Source
Vishne, Gerber, Knight and Deouell (2023), LIT-644, “Distinct ventral stream and prefrontal cortex representational dynamics during sustained conscious visual perception”, Cell Reports 42: 112752, read in full in NOTE-499. The general reading is the authors’ Discussion and Conclusion (NOTE-499 claims C4, a conditional interpretation, and C5, a proposal).
What is claimed, and by whom
The authors read their result as a general division of labour: the content of what is seen while an image stays on lives in a stable pattern across ventral visual cortex, whose position in an “experience subspace” fixes the content while its overall amplitude falls, and frontoparietal cortex registers the content only when it changes. NOTE-499 names the same claim as a candidate THEORY and notes that one source is not enough.
What the narrowed entry establishes toward it
THEORY-102 holds what was measured: in subdural high-frequency activity from ten epilepsy patients viewing clearly visible images for 300 to 1,500 ms, ventral temporal category and exemplar information stayed decodable by a time-invariant code for as long as the image was on while amplitude fell about 80%, and prefrontal and parietal decoding was a burst about 150 to 600 ms after onset that did not track duration. That is one method, one band, sparse and mostly right-hemisphere frontal coverage, and only seen images.
Evidence for and against so far
COGITATE (LIT-973, read in NOTE-812), an independent intracranial cohort with 576 prefrontal electrodes, finds a sustained posterior category representation and prefrontal category content confined to onset, as claimed. It does not hold for every dimension: face identity was not sustained in posterior cortex, and face orientation decayed after about 0.5 s while the image stayed on. So “the content” is so far supported for category, mixed for identity, and untested where percept and input come apart.
What would confirm or reject it
The promote_when, carried over from THEORY-102 unchanged: a second recording of the same dissociation by another method or cohort, and a paradigm where percept and input come apart (binocular rivalry, or a sustained image that is present but unseen) with the ventral code following the percept. It is refuted if that code follows the input when the percept changes, if an unseen sustained image is coded the same way, or if a sustained content-specific prefrontal code turns up in bands or sites the study did not sample.
What this does not say
- Anything about the theories’ formal content. The study tested the COGITATE paradigm predictions attributed to IIT and the global workspace, not integrated information or a workspace mechanism.
- That “representation” is causal. The authors define it as a pattern correlating with stimulus descriptors.