Previous studies have demonstrated a retardation in the rate of novel learning about previously blocked cues as compared to appropriate control cues. We report an experiment investigating whether this retardation in novel learning about a blocked cue is accompanied by a reduction in attention to this cue, as anticipated by attentional theories of associative learning. Consistent with these theories, eye gaze measures revealed a reduction in overt attention to the blocked cue both during the compound training phase of the blocking procedure, and also during novel learning with respect to new outcomes. Moreover, the extent of the bias in overt attention away from blocked cues was positively correlated with the subsequent reduction in rate of novel learning about these cues.