Source
- Block (2007), LIT-646, read in NOTE-497: §§13–14 (working memory and phenomenology in the brain; neural coalitions and the mesh argument). Target article only.
- Block (2005), Two neural correlates of consciousness, LIT-629, read in NOTE-496: the phenomenal and access NCCs, the MT/V5 and Supèr et al. evidence, Box 1.
- Block (2019), What Is Wrong with the No-Report Paradigm and How to Fix It, LIT-637, read in NOTE-495: the no-report evidence on both sides, Brascamp et al.’s design, “Anticipating an Objection” and the Concluding Remarks.
The claim, assembled
Two correlates. Block (2005), LIT-629, separates a phenomenal NCC, the minimal basis of an experience’s content, from an access NCC, the frontoparietal activation that broadcasts a content to reasoning and report. For one kind of seen motion the phenomenal NCC is recurrent activity in and around MT/V5. TMS to MT/V5 produces moving phosphenes, and a V1 pulse delivered within the feedback window after it leaves them mostly stationary (Pascual-Leone and Walsh), which “suggests that experiencing moving phosphenes depends not only on activation of MT/V5 but also on a recurrent feedback loop to V1 and back”. Access is a winner-take-all competition: in the attentional blink, visibility ratings are almost never intermediate. So a recurrent sensory representation can lose that competition and stay unaccessed. In Supèr et al.’s monkeys, V1 figure–ground modulation, which vanishes under anaesthesia and is “probably produced by recurrent processes”, averaged the same whether or not the monkey saccaded to the target once salience was lowered or catch trials raised. Block takes it as “an indication of, if not directly part of” a phenomenal NCC, and calls the whole case “a glimmer”.
The mesh argument. Block (2007), LIT-646, §13, places the bottleneck of working memory in dorsolateral prefrontal cortex: delay firing there maintains representations held in posterior areas, and damage impairs delayed but not simultaneous performance. He places the core basis of seen motion in a V1–V5–V1 loop (Silvanto et al.). Frontal lobotomies do not appear to dim basic perceptual content: “Frontal damage impairs access but it doesn’t dim the bulb” (§14). Kouider et al. (2007) found supraliminal words under distraction activating occipito-temporal cortex strongly without triggering frontal coalitions. The argument (§14): “If we assume that the strong but still losing coalitions in the back of the head are the neural basis of phenomenal states (so long as they involve recurrent activity), then we have a neural mechanism which explains why phenomenology has a higher capacity than the global workspace. If, on the contrary, we assume that the neural basis of phenomenology includes workspace activation, then we do not have such a mechanism.” Overflow (THEORY-103) shows only that the machinery of experience is partly distinct from that of access; “what the mesh argument adds is that the machinery of phenomenology does not include the machinery of cognitive accessibility”. Block leaves room for causal influence in both directions. Top-down amplification may usually be involved, but it is not constitutive.
The prefrontal question since. Block (2019), LIT-637, frames the dispute as whether certain prefrontal regions (dorsolateral, medial prefrontal, anterior cingulate, orbitofrontal) are necessary for perceptual consciousness. The global workspace and higher-order theories say yes, and recurrent processing and IIT say no. He reports evidence on both sides from no-report binocular rivalry. Frässle et al. found prefrontal differences minor once report was removed. Panagiotaropoulos’s group decoded rivalry contents from monkey prefrontal cortex with no report, in monkeys never trained on a discrimination. Block argues that neither settles the question, because uninstructed subjects still think about what they see (THEORY-094). In Brascamp et al.’s inconspicuous rivalry, switches were detected at chance by d′, so nothing in thought followed them, and prefrontal differences were “altogether undetectable” by fMRI. Block’s verdict is that “the Brascamp et al. experiment is evidence against prefrontalism, further experiments with ECog could support prefrontalism”, and that the article’s “main point … is to home in on the methodological issue rather than supporting one side”.
What this does not say
- It does not say the prefrontal question is settled. Block (2019), LIT-637, says it is not, by no-report designs or by the one no-cognition design he offers. That design is a single fMRI null, and Block grants that its changes “might require temporal and spatial resolution too fine grained for fMRI”. The no-report prefrontal decoding in monkeys is explained away by the bored-monkey problem, not refuted. It too awaits a no-cognition design.
- It does not say prefrontal cortex plays no causal role. The claim is constitutive: prefrontal access machinery is not part of the basis of perceptual experience. Block allows that top-down influence may usually be what makes sensory activity strong enough.
- It does not say recurrence is sufficient for experience in general. The recurrence evidence is for one kind of seen motion (MT/V5) and, more weakly, for figure–ground modulation in V1. Block (2005) calls different phenomenal NCCs for different contents the expected outcome, and leaves “exactly which brain areas are required” as background unknown. Block (2019) speaks only of “the back and middle of the head”.
- It does not cover the phenomenology of thought. Noncognitivists can allow that if thought has a phenomenology of its own, it is prefrontal (LIT-637). The claim is about perceptual consciousness.
- Every experiment is second-hand, and Block calls the neural details of 2007 “wrong, or at least highly incomplete”. The argument is an inference to the best explanation, and it inherits the contested step of THEORY-103, that the overflow is experienced.
Why it presupposes the overflow account
This account presupposes THEORY-103; it does not refine it. The mesh argument takes overflow as its psychological premise and supplies the neural mechanism for it, losing recurrent sensory coalitions. If overflow were shown to be an illusion, the mesh argument would lose its premise, and this account would rest on the 2005 and 2019 evidence alone.
The rival account
THEORY-096 (global availability by ignition, from Mashour et al., LIT-631) puts prefrontal hubs inside the basis of conscious perception, and treats prefrontal coding of the percept without report as unconfounded. Block’s 2019 paper (LIT-637) denies exactly that second step. The rivalry is declared on that file. Inconspicuous rivalry with intracranial prefrontal recording would bear on both.
Connections
- THEORY-102 (Vishne et al., LIT-644) is not a source. Its data fit this account: for a clearly seen image, content was carried in ventral temporal cortex for as long as the image stayed on, while prefrontal content was an onset burst that ignored duration. But three things keep it from bearing weight here. Every image was seen, so experience and access never came apart. The global workspace’s own pre-registered predictions were met too, as the authors report. And its no-report trials still required each image to be judged a non-target, which is post-perceptual cognition of the kind THEORY-094 says leaves the onset burst ambiguous. This account does not build on or replace that one; it presupposes it, since the argument against the no-report prefrontal evidence above takes its conclusion as a premise.
- THEORY-100. In crows, neurons in the nidopallium caudolaterale, the functional analogue of prefrontal cortex, followed the reported percept. On this account that is an access correlate. It is what the account predicts of a region that carries access, and it says nothing about the basis of the experience. No relation is declared: the two accounts are about different correlates.
- Block (2009), LIT-632, names this the “biological” theory and argues that on it broadcasting is “what consciousness does rather than what consciousness is”.
Note of 2026-10-14: COGITATE and sustained recurrence
COGITATE (LIT-973, read in NOTE-812) is consistent with this account in one respect: content in all three tested dimensions sat in posterior cortex, and prefrontal cortex added nothing decodable beyond category. But IIT’s proponents predicted sustained gamma-band synchronization between category-selective cortex and V1/V2 throughout the percept, and it was not found. The coupling was brief, low-frequency and mostly evoked (THEORY-296). That prediction is about recurrence between high-level areas and early visual cortex, so the data do not support sustained recurrent coupling as the carrier of the experience. V1/V2 was sparsely sampled in iEEG, and the test was of one proposed implementation. The statement is unchanged.