No Evidence that Predictions and Attention Modulate the First Feedforward Sweep of Cortical Information Processing

2019 
Predictive coding models propose that predictions (stimulus likelihood) reduce sensory signals across the cortical hierarchy, including in primary visual cortex (V1), and that attention (stimulus relevance) can modulate these effects. Indeed, both prediction and attention have been shown to modulate activity in V1. However, prior work either confounded prediction and attention, rendering their specific effects unclear, or used functional Magnetic Resonance Imaging (fMRI), which has low temporal resolution, leaving it unclear if these effects reflect a modulation of the first feedforward sweep of visual information processing and/or later, feedback-related activity. In two experiments, we exploited the high temporal resolution of electroencephalography (EEG) and orthogonally manipulated spatial predictions and attention to investigate if these top-down factors can modulate initial afferent activity (before 80ms). In each trial of the task, participants were cued to direct their attention towards one of two locations, while the likelihood of a stimulus appearing at these locations was manipulated block-wise. In both experiments, we found no evidence for early modulations of visual processing by either prediction or attention, as indexed by the amplitude of the C1 ERP component. This was confirmed by multivariate pattern analyses (MVPA), which furthermore showed no top-down modulations of spatial representations before 100ms. Prediction and attention did affect later stages of information processing. These findings indicate that neither prediction nor attention, separately or in interaction, may influence the earliest cortical stage of visual information processing in humans. This knowledge has important implications for theories of predictive processing and attention.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    73
    References
    4
    Citations
    NaN
    KQI
    []