Abstract
How do we form perceptual decisions about objects under conditions of little or no visibility? Two explanations have been proposed. For one, random noise fluctuations in sensory regions might determine the decision. Alternatively, networks for guessing distinct from the networks active during normal object viewing might determine the choice. Here we employed multivariate pattern classification analysis of fMRI data (searchlight decoding) to investigate the encoding of perceptual decisions about objects under conditions of little or no visibility. In experiment 1, subjects discriminated between noisy images of faces, houses and cars presented for 33ms. Performance was titrated to 50% correct (33% chance level). Correct choices(and thus correctly perceived stimuli) weredecoded from bilateral occipito-temporal cortex and left parietal cortex. Incorrect choices were encoded in left occipito-temporal cortex and medial parietal cortex (all results p<0.05, FWE cluster-level corrected). Experiment 2 implemented the same discrimination task as experiment 1 using pictures of faces, houses and cars in 1/3 of trials. In 2/3 of the trials, however, pure Fourier-scrambled noise images with the mean amplitude spectrum of the object images were presented. Category choices for pure-noise trials were found to be encoded in bilateral occipito-temporal cortex, bilateral parietal cortex and left DLPFC (p<0.05, FWE cluster-level corrected). Additionally, a reverse correlation analysis of the presented noise images and the category choices on the respective trials revealed that the averaged noise images were similar to the average face and car images. This indicates that subjects used the extremely limited visual information to make category decisions, even on pure-noise trials. In summary, our results implicate the involvement of ventral temporal, parietal and prefrontal regions in perceptual decision making when little or no visual information is available to inform the decision process.
Meeting abstract presented at VSS 2012