Figure 3.

Delay task. (a,b) Effectiveness of precommitment: proportion of achieved delayed LRs was higher in precommitment trials (black) than in standard trials (grey; a) and declined with increasing waiting requirements of the LR (b; triangles, predicted probability; circles, observed probability of LR achievement). (c,d) Choice behaviour: opt outs of an initial LR choice during the waiting period were rare (c) and proportion of LR choices (without or with previous precommitment) showed a similar pattern to LR achievement. (e–g) Comparison of the MM and WP precommitment models. (e) Average value of precommitment to the LR option at each waiting requirement level as determined by the MM and WP model. (f) The average obtained θ value reflecting sensitivity of participants' precommitment decisions to the model-predicted value of precommitment was higher for the MM model than for the WP model. (g) Observed precommitment rates (black circles, partially hidden behind grey diamonds) closely aligned with the predicted probabilities of precommitment by the MM precommitment model (grey diamonds), but showed the opposite pattern than that predicted by the WP precommitment model (white diamonds). Error bars represent s.e.m.; *** denotes a significant effect with p ≤ 0.001, (*) denotes a tentative effect with p < 0.1 (two-tailed).