We tested the idea that paying close attention to errors leads to improved performance on an inference task-We operationalized "attention to errors" as postfailure reflectivity, the ratio of the mean postfailure and post-success latencies on a self-paced inference task, where the interstimulus interval was subject controlled. Subjects were 9-, 10-, 11-, and 12-year-old children (A* = 378). Results reveal that (a) postfailure and reflectives show high cognitive performance during their initial exposure to the inference task; (b) postfailure impulsives have lower scores than postfailure reflectives on IQ, Raven, a test for attentional capacity, an arithmetic test, and teacher's evaluations in percentiles; and (c) "fast" versus "slow thinkers," as defined by Kagan's cognitive style construct, do not form homogeneous groups: fast-accurate children are postfailure reflective, and slow-inaccurate children are postfailure impulsive. from the comments of an anonymous reviewer, for which we are thankful.