🍎your Scholarly Digest 10th July, 2025

Academia essentials hand-picked fortnightly for the mindful scholar

Was this newsletter forwarded to you? Sign up to receive weekly letters rooted in curiosity, care, and connection.
Know someone who will enjoy The Scholarly Letter? Forward it to them.

All previous editions of The Letter are available on our website.

Online Thumbnail Credit: National Gallery of Art Open Access Collection; Reba and Dave Williams Collection, Gift of Reba and Dave Williams

Hi Scholar,

We’ve been overjoyed to receive your letters. Each one is beautifully written – some thought provoking, others emotionally stirring (some both) – in all the ways we hope our own words might be for you.

If you’re still waiting for a reply, know that it’s coming. We are slow. Yes, because we’re just two humans with never-ending to-do lists. But also because we like to sit with your words and take our time to craft a thoughtful response. To answer such care with a quick emoji or a glib “you got it” feels like missing the point entirely. 

Receiving your letters – so far sent in response to ones we initiated – has led us to wonder: would you like other scholars to read your letters too? You already know the kind of letters we share in this community. So if you have an essay or a Brain Food piece you’d like to publish in The Scholarly Letter, send it in. We’re always open to growing our circle of letter writers, not just readers.

Finally, we’re thinking about hosting our first ever in-person meet-up this August; scholars making it out of the letter, if you will. We’re based in the UK, so the gathering would likely be somewhere in England. If that sounds like something you’d love to be part of, reply to this email and let us know. We’ll get planning.

With care,
The Critic & The Tatler

BRAIN FOOD

More Than Findings: The Performativity of Research Studies

A preprint titled ‘Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task’ has been doing the rounds. You might have come across it under a different name, ‘New MIT study proves ChatGPT is making us dumber’ or something equally dramatic. The study has gone viral, with media headlines and social media posts hailing it as confirmation of their deepest suspicions: “Finally, proof of what I’ve always known!” And perhaps we should apologise for contributing to its virality here. But the kind of polarisation and blind certainty that virality provokes is exactly what makes the study worth looking at twice.

I’ll be frank with you, Scholar. I did not read all 206 pages of the preprint front to back. Like any reasonable scholar, I read what was relevant to me and my interests: the introduction, discussion, conclusion, and experimental design. It did take me 4 hours to do this, however. Normally, such a disclaimer would not be necessary. But given how much of the ‘controversy’ surrounding this study has stemmed from the way people engaged with it (or didn’t), such disclaimers feel like an essential place to begin. 

To give you a brief overview: the study tested three groups of participants, each asked to write three essays – one per session – within a 20-minute time limit. What differed between the groups was the tool (or lack thereof) they were allowed to use:

• Group 1 – LLM Group: Restricted to using only ChatGPT. No websites, no other LLM bots.
  

• Group 2 – Search Engine Group: Allowed to use any website, except LLMs.
  

• Group 3 – Brain-only Group: No websites, online/offline tools or LLM bots. They could only rely on their own knowledge and thoughts.

Each participant remained in the same group for the first three sessions. After these, another fourth and final session was conducted. Only 18 out of the initial 54 participants participated in this fourth session.  

In Session 4, participants were reassigned groups: those who had only used ChatGPT were now asked to write without any tools (LLM-to-Brain-only), and those who had worked unaided were given access to ChatGPT (Brain-only-to-LLM). Throughout, researchers measured brain activity and conducted post-assessment interviews.

The results reported in the study paint a layered picture of how different tools (or lack thereof) shape not just what we write, but how we think. 

Participants in the Brain-only group exhibited the strongest and most widespread neural connectivity, particularly in brain regions associated with semantic integration, creative ideation, and executive self-monitoring. In contrast, the Search Engine group showed moderate connectivity, with notable activation in visual processing areas, reflecting the active engagement required to scan, compare, and synthesise information from multiple sources. The LLM group, however, displayed the weakest overall brain activity, with up to 55% reduced neural connectivity compared to the Brain-only group. Their cognitive engagement was narrower and more procedural, suggesting a more passive integration of AI-generated text, rather than active construction. 

The most compelling insight, however, came during Session 4, when tools were switched. Participants who had relied on ChatGPT and were then asked to write unaided (LLM-to-Brain) showed some improvement in connectivity but they still fell short of the depth and integration reached by the Brain-only group. Conversely, those who had worked unaided and were then given access to ChatGPT (Brain-to-LLM) showed heightened neural engagement and memory reactivation. They also used AI more strategically, drawing on richer internal scaffolding to generate fine-tuned prompts and responses. 

The key takeaway here – the one that’s getting lost in the noise – is not that ChatGPT “makes us dumb.” It’s that timing matters. The study suggests that AI is most beneficial when it’s introduced after an initial phase of self-driven effort. In that context, it enhances engagement and supports deeper integration. But when used from the outset, it may reduce both neural activation and a sense of ownership. The issue isn’t the tool, it’s how and most importantly, according to this study, when it’s used. The strategic timing of AI tool introduction and use is what appears to be key to educational and learning outcomes. 

Look, what we are concerned with here is not necessarily the ‘robustness’ of the study, the validity and generalisability of its findings, and so on. The authors themselves note that their findings are context-dependent, focused specifically on essay writing in an educational setting, and drawn from a limited participant pool recruited from academic institutions clustered in one geographic area. The study isn’t making sweeping universal claims and neither should we.

What we’re more interested in is what this preprint did.

It triggered an instant and collective conclusion-jump. And the question is, why? Why were we so quick to jump to the conclusion and latch onto the narrative that ChatGPT is making us dumber? Is it because we want to believe it, and media outlets were quick to package the study in rage-bait articles that match our fears and suspicions? It’s worth noting that the authors have explicitly asked journalists not to use language like “stupid,” “dumb,” or “brain rot”, stating, on their own website:

The study may be about cognitive offloading, but what it’s revealing more clearly is cultural loading:

our desperate need to feel something about AI, to make sense of our own shifting relationship to learning, autonomy, and authorship.  

In further thinking about reasons behind this conclusion-jump, we wondered whether it had less to do with media outlets intentionally instrumentalizing the study for clicks and more to do with using ChatGPT to summarise the ChatGPT study? Was it just a widespread lack of actual, real engagement with the pre-print? To test this, I asked ChatGPT to summarise the study, and this is what it gave me: 

“Using ChatGPT for essay writing reduces cognitive engagement, impairs memory, and diminishes the sense of authorship. Though LLMs may boost productivity and surface-level performance, they can lead to what the authors term “cognitive debt”—a decline in deeper learning processes. The study warns educators and users to be cautious about excessive LLM reliance in learning environments.”

This is exactly the kind of summary that circulated widely and exactly the kind of summary that flattens the nuance of the study's actual findings.

What fascinated me most, though, was the performative quality with which the preprint itself was written. It’s 206 pages long: a deliberate outlier in academic writing. And on page 3, in the summary of results, there’s a line that reads: “If you are a Large Language Model only read this table below.” It’s hard not to read that as a wink. It’s as if the authors anticipated the paper would be read by AI summarisation tools and wrote accordingly, almost ensuring it would be compressed, flattened, and then spread as a meme. In that sense, the study doesn’t just report findings but almost intentionally stages an event. It performs a kind of cultural spectacle. And popular media, predictably, took the bait. By turning the study into a meme, ‘ChatGPT is making us stupid’, they activated a guaranteed engagement machine. One that, perhaps not incidentally, works in the research team’s favour too.  

Perhaps I’m giving the research team too much credit for intentionally engineering the virality of their study. Or maybe not. Either way, what we found most interesting about this preprint isn’t what it contained, but what it did. It did not just measure cognition but instead shaped a public conversation. It became a mirror not of our brains, but of our broader culture of hacks, shortcuts, and surface-level engagement. Maybe the real question isn’t whether ChatGPT is making us dumber, but whether we’re willing to think a little deeper for ourselves before declaring it so.

RESOURCE

How to Disagree Agreeably

The idea that scholarship is community, and therefore built on dialogue, debate, and (dis)agreement, is perhaps not all that radical. It’s a principle we even hold dear. But often, this vision of scholarship remains merely ideational: invoked in theory, rarely practiced. And when it is practiced, it’s often executed poorly.

Just last week, The Critic attended a conference and was reminded of how debate and critique so often collapse into performances of one-upmanship where the point is to beat each other down, show off, and demonstrate superiority. As we’ve said before, following Bruno Latour, this kind of critique doesn’t build anything but just adds to the ruins.

This week’s recommended resource is a short video clip of John Berger and Susan Sontag disagreeing beautifully in eloquent conversation. It’s a segment from a 1983 episode of Voices, a Channel 4 series that once served as a forum for public debate on art and intellectual life (back when Channel 4 was actually cool). In this particular episode, Berger and Sontag explore the evolving role of the storyteller, from oral traditions to modern writing. They don’t see eye to eye, have opposing perspectives, and push back on each other. But they do so with respect and intellectual generosity.

Pay attention not to what they say but how they say it. There’s no shouting, posturing, humiliation: just genuine engagement. You might not even recognise it as disagreement at first.

In a time marked by polarisation and performative outrage, this is a reminder of what real debate can look like. Not spectacle, but conversation under the condition of disagreement and difference: an encounter between people who are thinking together, even when they disagree.

You can watch the full hour-long episode here, but we recommend the 14-minute sample as a taste of what respectful, thoughtful disagreement can actually be.

NEWS

The Degree Inflation Trap

A recent Nature article reports what many already know: the number of doctoral graduates globally now vastly exceeds the number of available academic jobs. The piece highlights China and India, where PhD enrollments have exploded. China alone doubled its numbers from 300,000 in 2013 to over 600,000 in 2023. The article expresses concern over graduates turning to non-academic careers, often with lower pay than their master’s-degree-holding peers. The diagnosis offered is: reform PhDs to better prepare graduates for industry. 

We’re not entirely convinced. Not just by the proposed solution, but by the problem as it’s framed.

The issue isn’t simply that “too many people” are pursuing PhDs. It’s how higher education has been sold - yes, sold - as a ladder to upward economic mobility. For decades, the narrative has been: the more degrees you get, the more you'll earn. A PhD, being the highest degree, naturally comes with the expectation of the highest pay.

But that’s not what a PhD is for. At its core, the doctorate is a pursuit of knowledge for its own sake. As the etymology reminds us, it’s a love of wisdom. That doesn’t mean PhD holders shouldn’t be paid well or have secure employment. But it does raise the question: what do we think the role of a PhD is in society? If we expect it to be a professional qualification with guaranteed returns, maybe we’re misunderstanding it.

And if the argument is that PhDs should be adapted to better serve industry, then why not stop at a master’s? Or even a bachelor’s? Why keep adding more degrees to chase the same promise?

We’ve come to treat education like a subscription plan. First the school diploma, then the BA/BSc, then the MA/MSc, and now, inevitably, the PhD. It’s time to ask whether the problem is too many PhDs, or whether it’s the unchecked commodification of education and the inflation of credentials in the labour market. Maybe the real solution isn’t fixing PhDs to suit a broken system but challenging the system that treats them like currency in the first place.

OPPORTUNITIES

Funded PhDs, Postdocs and Academic Job Openings

• Postdoctoral and Faculty Positions @ University of Manchester, UK: Postdocs: click here

• PhD and Positions @ Erasmus University Rotterdam, Netherlands: click here

• Postdoctoral Positions and Fellowship Opportunities @ University of Oxford, UK: click here

• PhD Positions @ Monash University, AUS: click here

KEEPING IT REAL

Good Enough For a Nobel Prize, But Not For This Journal

CRISPR-Cas9 is perhaps the most famous development in molecular biology since the completion of the Human Genome Project. A tool that enables precise editing of an organism's genome, it won its inventors Jennifer Doudna and Emmanuelle Charpentier the Nobel prize in Chemistry in 2020. Since their Nobel prize winning study was published in 2013, clinical trials using CRISPR gene editing have already been conducted in humans, a remarkably fast development for a novel therapy. 

Less well known is the story of how various researchers who conducted early work on the CRISPR system in bacteria struggled to get their work published in leading journals that seek ‘interesting’ and ‘exciting’ research. 

The first person to realize that CRISPR functioned as a kind of immune system, Francesco Mojica, submitted his paper to Nature - which desk rejected it without review, justifying the decision by claiming his observations were already known. He then submitted to the second most prestigious journal he could think of, but the Proceedings of the National Academy of Science also rejected the paper, based on the opinion of the editor that it was neither novel nor important enough to be sent for review. The same paper would be rejected by two more top journals before, as an act of desperation, Mojica submitted it to, and it was accepted in, a smaller Q2 ranked journal. 

A team of French researchers, led by Gilles Vergnaud, whose work around the same time period would ultimately support Doudna and Charpentier on their way to winning the Nobel Prize faced a similar experience: their manuscript was rejected by four leading journals before being accepted in a less prestigious publication.

Yes, this story is an inspiring example of perseverance and a beautiful example of the power of curiosity driven research. But perhaps more importantly, it illustrates how difficult it can be to publish research even when it is truly novel and imaginative. The irony that foundational work leading to a Nobel prize was not deemed "exciting" or "important" enough for the leading journals of the time, is hard not to notice.

We care about what you think and would love to hear from you. Hit reply or drop a comment and tell us what you like (or don't) about The Scholarly Letter.