An increasing number of studies in the conflict/control and perceptual desirable difficulty literatures show memory benefits for information in high-conflict task situations. Recent work suggests that increased conflict does not produce a task-wide encoding benefit; rather, conflict must focus high-level attention on to-be-tested information to produce an encoding benefit. We used pupil dilation measures to directly assess this stage-specific model of conflict-encoding effects. We show clear performance costs of incongruency (slower RT and larger pupil dilation) with both semantic and response distractors, but show memory benefits only with semantic conflict. Further, when participants were encouraged to focus more (eliciting greater endogenous effort and control for all trials, not just incongruent trials), we observe larger and more similar pupil responses and reduced memory differences between high versus low semantic conflict conditions. These data confirm and extend a stage-specific model of conflict-encoding effects, with converging behavioural and physiological data.
