Network predictions sharpen the representation of visual features for categorization
Yan, Y.; Zhan, J.; Ince, R. A. A.; Schyns, P. G.
Show abstract
Models of visual cognition assume that brain networks predict the contents of a stimulus to facilitate its subsequent categorization. However, the specific network mechanisms of this facilitation remain unclear. Here, we studied them in 11 individual participants cued to the spatial location (left vs. right) and contents (Low vs. High Spatial Frequency, LSF vs. HSF) of an upcoming Gabor stimulus that they categorized. Using concurrent MEG recordings, we reconstructed in each participant the network that communicates the predicted contents and the network that represents these contents from the stimulus for categorization. We show that predictions of LSF vs. HSF propagate top-down from temporal to contra-lateral occipital cortex, with modulatory supervision from frontal cortex. In occipital cortex, predictions sharpen bottom-up stimulus LSF vs. HSF representations, leading to faster categorizations. Our results therefore reveal functional networks that predict visual contents to sharpen their representations from the stimulus to facilitate categorization behavior.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Different computations over the same inputs produce selective behavior in algorithmic brain networks 98%
- Non-feature-specific elevated responses and feature-specific backward replay in human brain induced by visual sequence exposure 96%
- Neural dynamics of visual working memory representation during sensory distraction 96%
Similar papers in this journal
- Disentangling Semantic Composition and Semantic Association in the Left Temporal Lobe 96%
- Joint population coding and temporal coherence link an attended talker's voice and location features in naturalistic multi-talker scenes. 95%
- Dynamic representation of multidimensional object properties in the human brain 95%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.