Attention Residue Accumulation

In the AI Story

Leroy's original experimental design was elegant in its simplicity. Participants worked on a problem, were interrupted before completion, then assigned a new problem. Performance on the new task degraded measurably. The effect was strongest when the interrupted task was perceived as unfinished — when closure was denied. The degradation was not caused by the difficulty of the new task but by the cognitive persistence of the old one, held in working memory as unresolved business.

Mark extended the finding from laboratory isolation to the field. Real knowledge workers impose far more switches per day than any laboratory study could ethically simulate, and the switches accumulate across intervals that laboratory sessions — constrained to an hour or two — cannot observe. Mark's data showed that the curve of executive function degradation was predictable: early in the day, full cognitive capacity absorbed residue without perceptible loss; by mid-afternoon, the accumulated residue had measurably impaired the cognitive capacities that knowledge work most requires.

The AI-augmented workflow creates a structural condition Leroy's original experiments did not contemplate: permanent incompleteness. A conversation with Claude Code has no inherent endpoint. Each response suggests refinements. Each refinement opens sub-tasks. The filing mechanism that allows residue to dissipate — the sense of done — never activates, because the conversation remains productive. The residue therefore accumulates not as an intermittent cost but as a continuous condition.

Mark's forty-seven-second measurement captures the behavioral signature. The cognitive consequence is the slow, silent erosion of the capacities — judgment, architectural intuition, taste — that Segal identifies as the remaining twenty percent after AI handles implementation. The cruel irony, in Mark's framing: the capacities most valuable in the AI era are the capacities most vulnerable to the attentional costs of AI interaction.

Origin

Leroy published "Why Is It So Hard to Do My Work?" in Organizational Behavior and Human Decision Processes in 2009. Mark's field extension came through the successive observational studies she conducted at UC Irvine, which documented the accumulation pattern across real workdays. The concept traveled rapidly from academic psychology into popular discourse on productivity, usually in simplified form. Mark's work restored the empirical rigor and identified the structural conditions under which residue accumulates most severely.

Key Ideas

Residue is involuntary. The worker does not choose to keep thinking about the previous task. The persistence is a structural feature of how working memory manages competing demands.

Incompleteness is the amplifier. A completed task can be filed and released. An incomplete task remains active, demanding periodic attention and generating low-level anxiety.

Accumulation follows a predictable curve. Early-day switches impose negligible costs; late-day switches accumulate residue that measurably degrades executive function.

AI interaction produces continuous residue. The absence of natural endpoints in AI conversations denies the closure that would release the cognitive allocation.

The worker cannot perceive the accumulation. Subjective energy does not track cognitive depletion in real time; the first sign of depletion is often a feeling of confidence rather than fatigue.