A reader opens a phone to check the train time. A notification appears, then a news alert, then a short video selected because the reader paused on a similar one yesterday. Ten minutes later the train time is still unknown. The story is familiar, but it is not evidence that the reader has a defective brain. It is a question about how attention is allocated when an environment repeatedly proposes new goals.
Attention sovereignty is the Institute’s term for a person’s practical ability to direct and redirect attention in line with purposes they can endorse. It is a normative design concept, not a validated psychological scale. It joins research on attention and digital behavior with questions about choice, interface design and institutional power. The term should be used carefully: a person can have strong attentional skills and still face an environment built to capture them.
Attention is selective, but environments supply candidates
Attention helps prioritize some information over other information. A phone alert can become a candidate for selection even when the person does not open it. Stothart, Mitchum and Yehnert found that notifications disrupted performance on an attention-demanding laboratory task without direct interaction with the phone. The experiment supports a narrow claim about that task and cue; it does not measure the full cost of all notifications in everyday life.
Salience matters, but so does meaning. An alert from a family member can be genuinely important. A timed reminder can support a person’s chosen plan. The question is not whether every interruption is harmful. It is whether the person can tell what deserves attention, whether the interruption matches their priorities, and whether the system gives them usable control over its timing.
In a persuasive environment, the candidates are not random. Interfaces can rank, repeat and personalize material to influence what a person sees next. Kozyreva, Lewandowsky and Hertwig describe several digital challenges, including persuasive choice architecture and distracting environments. Their review proposes cognitive tools, while recognizing that individual skills and structural rules address different parts of the problem.
From a cue to a new goal
A notification is a cue; the action that follows depends on the person’s task, expectations and surroundings. A badge on an app icon might signal an urgent message or merely an accumulated count. A feed can offer apparently relevant content without telling the person why it was selected. The device can shift from serving a chosen task to suggesting a fresh one, repeatedly.
That shift is not always conscious. A reader may tap a headline out of habit before deciding whether it matters. Yet calling every such action “addiction” would exceed the evidence. Researchers use different measures for problematic use, habit and attention, and a single redirected glance does not diagnose a condition. The immediate question is simpler: what did the person intend, what did the interface present, and could they return?
Wilmer, Sherman and Chein’s review of smartphones and cognition surveys evidence on attention, memory and related functions, while noting limitations in the literature. Much work is correlational, and technologies change faster than many study cycles. It is reasonable to investigate habits and distraction; it is not justified to claim that one phone feature has permanently reduced everyone’s capacity to think.
Persuasion is not automatically manipulation
An interface can persuade someone to take a medicine on time, finish a lesson or avoid an unwanted purchase. Persuasion becomes ethically troubling when the user’s goals are hidden, distorted or subordinated to the designer’s interests. A timer can accurately report a real deadline, or create false urgency. A reminder can honor a chosen plan, or repeatedly override a refusal.
Mathur and colleagues examined dark patterns across thousands of shopping sites and identified recurring categories of interface practices. Their crawl documents design patterns on the sites and pages sampled. It does not prove that each visitor was manipulated or quantify every individual outcome. The study is valuable because it makes visible design practices that might otherwise be dismissed as isolated annoyances.
The Federal Trade Commission’s dark-patterns report gives examples such as disguised advertising, hard-to-cancel subscriptions and obscured terms. This is a policy source about consumer-protection concerns, not an experiment measuring attention. It broadens the analysis from momentary distraction to the ability to make informed, reversible decisions. Different evidence types should be kept distinct.
Attention and autonomy are connected, not identical
A person can choose to spend an hour on a game and be fully aligned with their own priorities. Another can complete a serious article while feeling pressured to remain online for work. Time spent on a screen is therefore a weak proxy for sovereignty. We need to know whether the person can enter, pause, leave and return according to purposes they endorse.
Self-determination theory describes autonomy as volition and endorsement rather than independence. The Institute’s attention-sovereignty concept applies that distinction to the allocation of attention. A person may willingly delegate some filtering to a trusted tool; delegation can support rather than undermine agency when it is transparent and revisable.
The reverse also holds. A person may appear to exercise choice while a system makes leaving difficult, uses defaults that favor the provider, or asks them to reconsider the same decision repeatedly. The European Union’s Digital Services Act addresses certain manipulative interface practices for online platforms. Its text is a legal and policy boundary in a defined scope, not evidence that every attention problem has been solved by regulation.
Critical ignoring as a positive skill
Digital literacy is often described as learning to examine more information. Yet attention is limited. Kozyreva, Wineburg, Lewandowsky and Hertwig argue that critical ignoring is also a competence: deciding which material does not deserve further engagement and where to invest attention instead. The argument does not mean ignoring uncomfortable facts. It asks for better allocation of a scarce resource.
A person can use a simple rule: inspect the source before reading a provocative claim in full; leave a low-quality page without debating every sentence; or delay a nonurgent alert until the task is complete. Such strategies are examples, not clinically validated prescriptions for every person. Their value depends on the information environment, the stakes and access to more reliable sources.
Education can teach people to recognize cues and verify claims, but training alone does not change an interface that exploits uncertainty. A person should not have to win a contest against every feature of a product to exercise ordinary choice. The Institute’s synthesis pairs skills for selection with design requirements that reduce unnecessary capture.
What a more supportive interface might do
It might let people group nonurgent notifications, make the reason for a recommendation legible and provide a clear stopping point. It might remember a preference to decline a prompt. It might allow a reader to finish the original task before proposing another. These are testable design hypotheses, not a list of features already proved to improve well-being in every product.
There can be tradeoffs. A default silence setting may reduce unwanted alerts but hide an important message. A friction step before opening a feed may help one person and frustrate another. A study should examine actual tasks, user preferences and unintended effects. It should not infer success only from a lower engagement number or a higher satisfaction score.
The evaluation also needs an honest account of business incentives. A product rewarded for time-on-site may have reason to keep suggesting content after the user’s original purpose is complete. Transparency about that incentive does not automatically resolve the conflict, but it makes design choices easier to question and govern.
Measurement without reducing a person to screen time
One possible measure is goal completion: did the person accomplish what they opened the service to do? Another is recoverability: after an interruption, could they return to the task? A third is experienced control: did they feel able to pause or decline a recommendation? Each captures a different part of the concept. None alone establishes sovereignty.
Researchers could combine behavior logs with participant reports, while protecting privacy. A log can show taps and time but not why an action mattered. A survey can capture perceived control but be influenced by memory or expectations. A diary can reveal particular episodes, yet it samples only what a participant notices and records. Mixed evidence is more informative when each limit is stated.
Longitudinal work would be needed to claim that a design changes enduring attentional capacity. A short experiment can test whether one alert disrupts one task. It cannot establish a permanent developmental effect. Conversely, the absence of long-term evidence should not prevent designers from fixing an obviously obstructive cancellation flow. Different claims require different standards of proof.
Measurement should also attend to whose purposes count. A platform may report that a user stayed for twelve minutes and completed six recommended items. The user may report that they intended to check one fact and leave. Both descriptions can be accurate, but only one names the original purpose. An evaluation can ask participants to state their goal before entering and then compare the observed path with that goal. This is not a perfect test: goals change legitimately as people discover information. It is a way to make the user’s aim visible beside the platform’s engagement metric.
A useful result would report variation rather than one average. Some readers may welcome exploratory recommendations, while others need a direct route to a specific answer. A design that supports sovereignty can make exploration an intelligible option without making it the unavoidable path. That hypothesis needs testing with different users and tasks.
Who bears the burden of attention protection?
A parent coordinating care may not be able to switch off every notification. A shift worker may need rapid alerts from colleagues. A student without a quiet room faces different constraints from one who can study uninterrupted. Advice to “just turn the phone off” can ignore the roles that the same device serves. Attention policy should start with the conditions of the people expected to use it.
Employers and schools also shape those conditions. If a team expects instant replies at all hours, an individual’s focus settings will not solve the social demand. If a course requires multiple platforms with conflicting alerts, students may be asked to manage an avoidable burden. The Institute’s synthesis is to assign responsibility at the level where the interruption is created.
That can include quiet hours, clear urgency labels, fewer duplicate channels and a norm that delayed response is acceptable for nonurgent requests. These proposals need evaluation in specific settings. They should not be sold as a generic productivity cure or used to judge a person who still finds concentration difficult.
Designing an honest test
Suppose a news app replaces endless recommendations after an article with a clear end point and an optional next-reading choice. A test could compare goal completion, unplanned extra reading, satisfaction and the ability to recall why the user opened the app. It should include people who come for different purposes, such as research, leisure and breaking news.
The test should measure whether important information becomes harder to find and whether users work around the feature. It should record what the app optimizes: subscriptions, advertising, informed use or something else. A feature that reduces time-on-site may be judged a failure by one metric while improving the user’s intended task. The metric choice is part of the ethical question.
A comparative design can estimate the effect of the new interface. Interviews can reveal whether participants experienced the end point as helpful or paternalistic. Neither method alone settles the normative issue of whose goals the product should serve. That issue requires a stated design commitment and a way for users to contest it.
A useful institutional test would compare the same task under two interface conditions, record whether the user achieved their original purpose, and ask what they noticed or missed. It would separate interruption from deliberate exploration: an unexpected alert is different from a person choosing to open a relevant source. It would also include people who rely on notifications for accessibility, caregiving or urgent work. Silencing every prompt may protect one kind of focus while removing a needed signal for someone else. This is a proposed evaluation of attention sovereignty, not a result established by the notification experiment or the shopping-site crawl. The question is whether the person can understand and govern the competing claims on attention in their actual context.
What the evidence does not show
The cited studies do not validate “attention sovereignty” as a single score or establish that one feature causes a lasting loss of free will. A laboratory notification effect cannot be multiplied into a universal daily-life cost. A dark-pattern crawl identifies practices, not the complete experience of every visitor. Reviews of smartphone use contain mixed measures and causal limits.
Nor does the evidence show that all persuasive design is wrong or that individual skill is irrelevant. The practical question is whether cues, defaults and exit paths help people carry out purposes they can endorse. Both design and education may matter; their contribution should be measured separately rather than assumed.
Back to the train time
The reader finally sees the departure board. The lost ten minutes may be minor today. The more important question is whether the service made a simple chosen task easy to complete, and whether the reader could decline the detours it offered. That question can guide a concrete redesign.
Attention sovereignty names a standard for the relationship between people and persuasive environments. It asks for skill in selection, honest interfaces and institutions that respect the limits of attention. It does not demand perfect concentration. It asks that a person’s purposes remain visible in the systems that compete to shape them.
Research references
- Stothart, C., Mitchum, A., & Yehnert, C. (2015). The attentional cost of receiving a cell phone notification. Journal of Experimental Psychology: Human Perception and Performance, 41(4), 893–897. View source
- Kozyreva, A., Lewandowsky, S., & Hertwig, R. (2020). Citizens versus the internet: Confronting digital challenges with cognitive tools. Psychological Science in the Public Interest, 21(3), 103–156. View source
- Wilmer, H. H., Sherman, L. E., & Chein, J. M. (2017). Smartphones and cognition: A review of research exploring the links between mobile technology habits and cognitive functioning. Frontiers in Psychology, 8, 605. View source
- Mathur, A., et al. (2019). Dark patterns at scale: Findings from a crawl of 11K shopping websites. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), Article 81. View source
- Federal Trade Commission. (2022). Bringing Dark Patterns to Light. Staff report. View source
- Kozyreva, A., Wineburg, S., Lewandowsky, S., & Hertwig, R. (2023). Critical ignoring as a core competence for digital citizens. Current Directions in Psychological Science, 32(1), 81–88. View source
- European Union. (2022). Regulation (EU) 2022/2065, Digital Services Act, Article 25 and Recital 67. View source