The study investigates whether functional magnetic resonance imaging (fMRI) can decode the visual and subjective contents of the human brain. Researchers used statistical algorithms to classify brain states and found that ensemble fMRI signals in early visual areas could reliably predict the orientation of visual stimuli a person was viewing. The study also showed that attention to one of two overlapping orthogonal gratings biased fMRI activity towards the attended orientation, demonstrating that early visual areas, including V1, contain detailed orientation information that can predict subjective perception. The study highlights the potential of fMRI to read out fine-tuned representations in the human brain and their subjective contents. It addresses the question of how the human brain represents basic visual attributes, such as orientation, and how these are represented during conscious perception. The study presents a novel approach called "ensemble feature selectivity" to measure feature selectivity from an ensemble of neuroimaging signals, using population coding and multi-voxel pattern analysis. The study found that fMRI activity patterns in the human visual cortex are sufficiently reliable to predict stimulus orientation on individual trials. The accuracy of orientation decoding was quantified for four subjects, with RMSEs ranging from 17.9° to 31.2°. The study also showed that orientation selectivity was most pronounced in early visual areas (V1 and V2) and declined in higher visual areas. Human area MT+ showed no evidence of orientation selectivity, consistent with its sensitivity to motion rather than stimulus form. The study further investigated the physiological reliability of orientation signals in the human visual cortex by testing generalization across separate sessions in two subjects. The results indicated that these orientation-selective activity patterns reflect physiologically stable response preferences across the visual cortex. The study also explored the source of orientation information and found that it reflects actual orientation-dependent responses. The results suggest that ensemble orientation selectivity arises from the orientation information inherent in individual voxels, which can then be pooled together. The study also addressed the problem of mind-reading, determining a subject's mental state based on their brain state. The results showed that activity patterns evoked by unambiguous single orientations could be used to decode which of two competing orientations was dominant in a person's mind under conditions of perceptual ambiguity. The study found that early visual areas showed strong bias towards the attended orientation, indicating that top-down voluntary attention acts very early in the processing stream to bias orientation-selective signals. The study concludes that fMRI activity patterns in the human visual cortex contain reliable orientation information that allows for detailed prediction of perceptual and mental states. The results suggest that feedback signals to V1 and V2 may have an important role in voluntary feature-based attentional selection of orientation signals. The study provides a framework for the readout of fine-tuned representations in the human brain and their subjective contents.The study investigates whether functional magnetic resonance imaging (fMRI) can decode the visual and subjective contents of the human brain. Researchers used statistical algorithms to classify brain states and found that ensemble fMRI signals in early visual areas could reliably predict the orientation of visual stimuli a person was viewing. The study also showed that attention to one of two overlapping orthogonal gratings biased fMRI activity towards the attended orientation, demonstrating that early visual areas, including V1, contain detailed orientation information that can predict subjective perception. The study highlights the potential of fMRI to read out fine-tuned representations in the human brain and their subjective contents. It addresses the question of how the human brain represents basic visual attributes, such as orientation, and how these are represented during conscious perception. The study presents a novel approach called "ensemble feature selectivity" to measure feature selectivity from an ensemble of neuroimaging signals, using population coding and multi-voxel pattern analysis. The study found that fMRI activity patterns in the human visual cortex are sufficiently reliable to predict stimulus orientation on individual trials. The accuracy of orientation decoding was quantified for four subjects, with RMSEs ranging from 17.9° to 31.2°. The study also showed that orientation selectivity was most pronounced in early visual areas (V1 and V2) and declined in higher visual areas. Human area MT+ showed no evidence of orientation selectivity, consistent with its sensitivity to motion rather than stimulus form. The study further investigated the physiological reliability of orientation signals in the human visual cortex by testing generalization across separate sessions in two subjects. The results indicated that these orientation-selective activity patterns reflect physiologically stable response preferences across the visual cortex. The study also explored the source of orientation information and found that it reflects actual orientation-dependent responses. The results suggest that ensemble orientation selectivity arises from the orientation information inherent in individual voxels, which can then be pooled together. The study also addressed the problem of mind-reading, determining a subject's mental state based on their brain state. The results showed that activity patterns evoked by unambiguous single orientations could be used to decode which of two competing orientations was dominant in a person's mind under conditions of perceptual ambiguity. The study found that early visual areas showed strong bias towards the attended orientation, indicating that top-down voluntary attention acts very early in the processing stream to bias orientation-selective signals. The study concludes that fMRI activity patterns in the human visual cortex contain reliable orientation information that allows for detailed prediction of perceptual and mental states. The results suggest that feedback signals to V1 and V2 may have an important role in voluntary feature-based attentional selection of orientation signals. The study provides a framework for the readout of fine-tuned representations in the human brain and their subjective contents.