Forecasting and recursive Inhibition within a decision cycle — LessWrong