The Linear Mind

What it is

Carr's starting point is that a technology which extends the mind changes the mind that uses it. In his framework of intellectual technologies, the map, the clock, the alphabet, the book and the network each carry what he calls an intellectual ethic, an unexamined set of assumptions about how thinking ought to be done. The intellectual ethic of the printed book is sustained linear attention. The intellectual ethic of the network is retrieval and speed. His argument is that we adopted the second without noticing we were also giving up the first.

The historical half of the case is the strong half, and it is largely borrowed from Maryanne Wolf. Sustained reading is not a natural capacity but a built one, and it is recent. Early text was written in scriptura continua, without spaces between words, which made reading a decoding puzzle rather than a fluent act. Augustine records his astonishment at watching Ambrose read without moving his lips, because silent reading was rare enough to be remarkable. Word separation arrived by the thirteenth century, fluency followed, and only then could attention be spent on meaning rather than on decoding. On Carr's account the contemplative reader is roughly 550 years old. That is the move that gives the argument its force: if the capacity was assembled by a technology, another technology can disassemble it, and nothing in human nature guarantees it stays.

The weak half is the replacement claim, and the evidence Carr offers for it is behavioural rather than cognitive. His main exhibit is Jakob Nielsen's description of the F-shaped reading pattern: web readers sweep fully across the top of a page, then progressively less far down the left, producing a shape like the letter F. That is a real observation about where eyes go on a page. It is also commercial usability research, published on Nielsen's own consultancy website rather than in a journal. The endnote in The Shallows gives it as "F-Shaped Pattern for Reading Web Content", Alertbox, useit.com, 17 April 2006. The often-repeated detail that the work covered 232 people does not appear in that citation. The observation is probably broadly right and has been echoed by other eye-tracking work, and it should be labelled for what it is wherever it appears, because it circulates as though it were a study.

The stronger and better-behaved version of the claim is screen inferiority, which is measured, meta-analysed and bounded. Pooled across 54 studies and 171,055 participants, paper holds a small advantage over screens on reading comprehension, at Hedges's g of -0.21. The boundaries are where the finding becomes usable: the advantage appears for expository text, it appears under time pressure, and it does not appear for narrative at all. That pattern points away from the screen as an object and towards what a screen invites a reader to do when a clock is running, which is skim.

Between those two is the sweeping version: that a mode of thought responsible for several centuries of intellectual history is being lost. That is a historical thesis, not a finding, and this page treats it as Carr's argument rather than as a result. There is no instrument that measures a linear mind, no baseline from 1850 to compare against, and no study design that could establish the claim at the scale at which it is made. Saying so is not a dismissal. It is the difference between an idea worth arguing with and a fact that can be cited.

Carr's own intellectual honesty on this point is worth recording, because it is unusual in the genre. The most obvious counter-evidence to any claim that modern media are making people less able to think is the Flynn effect, the sustained rise in measured IQ across the twentieth century of roughly three points per decade. Carr meets it directly. He concedes the rise is real, accepts the figures, and says the gains should hearten us. His resolution borrows Flynn's own: the gains are concentrated in abstract, nonverbal, classificatory tasks, while vocabulary, general knowledge, arithmetic and reading scores are flat or falling, so the rise records a change in the kind of thinking rather than the amount. That is a genuine answer rather than an evasion, and it is also, unavoidably, an argument about interpretation rather than a test of the hypothesis.

In effect

The meta-analysis that measures the residue

Delgado, Vargas, Ackerman and Salmerón's 2018 synthesis in Educational Research Review, published under the title "Don't throw away your printed books", pooled 54 studies and 171,055 participants and found paper ahead of screens on comprehension by g = -0.21. Virginia Clinton's independent 2019 meta-analysis landed close by at g = -0.25. This is the measurable residue of Carr's much larger claim, and the comparison between the two is instructive. Carr's thesis is about centuries, civilisations and the shape of thought. What survives measurement is a fifth of a standard deviation on comprehension tests, for expository text, under time pressure, and absent entirely for narrative. The residue is real. It is also several orders of magnitude smaller than the argument it is used to support.

The article you meant to finish

The recognisable everyday version is the long argument abandoned in the third paragraph. You open something substantial, read enough to extract its position, open two more tabs from within it, and never come back to the first. Nothing was damaged in the process, and the behaviour has an entirely sufficient local explanation: the environment offers a cheaper option at every line break, and taking it is rational in the moment. This is where Carr's claim is at its most persuasive and at its least testable. Everyone recognises the experience. Nobody can say whether it reflects a changed mind or a changed situation, and the distinction between a capacity lost and a strategy adopted is precisely what none of the available evidence can settle.

Designing pages to be scanned, which is the F-shape put to work

Nielsen's observation did not stay an observation. It became the operating assumption of an entire profession. Front-loaded headlines, the important word first in every subheading, short paragraphs, bulleted lists, and the key claim placed top-left where the F begins are all standard practice in web publishing, and they are practice precisely because readers were seen to scan. This is the part of Carr's argument that is least deniable and gets the least attention. Whether or not the network has changed how people think, it has certainly changed how text is made for them, and text designed for scanning is easier to scan than to read closely. A publication that formats for retrieval is helping to produce the reader it assumed. It is worth adding that Nielsen Norman Group sells this expertise, which does not make the observation wrong, and does mean the finding has a commercial life independent of its evidential status.

What it does not say

It does not say anything that has been measured. There is no test for a linear mind, no historical baseline, and no experiment that could confirm or falsify the claim that a mode of thought built by print is being dismantled by the network. The vault tags this not-a-research-claim for that reason, and the tag is not a criticism. Frameworks are how arguments get organised. The failure mode is citing one as though it were a result.

It does not inherit the authority of screen inferiority. The two are constantly merged, because screen inferiority is the only measured thing anywhere near Carr's thesis, and borrowing it is a substitution: a small, bounded, well-evidenced effect standing in for a civilisational claim that has not been tested. The moderators are half the finding. An effect that disappears for narrative text and shrinks without time pressure is not evidence about the shape of the modern mind.

The Nielsen evidence should not be described as a study. It is commercial usability research published on a consultancy's website, and the participant count that usually accompanies it is not in the source Carr cites. The observation may well be broadly correct. It is not peer-reviewed science, and a page that quietly upgrades it is doing the thing this site exists to argue against.

And it does not stand unopposed. The Flynn effect is real, large and repeatedly confirmed, and it points the other way. Carr answers it rather than ignoring it, but his answer is an interpretation, and the honest summary is that measured cognitive performance rose for a century while the capacity he is worried about has never been measured at all. The effect has also slowed or reversed in several countries since the 1990s, which cuts both ways and settles nothing. The argument belongs in the same family as techno-panic, the recurring form in which a new medium is blamed for a decline. Belonging to that family does not make an argument wrong. It does mean the burden of evidence sits with the person making it, and on this claim that burden has not been discharged.

Who developed it

Nicholas Carr named the linear mind in The Shallows in 2010, drawing the historical account of reading largely from Maryanne Wolf's Proust and the Squid (2007) and the evidence for its replacement largely from Jakob Nielsen, a usability consultant and co-founder of Nielsen Norman Group.

---