Contribution
The paper turns the phenomenal/access distinction of Block (1995) into a claim about the brain: there are at least two NCCs, and a framework with two makes more sense of the data than one with a single NCC. It also proposes a method, signal-detection theory, for getting at experience that is not accessed.
Key insight
Access is a competition with one winner. A content that loses does not thereby lose the recurrent sensory activity that, in the clear cases, is the basis of experience. So the competition for access gives a mechanism by which experience can occur without access, and signal detection gives a way to see it: hold the sensory signal fixed and move the decision criterion, and report changes while the sensory correlate does not.
Assumptions
- Koch’s definition of the NCC, “the minimal set of neuronal events and mechanisms jointly sufficient for a specific conscious percept”, read as allowing more than one NCC per percept.
- Core versus total NCC (Shoemaker; Chalmers): the core distinguishes one experience from another; the rest of the total NCC is enabling background.
- The global workspace model of access (Baars; Dehaene and Changeux).
- The SDT model of report: a report reflects both the strength of the sensory signal and a criterion that rewards, penalties and catch-trial rates shift.
Key results
- MT/V5 as a core phenomenal NCC for seen motion. Seven lines of evidence: activation during motion perception; microstimulation biasing judgements by column direction; akinetopsia after bilateral damage; activation by the motion aftereffect and its disruption by TMS; activation by implied motion in stills; moving phosphenes from TMS to MT/V5.
- Recurrence. With TMS to MT/V5 and V1 overlapping in visual space, a V1 pulse 5–45 ms after the MT/V5 pulse made subjects report mostly stationary phosphenes (Pascual-Leone & Walsh), suggesting a feedback loop is needed.
- Winner-take-all access. In Dehaene et al.’s attentional-blink variant, at the 260 ms lag subjects almost never used intermediate visibility ratings.
- The exclusion paradigm. Visser and Merikle’s reward manipulation changed exclusion but not inclusion performance. On Snodgrass’s SDT reading, a lowered criterion lets weak conscious perception of “reason” block “-son” while subjects think they saw no word. So some 50 ms stimuli are weakly conscious without being broadcast.
- Neural SDT (Supèr et al. 2001). V1 figure–ground modulation tracks detection for salient (16-pixel) targets and few catch trials. Reducing the target to 4 pixels with 20% catch trials, or raising catch trials to 50% at 16 pixels, makes modulation the same whether or not the monkey saccades. Restoring 0% catch trials at 4 pixels restores the coupling. The modulation vanishes under anaesthesia and is probably recurrent.
- Box 1. Nakamura and Mishkin removed frontal, parietal and superior temporal areas in one hemisphere of monkeys; with limbic areas intact the animals made exploratory eye and head movements. Block suggests access areas may be removable with experience left intact.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Recurrent MT/V5–V1 activity is a good bet for the core basis of one kind of seen motion | moderate: converging lesion, stimulation and imaging evidence, all cited | text, Fig. 1 |
| C2 | The access NCC is frontoparietal activation by posterior areas | moderate: cited imaging of bistable perception | Fig. 2 |
| C3 | A two-NCC framework fits the data better than a single NCC | weak to moderate: “a glimmer of an empirical case” | Conclusion |
| C4 | The exclusion data show experience without access | weak: rests on a contested SDT reinterpretation | Visser & Merikle; Snodgrass |
| C5 | V1 figure–ground modulation is evidence of a phenomenal NCC uncoupled from access | weak to moderate: “some plausibility to taking it as an indication of, if not directly part of” | Supèr et al., Fig. 3 |
| C6 | Introspective report is not a gold standard for consciousness | moderate: argument by analogy (unobservable electrons, two computers with one input–output function) | text |
Concepts
- phenomenal NCC: the minimal neural basis of the content of an experience.
- access NCC: the neural basis of a content’s being broadcast in the global workspace.
- core vs total NCC: the part that distinguishes one experience from another, versus the whole sufficient condition.
- criterion (SDT): the threshold a subject implicitly sets for reporting a stimulus, separate from the strength of the signal.
Connections
- Block (1995), LIT-643. The conceptual distinction this paper gives two neural correlates.
- Block (2007), LIT-646. Restates the MT/V5 evidence nearly word for word (its §13) and adds the overflow argument and the mesh argument.
- Chalmers. Cited (refs 4–5) as accepting the distinction but holding that P and A coincide in the real world. “Facing Up” (LIT-406) states that coherence principle.
- Dennett, “The path not taken” (1995), cited as denying the conceptual distinction. Not held.
Bearing on the record
- THEORY-102. A sustained, content-specific ventral temporal code with an onset-only prefrontal burst is the pattern this paper’s two NCCs would predict if the first is phenomenal and the second access. That is my reading, offered as a candidate for that theory’s discussion.
- THEORY-100. NCL activity that follows the reported percept is, in these terms, a candidate access NCC. NCL is the bird’s functional analogue of prefrontal cortex, where this paper places access. On its terms the crow finding says nothing yet about a phenomenal NCC.
- No ML instruction.
Limitations
- Every empirical claim is second-hand and summarised in a few lines; the paper is an Opinion piece.
- The phenomenal NCC is identified from reported cases, the circle Block (2007) names “the methodological puzzle”. This paper does not yet face it.
- Exclusion results support “weakly conscious” stimuli only on one SDT interpretation; Visser and Merikle read their own result as attentional.
Open questions
- Box 2: does the fusiform face area in an extinction patient show recurrent activation, and figure modulation, when he says he sees no face? If so, that would be a phenomenal NCC without report.
- Why recurrent activation of MT/V5 should be the basis of seen motion rather than another experience or none: Block calls this a form of the hard problem.
Corrections
- none to a seeded skim (there was no seed)