According to the normative standpoint, decision-making under risk, as it is traditionally instantiated in choice among fully described probabilistic lotteries, shold not be affected by the disclosure of outcomes (feeback). Yet, contrary to this normative prescription, empirical studies have repeatedly reported that risk preferences are affected by immediate presentation of choice outcomes. The consensual and intuitive hypothesis is that repeated feedback affects risk preferences via a learning process, whereby experienced outcomes alter the representation of subjective values. However, despite a large agreement concerning its psychological mechanisms, available evidence does not allow to roboustly establish the direction the impact of feedback on key behavioral outputs, namely, risk propensity and maximization. Here, we ran a series of seven experiments, tailored to decipher the effects of feedback on risk preferences and found that the presence of feedback consistently increases risk-taking, without any detectable impact on expected value maximization. Crucially, the temporal dynamics of the effect of feedback and trial-by-trial analysis on outcome sensitivity directly falsified the learning hypothesis in general, and more specifically one of the currently most influential model of the role of experience in decisions under risk. These results challenge the learning account in favor of an attitudinal effect, induced by the anticipation of feedback information. Epistemic curiosity and regret avoidance may drive this effect in partial and complete feedback conditions, respectively.