Users lose their sense of elapsed time
Aliases: retrospective time estimation · time distortion during media use
What it is
In a continuously scrolling feed, users' estimates of elapsed time become markedly distorted — not a simple over- or underestimate, but a loss of the basis for estimating. Time perception relies on external markers and retrievable episodic structure; a homogeneous, boundary-less stream supplies neither, so the link between subjective and objective duration loosens as immersion deepens.
Why it happens
People estimate duration through two channels: prospective estimates during the activity depend on how much attention is allocated to time itself, and retrospective estimates afterward depend on how many retrievable events memory holds. Scrolling degrades both. Attention is fully captured by content, leaving the prospective channel nearly empty; and the retrievable, distinguishable events the retrospective channel needs are scarce in a homogeneous stream — items are structurally similar, unbounded, positionless, yielding no paragraph sense. Interfaces typically hide clocks and progress state as well. With both sources disabled, only the felt immersion remains, and immersion is a dull indicator: "just browsed a bit" coexists easily with forty minutes.
Studying it
Classic time-estimation paradigms transfer directly: prospective/retrospective separation (in-task probes vs. post-session reports), verbal estimation and time reproduction, and experience sampling of momentary absorption and time sense. Independent variables include segmentation density of content, presence of on-screen time markers, and task type; dependent variables are the signed error of estimates against logged duration. In interface research this measures how much a feed distorts time perception. Methodological cautions: knowing time is being measured changes attention allocation and contaminates prospective estimates; everyday-use studies must be anchored to objective logs, and the self-report/log discrepancy is itself a stable, informative dependent variable rather than noise.
Where it stops holding
Distortion is a byproduct of immersion, not something unique to infinite scroll — deep reading, games, and flow work do the same; the claim here is only that continuous feeds maximize the preconditions (no event boundaries plus hidden clocks). Time awareness varies widely across individuals, and distortion does not equal loss of control. Users browsing with a self-imposed deadline build their own external markers, partially offsetting distortion. The direction of retrospective bias is not fixed: highly distinctive event streams can be overestimated, so directional claims depend on event structure.
Applying it
Give browsing feeds periodic self-awareness anchors: fixed-interval session-length notices, a time watermark alongside the scroll, or segmentation by natural content units ("you've seen everything from today"). Avoid stacking distortion-amplifying designs on top of immersion (autoplay chaining, seamless switchover to another source). Verification: compare the distribution of retrospective time-estimation error against logs between groups with exposed versus hidden time markers — the anchor group's error should narrow substantially.