Unfinished progress creates persistent psychological tension

What it is

The Zeigarnik effect: tasks interrupted mid-course occupy the mind more persistently than completed ones — they surface in thought more often and are harder to put down. In interfaces it shows up as the half-filled form, the film stopped halfway, the loyalty card three stamps short, returning to mind after the user has left. This is not the same explanation as "visible progress raises completion": that is a phenomenon of effort scaling toward a visible goal. Tension answers the other half — why the unfinished refuses to leave the mind, and how an interface amplifies or settles that refusal.

Why it happens

Tension comes from suspension: starting a task stands up an unpaid psychological commitment. Completion pays it off; interruption leaves it charged, drawing on attention to remind that something is outstanding. The reminder takes the form of occupancy — intrusive thoughts, the task floating up during other work, cognitive resources leaking away from the task at hand. One level deeper: the intensity of occupancy tracks not how much of the task remains but how close closure is. Long-suspended, far-from-goal tasks sink into the background; the closer to closure, the stronger the pull — "almost done" torments more than "barely started," because near closure every recall carries a nearly redeemable gap. This is also why the interruption point matters: stopping at a clean boundary (end of a chapter) leaves weak tension; stopping mid-action (a half-written sentence, a half-filled field) leaves strong tension.

Studying it

• Paradigm: the classic interrupted-recall paradigm — participants do a series of small tasks, half allowed to complete and half interrupted, followed by a surprise recall test comparing how often unfinished and completed tasks are remembered. Modern versions measure occupancy instead: probe reaction times, reading comprehension, and working-memory tasks index how much an unfulfilled goal interferes with ongoing cognition. A third design substitutes planning — participants write a concrete continuation plan, and occupancy is observed to fade without completion. Interface studies use logs and experience sampling to relate interruption point to later resumption.
• Variables: manipulations include interruption point (distance to completion), task commitment, and whether a save or plan exists at interruption; outcomes are recall frequency, probe interference, self-reported intrusive thoughts, and resumption rate and delay.
• Methodological cautions: a recall advantage is not harm — being remembered can be an adaptive reminder, so harm needs interference measures of its own. Lab interruptions are experimenter-imposed while interface interruptions are chosen or accidental; commitment differs, and tension strength does not transfer directly. Effects depend on the goal being endorsed: unfinished tasks the user rejects are also remembered, but with irritation rather than motivation to return — measure "being remembered" and "wanting to return" separately.

Where it stops holding

Evidence first: the classic Zeigarnik findings are old and their replication record is mixed — the recall advantage is sensitive to task type, interruption style, and participant motivation, with failed replications concentrated on low-commitment tasks irrelevant to the participant. The modern occupancy version (unfinished goals interfering with ongoing cognition, interference fading once a concrete plan is made) has newer experimental support, but effects vary widely across conditions and no specific number travels out of its setting. In use: tension is resource occupancy, not a guarantee of motivation — it pushes "return to the task," not "do the task well," and never asks whether the task deserves returning to. For flows the user never wanted, tension is no license for a hook: what it manufactures is annoyance and avoidance, and avoidance at its most thorough is uninstalling the app.

Applying it

• Autosave on interruption and say what was saved and where to resume — rewriting the suspended state from "about to be lost" into "a held to-do" turns tension from an anxiety source into a return motive.
• Name the concrete re-entry point ("step 3 of 5," "400 words in"): the more specific it is, the lower the reconstruction cost of resuming.
• Offer an optional pause point beyond clean boundaries (end of chapter, end of step) so users can park the tension where occupancy is low.
• Give quitting a formal entry (archive, remove, mark done): an acknowledged unfinished task occupies less than a suspended one.
• To validate: compare resumption rate and interruption-to-resumption delay with and without a save notice; sample experience diaries for how often users report "remembering something unfinished."

Related

• Same group: P2.11.2 Remaining steps predict abandonment better than percentage completed · P2.11.3 Fine-grained progress exposes the length of the path · P2.11.4 Completion drive can be aimed at tasks unrelated to the user
• Nearby: P2.03.2 Endowed initial progress strengthens completion intent · P2.06.5 The user's own investment becomes sunk cost
• Search terms: Zeigarnik effect · Ovsiankina effect · unfinished task