Multitasking

The hardware does not multitask. It task-switches — and each switch carries a cost.

The Attention entry established the principle: attention is a single resource. The system cannot allocate full processing capacity to two cognitive operations simultaneously. What feels like multitasking is the system rapidly alternating between tasks — giving each partial processing capacity and switching between them at a rate that creates the illusion of simultaneity.

---

The cost of task-switching is measurable. Each switch requires the system to unload the context of the current operation, load the context of the new operation, and begin processing from a non-optimal starting point. The Interruption entry’s reloading cost applies with each switch. The organism “multitasking” across three operations is paying three reloading costs repeatedly, running all three operations at reduced capacity, and producing lower-quality output on all of them than it would produce on any single one with full allocation.

The exception: tasks that use different processing channels can run simultaneously without significant switching cost. Walking and talking. Listening to music and doing physical work. The cognitive channel and the motor channel can operate in parallel because they use separate processing infrastructure. Two cognitive tasks — reading and writing, listening and planning, processing email and participating in a meeting — cannot.

From the chair: the system performs better on focused sequential operation than on simultaneous partial allocation. One thing at a time, fully processed, then the next. The system resists this because the stimulation of task-switching produces its own small reward hit. The output quality doesn’t care about the reward hit.