups and completed three essay-writing sessions under the same condition; in a fourth session, LLM users wrote without tools and brain-only users switched to the LLM. 54 people took part in sessions 1 to 3 and 18 completed session 4. The researchers recorded EEG during writing and analysed the essays with NLP, human teachers, and an AI judge.
"Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed three sessions under the same condition. In a fourth session, LLM users were reassigned to Brain-only group (LLM-to-Brain), and Brain-only users were reassigned to LLM condition (Brain-to-LLM). A total of 54 participants took part in Sessions 1-3, with 18 completing session 4. We used electroencephalography (EEG) to assess cognitive load during essay writing, and analyzed essays using NLP, as well as scoring essays with the help from human teachers and an AI judge."
Available in