ouldn't you just know it: the first time we get a bona fide philosophical
problem, the philosophers give up and hand it over to neuroscience. The
neuroscientists mindlessly apply the reductionism that worked so well
before, and—mutatis mutandis—both sides get nowhere.
We have a good metaphor at hand: microfractures are already starting to appear in dogma in contemporary thinking about AI and consciousness. NIH just announced that grant reviewers must not use ChatGPT to review their grants for them. It is, says NIH, a breach of confidentiality. They even hinted that they want us to actually “think” about the grants and—get this—ChatGPT is not actually able to do that.
It's encouraging to hear somebody say that, but it's also fair to say that, when it comes to AI and consciousness, it's one of those rare times when scientific reductionism leads us down the garden path. Without questioning our basic assumptions, we will get nowhere. Or worse, we will get an answer without realizing that we have merely redefined the question to fit the answer we wanted.
I mention NIH because they're the organization that punishes scientists for asking questions they don't already know the answer to. And what those celebrity scientists really want is to impress each other and to convince NIH their ideas are fundable.
Is asking where consciousness is generated in the brain the right question?
A popular idea being that consciousness is something, like electricity, that the brain ‘generates.’ fMRI studies haven't pinpointed exactly what part of the brain is activated when a person is ‘conscious’ of something, but with all due respect to the existentialists we can't just assume that all consciousness is consciousness ‘of’ something.
Nor can we just assume that the brain creates it. Soon enough an fMRI study will get a result; the usual celebrity scientists will congratulate themselves about how science's relentless reductionism has solved another major problem. Then, just as the phony AI that industry is peddling is guaranteed to implode the first time it gets out of its depth, people will suddenly realize that they still don't really know the answer, and science will have proved once again what every scientific paper is designed to prove: that more funding . . . I mean research . . . is needed.
One reason it's taking so long is that no two people agree what the question is. Neuroscientists think it means awareness of oneself. Perhaps, just as the fusiform cortex handles identification of faces, there's another brain region that handles awareness of awareness. In other words, consciousness is perception.
It would certainly be nice to know more details about how perception occurs, but knowing the brain connections that underlie perception might no more explain what consciousness is than knowing where the filament is in a lightbulb tells us about the nature of light.
So, you're probably saying, okay Mister Smartypants, what question should we be asking, then? I'm glad you asked! Are we asking (1) how perception occurs, (2) why sometimes we're awake instead of asleep, or (3) why we feel like we're one person instead of another? All three definitions show up in the literature. Question (3) is closest to what David Chalmers famously called the “hard question,” whereas the neuroscientists seem content to answer questions (1) and (2). Nobody seems to have a clue about the right way to ask question (3), let alone how to solve it.
Questions (1) and (2) are susceptible to a rigorous reductive approach: stain neurons and use popular techniques like optogenetics and reporter viruses, and buy fancy two photon microscopes and fMRI machines. Answering question (3) is a matter of thinking hard about the question you want to answer. The problem science faces is that NIH will fund one but not the other. In so doing, it guarantees that we will fill the textbooks with answers that may be technically correct but which are in fact answers to the wrong question in the wrong domain.
Philosophy tried hard, but came up with too many answers
Bertrand Russell's monism, which proposed that conscious states are the intrinsic nature of brain states, was mostly a circular argument: what is the nature of brain states that makes them conscious if not their ability to exhibit consciousness? And how could we even test that?
Thomas Nagel famously described consciousness as “what it is like to be x,” where ‘x’ is a person, a dog, a cat biting a dog, or whatever. But this leads to a discussion of qualia and associations among memories of different sensory modalities. We look at a waterfall and experience a variety of sensations and associations within our memory. We may call this being conscious of the waterfall, but plainly it is actually a combination of two separate things: perception and subjectivity, one of which is a lot easier to describe physiologically and phenomenologically than the other.
Another approach is to assume that consciousness is an emergent property, which means that only complex entities with a specific architecture can be conscious. While some version of emergentism may sound intuitively appealing—we have difficulty ascribing consciousness to a mound of dirt or a cell in a petri dish—the difficulty we encounter is that it is unclear what that architecture might be other than something the human brain has that an individual neuron does not.
A third argument is to say that consciousness is only an illusion. This comes from the idea that every physical process is ‘causally complete,’ which is to say reducible to basic physical and chemical laws. Therefore, the reasoning goes, consciousness cannot exist. Critics counter that this conflicts with our daily experience. Science, they say, is designed to exclude any subjective components, so it would need an entirely new language to account for non-objective phenomena. Taken to an extreme, the illusion theory ends up as what philosophers would call dispositional essentialism, which is the view that the physical properties of, say, an electron or of fire, describe all that can be known about it.
Panpsychism: everything is conscious
At the other extreme is panpsychism, which proposes that the intrinsic nature of matter is to be conscious. Thus, a subatomic particle is conscious, but only at an elementary level. An objection to this is that it requires yet another mechanism by which the tiny consciousnesses in electrons can combine to form what appears to be a unitary one in the brain.
One possible solution might be to think of consciousness as a property of information. Information has most of the properties we need: it can be experienced by one person and not another; it can exist in different modalities, which allows for qualia; and it can grow in complexity as one ascends the scale from a simple mechanical relay to a human.
The challenge science faces is how to study that without getting bogged down in tracing neuronal pathways and making patients glow in the dark, which is where the money and prestige are. The other question is: is understanding consciousness really something that people want? After forty years of studying fatal brain diseases and seeing how enthusiastically academics short-circuit their own thinking in their drive to become famous and well-funded (which nowadays are virtually synonymous), I am doubtful.
Nevertheless, the question is fascinating. The answer will tell us the basic nature of our existence and maybe even about spacetime itself. Why are we separate beings? Why am I in one specific place with a unique reference frame, and you are in another? If Einstein had thought about this question and stopped all that nattering about trains and beams of light, maybe we'd know by now.
jun 29 2023, 8:20 am
