Recent research showed that people recall past-oriented, evaluative feedback more fully and accurately than future-oriented, directive feedback. Here we investigated whether these memory biases arise from preferential attention toward evaluative feedback during encoding. We also attempted to counter the biases via manipulations intended to focus participants on improvement. Participants received bogus evaluative and directive feedback on their writing. Before reading the feedback, some participants set goals for improvement (experiments 1 and 2), or they wrote about their past or future use of the writing skills, and/or were incentivised to improve (experiment 3); we objectively measured participants’ attention during feedback reading using...