Confidence inflation from confirming post-identification feedback is greater when the eyewitness is inaccurate than when the eyewitness is accurate, which is evidence that witnesses infer their confidence from feedback only to the extent that their internal cues are weak. But the accurate/inaccurate asymmetry has alternative interpretations. A critical test between these interpretations was conducted by including disconfirming feedback conditions. Student participants (n = 404) witnessed a mock crime, had either a strong or weak ecphoric experience when making their line-up identifications, and subsequently received no feedback, confirming feedback, or disconfirming feedback. Consistent with a cues-based conceptualization of the feedback effect, disconfirming feedback influenced witnesses with weak ecphoric experiences more than witnesses with strong ecphoric experiences, ironically increasing the confidence-accuracy relation. Copyright © 2011 John Wiley & Sons, Ltd.