Notifications are engineered external cues — pings, badges, banners, buzzes — designed to interrupt whatever you are doing and pull you back into an app. In behavior-design terms they are the classic external trigger: in B.J. Fogg’s model, behavior occurs when motivation, ability, and a prompt converge, and notifications supply the prompt on the platform’s schedule rather than yours. Nir Eyal’s Hook model treats the external trigger as the entry point of the engagement loop; the notification’s job is simply to start the cycle that the feed then sustains.
The manipulative versions withhold information to manufacture a curiosity gap — “someone commented,” “you have 3 updates” — so the alert cannot be resolved without opening the app. Red badges borrow the salience of danger; escalating cadence (“we noticed you haven’t…”) ramps pressure on users who ignore the first prompt. Because each interruption also fragments attention and imposes a switching cost, the harm compounds across the dozens or hundreds an average device delivers daily, even though any single ping is trivial.
This is the rare conditioning tactic that is both easy to recognize and easy to neutralize, which is why it scores lower than the schedules it feeds. The honest test is whom the alert serves: a message from a person, a genuine alarm, or a delivery you’re waiting on serves you and survives verification; an alert whose only function is to reopen the app serves the platform. Turning the second category off costs nothing and removes the doorway before the deeper hooks — variable rewards, infinite scroll — ever get a turn.