Autopilot Thinking: System 2 is just System 1 with working memory

In a recent LessWrong post, user XelaP explores the idea that System 2 (deliberate, conscious reasoning) is actually just System 1 (automatic, intuitive processing) combined with working memory. According to this model, each step of deliberate thought is an atomic System 1 operation that reads from and writes to working memory. Examples include calculating chess moves or composing sentences: each micro-decision (which move to consider, which word to write) feels like an intuitive snap judgment. The implication is that these atomic steps can occur on autopilot, even when working memory is impaired due to tiredness, intoxication, or cognitive load.

In practice, XelaP reports that when feeling cognitively impaired, simply directing one's autopilot toward a problem still yields useful output. For instance, when down, asking 'what should I do next?' triggers ingrained heuristics like opportunity cost and the sunk cost fallacy without deliberate effort. The author also suggests using this technique to overcome task aversion: start an activity in 'improv sketch' mode instead of trying to load everything into working memory. While full engagement may follow, even a few minutes of autopilot can break the paralysis. The epistemic status is noted as medium certainty with weak lines of evidence.