Consciousness Science Can Measure Access Without Measuring What It Is Like

Joseph Levine begins with an invoice nobody knows how to pay.

He grants the physicalist almost everything. Perhaps pain really is a brain state, just as lightning is atmospheric electrical discharge. Fine.

But why should that state feel like pain?

More circuitry, more chemistry, more microscopic enthusiasm. None of it seems to entail the hurt.

Levine does not declare physicalism dead. He merely leaves it with an unpaid bill.

A description of what a system does is not yet a description of what it is like.

Crick and Koch arrive with the sensible response: then ask a question laboratories can actually touch.

They propose the neural correlates of consciousness. The minimal neural events and mechanisms jointly sufficient for a given conscious experience.

This is not surrender. It is a change of job description.

The laboratory can vary a stimulus, collect a report, compare a brain state, and look for the machinery that travels with the difference. The explanatory gap remains parked outside, wearing a raincoat and refusing to leave.

But science has work to do indoors.

Their follow-up makes the method concrete. Primary visual cortex is active during vision without that activity itself constituting the content of visual awareness.

Activity is not automatically appearance. The brain contains many busy departments, and not every one gets to be consciousness.

Ned Block supplies the warning label.

Access consciousness is information available for reasoning, report, and behavioral control. Phenomenal consciousness is the qualitative feel.

The same word had been doing two jobs and collecting one salary.

Notice the difference between being able to report a word flashed at you and asking what, if anything, its appearance was like before the report.

The first question has a response, an accuracy rate, perhaps a button. The second does not dissolve because the button works.

So the shorter route presents itself. Find the neural signature of a report, map the computation that makes it available, and declare experience explained.

The phenomenal ghost has received an appointment and, one assumes, a parking pass.

It has not.

A correlate can be a genuine correlate. An access mechanism can be a genuine mechanism.

Neither claim says why the event is accompanied by a particular feel, or establishes that access and feel are identical. Block’s distinction does not make experiments futile. It keeps them from answering a smaller question while wearing the larger question’s coat.

The field has learned to make that smaller question remarkably exact.

In binocular rivalry, incompatible images shown separately to each eye produce alternating percepts rather than a fusion, while retinal input stays constant.

Logothetis’s monkey studies find that early visual neurons fire regardless of which image dominates, while inferior temporal neurons track the animal’s reported percept closely.

A fixed stimulus and a changing report can be made to sit still long enough for the instruments to interrogate them.

Visual masking performs a different act of mischief. A target is flashed, then quickly followed by a mask that prevents conscious report, and the target can still influence behavior under controlled conditions.

This is an operational foothold. It is not a tiny trapdoor into somebody else’s appearance.

Global workspace theory makes that foothold into an architecture.

A content wins a competition, enters a fronto-parietal workspace, and is broadcast for report, reasoning, and behavioral control. The predicted competition runs on a timescale of about a tenth of a second, and a late widespread scalp response becomes one candidate signature.

It is an elegant account of access. The fact that elegance has a clipboard should not make it a theory of redness.

Integrated information theory takes a rather grander route.

Tononi begins with claimed axioms of experience, that it is intrinsic, structured, specific, unified, and definite, and derives physical postulates from them.

Its quantity, phi, measures how irreducible a system’s cause-effect structure is to its parts. The theory identifies consciousness with that integrated information, and an experience’s quality with the structure’s shape.

It predicts a posterior cortical hot zone rather than a necessary prefrontal role. It also permits sufficiently integrated non-biological systems some nonzero experience, which is how an argument about cortical mechanisms finds itself contemplating panpsychism before lunch.

The theory does not ignore phenomenality. It builds its front door out of it. Whether the construction holds is another matter.

The COGITATE collaboration makes the disagreement earn its lab coat.

Dehaene represents the workspace theory. Tononi and Koch represent integrated information. The parties agreed beforehand on methods and criteria, then ran the tests across several imaging techniques.

A disagreement is much healthier when nobody can move the goalposts after lunch.

Neither theory came out fully vindicated. Both have disconfirmed predictions, and both camps dispute what the dataset says.

This is not science failing to choose a winner. It is science declining to rent the trophy case before the game is over.

Notice what every one of these instruments has in common.

Rivalry, masking, workspace competition, integration measures. Each is built around a contrast between a condition in which something is reportable and a condition in which it is not.

That is exactly the right design for access. It is a design that cannot address phenomenality without first assuming the two are the same thing.

The other frameworks in cognitive science do not escape the shape. A computational account says what a system can represent. An embodied account says which sensorimotor resources a concept recruits. Predictive processing says how a signal is constrained by expectation.

Each is a real explanatory programme, and each explains a capacity.

There may be one framework that eventually closes the gap. That is speculation.

The current candidates make rival claims about representation, learning, action, broadcast, integration, and neural location. None of them makes the gap vanish by standing near it.

What survives is not that first-person life defeats measurement, nor that measurement has finally evicted first-person life from the building.

Consciousness is both a capacity a system displays and a fact of appearance.

Report, discrimination, broadcast, and behavior give science handles on the capacity. What it is like remains the fact those handles do not simply become.

The instruments can tell us when a representation enters the room. They cannot, merely by counting its footsteps, tell us what it was like to arrive.