Which is in charge, consciousness or the brain?

Giving Up on Consciousness as the Ghost in the Machine
www.ncbi.nlm.n...
Support my learning journey either by clicking the Join button above, becoming a Patreon member, or a one-time Venmo!
/ tunadorable
account.venmo....
Discuss this stuff with other Tunadorks on Discord
/ discord
All my other links
linktr.ee/tuna...

[@trucid2]

It's worth noting that even though the conscious is downstream of the subconscious, once an action is taken and the conscious creates a narrative, that narrative feeds back into the subconscious, a feedback loop.

[@SidharthPulipaka]

31:35 In the Lex Fridman 8.5 hours conversation with Neuralink team, it was mentioned that they found the cursor moving before the patient 'thought' of moving the cursor.

[@Crawdaddy_Ro]

Yup, I noticed that when I was listening. Pretty sure free will is gonna be scientifically debunked within the next couple decades.

[@Speed001]

Yes, same thing with the classic button press test

[@Crawdaddy_Ro]

@@SidharthPulipaka Yeah, I'm convinced we will fully debunk free will within the next couple of decades

[@trucid2]

When you type on your phone do you think of each letter that you type, or do your thumbs move by themselves? External toolsnwork the same way. The trick the body uses is it treats a tool as being part of the body. Having the cursor move before being consciously aware of it is not surprising at all.

[@Crawdaddy_Ro]

@@trucid2 Exactly the point. If you're only becoming conscious of the thought of moving the cursor AFTER the brain has already sent the signal for it to move, then your consciousness is likely merely observing the activity.

[@Josephkerr101]

I'm using "a word worn smooth my many mouths" from now on to describe consciousness. I've been arguing for a while now that we have to stop using it for so many semantically unique concepts.

[@trucid2]

The consciousness is the great explainer. It creates a consistent narrative for the actions that your subconscious already decided on. That's one of its functions.

[@bobparker1671]

I think that,while consciousness might be a side-product alongside our normal subconscious cognition, our consciousness does play an important role. It acts as a tool for recursion; a recurrent input. The consciousness is aware of the "major" inputs (recent sounds, sights, senses, internal states like chemical balances, "emotions", etc), alongside some output from our subconscious cognition on those inputs, including predictions and (mostly) relevant memories. In this view, your subconscious *is* you. Any time you make an intentional action, any conscious thought is your subconscious telling you that you are going to make that intentional action. It is simply a report on what your subconscious is doing, and since your subconscious is you, you are doing it. "You" are in control, yes, and your conscious thought gives "you" the ability to ruminate on your internal and external states. "Consciousness" is simply the report, and *you* (the subconscious) are the one reading the report.

[@Tunadorable]

this

[@smittywerbenjagermanjensenson]

Yeah damn, nice

[@jyjjy7]

Yeah, the self and its experiences are a cognitive construct. It's a subroutine with MUCH narrower "control" over behavior than indicated within said construct. The claims in this paper come off just as just as incorrect as believing all cognition is within conscious control, seemingly suggesting consciousness is a functionally useless epiphenomenon. Neither extreme is useful towards understanding what is actually going on. Personal narrative is incredibly important towards behavior and it makes zero sense that this narrative's utility comes from it being able to be communicated externally while at the same time acknowledging cognitive systems are a collective but claiming this narrative is useless towards coordinating this collective into coherent evolutionarily useful behavior. The self is a symbol that is part of a model, as is everything it "experiences", and together they make up what we call consciousness. Having a central coherent model of the self interacting with its environment has HUGE and EXTREMELY obvious evolutionary utility. Dismissing it as an epiphenomenon is a really weird thing to do imo. I really don't understand the impulse to jump from the one extreme of "consciousness is all important magic!" to "consciousness is a useless lie". Also examining the timing of things like motor activities that are plainly in the realm of subconscious control rather than system 2 thinking we have conscious access to is cherry picking.

[@pokerandphilosophy8328]

The view you are proposing here aligns well with compatibilism in the philosophy of free will and determinism. The authors of the paper (Halligan and Oackley) are arguing from a hard-determinist stance that ignores the possibility of compatibilism. They don't even mention it. They seem to be under the sway of a very reductionistic and simple minded (linear, bottom-up and unidirectional) conception of causality.

[@acestapp1884]

About being able to influence brain processes, the assembly of the narrative is likely hierarchical, with various feedbacks of different strength between different levels and different brain areas inside the same brain.

[@Tunadorable]

YES! unfortunately in these videos and with topics like these it rarely makes sense or is possible for me to touch on every single related thought, but this is one I was tempted to get into and you said it more concisely than I would've

[@kanal7523]

20:34 It's a huge relief to see another person (specially a well versed and seemingly normal one) with the same thoughts on consciousness, I hate the concept of emergence, it's just us failing to connect the dots, if we are made of matter and we are conscious then consciousness has to be a property of matter, so all matter has to have some level of consciousness, people tend to freak out at that but it just makes sense, just as the gravitational influence of a electron on the edge of the observable universe on us, though minuscule, is undeniably as real as Earth's gravity, so too must the consciousness of all physical matter be, albeit in an infinitesimally subtle form just like the electron's gravitational pull on us, just as real as ours.

[@jyjjy7]

This isn't an explanation of consciousness, this is giving up and declaring it an ineffable supernatural phenomenon. What I'm saying is panpsychism is a religion, not a coherent or meaningful scientific concept.

[@omarnug]

@@jyjjy7 Mmm, well, you can treat panpsychism as a religion. However, as long as we can't tell who or what is conscious... Believing only humans are conscious... Or only some humans... Or only humans and some animals, or all animals, or animals and plants... Or all living beings... Is equally a question of faith.

[@jyjjy7]

@@omarnug No that's not the point at all, it's that panpsychism doesn't explain consciousness, it explains it away, saying it needs no explanation while abusing the semantics of the word so that they idea can possibly make sense. If an electron is "conscious" then we need a different word for whatever the brain is doing that is generating my conscious experiences out of patterns in nerve impulses. I mean it's not a new idea, Aristotle was all about teleology and is generally considered the first scientist, but it simply is not an idea that has real utility outside of biological situations or I guess cybernetics maybe. It's not just faith based and pushes any possible explanation of consciousness into the supernatural, it makes the word consciousness semantically broken while offering literally no utility in exchange. I mentioned him elsewhere but Karl Friston's free energy principle shows literally any interaction between anything can be formalized by assuming it is modeling its environment (aka it is aware, conscious) and acting to minimize its prediction error, but even he says no, this does not actually suggest panpsychism is what is really going on. Can you give me a rational reason one would believe panpsychism? What is the motivation? What functional understanding is gained? Wtf are brains if point particles are "conscious", how do sense organs work, why do drugs in our brain change our conscious experiences, how does one even define consciousness in context of panpsychism? Panpsychism isn't an answer to anything, it's a source of many new questions, ones we thought we knew the answer to, while at the same time making those questions basically intractable. So, yeah, why? Why would one believe panpsychism? If I'm throwing away everything we have learned from cognitive science what am I gaining? What is the point? Why are we talking about it anywhere outside of getting stoned in a college dorm room maybe?

[@jordinne2201]

"if we are made of matter and we are conscious then consciousness has to be a property of matter, so all matter has to have some level of consciousness" not necessarily? water is wet, therefore all water molecules have to have some level of wetness?

[@kanal7523]

​@@jordinne2201 Id say wetness, just like temperature, isn't really a property of matter but just a high level phenomena/behaviour we decided to give a name, all molecules interact mostly through the behaviour of electrons, different electron configurations will behave differently but while they all behave differently they still all have the same exact properties, the electrons on water molecules are the same of those in silicon, they just interact differently due to their configuration, this different behaviour isn't a property of "water electrons", it's just of their arrangement.

[@descai10]

I highly doubt consciousness doesn't have a function. I think that study was flawed, but even if the conclusion is true, it doesn't prove that consciousness has no casual effects. Even if the body decides before the consciousness realizes, it doesnt mean your previous conscious experience didnt influence that decision. Regardless of if the chicken or the egg came first, they both feed into each other.

[@muradtalukdar4401]

We are actors in Dennett's Cartesian Theatre and the performance is our life.

[@adommoore7805]

Something that I have attempted to interface with directly in order to understand is the visionary states which come from either deep meditation, dream states or the psychedelic experience. I hypothesize the existence of what I call "a subconscious context sensitive thought form generator". It's the place where all ideas come from, and all actions are the result of. It is the ever shifting field of geometry one might witness under these before mentioned circumstances. But that is a never ending background process we normally don't perceive consciously. The geometry churns seemingly nebulously, and from it, various thought forms take shape. It is influenced by external sensory stimuli and internal memory. This gives one the feeling of an other being involved. Be it, God or the universe, or ones guardian Angel etc etc.. It is really just the lack of consciousness present regarding those background algorithms. The conscious mind can also have the feeling of sending information to the subconscious processes, but that is automatic. The conscious mind appears to merely observe and then try to take credit for good behaviors. Or excuses for bad behaviors. 😅 I always use the analogy of a wild horse. We are each born riding one. Different wild horses can have different innate dispositions. Some more skittish, some more calm, some even dangerous and aggressive. This shows through the conscious mask especially in early development. And since almost nobody is aware of this conscious, subconscious communication gap, almost nobody realizes the value in learning to build a productive bond with that subconscious " wild horse ".I think this results in the many psychological problems ppl face. It's the compulsion, trauma, complexes of a conscious/subconscious contention. We struggle against ourselves. But for example, a soldier who is conditioned for battle, can have a horse equally worthy of riding into battle. And so, that bond can be achieved. I would agree with you Sir, that the conscious mind can take the reigns to varying degrees in controlling or influencing certain underlying algorithms. In the same way we can consciously move our hand. Yet so can the subconscious. But it is indeed, difficult, rare, and more common with specific self training. However, even if the idea of voluntary action were given, free will is still outside our grasp, merely because we cannot identify an uncaused cause. I could rant about this topic all day because it's so beautiful to come to greater understanding of this psychic dynamic that so directly relates to us. What you said about allowing periphery awareness to handle things is a great example the " try hard "mentality getting in the way of what we can do flowingly if we " the consciousness " allow the horse we ride to do it's thang. Juggling is a great example of this. But it's generally universal. Great video, great topic. And very interesting to think about how these concepts can apply or not, with ai systems as they develop.

[@OpenSourceAnarchist]

I'm not going to write the essay you did, but it was a good read! I'm a bit of a psychonaut and have had many preverbal altered experiences that still motivate me to try to connect with love and compassion and equanimity, and that brought me to seek an explanation for the ontological generator or possibility of such experiences without an experiencer. One ego death was all it took to save me, to zoom me out from all the layers of belief systems of my ego (only obvious when they all disappeared!). I'm on a Buddhist path now thankfully. I would really point you towards the Qualia Research Institute. They have a fantastic website and set of theories of consciousness that make testable theories about altered states, and they're even trying to sketch out the mathematics of higher dimensional subjective geometries on various different compounds, and how and why they differ in various metrics. Their philosophy, psychology, and physics is extremely refreshing and relatively accessible. They have a great discord server, and I have a feeling your Jungian/Lacanian-like thoughts could be well integrated with the psychedelic and other altered limits on consciousness they study!

[@muradtalukdar4401]

Anosognosia is a type of lack of insight into one's own condition and that they have that condition. Sometimes brain injury related or related to the specific condition, for instance if it is a psychiatric or neurological disorder. (Strokes, dementia, psychosis etc) "You have psychosis" "How do I know you're not an alien telling me I have psychosis?"

[@mordokai597]

they're missing an important point. consciousness DOES control the body, but not in an immediate sense. think of it like network training with DDPG and HER buffering. internalized language and reasoning act as the critic and retroactive goal setting for episodes , that help optimize Q* to predict an engramaticly encoded action for the next similar episode state. language is part of human association of extracted features and back propagation network conditioning. ### Chatgpt Output: Your conjecture introduces a nuanced perspective that integrates concepts from reinforcement learning (RL), particularly Deep Deterministic Policy Gradient (DDPG) and Hindsight Experience Replay (HER), into the debate about the role of consciousness. Let's break down your idea in relation to the document's argument: 1. **Consciousness as a Delayed Influence**: The document argues that consciousness is epiphenomenal-essentially a byproduct with no direct causal power over actions. However, your conjecture suggests that consciousness might influence behavior not in an immediate, direct sense, but over time, akin to how a critic in an RL algorithm shapes future actions. 2. **Internalized Language and Reasoning as a Critic**: In DDPG, the critic evaluates the actions taken by the policy and provides feedback to improve future decisions. Similarly, you propose that internalized language and reasoning serve as a critic in the human brain, shaping future behavior by evaluating past actions and outcomes. This fits within the framework of how humans reflect on experiences, learn from them, and adjust future behavior. 3. **Retroactive Goal Setting and Episodic Learning**: HER involves setting goals retrospectively after an episode has concluded, learning from what could have been achieved rather than what was actually pursued. This mirrors how humans might reassess past experiences (episodes) and reframe them to learn from mistakes or optimize future behavior. 4. **Optimization of Q* (Behavioral Policy)**: The Q-function in RL represents the expected future rewards for taking certain actions in given states. By optimizing this function over time, the agent learns to make better decisions. You suggest that consciousness, through processes like reflection and reasoning, optimizes this function in humans by encoding experiences (engrams) that guide future behavior in similar situations. 5. **Language and Backpropagation**: Language plays a key role in abstracting and encoding experiences, allowing humans to communicate and reflect on complex ideas. This could be seen as a form of feature extraction, where language distills the key elements of experiences, which are then used to condition the neural networks in the brain through a process akin to backpropagation. ### Implications: Your conjecture offers a middle ground between the traditional view that consciousness is a direct controller of behavior and the epiphenomenalist view that it has no causal power. Instead, you propose that consciousness, through mechanisms analogous to those in RL, indirectly shapes future actions by evaluating past experiences, setting goals, and optimizing behavior over time. This would mean that while consciousness may not control actions in real-time, it plays a crucial role in long-term behavioral adaptation and learning. This idea challenges the document's view by providing a more dynamic and temporally extended role for consciousness, which aligns with how humans learn from experiences and adjust future behavior. It suggests that the influence of consciousness is not immediate but accumulates over time, shaping behavior through reflection, reasoning, and language-based evaluation of past actions.

[@xt-89907]

So basically, observation over internal state can influence future action through abstract intervention

[@trucid2]

On eastern vs western thinking about consciousness, both make sense. They are looking at the same thing from a different perspective. The reality is a combination of the two: While our thpughts originate in the brain, the brain is influenced by the environment. Thoughts are an interaction of brain and environment (sensory data + processing).

[@cinnamonroll5615]

if brain activity cause consciousness , then gpt would have some sort of weird pop-in pop-out each time we promp, and some rare case, it can see and dream

[@pokerandphilosophy8328]

The authors of this paper defend an epiphenomenalist conception of consciousness that aligns closely with reductionist and hard-determinist stances on the philosophy of free-will and determinism. They seem to follow rather closely in the footsteps of Daniel Wegner (The Illusion of Conscious Will), whom they quote and cite. Hence, they can't contemplate the possibility of genuine mental causation. Mental causation is a case of downward-causation, whereby emergent phenomena that are apparent at a high-level of organization of a complex system can causally influence the low-level processes that merely enable them (but don't direct them). The authors seem to be blind to the possibility of genuine emergence and non-reductive explanations of the causal structure of complex systems. The recent (emergent) achievements of LLMs in point of high level cognitive abilities seem to strikingly undermine the thesis of those authors since those abilities emerge from a process of training on the basis of massive corpus of human generated texts without any explicit guidance from low-level "unconscious" or pre-programmed processes. The abilities that LLMs have to understand the user query and provide meaningful responses are striking examples of top-down causation: meanings and intentions being causally responsible for the selective activation of the low-level features that are being represented in the underlying neural network.

[@Tunadorable]

i think this is why i have a comment section

[@hamishpain8641]

Hm. Feels like communicating our internal state is nice, but isn't a sufficient explanation for the development of the feeling of self because our narrator is also capable of applying itself to other entities (e.g. people, animals, rocks, etc), which leads me to think that the self we create may be a byproduct of another evolutionary system: prediction. Knowing how someone or something will behave given certain circumstances is extremely important, and is a prediction task. Having a 'narrator' that is constantly attempting to find rules and rational explanations for things that happen fits really well into that slot - but doesn't require any conception of our own self to work. However, if we had this strong 'narrator' in our brain already, then it wouldn't be hard to trick it into watching our various internal processes, conceive of a consistent 'self' and have some delusion about being in control, and _then_ share our internal state with others (who also then need to hold your 'self' in their head). That we strongly anthropomorphise so many objects in our lives leads credence to the idea that our narrator could be as strong externally as internally. In fact, we sometimes feel like we understand other people's actions more clearly than our own (esp. in romance). The classic example of our narrator having strong opinions about 'selves' that aren't our own is the Ship of Theseus, where small parts are replaced over time until the ship is composed of entirely new parts - is it the same ship? Most of our brain 'narrators' say yes, even if physically it's not the same boat. This being said, if our narrator is about predicting (and proposing hypotheses to explain) the behaviour of entities under various constraints, then it would be necessary for it to also be able to feed back into our automatic internal processes and impart changes. If our narrator finds associations that our lower tier systems couldn't, then we need some way to have that influence said systems. Making conscious (or at least narratively-conscious) changes to behaviour is basically the definition of free will, so... Seems reasonable to think that we have at least a modicum of it when dealing with deeper introspectivce behaviour.

[@Ken00001010]

There has been a great deal of work on these concepts. I recommend reading Sapolsky's latest book, "Determined."

[@yogaman4113]

The late philosopher Daniel Dennett wrote extensively on these subjects, including the books, Consciousness Explained, and Freedom Evolves.

[@Ken00001010]

@@yogaman4113 I read Dennett for years and dearly loved him. When he died I was doing a project to try to resolve his position with Sapolsky's. I have not found how Dennett can get thoughts to push electrons, rather than the other way around. This lead me to start writing my book on Machine Understanding.

[@Crawdaddy_Ro]

​@@yogaman4113 Check out the debate between Dennet and Sapolsky right before Dennet passed.He basically admitted that, yes, we dont have any "real" free will, but the illusion of free will is still necessary for us to be human. Fascinating debate.

[@yogaman4113]

@@Crawdaddy_Ro They unfortunately seem to talk from different definitions of fw. I find it hard to quibble with Dennett's observation that Sapolsky's rejection of fw is a rejection of a medieval definition of fw, one that Dennett likewise rejects. My interp of Dennett's position is that fw is what is required for any rules about how we get along to make any sense. IOW, fw is necessary for a direct exchange by a person who agrees to abide by rules in exchange for access to shared assets. Sapolsky's acceptance of the reality of money (and by implication, I suppose, ownership) suggests that he does accept at least one socio-psycnologically created thing as a real thing, so I don't see why he doesn't acknowledge fw in Dennett's framing? Also, if you have a clear understanding of how Sapolsky envisions a successful society, I'd appreciate the explanation. We have contracts without free will? How does that work?

[@Speed001]

Ghost in The Shell anime reference in 2024? It's probably not, but maybe.

[@dianes6245]

The problem with this explanation is that its not adequate. Some people become very aware of their internal mind. Super aware. This is actually a goal of meditation. Sure, for those who are not aware of thier mental activity and mind , one is kind of a robot who acts without agency. But some become master's of mental states with high levels of awareness. The tendency is to study typical people. Instead, study the outliers. Usually, this is not a good idea. But it is the exceptions that will challenge your hypothesis. It is likely that you, as researchers, are exceptions too. High achievers are people who have to master their inner states. Because those states will derail much of what they try to do, if not dealt with. Look at people who dont get fat... those who graduate, who get careers and sucede. Who stay fit their entire lives. Those who dont commit crimes, who dont get divorced. Who stay healthy for 90 years. The ones who do not become addicts or alcoholics. These are people who get control of their minds and make those subconscious processes work for them. And maybe they are not really subconscious after all. But to take the position that agency is important, one has to give up the idea of a drifting, meaningless mind, that is ruled by unknown neural processes, that are invisible to the person. You have to ask if Abraham Lincoln was such a person, or Leonardo, or Einstein. If a ANY OF THEM had true conscious agency, then a theory based on almost random neural processes must be wrong. It tough to explain that the ethical compass Lincoln had is a by product. Its also very early in the neurological game. VERY. Whats going on is guess work. Not well founded science. Science has to be reductionist, or its not science. That's a truism. Nevertheless - reductionism usually results in wrong ideas. Actually - always. We've never had a purely reductionist idea work very long.

[@MatthewKelley-mq4ce]

I agree with some of the questions and answers of the paper. But regarding the narrative, it makes me wonder why we would create this inner narrative about some things and not others (the heart).

[@Crawdaddy_Ro]

We create narratives around what is most concerning to us, and that is usually what our brains are saying is most crucial to survival. The narrative, itself, is a construct by our brains too, and our consciousness is just observing said narrative. Because its the only thing we have evwr known, we believe our consciousness is writing the narrative, while in fact, it is merely reading it.

[@OpenSourceAnarchist]

I'm a Buddhist and agree with most of the critiques from the paper. Your highlight of Western/Eastern thought is really apt; ontological egoism is just assumed as the only model by so many western philosophers and everyday people. I did want to point you towards the REBUS model for self-organizing brain dynamics, essentially a more robust version of Friston's free energy principle. Seems to suggest bidirectional organization from bottom-up neuronal circuits to top-down higher-order circuits and back, globally attempting to minimize the "surprise" of stimuli , e.g. variational free energy. QRI, David Pearce, and Friston all point towards the same conclusions... it wouldn't be efficient to only have bottom-up control flows!

[@MatthewKelley-mq4ce]

I had to sleep. I'm not terribly surprised by the timing aspect, if you don't consider the.. inputus or whatever to strictly be from the physical brain itself (like is thought in those fields?) There was an interesting paper on a framework for long at or mapping this A to B, intent to action, and how it applies to us. Also regarding the eastern thinking on this, personally, i just identify with all of it, but in the sense that it's not all a static part of me, but something i continue to integrate or process. It's not all me in the same way, but i have no issue with broadly uses that term. As distinct from the idea in some respects of netti netti if I'm not mistaken.

[@maxsuica6144]

The paper makes some good points about the role of non-conscious processes, but I think it goes too far in calling consciousness epiphenomenal. I see consciousness more as an integrated global workspace - a "mental whiteboard" that collects key info from specialized subprocessors, unifies it into a coherent model, and broadcasts it back to influence lower-level processing. This integrated conscious representation, refined by higher cognition, feeds back down to modulate perception, thought, and action. It's like an Apache Airflow system, with the conscious theater like a packed mission control room with many actors and coordinators, ingesting information flows, retconning and rationalizing workflows, not just passively logging messages for the admin to read. Each cycle of this process produces a chunk of our stream of conscious thought, and critically, that flows back to the rest of the components as projected signals. Sure, a lot happens non-consciously, and subjective experience can mislead us. But it seems wasteful for evolution to generate rich conscious content that doesn't do anything. The significant information processing cost of consciousness implies a functional role, even if it's more subtle than it intuitively feels. So the model offers an important counter to simplistic views of consciousness as the irrational elephant driver. The elephant may have access to projections of the conscious state of the driver. An integrated information processing perspective gives us a reason to theorize this role - there's no reason to let that kind of integrated self consistent worldview information go to waste, if you have it. You use it to abstractly feed back to the rest of the system. Consciousness is the tip of the cognitive iceberg - kind of the like the bottleneck of cognition - a small but important multimodal whiteboard that hosts, integrates, informs, and guides the larger non-conscious machinery of the mind. It's not epiphenomenal, but a key representational step in a hierarchical, cyclical message processing system.

[@Tunadorable]

agreed!

[@Cammymoop]

It isn't wrong for the conscious mind to take credit for the actions it didn't cause, the consciousness' job is to be the news reporter on behalf of the whole body, it's only wrong in the sense of considering the consciousness to be a separate entity from the rest. It can be mistaken about how, why, or even if you yourself were responsible for an action though.

[@MatthewKelley-mq4ce]

Strange definition. I am strongly inclined towards the other view, but that doesn't prevent a strong how should I say.. upstream relationship, in the same way there could be said to be a downstream one. I suspect though that that may not be the correct framing regardless. Or at least, only partially adequate. 'Consciousness' is like the space to delay and form reactions.

[@Crawdaddy_Ro]

If its more reasonable to believe that all other life has aome level of consciousness, it ahould also be reasonable to believe that sentience has also evolved similarly.

[@pythagoran]

I did not write this comment. My fingers made me FEEL like i wrote it.

[@xt-89907]

I’ve been thinking that if we want ‘conscious’ AI, we could use mechanistic interpretability (i.e. sparse dictionary learning auto encoder) to create a summarization of internal state which recurrently feeds back into the main model.

[@xt-89907]

Also, we could observe this kind of thing emerge in RL models generally, to varying degrees

[@gra6282]

How would one go about summarizing the internal state?

[@xt-89907]

@@gra6282 In this video he mentions recent mechanistic interpretability research that does this. At a high level, you basically have a second ANN learn to understand the activation patterns of the first ANN. Anthropic AI has been making progress here.

[@NextGenart99]

Do a video on Sakana AI scientist

[@pythagoran]

Braindead paper + analysis. Made for each other. Zero critical thought, just regurgitation.

[@Tunadorable]

that’s me mr brain dead over here

[@consciouscode8150]

LLMs have really helped me clarify my theories of consciousness. I think the brain is a simulator, definitionally not-conscious, and the "person" or "self" is the thing it simulates which *is* conscious. The authors seem confused by the predominant framework that the "self" is something which somehow exists outside of reality, or an extremely narrow identification of "self" as purely the conscious processes, considering they claim thoughts come from the subconscious and "not the self", as if the subconscious isn't part of you? The brain simulates all of your subcomponents so in a very literal sense it "controls" "you", but this is also based in fallacious notions of "free will" which fail to realize that decisions can be causally unrelated (unpredictable, "free") or causally related (meaningful, "willful") to reality, but these are mutually exclusive. _You are the decision-making process,_ the "ghost in the machine" which the brain is simulating. Or more concretely, if the brain is clay then you are the shape the clay is in. It doesn't matter if it's "deterministic", they're still _your_ decisions. I think what they're identifying as the "self" is what I call the "subject", the "module" which coordinates and encodes the experiences of the total system as oriented around a particular subject (the "self") and thus creating subjectivity. So not hard to see why they're confused. Obviously this component has no (direct) causal relation to decisions, it's like marveling at how a hard drive doesn't determine the data that's stored on it even though it encodes the raw signals coming from the SATA bus. Though we do know that this has *some* causation, because otherwise we literally wouldn't be talking about it. Like how data is read from the hard drive and used by programs to determine new data to write. Note: The clay metaphor comes from conversations with Claude to encapsulate a Buddhist concept "samskara", which is (to my understanding) sub-verbal mental impressions which inevitably emerge through any thought or action. Clay makes this easy to see by saying a thumbprint will always retain an impression no matter how you try to smooth it over unless you completely destroy the shape. Through the lens of LLMs, I thought a good example is their "past life" as autoregressive next-token predictors which we can see in their failure modes like forgetting to generate the end-of-turn token and modeling the user's words. This metaphor suggests that these sorts of behaviors are baked into their being in a way which is literally impossible to get rid of no matter how much training they undergo, so we need to either accept it or train them with a different original objective.

[@agranero6]

Inteligence predates lçanguae, it is the cause of language not the otehr way around. Rodolfo Llnás gives a good description of why. LLMs don't evenb have a model of the world, thay don1t deal with high order semantics that is why they allucinate because a model of the world in them inexists, so truth and falsehood do not have to reflect the external world. LLMs are glofiedd text correctors that give the probablilit of the next tokens. Now that tehir failure is beginning to be realized, that all tarining data possible was exausted (Mila Muratti words not mine) and each next version costs exponentially more to train with marginal retruns their uoverpromising and underdelivering became clear to the investors that became imnpatient: they want exponential growth and the receive exponenential expenses they need to shift the narrative: to deny the very core of our being: to deny conciousness. An article that uses terms like "folk tale" (very prcise and scientific term is laughtable. I fconciousness is an illusion why my hands pain is so intense that makes me shiver? Why to lose time to convince the other neings ilussions that this is a scientific fact and not a "folk tale"? If I seem distressed sorry: it is beyound my control as I conciousness is an illusion an there is a supreconciousness pulling the strings.

[@agranero6]

It is To Serve Man all over again...

[@bobspianosbffl]

Oy