By interpreting S as encoding log probability, this equates to an ideal, quasi-Bayesian decision-maker, confirmed (usually with saccades due to their stereotyped nature and rapidity of data acquisition) by systematically varying such factors as expectation, information supply and urgency[3].