The predictive coding hypothesis is a theory on the brain, that is very interesting from a psychonauts point of view.
Neuroscientists are working with it, but it's also being used in the field of A.I.

What the theory states, is that perception is not simply a matter of making observations.
The theory suggests that the brain is actually making predictions all the time. These predictions are something like a conditioned response, based ofcourse, on previous experiences.

According to the theory, the brain is a complex structure, that consists of several layers.
At the bottom layer, the processing of single details is taking place. It deals with the very rudimentary aspects of, when it concerns visual processing, a picture.
The next layer will be putting these atomic details together, to form a bigger picture, that would still be just a mere detail of the whole picture. And so, layer by layer, a whole picture is being formed step by step, until eventually on the top layer, everything, from seeing, hearing, smelling, sensing, is being put together, to create the whole of counscious experience.

But it's the way these layers are doing that, that's the essential part of the theory.

The theory of predictive coding rests on the assumption that the brain is not directly making predictions about the environment, but rather, that it is constantly making predictions about itself. Predictions about how it is going to respond to the data being fed to it.

It is doing so, by letting each layer making predictions about the layer right below it.
So on a linguistic level this would mean for instance, that the top layer would be fed the words: "to be or not to be, that is the...", and, having heard these words before, it's going to predict that the layer below it will continue this sequence of words with"..least little thing i care about when i'm drunk". And that layer would have been processing just chunks of sentence, so it is going to predict that on a layer below it, there will be a branch of braincells producing:"when i'm drunk". And yet a level below that layer, there will be a branch of braincells producing "drunk". And finally, there will be, as a result of this prediction,a layer, on it's turn predicting the letter "R" somewhere.

A crucial part of this model is though, that the brain is being corrected whenever errors occur. So for instance, the last word of the sentence could actually be "dead" instead of "drunk". In that case, error detection will occur in the branches doing "R" and the following letters, and from there on, it will, as a sort of feedback mechanism, go all the way up again.

If this theory is correct, i think it would be sort of abvious to assume that hallucinogens are somehow disrupting this error detection system. I think it would be a neat explanation for what we experience when we're on hallucinogens.

I see how it applies to some areas of human perception and cognition, but can you elaborate on how it explains what we experience on psychedelics? It seems quite vague. I'm not sure how the whole alien insect motif, for example, could be explained by such a theory.

Well, i suppose that, without error detection, the brain would be just running it's own script. The error detection is important because it helps the brain to learn and make inceasingly better predictions, but the most important reason for it's existence is ofcourse to correct the brain directly. The brain is not clairvoyant. Sometimes it will simply make a false prediction, and that's why we do need sensory imput to correct for errors. If this error detection system is being shut down, all we would be perceiving is what's being made-up by the brain itself, uninhibited by sensory imput, 'reality'. In other words..'fantasy'.

We spoke about this in another topic, and I generally think you're on the right track, at least where error-detection is concerned, although I think the nitty-gritty of predictive coding is making a few too many leaps in logic for me. I know it works well for comp sci people and AI researchers, but just because an algorithm successfully mirrors human behavior does not mean that humans are using that algorithm.

I think a simpler, and more elegant, predictive explanation is that the at time t, the brain tries to predict what kind of sensory input it's going to get at time t+1, which is similar to the predictive coding theory you've outlined, but it removes the extra layer of the brain trying to predict how it's going to respond at t+1.

They may actually be the same thing (each 'layer' predicting what's coming down the pipe from the layer below it), but I think it makes more intuitive sense this way and removes the need for the self-referential stuff, which can be confusing.

The interesting question for me is: where does this error detection occur? There's evidence that the cingulate cortex is activated in error-detection task paradigms (and as a bonus, is suppressed by psychedelics) and is overactive in OCD patients (who I hypothesize have over-active error-detection circuits), or if this is a more decentralized process.

I'm inclined to think that it's more localized.

Blessings
~ND

Yeah, it was the 'serotonin/depression' thread. I think that the theory is interesting enough, from the perspective of anyone who's interested in the psychedelic experience and possible explanations for it, to have a little thread of it's own.

You're right that the theory isn't perfect. I think there's probably a lot more two-way traffic going on between differen layers. After all, to have a good understanding of what's going on around us, we also need to 'look back' every now and then. For instance when sensing motion, like objects moving with great speed. If we could not 'go back in time', then, as our brains aren't that quick, we would probably perceive such objects as not one object but as a series of unrelated objects 'popping in and out of existence' on a series of different locations.

There actually is some sort of brain dissorder that causes people to experience motion that way. Must be confusing.

What you're suggesting probably comes down to the same thing indeed. I think the reason why the predictive coding theory focusses so much on this 'layer thing' is because of it's application in A.I.
But the theory can also be seen as an integration of what you're suggesting, and the 'modular thinking' that has become fashionable in neuroscience.

An interesting aspect of predictive coding, is that the predictive coding mechanism in a sense also encompasses itself, like in a weird escher loop. The brain also learns to predict how accurate it's predictions are going to be in different circumstances: when you're in your own home at night, then predictions are probably going to be very accurate. The need for error-detection is less great then. When you're in a dark place you've never been before, the error-detection 'device' is probably becoming a little more active.

I don't know if the implications of this are that, the more often you take hallucinogens, the deeper you'll be able to delve into the experience. I suppose that most people here probably do experience it that way.

The disorder you mention is called akinetopsia, and the existence of such disorders is actually a really strong argument for 'modular processing' since they almost always emerge after a stroke to a particular brain region. There is definitely feedback from 'higher' levels to 'lower' ones...I just read a paper about that, but I can't remember the citation off the top of my head. It was in the visual cortex, I think. The existence of sensory gating is a pretty good example though, where a signal can propagate back to suppress a function.

I'm not an AI person, I'm a neuroscientist, so I can't speak to the AI side of things with any real expertise but to say, again, that just because it works in computers doesn't mean that's how it works in the brain.

The existence of the escher-loop is actually kind of cool, given how clearly self-reference is implicated in consciousness, I wonder if there's something there. Maybe time to dust off my copy of GEB and give it a re-read.

Blessings
~ND

You know, ibogaïne causes a visual phenomonon that is very close to this akinetopsia disorder. It's very odd. First you see just a series of objects. Then, sometimes even e few SECONDS later, you see motion. The perception of motion is not being shut down completely, but it's delayed significantly. It's a realy confusing and also quite spectacular visual effect. The odd thing is...when motion finally occurs, it happens in a very complex way. Parts of the object in it's previous position are very rapidly being moved towards their new location and when they merge with the corresponding parts of the 'new picture', a bright flash of white light is seen. And this is happening very quickly, but also piece by piece, as if the visual system is literally 'putting the pieces together'. So you constantly see a series of bright flashing lights, wich makes it even more disorienting.

And the thing is, everybody who takes ibogaïne will experience this odd visual efect.
Ibogaïne is probably a very interesting substance for a neuroscientist. If it wouldn't be such a heavy drug with significant risks attached to it, i would definately reccomend it.

Part of what I like and don't like about your theory both stem from how DMT hyperspace presents itself to me as being systematic. On one hand, the sort of pyramid layering you describe as with the example you present of the sentence with the word "drunk" in it, could lend itself well to a systematized kind of processing. I kind of start to lose you on the ground of error detection however. I feel like this is a kind of analogous argument to the DMT experience being the result of neurons misfiring or the stimulation of the visual cortex or something along those lines. I feel like when you look at the experience like this, it just doesn't add up because DMT hyperspace is so rich and complex, but most importantly, it is interconnected in a systematic fashion.

I experience DMT as ethereal energy that is localized around my surroundings, which coalesces and arranges itself into the entities/architecture/objects/landscape of hyperspace. The result is quite frequently holographic and multidimensional. This energy, which is the composition for everything is affected by/affects a variety of elements, including light, sound, heat, vibration, breath and focus. Any one of these elements can affect the energy (including what I've mentioned the energy composes), which can subsequently affect perception of the other elements. An example would be sound in the form of music, entraining the motion of entities (made of energy) that appear as visual light. In some cases, artifacts of light are perceived as a result of hearing the sound.

Then there's also the matter of spiritual/religious themes including the cultural themes of the Egyptians, Mayans, and Hindu, which are commonly encountered by users. Why in your philosophy, would these be encountered, as opposed to say Spider-Man or video game characters or other pop cultural elements? I suppose the point I'm trying to make here is that I have a hard time attributing this very interconnected system as a result of error detection.

It's not realy like the idea of 'neurons misfiring'. I think you should rather compare it with dreaming or fantasizing. It's like the brain is actually creating dream-like imagery all the time, but this imagery, this 'dreaming' is normally being 'guided' by sensory imput.
Whenever we see something, it's not realy the senses, but rather the brain, creating the image. The senses merely 'steer' the imagery in the right direction.
So when this 'guiding mechanism' is being supressed, the brain can run it's own program. It can fantasize freely, uninhibited by any kind of correction mechanism. That's the idea.

I think that in no way contradicts what you're saying. We all know from experience how rich, how detailed and how convincing dreams can be.
But when you look at the famous 'A.I. dream pictures' that have been the result of a simmilar proces as the predictive coding mechanism, then it's clear that if pattern-recognition is not being inhibited by some kind of correction mechanism, you can easily create very psychedelically looking, multidimensional images that don't seem to bear any resemblance to everyday reality as we know it.

And the content of the experience...i have no idea why that is either. But when i sleep at night, i don't dream about spiderman either. Maybe many pop-cultural elements aren't just interesting enough for the brain to dream about.

I really like the way you put that! It reminds me of the feeling of coming back from a dissociative salvia state of mind, where the first sensory signals that you (re)percieve are profoundly influencing your perception.

But how does this fit together with the range of focus that human minds usually have? This could also imply, that we shift the point of attention mostly by our senses, but what about psychological guiding of "dreaming"?

Attention is something mysterious. I think that if we would know what attention actually is, we would probably understand a major part of how counsciousness works. Attention and counsciousness are weird things: we all know what the words refer to, in the sense that we understand the meaning of a phrase like 'pay attention', or 'it's an altered state of counsciousness'. But we don't realy know what counsciousness or attention are.

I wonder btw, whether it would be reasonable to assume that serotonin is responsible for error-detection, and dopamine for the actual proces of predictive coding istelf? It's probably not that simple, but i mean, would it be reasonable to assume that these neurotransmitters are playing a major role in those processes?

From a laymans perspective it would seem Obvious. But when it comes to the brain and neurochemistry, things hardly ever realy áre Obvious. Wich is Obvious, ofcourse, because obviously we would all have to be so obviously dumb then, that nothing realy could be Obvious to us. Obviously.

This has been a helpful simplification of what I was trying to understand about google deep dream a while back. Thanks. I wonder if this explains the openness of the experience too. By not error correcting based on social programming does each thought then lead to higher fantasy? Is this how it essentially "deprograms" us? I feel there must be some separation here between learned correction and innate correction because it seems that some aspects of psychedelia remain fundamentally consistent and perhaps this just disrupts those learned pathways that usually lead to predictable outcomes, allowing us to see the broad array of potential outcomes. I dont know. It's really cool to see the science of psychedelia emerging in our lifetime

I think suggesting a single neurotransmitter system might underlie such a massive cognitive process is a little simplistic. I'm confident that 5-HT *mediates* error detection, probably via the 2Ar, 2Cr, and maybe a few others, but given how few neurons in the brain are actually serotonergic, I doubt they carry the whole load. I imagine that a lot of this is encoded structurally, rather than chemically.

I disagree with the association between attention and consciousness - I think the systems of attention may play a role in the particular quality of our consciousness, but I think, globally, the phenomena of consciousness is largely independent of attention (or any one particular cognitive process). I subscribe (broadly) to Tononi's theories so consciousness has a lot less to do with specific interacting cognitive tasks and more to do with how information is integrated on a cellular level.

Blessings
~ND