Abstract
Behavioural nudges are often criticised because they "work best in the dark". However, recent experimental evidence suggests that the effectiveness of nudges is not reduced when they are delivered transparently. Most people also endorse transparent nudges. Yet, transparent nudging may undermine human autonomy-a minority may oppose to being nudged and feel manipulated, even if they know what is happening. We propose an alternative way of maintaining autonomy that is not reducible to transparency: individuals can be asked if they consent in advance to being nudged. To assess whether consensual nudges are effective, we ask consent from 1518 UK citizens to be nudged. Subsequently, we default all participants into donating to a charity of their choice, irrespective of self-reported consent. We find that the default nudge is equally effective for both consenting and non-consenting individuals, with negligible difference in average donations. However, non-consenting individuals report higher levels of resentment and regret and lower levels of happiness and support compared to the consenting group. Based on these findings, we argue that ignoring consent can have serious ethical ramifications for policy-making with nudges.