Figure 2.

Schematic illustration of how the free-choice paradigm produces artificial spreading of alternatives in the simulation study. Each normal curve represents the distribution of participant’s preference (true preference + random noise) for item A (blue) and B (red). In our simulation study, participants’ temporal preference for each item was randomly drawn from each distribution, and these numbers are rounded to the nearest integer that represents participants’ reported preference for each item. In this example, reported preferences for both items are 7 (A1 = B1 = 7). When a participant is asked to make a choice between A and B in the choice task, B is more likely to be chosen because true preference for B is higher than that for A. Then, when participants’ temporal preference for each item was randomly drawn for the second time from the same distribution (preference rating task 2), temporal preference for A2 (rejected item) is likely to decrease, while temporal preference for B2 (chosen item) is likely to increase purely by chance. Accordingly, reported preference for A is also likely to increase, while reported preference for B is likely to decrease (A2 = 6, B2 = 8 in this example). Note that distributions for each item A and B stay the same across two preference rating tasks, indicating that there is no change in participants’ true preference.