A hallmark approach in consciousness research has been to isolate the neural basis of consciousness by sorting measurements based on whether observers report a target stimulus as subjectively visible (seen response) or subjectively invisible (unseen response), a method known as post hoc sorting. However, decisions about sensory input are not only driven by sensory information, but also by other information, such as the expected rewards of a decision (known as the payoff matrix) or by information about statistical regularities in the environment (cognitive priors or predictions). To investigate the effect of these different types of information on conscious experience, we used three carefully matched manipulations that typically result in behavioral shifts in decision criteria: a visual illusion (Müller-Lyer condition), a punishment scheme (payoff condition), and a change in the ratio of relevant stimuli (base rate condition). To gauge whether these manipulations are accompanied by changes in conscious experience, we introduce a novel task in which participants not only make decisions about what they have just seen, but are also asked to reproduce their experience of a target stimulus. Using Bayesian ordinal modeling, we show that although each of the three manipulations affects the decision criterion, only the visual illusion affects conscious experience. Follow-up experiments show that non-perceptual criterion shifts also affect confidence scores during perceptual decision making, even when they do not affect conscious experience. 

In the second half of this talk I show the effect of non-perceptual criterion shifts on the neural correlate of conscious and unconscious processing under post hoc sorting. I use a signal detection theoretic simulation to show that – counterintuitively - a conservative response criterion not only inflates the correlate of 'conscious' but also the correlate of 'unconscious' processing, while a liberal criterion does the reverse.  Next, I show that this problem is not just hypothetical, by decoding conscious and unconscious conditions in two EEG experiments that employ two common subjective measures. Follow up simulations explicate how experimental context determines the degree to which neural correlates of either conscious and/or unconscious processing are affected by the criterion. Together, these data show that criterion placement threatens the construct validity of subjective measures in the neuroscience of consciousness.