CONCEPT
Context Reconstruction
The cognitive labor of reloading a project's goals, constraints, and evaluative criteria into working memory after a switch — expensive and never fully complete.
Context reconstruction is the first of three cognitive operations the monitoring builder must perform at each evaluation event. When an AI agent's output arrives, the builder must reload that project's context into working memory: its objectives, current state, constraints, and the criteria against which the output will be judged. This context was displaced by whatever the builder was working on when the output arrived. Reconstructing it requires retrieval from long-term memory, reactivation of the project's task-set, and re-establishment of evaluative standards. Each operation consumes working memory capacity and executive control resources that are already partially occupied by residue from the previous task. The reconstruction is cognitively expensive and rarely complete — some context elements will have decayed, some associations will need re-derivation, and the rebuilt context will be thinner than the original.
In The You On AI Field Guide
The expense is measurable. Studies of task resumption after interruption consistently show a 'resumption lag' — the interval between returning to a task and achieving the performance level that characterized work before