When the Clock Lies: Intention Reports Can Be Biased by Authoritative Suggestion

Since the 1980s, neuroscientific studies of volition have examined participants’ subjective experience of when the intention to move arises. A common approach involves participants watching a fast-rotating clock, performing a spontaneous movement (e.g., a button press), and then reporting the clock position when they experienced the onset of intention (W time). However, previous evidence suggests that W times might be inferred rather than reflect the true onset of conscious experience. We tested 240 participants in an online version of the clock task using an order (O) mode of recall. After each movement, a stationary clock hand appeared at a predetermined position, and participants indicated whether they had decided to move before, after, or right now relative to the displayed time. Crucially, participants in the deception group were told that the stationary hand reflected an AI system’s prediction of their decision, whereas those in the control group were told the clock position was random. Participants in the deception group selected right now more often than those in the control group, indicating that W reports can be easily biased and may not correspond to a distinct introspective event. This effect was strongest when the bias was small (75 ms) and positive (i.e., when the clock hand appeared slightly later than the participants’ “veridical” W estimate), which we interpret as a symptom of shortcomings of the clock method in general. Overall, our findings suggest that participants can adopt deceptively biased cues, supporting the interpretation that W times are, at least in part, inferred judgments rather than direct readouts of conscious intention.