Resumption cost has two components — re-executing the primary task and remembering the interrupting task — that must be accounted for separately

What it is

The total cost of an interruption is usually lumped together as "the time it takes to get back to the primary task," but it actually splits into two separate accounts. One is the cost of rebuilding the primary task's goal state — recalling where things were left off and what comes next. The other is the cost of remembering that the interrupting task itself is unfinished — if the interrupting task was only half done when it got cut off, the person still has to recall, at some later point, "I still owe that." This split comes from memory-for-goals theory.

These two accounts are frequently collapsed into one vague "interruption cost," but they are fundamentally different in kind: the first is reactivating an existing goal state; the second is a prospective memory reminder about a future action. Merging them hides the hidden cost carried by the second account.

Why it happens

Memory-for-goals theory treats the cognitive system's tracking of tasks as maintaining a set of "goals" and their states: when a goal is activated, when it is completed, and when it gets suspended by an interruption. When an interruption occurs, the primary task's goal is not deleted — it is marked suspended, while the interrupting task's goal is activated. The suspended state itself carries a timestamp and contextual cues, and it is precisely these cues that pull the primary task's goal back from suspended to active during resumption — this process corresponds to the part of resumption lag that gets rebuilt.

But if the interrupting task itself gets interrupted before it's finished (an interruption within an interruption), it too gets suspended — and now the person must maintain not one suspended goal but two or more: not just resuming the original primary task, but also remembering that they still owe an unfinished secondary task. This latter part is a prospective-memory burden: not "how do I continue right now," but "at some future point I need to recall there's still something left undone." These two burdens correspond to different failure modes — failure of the first shows up as slower resumption and more errors; failure of the second shows up as forgetting entirely that something was left unfinished, a failure of completeness rather than merely of speed.

Studying it

These two components are typically studied with a nested-interruption paradigm: participants perform a primary task, a secondary task is inserted midway, and the secondary task is itself interrupted or deferred before completion. Researchers then track primary-task resumption performance (corresponding to the first component) and whether the secondary task is eventually remembered and completed (corresponding to the second component).

Common independent variables: the secondary task's completion status (fully completed vs. interrupted midway), the time interval between resuming the primary task and needing to recall the secondary task, and whether an external reminder cue is provided. Common dependent variables: the reaction-time cost after primary-task resumption (the marker for the first component), and the eventual completion rate or forgetting rate of the secondary task (the marker for the second component).

This distinction is useful for evaluating multitasking workflows and reminder systems: a system might make "returning to the main screen" very smooth (handling the first component well) while having no mechanism at all to remind the user of a shelved secondary task (neglecting the second component) — such a system is prone to real-world task omission.

A methodological caution: failures of the second component often only become observable much later (a user remembers something was missed days afterward), so short lab time windows likely underestimate how often this kind of prospective forgetting actually occurs.

Where it stops holding

• The two components do not always both exist. If the interrupting task was fully completed before the interruption ended, there is no "remembering an unfinished task" component left — only the cost of rebuilding the primary task's goal state remains.
• The probability of prospective-memory failure rises with delay. The longer something needs to be remembered and the more other events occur in between, the higher the risk of the second component failing (total forgetting) — this is not simply linear; more interference means higher risk.
• External reminder cues significantly reduce the failure rate of the second component but not the first. A reminder can make someone recall "there's still something left" — but after recalling it, they still have to go through goal-state reactivation.
• This describes what the two components of resumption cost are, not which factors amplify or shrink either one's size — that belongs to the level of modulating factors.

Applying it

• Treat "helping the user resume the primary task" and "reminding them of an unfinished secondary task" as two separate design goals with separate solutions. Good state preservation (which helps primary-task resumption) does not automatically solve secondary-task forgetting — they are different problems.
• For secondary tasks that might get interrupted midway, proactively generate an external reminder (a to-do entry, a badge count, a prompt at next login), moving the prospective-memory burden out of the user's head and into the system, instead of relying on the user to recall it unaided.
• Be especially careful with deeply nested interruptions (an interruption occurring within an interruption): each additional layer of suspension compounds the risk of total forgetting, and such scenarios should have a mandatory completion checkpoint rather than relying entirely on the user remembering.
• How to check: design a usage scenario that includes "the secondary task gets interrupted midway," follow up with users days later, and tally the secondary task's eventual completion rate. A markedly low completion rate indicates the system lacks support for the second component and needs an external reminder mechanism.

Related

• Same group: A5.08.1 Interruption timing should be chosen at subtask boundaries · A5.08.2 The system should preserve the state and position from before the interruption · A5.08.3 A measurable resumption lag exists before performance returns to pre-interruption levels · A5.08.5 Self-initiated interruptions have lower resumption cost than externally imposed ones · A5.08.6 The more information an interruption shares with the primary task, the more reconstruction work resumption requires
• Nearby: A6.21 External memory and cognitive offloading · A9.16 Cognitive offloading and external representation
• Search terms: memory for goals · prospective memory · dual-component resumption cost · interruption