Decoding region-level visual functions from invasive EEG monkey data
Zhang, X.-Y.; Lin, H.; Deng, Z.; Siegel, M.; Miller, E. K.; Yan, G.
Show abstract
Decoding vision is an ambitious task as it aims to transform scalar brain activity into dynamic images with refined shapes, colors and movements. In familiar environments, the brain may trigger activity that resembles specific pattern, thereby facilitating decoding. Can an artificial neural network (ANN) decipher such latent patterns? Here, we explore this question using invasive electroencephalography data from monkeys. By decoding multiregion brain activity, ANN effectively captures individual regions functional roles as a consequence of minimizing visual errors. For example, ANN recognizes that regions V4 and LIP are involved in visual color and shape processing while MT predominantly handles visual motion, aligning with regional visual functions evident in the brain. ANN likely reconstructs vision by seizing hidden spike patterns, representing stimuli distinctly in a two-dimensional plane. Furthermore, during the encoding process of transforming visual stimuli into neuronal activity, optimal performance is achieved in regions closely associated with vision processing.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- The representational hierarchy in human and artificial visual systems in the presence of object-scene regularities 95%
- Increasing neural network robustness improves match to macaque V1 eigenspectrum, spatial frequency preference and predictivity 95%
- Pre-training artificial neural networks with spontaneous retinal activity improves motion prediction in natural scenes 95%
Similar papers in this journal
- Global motor dynamics - Invariant neural representations of motor behavior in distributed brain-wide recordings 94%
- Decoding of cortex-wide brain activity from local recordings of neural potentials 94%
- Unsupervised, piecewise linear decoding enables an accurate prediction of muscle activity in a multi-task brain computer interface 94%
Similar papers in this journal
"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.