Source
- Block (1995), On a confusion about a function of consciousness, LIT-643, read in NOTE-493: §4 (the definitions of P-consciousness and A-consciousness, and superblindsight), §6 (the fallacy) and §7. Target article only.
What was actually shown
Two concepts. Block (1995), LIT-643, §4, separates phenomenal consciousness (P), what it is like to be in a state, from access consciousness (A). A state is A-conscious if, in virtue of having it, a representation of its content is poised for use as a premise in reasoning, for rational control of action and for rational control of speech. A is a functional notion. Block takes P not to be one.
The argument form. Marcel’s thirsty blindsight patient does not reach for a glass of water in his blind field, though he can “guess” about it. From this, Marcel, Flanagan, van Gulick and Schacter conclude that a function of consciousness is to let represented information guide action, and Searle draws the same conclusion from Penfield’s petit mal patients. Block’s diagnosis (§6): “in the blindsight patient, both P-consciousness and A-consciousness of the glass of water are missing. There is an obvious explanation of why the patient doesn’t reach for the glass in terms of the information about it not reaching mechanisms of reasoning and rational control of speech and action, the machinery of A-consciousness.” The step from there to experience is “sliding from an obvious function of A-consciousness to a nonobvious function of P-consciousness”. For petit mal, Penfield’s descriptions are of deficits in thinking, planning and deciding, and “No mention is made of any sensory or phenomenal deficit” (§6), so the argument does not even establish that P is missing.
What survives. Block is explicit that the conclusion may be true. “If that is the argument in any of these cases, I do not say ‘fallacy.’ The idea that the lack of P-consciousness is responsible for the lack of A-consciousness is a bold hypothesis.” P may be the gateway to access, as in Schacter’s model, or P and A may “amount to much the same thing empirically even though they differ conceptually”, and in either case P would have the function. What the deficit cannot do is choose among these and the alternatives, that P is a product of A or that both have a common cause. “The fallacy is failing to make the distinction, rendering the alternatives invisible.”
Why the claim holds even if blindsight lacks experience. Block takes the patient’s word that P is missing, and notes that if it is not, the case reduces to the petit mal one: only access is missing, and nothing follows about P at all. Either way the deficit does not license the inference.
What could have come out otherwise: Block’s diagnosis would fail if access were intact in these deficits, that is, if blind-field information were available to reasoning and the control of action and still went unused. The case he imagines, superblindsight (A without P), would separate them. Block reports that such cases “do not appear to exist” and that each training lead he pursued failed.
What this does not say
- It does not say phenomenal consciousness has no function. It says a deficit in which both are lost cannot show which has it. That P causes A is left open as a hypothesis, one Block calls bold.
- It does not bind a functionalist who identifies P with A. For someone who takes access to be the whole of phenomenal consciousness, the function is consciousness’s by definition. The deficit still does not show that. The identity must be argued separately, and the argument then rests on that premise and not on the deficit.
- It does not show that P and A come apart in fact. Block’s cases of P without A (the drill noticed at noon, Sperling’s arrays, anaesthesia) “don’t show anything on their own” by his own account. His later papers argue for overflow instead (THEORY-103).
- The definition of access is chosen to suit the argument. Block sets A so that blindsight guesses do not count: “There is no right or wrong here … my choice here is mainly determined by the needs of the argument.” On a weaker notion of access, blindsight would have some, and the confound would have to be stated in terms of the access that is missing.
Why the record needs it
Two Proposed accounts in the record assign a function to a felt state,
and both meet this confound. THEORY-090 (ALARM) says the first
function of basic affective consciousness is alarm, with generalised
one-shot learning among its roles. Its promote_when asks for lesion
evidence that such learning disappears when basic arousal is abolished
while reflex withdrawal survives. On this account such a lesion would
show the learning depends on whatever was abolished. It would not show
that the learning depends on the feeling rather than on the feeling’s
content reaching the systems that learn. THEORY-101 makes felt
valence “a common currency” for sharing evaluative information across
control systems in action selection. That is close to a description of
access itself. Veit identifies felt valence with the currency, which is
the functionalist identity of the second bullet above. So the function
there is the felt state’s by that premise, and not by any deficit. No
relation is declared to either. This is a claim about evidence, not a
rival account of what consciousness is for, and it does not carry
either account further.
Connections
- THEORY-094 is the same confound in correlational evidence. A correlate found with report, or without report but with thought about the percept, may be a correlate of access. Here a deficit with access missing may be a deficit of access.
- THEORY-100 states, for crows, that a report paradigm cannot separate access from phenomenal consciousness. That limit is the distinction drawn here, applied to report.