I admit to having been averse, for a long time, to thinking about “consciousness.”

The very topic felt “woo” somehow; I was even suspicious of the motives of people who wanted to study it.

But in a literal and narrow sense, consciousness is a perfectly normal, mundane thing, with a well-defined meaning in medicine and neuroscience. There is no ambiguity about what an anesthesiologist means by checking whether a patient is conscious.

Eric Hoel recently wrote that pretty much all psychologists, philosophers, and neuroscientists who study consciousness define it the same way, and allude to the same features:

• consciousness includes all subjective experiences

  • including thought, emotion, sensory perception, etc

• consciousness is centered around a self; it is you who experiences things

• consciousness is present during waking life, absent during dreamless sleep and anaesthesia

One can talk about consciousness without being committed to any particular philosophical belief. The belief that nothing in ordinary experimental science can “explain” consciousness, aka the belief that the “hard problem of consciousness” is indeed hard, is only one philosophical approach. You could also be a scientific materialist who thinks consciousness is some particular, as yet unknown, process in the brain.

Hoel’s consensus criteria for consciousness suggest three neuroscience-based approaches to figure out what’s going on in human brains when we’re conscious:

• Neural correlates of perceptual awareness

  • What’s the neurological difference between what happens when we consciously report perceiving something, vs. when we’re exposed to a sensory stimulus but aren’t aware of perceiving it?

• Neural correlates of selfhood

  • What’s the neurological difference between the normal state of a “sense of self” vs. altered states like dissociation or meditative ego death that don’t feel like “having a self”?

• Neural correlates of arousal

  • What are the neurological differences between the states of waking, dreaming, deep sleep, delirium, and coma?

I’m going to try to learn about this stuff; I’m by no means the first person on this path, but I find it useful to log my journey.

Blindsight

One crisp way to allude to conscious sensory perception is the “difference between sight and blindsight.”

Blindsight is the phenomenon by which some people with damage to their primary visual cortex (V1) report seeing nothing at all, but if asked to guess the locations of objects, are able to predict them accurately. In other words, visual perception is happening, but the person isn’t consciously aware of seeing anything.

How does this work?

First of all, anatomically, the retina connects to many areas of the brain, not just V1.

You can even completely remove V1 from monkeys and get “blindsight” effects where they still respond correctly to (some) visual cues and are able to move their eyes to track moving objects.1

So that tells us how people with damage to V1 are still able to process visual information — they’re getting it through other parts of the brain.

The remaining question is “why is blindsight blind?” Why does loss of V1 impair the conscious awareness of seeing things?

Some blindsight patients report “knowing” where a visual cue is — not only do they “guess” right, but they feel confident that they’re going to guess right — but they still don’t report that they can actually see it.

So, is V1 essential for conscious visual perception?

No, actually! Transcranial magnetic stimulation of blindsight patients can make them see phosphenes (bright lights that don’t correspond to anything in the physical world). That’s a conscious visual experience — the patients report seeing the phosphenes — that you can still have even with severe damage to V1.2

Ok, then why do V1-damaged blindsight patients not have conscious awareness of their residual vision?

One hypothesis is that there’s less conscious awareness of visual perception when there’s less input to the extrastriatal cortex, a region adjacent to V1 in the occipital lobe that includes regions dedicated to various “higher” levels of image processing, including shape, color, and motion recognition.3

Some extrastriatal visual processing areas like V4 can show attentional modulation; that is, changes in attentional focus can change neural firing rates. This is suggestive that these areas are relevant to conscious perception; after all, you are more aware of perceptions that you pay attention to.

When V1 is working, V1 is one of the inputs to the extrastriatal cortex; when V1 is damaged, there are other pathways from the retina to the extrastriatal cortex (which is why blindsight works at all — visual input is still “getting in” to guide action) but maybe when the input is lessened or coming from different locations, it doesn’t result in conscious awareness of vision.

Blindsight patients can both respond to and report some awareness of simple forms (which they still don’t report “seeing”), but when it comes to complex textures, they can only respond to them through action; they don’t report having this strange nonvisual “sense” or “feeling” that they’re there.4

This is suggestive, because we know that “first-order visual gratings” (aka a pattern of light and dark stripes) are detected by different brain regions than “second-order visual gratings” (aka a pattern of stripes alternating between low-contrast and high-contrast random noise).

First-order gratings are primarily detected by V1; in the absence of V1, there’s also other areas receiving direct input from the lateral geniculate nucleus (LGN), like V5, which would also be sensitive to first-order visual stimuli. But second-order stimuli are very dependent on the V1→ V2 pathway that allows nonlinear processing of differences-between-differences. And in blindsight, V1 is gone! So while you can still detect the 2nd order gratings a bit (you get a little V2 input directly from the LGN, which remains partly intact), it’s sparse; the 2nd-order-detecting regions get a lot less direct input from the LGN than the 1st-order-detecting regions.

So is “conscious awareness” then a function of the intensity or volume of stimulation? A little input is enough to give you better-than-chance performance on a forced-choice test, but you’ll only be aware of the perception if there’s more input?

If so, this still leaves a mystery.

Let’s say something new kicks in — something we call “conscious perception” — when sensory input is “strong enough.” But where is this “something new”? It certainly isn’t going to be isolated to a specialized visual processing region, because we can also be conscious of other senses than sight. Is it a response, somewhere else in the brain, to the strong signals in other sensory-processing regions? Or is the strength of the signal itself somehow “producing the awareness”, locally?

Both V2 and V5 project back down to the pulvinar nucleus in the thalamus, which is a huge hub for multisensory processing (auditory and tactile as well as visual). The pulvinar is responsible for attentional control, and lesions to it can cause hemispatial neglect (the phenomenon where, often after stroke, a patient might be wholly unaware of a whole half of their body or any sensations coming from that side.)

We’ll come back to this, but it’s a tantalizing candidate for what’s governing conscious awareness.

“Deaf Hearing” and “Numb Touch”

Centrally deaf people (with damage to the central nervous system rather than the ears or peripheral nerves) may have an analogous phenomenon to blindsight called “deaf hearing.” They can have reflexive responses to sounds, and perform accurately in forced-choice tests, while still not reporting hearing any sounds.

Central deafness is much rarer than central blindness. Deafness only results if there’s bilateral damage to the auditory cortex, and sometimes not even then; so we have much less data about “deaf hearing” than about blindsight.

There’s also a phenomenon of “numb touch.”

A patient who was completely numb on one side after a stroke, unable to feel even burning or pain, could still point fairly accurately to where an experimenter touched her on the numb hand. 5

Like some of the blindsight patients who had a “sense” or “feeling” for the location of a visual stimulus but couldn’t “see” it, this numb-touch patient reported being unable to “feel” a touch, but to “know”, “see”, or “hear” where it’s occurring — in other words, there’s some kind of conscious awareness but not experienced as a tactile sensation.

Patients with “numb touch” or “numbsense” generally have lesions to the somatosensory cortex (S1 and S2) but have a different sensory-detection pathway intact, from the posterior region of the thalamus to the posterior parietal area.

Another patient6 who was completely unable to feel touch, heat, or pain on one side due to a stroke and damage to S2, and even unable to guess better than chance as to whether he was being touched or not, could still do better than chance in selecting 3D shapes and objects by touch alone.

What’s weird is that normally S2 is “for” tactile object recognition, while S1 (which was intact in this patient) is “for” simple perceptions of pressure, texture, and vibration. But there may be a preserved information stream from S1 to the intraparietal sulcus, skipping S2, which allowed the patient to recognize shapes and objects.

Again, as with blindsight, we see a pattern that there are many “paths” of neuronal connections to transmit sensations, such that some perceptual abilities may survive even severe damage to the “main path”; but weaker, sparser, or rerouted signals may result in loss of conscious sensation.

You can also induce numb-touch by using rTMS stimulation to inhibit S1. Subjects with S1-inhibiting stimulation had much less touch-sensitivity (they reported “no” more to being asked “can you feel this?”) but still performed well above chance when required to guess which hand was being touched.7

Subthreshold Sensations

One critique of the concept of blindsight is that blind people may simply be uncomfortable declaring that their poor vision “counts” as vision at all, even if they do in fact have conscious visual perceptions of some kind.8

Sighted as well as blind people will refuse to verbally admit to “seeing” things that are faint or ambiguous. There are psychophysical perception tasks performed on people with normal vision where people will report that they “don’t see” an indistinct target, but will still guess right on a forced-choice task. This sort of “blindsight in normal subjects” has nothing to do with V1 damage; it’s simply an artifact of the difference between the very subtle perceptions that people can pick up, and their unwillingness to say something “counts” as perception unless it’s a much more unambiguous signal. Daniel Kahneman refers to this as “criterion content.”9 People differ on their “criteria” for what perceptions “count” as something they’re comfortable articulating; two people may perform equally well on a perception task, but one will claim to be able to see the target more clearly.

On this alternative theory, blindsight isn’t a special, clear-cut example of “vision without conscious awareness”. It’s just an example of the usual thing where we claim not to be aware of any faint, subtle, or ambiguous perception (even if we can use it perfectly well to guide action.)

Anecdotally, on a few occasions I’ve been able to improve my sensory perception (of odors and of internal body sensations) simply by admitting to myself the existence of perceptions that I’d had all along but on some level thought were too faint to “count”. Attention really can improve sensory discrimination; but the subthreshold sensations weren’t exactly absent from my consciousness before I started paying attention. They were just “in the background” in some sense, going unacknowledged.

So, should a subthreshold perception count as an “unconscious” perception?

Certainly if you can see something but don’t feel comfortable saying so out loud, that’s still a conscious perception, albeit one that would be difficult for experimenters to detect.

But it’s also possible for subthreshold perceptions to be things you habitually “shut out” of awareness. And you become more aware of those subtle perceptions — they become “more conscious” — when you’re actively paying attention to them.

So, we might ask, what’s different in the brain between subthreshold perceptions and perceptions we actually notice?

Healthy people have blindsight-like subthreshold effects in lots of senses, not just vision.

Odors can affect people’s emotions, and people can do better than chance at odor-discrimination tasks, even when they report not being able to smell anything.10

Subliminal speech sounds (fainter and faster than normal speech) can effectively “prime” subjects’ responses to audible speech sounds, which means that sound as well can affect our behavior even when we can’t consciously perceive it.11

For auditory stimuli, fMRI shows less (but still significant) auditory cortex activation for sounds too faint to hear, compared to sounds that subjects report hearing.12

This is consistent with the hypothesis that conscious sensory perception requires more intense activation of sensory processing regions than unconscious (but still perhaps behaviorally relevant) sensory processing.

Some interesting evidence also suggests that “unconscious” subthreshold sensory perception…kinda isn’t.

When subjects are asked to perform a forced-choice task on very indistinct visual stimuli, and also to rate how confident they are in their performance, they claim to be very unconfident over a wide range of actual performance, and only start to claim to be 50% accurate when they’re actually over 90% accurate.

However, if you ask them to bet on whether they got the discrimination task right or wrong (a sort of forced-choice test of confidence) they have a nice clean positive correlation between their confidence level and their accuracy level, starting at the very lowest accuracy & confidence levels.13

In other words, people don’t explicitly think (or aren’t willing to say) that they perceive indistinct sensory stimuli, but they’re aware enough of their subtle perceptions to bet successfully on how accurate they are.

Is this also true of blindsight? No!

If you ask a blindsight patient to bet on whether he got a visual discrimination task right or wrong, he’ll bet as though he believes he’s performing at chance, even though he actually performs significantly better.14

So maybe blindsight does, after all, work differently from ordinary subthreshold sensory processing.

One way of thinking about this is that subthreshold sensations in healthy people involve a weak stimulus through all the normal sensory processing pathways, whereas blindsight involves a stimulus that’s weak because the primary sensory processing pathway is damaged and only the backup pathways are preserved. That difference might matter.

Up Next: Attention

Inattentional blindness is a well-known phenomenon — we don’t notice sensations as much when we’re distracted, and we won’t report being aware of them.

They can still sometimes affect behavior, and we can detect brain activity signatures that show some sensory information is reaching the brain — but there are also brain differences in activity between stimuli we consciously notice and stimuli we don’t, in particular greater prefrontal cortex activity when we’re aware.

There are also qualitative differences in psychophysics about how sensory effects on behavior (like priming) work when the sensation is consciously perceptible vs. not.

Roughly, the measurable consequences of consciously attended-to perceptions stick around longer and affect more parts of the brain. More about that in a follow-up post.