Hannula, Simons, and Cohen1 advocate several changes for future imaging research of implicit perception. First, a reliable display technology should be used, and behavioural and imaging data should be acquired for the same participants under identical conditions. Because a valid masking set-up is now available for imaging2,3, future studies can realize these recommendations. Second, in terms of assessing unawareness, the authors maintain that, as in behavioural research, subjective (self-report) measures should be abandoned in favour of objective measures that demonstrate an absence of discrimination ability. That is, it is irrelevant whether participants claim that they are not aware of the pictures (subjective unawareness); instead, participants should not be able to discriminate them beyond chance (objective unawareness).

Although it is true that most imaging studies might have inadvertently established subjective unawareness, behavioural researchers do not generally agree that objective measures are more valid measures of awareness4,5,6. The debate persists because there is no conclusive evidence that either subjective or objective measures satisfy the requirements for a valid index of awareness ? that they capture all aspects of conscious processing (exhaustive) but no non-conscious processing (exclusive)7,8. Because stronger masking is typically required for objective unawareness than subjective unawareness9, this finding is often interpreted as evidence that subjective measures are affected by response biases10 and, therefore, are less sensitive than objective measures11. However, the principal drawback of objective measures is that they ignore the principally subjective nature of awareness. That is, because awareness is a subjective experience, it is more relevant to index what people notice subjectively than what they can discriminate objectively5,12. In analogy, the experience of pain cannot be indexed in terms of whether people can discriminate stimuli objectively, but whether they experience them subjectively as painful. Therefore, there is no evidence that objective measures fulfill validity requirements of exhaustiveness and exclusiveness better than subjective measures, so the discussion cannot be considered resolved in favour of objective measures. Furthermore, whereas the authors imply that methodological issues will disappear once imaging studies use objective measures, the results of behavioural research suggest that this conclusion is unwarranted. First, even for signal-detection measures, it is unclear what objective measure should be used (for example, face detection versus discrimination). Indeed, if discrimination ability per se is considered proof for awareness, implicit perception (for example, blindsight) is logically impossible5,12. Second, because unawareness is demonstrated by null sensitivity, this procedure attempts to prove the null (absence of awareness), and so depends on power13. However, there are no generally accepted criteria (such as number of trials, or alpha level). Third, if pictures are presented below subjective awareness, participants might have no motivation to perform the task14. If so, they might respond randomly, and, as a result, an objective measure would assess only subjective unawareness13. So, objective measures ignore the subjective nature of awareness and have additional problems.

Until the debate concerning a valid index of awareness is resolved, researchers are advised to adopt an eclectic approach using signal-detection measures to characterize unawareness comprehensively in terms of subjective and objective unawareness15.