Figure 1. Experiment procedure (illustrated with the number estimation task). Inter-trial intervals were indicated by a fixation cross, the duration of which varied randomly between 1350 and 1450 ms to prevent participants from forming expectations. Each trial began with the presentation of a dot array for 250 ms. Subsequently, participants were required to estimate the number of dots in the dot array as accurately as possible. The experiment would automatically advance to the next trial if no response was received within a 15-second period.

Figure 2. (a) Regression coefficients (β) of the respective variables for the full model of the two tasks in Experiment 1. CurRF (Current Relevant Feature) is the task-relevant feature in current trial, e.g., the number of dots of current stimulus in the number estimation task; PreRF (Previous Relevant Feature) is the task-relevant feature in previous trial, e.g., the number of dots of previous trial in the Number Estimation Task; CurIRF (Current Irrelevant Feature) is the task-irrelevant feature of current trial; PreIRF (Previous Irrelevant Feature) is the task-irrelevant feature in previous trial; PreRes (Previous Response) is the estimated value of previous trial. The error bars represent the standard error (SE) of the β-value; (b) Regression coefficient (β) of the respective variables for the full model of both tasks in Experiment 2.