Prior expectations about the visual world facilitate perception by allowing us to quickly deduce plausible interpretations from noisy and ambiguous data. The neural mechanisms of this facilitation remain largely unclear. Here, we used functional magnetic resonance imaging (fMRI) and multivariate pattern analysis (MVPA) techniques to measure both the amplitude and representational content of neural activity in the early visual cortex of human volunteers. We find that while perceptual expectation reduces the neural response amplitude in the primary visual cortex (V1), it improves the stimulus representation in this area, as revealed by MVPA. This informational improvement was independent of attentional modulations by task relevance. Finally, the informational improvement in V1 correlated with subjects' behavioral improvement when the expected stimulus feature was relevant. These data suggest that expectation facilitates perception by sharpening sensory representations.

Perception and perceptual decision-making are strongly facilitated by prior knowledge about the probabilistic structure of the world. While the computational benefits of using prior expectation in perception are clear, there are myriad ways in which this computation can be realized. We review here recent advances in our understanding of the neural sources and targets of expectations in perception. Furthermore, we discuss Bayesian theories of perception that prescribe how an agent should integrate prior knowledge and sensory information, and investigate how current and future empirical data can inform and constrain computational frameworks that implement such probabilistic integration in perception.

Early visual areas contain specific information about visual items maintained in working memory, suggesting a role for early visual cortex in more complex cognitive functions [1-4]. It is an open question, however, whether these areas also underlie the ability to internally generate images de novo (i.e., mental imagery). Research on mental imagery has to this point focused mostly on whether mental images activate early sensory areas, with mixed results [5-7]. Recent studies suggest that multivariate pattern analysis of neural activity patterns in visual regions can reveal content-specific representations during cognitive processes, even though overall activation levels are low [1-4]. Here, we used this approach [8, 9] to study item-specific activity patterns in early visual areas (V1-V3) when these items are internally generated. We could reliably decode stimulus identity from neural activity patterns in early visual cortex during both working memory and mental imagery. Crucially, these activity patterns resembled those evoked by bottom-up visual stimulation, suggesting that mental images are indeed "perception-like" in nature. These findings suggest that the visual cortex serves as a dynamic "blackboard" [10, 11] that is used during both bottom-up stimulus processing and top-down internal generation of mental content.

Predictive coding models suggest that predicted sensory signals are attenuated (silencing of prediction error). These models, though influential, are challenged by the fact that prediction sometimes seems to enhance rather than reduce sensory signals, as in the case of attentional cueing experiments. One possible explanation is that in these experiments, prediction (i.e., stimulus probability) is confounded with attention (i.e., task relevance), which is known to boost rather than reduce sensory signal. However, recent theoretical work on predictive coding inspires an alternative hypothesis and suggests that attention and prediction operate synergistically to improve the precision of perceptual inference. This model posits that attention leads to heightened weighting of sensory evidence, thereby reversing the sensory silencing by prediction. Here, we factorially manipulated attention and prediction in a functional magnetic resonance imaging study and distinguished between these 2 hypotheses. Our results support a predictive coding model wherein attention reverses the sensory attenuation of predicted signals.

In addition to bottom-up input, the visual cortex receives large amounts of feedback from other cortical areas [1-3]. One compelling example of feedback activation of early visual neurons in the absence of bottom-up input occurs during the famous Kanizsa illusion, where a triangular shape is perceived, even in regions of the image where there is no bottom-up visual evidence for it. This illusion increases the firing activity of neurons in the primary visual cortex with a receptive field on the illusory contour [4]. Feedback signals are largely segregated from feedforward signals within each cortical area, with feedforward signals arriving in the middle layer, while top-down feedback avoids the middle layers and predominantly targets deep and superficial layers [1, 2, 5, 6]. Therefore, the feedback-mediated activity increase in V1 during the perception of illusory shapes should lead to a specific laminar activity profile that is distinct from the activity elicited by bottom-up stimulation. Here, we used fMRI at high field (7 T) to empirically test this hypothesis, by probing the cortical response to illusory figures in human V1 at different cortical depths [7-14]. We found that, whereas bottom-up stimulation activated all cortical layers, feedback activity induced by illusory figures led to a selective activation of the deep layers of V1. These results demonstrate the potential for non-invasive recordings of neural activity with laminar specificity in humans and elucidate the role of top-down signals during perceptual processing.