Nicholas Carr’s The Shallows: What the Internet Is Doing to Our Brains is not a book about whether the internet is useful. Carr repeatedly acknowledges the enormous value of search, digital archives, communication, and rapid access to information. His argument is more specific: every medium trains habits of attention, and the internet’s dominant habits—scanning, clicking, switching, monitoring, and responding—may make sustained concentration harder to practice. The book therefore asks readers to judge technology not only by what information it delivers but by the mental routines that repeated use encourages (Carr, 2010).
Carr builds this argument from intellectual history, psychology, neuroscience, media theory, and his own experience of struggling to read long texts after years of intensive online work. Some of his claims were necessarily provisional when the book appeared in 2010. Sixteen years later, the evidence is more nuanced than the book’s strongest warnings, but several of its central questions remain highly relevant. Recent research on digital reading and media multitasking suggests that outcomes depend on design, task demands, reading purpose, and user behavior rather than on a simple opposition between “good paper” and “bad screens” (Wang et al., 2024; Drody, Pereira, & Smilek, 2025). Reading The Shallows today is therefore most useful as an argument about attention and intellectual habits rather than as a prediction that digital technology inevitably makes people less intelligent.
Carr’s Central Claim: A Medium Becomes a Habit of Mind
The foundation of Carr’s argument is neuroplasticity. The adult brain is not fixed; repeated experience can strengthen particular patterns of perception, memory, and action. Carr uses this principle to challenge the assumption that information technologies are neutral containers. If a medium repeatedly rewards one form of mental activity, users may become more practiced at that activity. Long-form reading rewards sustained attention, retention of context, inference, and reflection. Networked digital environments often reward navigation among fragments, rapid filtering, search, response, and task switching.
Carr does not argue that these newer skills are useless. Scanning and search can be extraordinarily efficient. A researcher who needs one statistic, a traveler checking a route, or a physician consulting a reference may benefit from speed. His concern is that a skill appropriate for information retrieval can become a default mode even when interpretation requires slower attention. The distinction is between being able to find information and being able to build a rich mental structure from it.
Working memory is central to this argument. Readers must temporarily hold information while connecting sentences, integrating ideas with prior knowledge, resolving ambiguity, and constructing a representation of the whole text. Interruptions consume some of that limited capacity. Hyperlinks, messages, alerts, advertisements, and parallel tasks can therefore create cognitive competition even when each individual interruption seems small.
Recent experimental evidence supports part of this concern. Earlier meta-analytic evidence also found a modest overall paper advantage for reading performance under some conditions (Kong et al., 2018). Delgado and colleagues’ earlier meta-analytic work found a modest paper advantage under some reading conditions, and Jensen, Roe, and Blikstad-Balas (2024) reported shallower processing and lower comprehension on screens in their student sample. Yet a broader 2024 meta-analysis of 37 experimental studies found no significant overall difference in reading comprehension between digital and paper formats once variation across tasks and readers was considered (Wang et al., 2024). The evidence therefore suggests that the problem is not the screen itself. Interface, scrolling, navigation, distraction, time pressure, metacognition, and reading strategy all matter.
Deep Reading, Memory, and the Difference Between Access and Knowledge
One of Carr’s strongest ideas is that memory is not simply a storage burden that technology can safely outsource. Long-term memory provides the organized knowledge through which new information becomes meaningful. A person who already understands economics can recognize why a new inflation figure matters; someone with no internal framework may locate the same number but lack the conceptual structure needed to interpret it. Search can retrieve facts, but expertise depends on relationships among facts already learned.
This point explains Carr’s defense of deep reading. A long argument or complex novel requires more than decoding sentences. Readers must retain characters, propositions, evidence, themes, and contradictions over time. They may need to pause, reread, compare, and revise earlier interpretations. The apparent inefficiency is part of the intellectual work. A medium optimized only for rapid retrieval can therefore conflict with tasks whose value depends on remaining with uncertainty.
Carr places this argument within a longer history of media change. Writing externalized memory; maps encouraged abstract spatial representation; mechanical clocks changed experiences of time; printing expanded access to books and standardized texts. Each technology created new cognitive possibilities while weakening or transforming older practices. His discussion of Socrates’ concern that writing would weaken memory is especially important because history partly vindicated and partly disproved the criticism. Writing did reduce the need to memorize some information, but it also enabled forms of scholarship impossible in an oral culture.
The same caution should be applied to Carr himself. Digital media has enabled searchable archives, collaborative scholarship, accessible texts, interactive visualization, translation, remote education, and immediate access to primary sources. These gains are not trivial exceptions to his argument. They show that media change produces trade-offs rather than one-directional decline. The important question becomes whether users and institutions can preserve demanding forms of attention while exploiting digital advantages.
Media multitasking makes this issue sharper. Drody, Pereira, and Smilek’s 2025 review argues that multitasking behavior emerges from interactions among cognitive architecture, personal dispositions, metacognition, task value, and environmental design. In other words, people do not multitask simply because a device forces them to. Digital platforms create opportunities and incentives, but goals, habits, and context influence whether those opportunities become interruptions. A 2024 experiment by Salmerón and colleagues found that media multitasking impaired processing and comprehension of multiple documents, while asking students to summarize main ideas reduced some of the damage. This supports a more practical interpretation of Carr: attention can be designed and trained, not merely lost.
Where Carr’s Argument Is Strongest—and Where It Overreaches
Carr is most persuasive when he questions the ideology of efficiency. Search engines and digital platforms are designed to reduce friction: locate the answer quickly, keep the user moving, surface the next item, and minimize delay. Human understanding does not always benefit from the same objective. Intellectual discovery can require confusion, rereading, wandering, note-taking, contradiction, and time away from the text. A system can become more efficient at delivering information while a learner becomes less effective at understanding it.
The book is also prescient about attention as an economic resource. Many digital services make money by keeping users engaged, returning, clicking, viewing, or generating data. This does not mean every designer intends to damage concentration. It means commercial incentives can favor notifications, recommendation loops, infinite feeds, and rapid novelty even when a user’s separate goal is sustained focus. Carr’s concern has become more relevant as smartphones, short-form video, real-time messaging, and algorithmic feeds have become integrated into everyday life.
Where the book is weaker is in its tendency to contrast “the internet” with “the book” as if each were a single cognitive environment. A distraction-free e-reader displaying a long novel is very different from a social-media feed interrupted by messages. A printed textbook can be skimmed superficially, while a digital article can be annotated carefully for an hour. Recent evidence increasingly treats device, interface, motivation, prior knowledge, text length, task instructions, and interruption as separate variables.
Carr also sometimes moves too quickly from laboratory findings about attention to broad claims about culture. Short-term experimental effects do not automatically prove permanent cognitive decline. Some newer work even suggests that heavy media multitaskers may develop specific adaptive response patterns under particular tasks. A 2025 neuroscience study, for example, reported faster anticipatory responses among heavy media multitaskers without reduced accuracy in its experimental setting. This does not overturn concerns about distraction, but it shows why cognitive change should not be described only as deterioration.
The strongest contemporary reading of The Shallows is therefore conditional. Digital environments can support deep thought, but they often contain more opportunities for interruption than print. Users can develop effective digital reading strategies, but those strategies require awareness and sometimes deliberate constraint. The brain is plastic, but plasticity allows both adaptation and retraining.
What the Book Means for Education and Work Today
For education, Carr’s argument suggests that digital literacy and deep literacy should not be treated as competing goals. Students need search skills, source evaluation, navigation, and AI literacy, but they also need the ability to read difficult material without constant switching, retrieve important knowledge from memory, build arguments, and sustain attention long enough to understand ambiguity. The 2024 paper-versus-screen meta-analysis is useful precisely because it rejects a universal prescription: medium effects vary, so educators should choose formats according to learning goals rather than ideology.
Practical design matters. Long readings can be presented in clean interfaces with fewer distractions. Students can be asked to annotate, summarize, retrieve key ideas from memory, or explain connections among sources. Main-idea summarization has been shown to mitigate some comprehension losses during media multitasking. Notifications can be silenced during complex tasks, while fast search can be used when the goal really is retrieval.
The same principle applies to work. Modern organizations often create environments in which employees are expected to monitor email, chat, dashboards, tickets, and meetings continuously while also performing analysis that requires concentration. Treating responsiveness as a constant requirement can fragment work even when no individual message is important. Protected focus periods, clearer escalation rules, fewer unnecessary alerts, and asynchronous communication can reduce the conflict between coordination and concentration.
Generative AI adds a new version of Carr’s question. AI can summarize, search, draft, and explain rapidly, but the intellectual effect depends on how it is used. If it replaces every difficult act of interpretation, users may learn less. If it helps compare sources, generate questions, or receive feedback while the user still evaluates and reconstructs the reasoning, it can support deeper work. Carr’s underlying principle still applies: evaluate not only the output of a tool but the habits the tool trains.
The Shallows remains valuable because it refuses to equate more information with more understanding. Some of its empirical claims now require qualification, and current evidence does not support the idea that digital reading is inherently inferior to print. Yet Carr’s broader warning survives: attention is shaped by repeated practice, and environments designed for rapid switching can make sustained thought harder unless users actively protect it. The book’s enduring question is therefore not whether society should reject the internet. It is whether people can preserve the capacity to choose the mode of thought appropriate to the task instead of allowing the most attention-demanding medium to choose for them.
References
Carr, N. (2010). The Shallows: What the Internet Is Doing to Our Brains. W. W. Norton.
Drody, A. C., Pereira, E. J., & Smilek, D. (2025). Attention in our digital ecosystem: The five interactive components that drive media multitasking. Psychonomic Bulletin & Review, 32, 2454–2471. https://doi.org/10.3758/s13423-025-02722-5
Jensen, R. E., Roe, A., & Blikstad-Balas, M. (2024). The smell of paper or the shine of a screen? Students’ reading comprehension, text processing, and attitudes when reading on paper and screen. Computers & Education, 219, 105107. https://doi.org/10.1016/j.compedu.2024.105107
Kong, Y., Seo, Y. S., & Zhai, L. (2018). Comparison of reading performance on screen and on paper: A meta-analysis. Computers & Education, 123, 138–149. https://doi.org/10.1016/j.compedu.2018.05.005
Salmerón, L., et al. (2024). Effects of media multitasking on the processing and comprehension of multiple documents: Does main idea summarization make a difference? Contemporary Educational Psychology, 77, 102271. https://doi.org/10.1016/j.cedpsych.2024.102271
Wang, X., et al. (2024). Which reading comprehension is better? A meta-analysis of the effect of paper versus digital reading in recent 20 years. Telematics and Informatics Reports, 14, 100142. https://doi.org/10.1016/j.teler.2024.100142
Academic Master Education Team is a group of academic editors and subject specialists responsible for producing structured, research-backed essays across multiple disciplines. Each article is developed following Academic Master’s Editorial Policy and supported by credible academic references. The team ensures clarity, citation accuracy, and adherence to ethical academic writing standards
Content reviewed under Academic Master Editorial Policy.
