A new proposal for how mind emerges from matter
noemamag.com
noemamag.com
Add to that the typical journalistic fault of forcing one to read through paragraph after paragraph of narrative before actually explaining what the thesis they are presenting is. I’d much prefer to read a journal article where they state their central thesis upfront
The article is pretty good making you rethink all your preexisting concepts of Mind/Intelligence based on what we know from Biology today. It is not an article on various theories of Mind but on how scientific research (conveyed through pointers from various researchers) is advancing rapidly on so many fronts that we are forced to confront our very fundamental beliefs.
Absolutely worth reading at least a couple of times.
And an article or book that uses the word "communicate" is even more annoying, since "communicate" seems to mean virtually anything.
End of my rant...
It's language Jim, but more as Chomsky said. or maybe Chimpsky.
I 100% agree with your rant. This time fly likes your arrow.
This metaphor of the pachinko machine (or Plinko game) is exactly how I explain LLMs/ML to laypersons. The process of training is the act of discovering through trial and error the right settings for each peg on the board, in order to consistently get the ball to land in the right spot-ish.
People also can lose your time or resources (first line support, women’s shopping, etc) and the reaction is the same.
I don’t know why there’s still no LLM with a constant reality feedback loop yet. Maybe there’s a set of technical issues that prevents it. But until this happens, pretrained AI will bullshit itself and everyone cause there’s nothing that could hit it on the head.
And fruit flies like bananas. :)
Really the way people use the Chomsky hierarchy in practice (both in linguistics and computer science) is “let’s agree to ignore the finitude of language” - a move borrowed from classical mathematics
I don't believe it is. Somewhere inside our brain there is some perception of self related to the outside world. So we can project self into the future using information from the now and make better choices and survive better ("Information is that which allows you [who is in possession of that information] to make predictions with accuracy better than chance"- Chris Adami). Why do we need all these difficult words?
I bet animals also have some image of self inside there somewhere, and make decisions based on simulated scenarios. Perhaps to a lesser degree, perhaps because of a lack of language they experience it in a different way? Not being able to label any of the steps in the process... Who knows?
Perhaps when we get to simulate a whole brain we can get some idea. But then there is the ethics. We do attribute great value to organisms that have this image of self.
And I don't get why, it seems to me quite self evident that you yourself can experience reduced-consciousness states, be it being half-asleep or quite drunk.
If we create an analogy with audio - audio can have 2 properties: frequencies and volume. Frequencies are the content of the audio and volume is how "present" the sound is.
Well the same could be applied to consciousness. When we say reduced consciousness do we mean that the mind experiences less content (frequencies) or do we mean that all the frequencies are there but at a reduced volume?
In any.
C. is a complex thing, and these can fail to lower level in myriads of ways. You have to align millions* of things to make it work. The biological (in our case) nature of it allows for some slack rather than immediate breakdown, so there are thousands of parameters to play with.
* numbers arbitrary
E.g. I have aphantasia - I don't picture things in my "inner eye". Through discussions with people about that, a lot of people described own differences from the perceived norm, and it is clear a not insignificant numbers also have other differences, such as not thinking consciously in words.
A lot of people then tend to express disbelief, and question the whole thing on the basis of a belief that if people's inner life does not match theirs, people couldn't possibly be conscious or reason.
People make far too many assumptions about the universality of their own experience of consciousness.
If you could wouldn't that involve you imagining what the building looks like from the inside and how the rooms are connected together?
Or would do it some other way?
(I'm assuming you could do it, since if you couldn't I think that would be a majorly debiliating condition rather than just some amusing fact)
I can "imagine" what a building looks like, but I can't see it. Even our language makes this hard to describe, and the reason why I didn't realise that this isn't how other people do it, because most of our language for remember how something looks uses terms that implies we see them, so I just assumed until a few years ago (and I turn 50 this year) that that was just a metaphor to everyone else too.
How do you imagine (there's that word again...) blind people remember places they can't see?
To me, the notion that I'd need to "see" what the building looks like to draw it is bizarre because I know where everything is in relation to each other, and their shapes, so why would I need to see it?
EDIT: increasingly, and in part drive by split-brain experiments, I tend to think that a whole lot of what we see as our conscious decision-making, are instead shallow retroactive attempts by parts of the brain at rationalising decisions largely already taken autonomously into a cohesive self/ego.
As a molecular biologist I often deal with minute quantities of substances in small containers, but I just picture them there, the molecules, what happens to them as I dilute a sample or add some enzyme. Reading a book is like watching it play out.
I have a colleague who is also much more text oriented. I really wondered how she can function without seeing all the departments in an overview and the connections between them, and how governance structures overlap with those departments, etc. For me it's a lot to hold in, overwhelming at times. But she's absolutely great in these types of things, just masterful, very structured as well. Her mind must be very alien to me, I always wonder what her experience of life looks like (or not "looks like", apparently... But how it is to her...)
But with some works - like Tolkien's, I will still enjoy the descriptions because the language itself is beautiful.
Just to make this weirder: I remember many things by appearance. E.g. I can find my place in a paper I've read years ago by what the page looks like. But I can't see it.
I do however see things when I'm dreaming, and I have one solitary experience I think of seeing things awake, during meditation. I say "I think", because there is the possibility that I fell asleep even though I don't believe I did, and the imagery was far clearer than during my dreams.
But I also can't recall images from my dreams while awake, and usually don't remember dreams at all past the first 30 seconds or so awake.
If you wanted me to draw an abstract floorplan of my house, or the layout of my town, I can do that easily. I can navigate an extended road trip unassisted by simply looking at a map ahead of time and mentally storing an abstraction of where I need to go. But if you wanted, say, an image of my house, or the main intersection in the middle of town, or anything like that -- I'm no good. I can't "see" it, and I would not be able to draw anything remotely accurate.
To circle back around too your "imagine" vs. "see" problem -- my response to the often-referenced "imagine a ball rolling on a table" exercise often elicits some confusion from people. I can imagine a ball rolling on a table just fine. What color is the ball, though? It has no color. What size is it? It has no size. It is just "a ball", in the abstract -- its only property is shape (spherical), which is ultimately all that is necessary for imagining rather than seeing. If you wanted me to visualize a large red ball on a green table, though, that's beyond me.
E.g. if I draw a fantasy drawing, it will be closer to impressionism in style, the same way as if I draw something in front of me. It'll be messy. If I draw from memory, the lines are clear, and stylized.
The clearest example of that was an art class at school where we were asked to draw our shoes first from memory and then while looking at it, and both were highly detailed, but without any conscious decision, the line-work was entirely different.
Same as you when it comes to the "ball rolling" scenario. It "has" the properties people add to it, but if they're not affecting the behaviour, they're just verbal labels attached to an abstract concept - I'll remember them, but they won't change anything about the "imagined" scenario.
It’s as though the connections among things are dimensional rather than visual. I can “render” blueprints as this thread discussed, or know which way I’m facing deep in an intricate tube station, but am not “picturing” it, it’s just, this is how these things relate in space time. Although I can do the same with exterior architecture (is that an image?) or street corners. I don’t think I’ve ever “seen” anything: picture yourself on a beach, counting sheep, always assumed were metaphorical expressions.
I also feel concepts are dimensional, and can tell if a piece of a concept is missing because where it should be is empty. Think how periodic table of elements worked.
A red ball on a green table maps to billiards in dimensional concept space so I can render that too.
Also fifty. Realized this when I was about seven and it hasn’t materially changed since. I have no problem drawing things like blueprints of my house but when it comes to more curvy objects I’m a terrible artist.
My impression is we attribute great value to organisms that can effectively push back against us.
respect based on force.
Not saying I want things to be that way.
Imagine this: language is an innate ability that all speakers have mastered, yet none are experts in—unless they are also linguists. And what, according to experts, is the source of such mastery? A rigid set of rules capturing the state of a language (langue) at a given time, in a specific geographical area, social class, etc., from which all valid sentences (syntactically and beyond) can supposedly be derived. Yet this framework never truly explains—or at best relegates to the background—our ability to understand (or recognize that we have not understood) and to correct (or request clarification) when ill-formed sentences are thrown at us like wrenches into a machine. Parsers work this way: they reject errors, bringing communication to an end. They do not ask for clarification, they do not correct their interlocutors, let alone engage in arguments about usage, which, under the effect of rational justification, hardens into "rules."
Giving in to the temptation of an objective description of language as an external reality—especially when aided by formal languages—makes us lose sight of this fundamental origin. In the end, we construct yet another norm, one that obscures our ability to account for normativity itself, starting with out own.
Perhaps this initial concealment is its very origin.
Edit: I'd expect this thinker/school also to argue that the being needs to be able to experience its intention as an intention (as opposed to a dumb, inarticulate urge); in other words, intelligence would require an agent to be aware of itself as an intelligent agent.
Edit 2: I strongly recommend Peter Watts' Blindsight to anybody who's on the market for sci fi that deals with these issues.
The article is really about conceptual framing — how clinging to outdated or vague definitions prevents progress in understanding biological and cognitive processes.
People keep forcing everything into the vague, overloaded concept of "intelligence" instead of just using the right terms for the phenomena they’re studying. If we simply describe behaviors in terms of adaptation, computation, or decision-making, the whole debate evaporates
https://chatgpt.com/share/67c07325-3140-8007-8177-c56a89b257...
Almost all the traditional criteria of intelligence - reasoning, planning, decisionmaking, memory etc - are exhibited pretty trivially by standard computer programs. Nevertheless no one would think of them as "intelligent" in the sense that humans or animals are.
On the other hand, we now have LLMs, that sent the entire tech world into a multi-year frenzy, precisely because they appear to possess that human-like intelligence.
And that is even though they perform worse than classical programs in some of the "intelligence" measures: For the first time, we have to worry that a computer program is "bad at math". They cannot reflect on past decisions and are physically unable to store long-term memories. And yet, we're much more likely to believe that an LLM is "intelligent" than a classical program.
This makes me think that our formal decisions of "intelligence" (the ones that would also qualify fungal networks, swarms, cells, societies, etc) and what we intuitively look out for, are really two different things.
The code needed create, train, and perform inference on a Transformer is quite short. How short depends on how you count the `import` statements in https://github.com/openai/gpt-2/blob/master/src/model.py and https://github.com/openai/gpt-2/blob/master/src/sample.py etc.
Spreadsheets performing linear regression etc., — Do they learn? Sure!
If you accept that Transformers adapt and learn then you must accept that a spreadsheet also does, because someone implemented GPT-2 in Excel: https://github.com/ianand/spreadsheets-are-all-you-need
Do polymorphic computer viruses adapt? Border Gateway Protocol? Exponential backoff? Autocomplete? And that's aside from any algorithmic search results or "social" feeds, which are nothing but that.
What differentiates a “traditional computer program” from a single matmul? How about a bunch of matmuls? How about a bunch of matmuls with some non-linear arithmetic thrown in between them? And if we are adding in sampling from a distribution but set temperature to zero and use random seeds so that it is fully deterministic?
What I said is thst a traditional computer program (say Excel) is not intelligent in itself, as it cannot, itself, without further input from a human, adapt.
Excel is as capable of adapting to new input as any AI, and the input need not come from a human specifically: the act of adapting to input is an automatic consequence of being able to perform linear regression.
In fact, this is the real value of a spreadsheet over a piece of gridded paper where the sums are done by hand.
If "adapting to input" is your test for intelligence, however, it is a test which a thermostat will pass.
I said 'sensory input', as in, "any primary, real world data". Can I just download excel, and without feeding it machine learning algorithms, feed it a bunch of videos or text and have it learn history?
>If "adapting to input" is your test for intelligence, however, it is a test which a thermostat will pass.
Not just adapting to specific input the system is designed for, adapt it to ANY input. A thermostats function will never change on its own without modification, just like excels, or whatever other traditional program.
It has some built-in.
Linear regression is a machine learning technique.
Likewise, any symbolic mathematical system that accumulates theorems that speed up future work as it solves current tasks, seems like a high intelligence type of activity.
Deep learning is “just” structured arithmetic.
I think different kinds of intelligence can look quite different, and they will all be “structured” or “tropic” at their implementation levels.
—
Stepping away from the means, I see at least four “intelligence” dimensions:
1. Span: The span of novel situations for which it can create successful responses.
2. Efficiency: The efficiency of problem solving.
I.e. When vast lookup tables, exhaustive combination searches, and indiscriminate logging of past experience can be matched instead by more efficient Boolean logic, arithmetic, pattern recognition and logging, we consider the latter more intelligent.
3. Abstraction: The degree to which solving previous different novel situations improves the success or efficiency in solving new problems. I.e. generalizable, composable learning.
4. Social education: Ability to communicate and absorb learned information from other entities.
Plants, and I expect all surviving life forms, are very high in intelligence types 1 and 2.
Adaptive nervous systems and especially brains excel at 3.
Many animals, but most profoundly humans (whose languages for communicating are themselves actively adapted for compounding effects), excel at 4.
Today’s humans are effectively more intelligent than 10,000 year ago humans, not because of 1-3, but because of 4. Learning as a child to read/write, do arithmetic, understand zero and negative numbers, and countless other information processing activities and patterns, from others, profoundly impacts our intellectual abilities.
Deep learning, as with the human species, non-trivially spans, and continues to improve, on all 4 types of intelligence.
Alternatively, neural networks are also just doing explicitly what they are designed to do… sure there is a larger computational graph with lots of operations, but it’s all deterministic… backprop is not really much different on a procedural level than the simple fitting algorithm that I outlined above, in as much as it is just a specific well-defined algorithm or sequence of steps designed to compute some parameters from data.
Adapting and learning means it chose to concentrate on packing the world into retrievable storage.
When do we not adapt and learn? When we ignore our inputs and do nothing with it (don’t store it, don’t retrieve it).
In the example you gave, those classical programs cannot concentrate, it’s a one and done.
In the example I mentioned, the program clearly is taking inputs (load), storing them, querying for previous values that satisfy certain conditions (the value at a certain timestamp for each of the last 7 days), running computation (computing a mean), and operationalizing the result (pre-emptively scaling) to achieve a goal (avoiding having insufficient capacity) that affects the world (whoever interacting with the system will have different experiences because of the actions taken by the system). That seems like it satisfies your criterion to me.
Not sure I follow your usage of discontinuous: clearly the load inputs are continuous variables (albeit discretely sampled in time and magnitude, but I don’t see the significance of either of those sampling frequencies, each of which can be chosen to be effectively arbitrarily small). Also don’t understand the relevance of continuity at all since you didn’t reference it in the post I replied to.
> A few null or unknown inputs that you didn’t even consider will break it.
Null handling should be assumed to be a trivial problem here (ie just use the last 7 valid values for the time step of interest), and I’m not sure what “unknown” values would be here, can you give an example? I think it’s safe to assume that the inputs have already been normalized to some meaningful scale as an implementation detail of the load sensors. Even if the telemetry scheme changes because the instrumentation changes or the infrastructure changes, so long as the instrumentation produces values at the same scale and the actuators can still respond to the outputs, the core kernel of sense/compute/actuate can still be left unchanged.
I’m assuming that you are citing the failure to respond to null or unknown conditions as a lack of adaptation capabilities, especially as you are referencing its inability to last 100 years - though good luck finding any piece of software that can last 100 years, especially more robust machine learning algorithms that have to deal with problems like data distribution drift, hardware changes, etc etc - same goes with wetware (besides turtles?). I’m strawmanning you a little bit on the appeal to time though, since I assume your main point was to just emphasize that “it won’t deal with changing conditions/new scenarios well” which I think is somewhat fair. On the other hand, I think it is safe to say that what I described does at least respond to the world and in turn affect it in a kind of cybernetic feedback loop, which includes adapting to the shifting conditions to meet some desired state. So maybe it would be helpful if you could define more precisely what you mean by adaptation? Not trying to be snarky, just trying to genuinely push you to try to make a clear definition of what you mean by adaptation.
A calculator is discontinuous graph/program because there is no point to plot when you divide by 0 (undefined). This is true for your classical program. Over time it won't know how to adapt its f(x) function to handle something as crazy as dividing by zero (it can never do it, but unless you put in an error output for that input, the program will never do so on its own).
The belief is these AI programs will be able to make that adaptation. All the changes you bring up that your program can apparently handle are still under your conceptual control (you can map in your mind if this happens, then that happens). You can prove this to yourself when you read your own statement "inputs have already been normalized" - sure, in a world you control.
I'm suggesting a world where your telemetry monitoring function can change into a Reddit function if need be, so yeah, kinda batshit. Why would it need to do that? Remember, we can't imagine why, that's the whole point.
That's how a program could possibly live a hundred years in a world that is constantly changing. Your program can only exist in a very static and predefined world.
There are some that will tell you that you are an ever-changing function. I try to stay out of that simulation :)
https://en.wikipedia.org/wiki/Matchbox_Educable_Noughts_and_...
I wouldn't say that's more "intelligent" than, I don't know, a workplace scheduling system.
The species property of intelligence encompasses stupidity.
Just two? You can name so many more terms in this concept cloud, e.g.: personhood, moral agency, consciousness, self-awareness, processing power, wit, autonomy, feeling-and-experiencing capacity, and so on… We don’t seem to agree on what’s separate from what, and yes, it would be useful.
That's true. We shouldn't casually conflate the two but also shouldn't make the assumption that they're independent.
AIXI is a definition of the optimal agent and is hence uncomputable but LLMs are approximations which are approaching AIXI. I recommend Fridman's interview with Hutter.
In my experience it comes down to speed of processing and response
Nobody views trees as intelligent because despite having extremely complex interdependencies with their ecology, including fungus and all kinds of other organisms, they don’t move quickly or appear to respond to input (even though they do, just slowly).
Meanwhile the mantis shrimp because of it speed, superhuman vision and flexibility is considered extremely intelligent.
The only thing that humans will accept as more intelligent than them is (something) that they cannot control and can dominate them.
This is why AI is “the things we haven’t done yet” - once a technology is pervasive and integrated it is “just computing.”
Because we take freely from trees. They must be stupid or something because they give things freely. It's kind of why many employers view labor as stupid. They give their labor for very little. Humans dabble in arrogance.
Of course, there is some intelligence in any life form behaviour, but if you want to say that a tomato plant is intelligent than you need to use another word or set of words for more advanced life forms. Putting in the same bag a tomato plant and a dolphin clearly makes the word intelligence so vague it loses almost every practical meaning.
To clarify also the part where it talks about earth being an organism. I've thought of this as well, the whole universe could be a life form, each planet and star in it a cell in its body. It's a possibility. Or maybe even our whole universe is just a cell in someone's body.
It's possible, but I fear there is little science in the people of that article and just old fashion Animism and protect "the mother earth" natural spiritual thinking that has existed for thousands of years. Those people see the world as if they are druids in a fantasy novel. I see myself more as a wizard. We might have different opinions. I will take them seriously when they can LITERALLY speak with animals daily in useful manners and tell vines to move. Until then is just (a) Fantasy. I can summon electricity and fireballs (with technology), if they want to say they are druids they better step up their enchantments. Telling "sit" to a dog and then writing long articles on how dogs are so intelligent doesn't cut it for me.
[0] https://www.frontiersin.org/journals/systems-neuroscience/ar...
Gotta watch The Happening again.
Ending on a positive note, many scientists both love nature and are rationally scientific (rather than romantic). All the awesome work being done with using LLM to speak with whales and recognize better dogs emotions and pain levels is very inspiring.
Some people over time cannot put down the fact that there’s just way too many coincidences. If you just pile up what it took for you to be here (let’s include your one in a million chance of out swimming the other sperm along with a perfect planet, possibly perfect parents, list goes on), I think you’d at least agree it’s one miracle after another.
I can sit here and measure the anger on your face based on the metric of how red your face is but I’ll never know why you are angry. You can tell me, but even then, is it true? That’s the issue with science, that measuring and collecting observations won’t reveal the true nature. So it must be put aside, at least to investigate the other possibility.
I don’t see how people aren’t open minded about there being more than a clinical understanding of this experience.
From my other comment:
https://www.sciencedirect.com/science/article/abs/pii/S01676...
Is it so hard to believe there is a global frequency?
What if it isn't?
What if it's many voices and they compete for control?
If a person argues that an LLM is conscious or intelligent based on how it responds, is the United States conscious or intelligent?
Assuming the United States react to stimuli and exhibit anger, and assuming that the definition of being conscious is reacting to stimuli and exhibiting anger, yes, the United States are conscious.
Also, considering who the current president is, the US is quite the opposite of intelligent.
Nitroglycerin comes to mind.
Discussed previously on HN:
July 2015 (210 comments): https://news.ycombinator.com/item?id=9905847
Feb 2021 (78 comments): https://news.ycombinator.com/item?id=26217834
Dec 2023 (1 comment): https://news.ycombinator.com/item?id=38814769
Maybe, but this strikes me as a borderline category error, like saying, "If databases can order results, then they are probably sorting algorithms". The argument makes assumptions that consciousness is transitive or "sticky" in some sense, where if a property applies to a part of the system it must also apply to some aggregate of those parts.
No, this doesn't follow, that's my point. This depends on further constraining materialist consciousness to have the specific kind of property I described. It's possible that consciousness is like "sorting algorithm", a specific property of a specific kind of system, and aggregates of such systems don't necessarily have that same property. They might, but it's too strong to say they definitely or probably do.
I think the argument is really about substrate-independence. If consciousness is just about functional properties, then why can’t a social collectivity exhibit those functional properties?
Many materialists do in fact endorse substrate-independence - the common belief that an AI/AGI could in principle be as conscious as we are (even if current generations of AI likely aren’t) depends on it - and I think substrate-independent materialism likely does fall victim to this argument. Now, maybe not, if there is some functional property we can point to that individual humans and animals possess but which their social collectivities lack-but then, what is that property?
Other viewpoints don’t endorse substrate-independence. For example, Penrose-Hameroff’s orchestrated objective reduction, if you would call that materialism - I think you can interpret it as a materialist theory (e.g. the in principle empirically testable claim that neurons have a physical structure with certain unique quantum properties, plus the less testable claim that those properties are essential for consciousness) or as a dualist theory (e.g. these alleged unique quantum processes as a vehicle for classical Cartesian interactionism). The more materialist reading could be viewed as a substrate-dependent materialism which escapes Schwitzgebel‘s argument. But I don’t think most materialists want to go there (seems too dualism-adjacent), and the theory’s claims about QM and neurobiology are unproven and rather dubious.
Computations are mechanistically clear, deterministic and substrate independent, but they still have properties that we use to classify them. Just because some component of a larger computational system has a property (or classification), it does not entail that the larger system has that property (or classification). Consciousness could be like this.
Rather what he is saying is this: given many candidate substrate-independent materialist definitions of consciousness, if the property is true of individual humans and animals, it will true of their social collectivities. But, he says “many”, not “all”-he’s not claiming you can’t define properties of consciousness for which that is false, he’s simply putting the onus on the proponents of the “materialism can explain consciousness” project of explaining in detail how, and justifying such a definition.
Furthermore, he’s not claiming that properties of consciousness are transitive from individual organisms to any arbitrary grouping of them. Rather, he’s pointing to social collectivities such as countries or governments as being so coordinated they sometimes act like they have will of their own emerging from the coordinated wills of their individual members. This isn’t true of arbitrarily defined wholes of which those individuals are part, e.g. the set of all humans (anywhere on earth) whose first name starts with the letter A. You are interpreting his point in terms of part-to-whole transitivity in general, but that’s not the case-the emergent properties of social groups are far more complex than simple part-to-whole transitivity. And it is those emergent properties he points to as evidence they may have a consciousness distinct from those of their individual members
The complexity of this behaviour could be at the level of an amoeba rather than a human mind. It's just not persuasive without knowing more about the specific characteristics of what materialist consciousness requires.
> From a snarl of roots that grip dry, shallow soil, the knobbly trunk of an ancient olive tree twisted into a surprisingly lush crown of dense, silvery-green leaves. Far above, the retrofuturistic pattern of a geodesic dome framed the blue sky outside. Dan Ryan considered the tree: “It’s probably close to 1,800 years old.”
Why do articles always have to start this way? These writers are writing the blandest and most boring clichés imaginable and everybody hates it, yet they keep doing it. Is it so fucking important for their egos to be perceived as some kind of antiquated stereotypical writer that they have to continue to humiliate themselves and their readers with this, just to impress who?
No wonder everybody is watching short form stuff on TikTok, that gets to the point.
No wonder people are listening to three hour podcasts, where the people who are being interviewed can talk uninterrupted, instead of having their quotes salted and peppered between some musings from the writer.
Where are the writers that make articles for normally intelligent people to read, that aren't filled with fluff, but aren't filled with equations either? Are these all making YouTube videos now?
Consciousness, on the other hand, is (even) less well-defined and is usually considered to be subjective. Being subjective, it tends to resist all of the usual objective approaches of description and analysis. Hard problem and all that.
Similarly to how computer is computer and half sized computer is half of its bigger friend – you can keep halving it until there is no "computer" left in it.
Or pencil – you have pencil that you call pencil; what about pencil half size of it? and so on until you hit single atom. You had pencil, now you don't, where on this line there was pencil and then there wasn't?
Edit: you can’t measure “pencilness,” but you can’t help but know whether or not you’re in pain.
These are all proxy measures, but all of science is done by proxy.
What is mind emerging into? When a video game experience emerges from the combination of processing, display, sound, and controller input, it emerges into a level of organization that a mind can participate in. It emerges into a system of organization emanating downward from the mind experiencing it. It can't just "emerge" into existence on its own. If a game falls in the woods, its not a game.
If you call the mind an emergent phenomenon but can't describe the context into which it emerges, you've added nothing to our understanding.
The emergent experience exists at the level into which it emerges. It's constructed at a lower level of organization, but not decomposable to them in a way that's meaningful without their recomposition—that's what makes the phenomenon emergent. The qualitative experience of a film does not meaningfully break down to the bits in the video stream, the compressed sound waves carrying the dialog, the photons hitting your eyes' rods and cones, or the biochemical signals in your brain.
The mind is not in the brain, but on the brain.
If a car is hurtling towards you, you don't perceive its handle. But if you're trying to go somewhere, you have to open the door. "Threat", "vehicle", or "handle" aren't just convenient framing device, but an accurate depiction of the object within your perceptual/motivation systems based on the current level of organization and analysis you're participating in.
We choose to see the mind because we are minds. Consciousness is. There is something which it is like to be. Denying our emergent experience of it, or reducing it to a "convenient framing device" tosses out the most fundamental empirical experience we have: to exist.
A video game is relatively easy, at least seemingly, to reduce down to its underlying principles. The content dissolves the more closely I look at the game. The issue here isn’t whether the game still exists (it does, in the place I’m no longer looking), the issue is in seeing why the game arises from its component parts, and not something else. Easy-ish for the game, it follows directly from what we know about physics and such, but hard for the mind. Why do neurons together produce pain that exists, rather than pain as a purpose-driven internal signal to help organize the escape from a predator? Emergence doesn't tell us why one or the other, just that whatever it is must emerge from constituent parts.
The game is certainly easier than the mind—I like it as an example because most of us have a hands-on knowledge of what the qualitative experience of "playing a game" is like. But the game still only emerges because the game developer, computer manufacturer, and player jointly give it a emanating system into which it can emerge. On its own, the raw game data doesn't really mean anything at all—if the bitstream of Diablo IV washed up on the beach, there's nowhere in that data encoding the experience of killing Diablo for the first time. One wouldn't even recognize it as something that could be decoded into such an experience [2].
I agree with you that the "why" is tough. Why have a conscious experience? Why have a sense of self at all? Why experience emotions rather than have them be—like you described—a purpose-driven internal signal? And then you get into theories like Internal Family Systems, which has empirical support at least within a prescriptive context if not necessarily a descriptive one [3].
The whole thing is a mess. A great, big, beautiful mess.
[1: https://www.amazon.com/Case-Against-Reality-Evolution-Truth/...] [2: https://benjamincongdon.me/blog/2021/02/21/Three-Layers-of-I...]
Consciousness is the emergent, graduated phenomenon of an information processing system when that information processing system has achieved significant complexity to model itself with relation to the various systemic inputs.
It's not binary, it's a gradient. I have more developed consciousness than my dog, which has more developed consciousness than a rat, and then a fish, then an insect, etc.
Somewhat disturbingly it also goes the other way, AIs may achieve a more profound conscious experience than humans - same for aliens. What does it mean for inferior forms of consciousness that have always placed themselves on a pedestal in relation to the rest of the animal kingdom simply because they have the most developed consciousness?
What you're describing is what's been proposed by Giulio Tonini as "Integrated Information Theory" (IIT) [1]. I quite like the framework and the math behind it is beautiful. Unfortunately, it hasn't been supported well empirically.
Re: AI, IIT actually gives basis for AI not being conscious. Not to mention that all conscious systems we can currently observe are dynamic/continuous, not discrete. The difference there is qualitative—there's no reason to assume that because a dynamic system is conscious that a discrete system approximating it is conscious too.
The article you cite [0] is labelled as "opinion". The standards for peer review of opinion articles in scientific journals are a lot lower than those for ordinary research articles. While precise standards vary from journal to journal, for opinion articles peer reviewers often see their role as simply excluding egregious misinformation and blatant errors, as compared to research articles where their role is to make sure the article is presenting high quality evidence in support of its conclusions. [1]
[0] https://www.nature.com/articles/nrn.2016.44
[1] https://ecologyisnotadirtyword.com/2021/02/24/lets-talk-abou...
https://journals.plos.org/ploscompbiol/article?id=10.1371/jo...
Scott Aaronson’s attempted refutation of IIT - https://scottaaronson.blog/?p=1799 - I think is better in that he actually tries to relate IIT to some of the philosophical literature (e.g. his distinction between Chalmers’ “Hard Problem” and the distinct “Pretty Hard Problem” which he sees IIT as trying to address)
I think it is a pity that Aaronson has never (to my knowledge) published his criticisms of IIT in a more formal setting-and I don’t know if Tononi has responded to them anywhere. I think Aaronson is probably right - that IIT fails as a mathematical model of what we intuitively consider conscious, since even though it excludes many common electronic devices we wouldn’t “conscious”, it is possible to mathematically construct an algorithm, capable of being physically implemented in electronics, which would be conscious per IIT but not per our intuition. And even if Tononi patches his mathematics to solve a particular case of that problem, someone with Aaronson’s skillset may just be able to construct another.
Tononi might then argue that if there is no mathematical model of our intuitions about consciousness lacking in special pleading, that’s a sign our intuitions are flawed. Okay, but then if we accept our intuitions can be flawed in some cases, why not in more cases? One could decide the intuition of consciousness is completely erroneous and become an eliminativist about it. Or, if IIT forces you to accept (contrary to our intuitions) certain (special cases of) simple electronic devices or computer systems as just as conscious as humans, why not violate those intuitions further and insist on that for even more cases?
My personal belief is that consciousness requires dynamic continuity. I don't think an algorithmic system is conscious because it's "cognition" is discrete and the information isn't integrated across frames. I don't have a "why that works"—its just a gut belief.
Stanislas Dehaene's book Consciousness and the Brain does a great job of describing this, though it's 10 years old now.
consciousness = meta-attention
A similar idea is present in Herbert Simon's 'The sciences of the artificial', where he describes a sentient city.
In both cases, defining the actual thing under discussion is hard. If you can accurately predict a decision in advance, is that "free will"? If an organism reacts to a stimulus in an appropriate manner, is that "intelligence"?
In both cases, we're complex chemical organisms doing all of this with complex chemistry. If we rule out souls and spirits and similar, then it's just chemistry.
If we're following a predefined set of chemical rules in response to a set of stimuli, then how is that "free willed" or "intelligent"? The arbitrary line between tropism and intelligence seems very arbitrary.
But on the other had, we are made of meat. We experience free will and intelligence. We think, and make decisions, seemingly unrestricted by the method we use to do that. We are clearly intelligent, and clearly make decisions that we are apparently free to make.
“Free will” does not exist because the world is deterministic. In other words, if you have made a decision, you couldn’t have made any other decision, so there wasn’t any choice in the first place. A persons IQ has nothing to do with this.
Here is a proof that quantum mechanics is deterministic:
Let there be a time `t`.
Let there be a qubit `q`.
At the time of `t`, we measured the qubit `q`, and the qubit `q` collapsed to the state of `a`.
At the time of `t`, we measured the qubit `q`, and the qubit `q` collapsed to the state of `b`.
`a` equals `b`.
Quantum mechanics is deterministic.
Quantum mechanics is indeterminate and probabilistic. Some of the most brilliant physicists like von Neumann and Bell, who won the Nobel Prize for this work, have proved it. Unless you have something that will win the Nobel Prize, your intuition is wrong.
Check out https://en.wikipedia.org/wiki/Quantum_indeterminacy
It's like being able to predict what someone is going to do, because you know them and how they think, and what decision they will make when presented with the choice, doesn't stop that from being their choice.
The universe may be deterministic, but my personality is still my personality. It is encoded into the chemical make-up of my brain, and so that complex chemistry behaves in ways that align with my personality. My personality is shaped by my previous experience, but it's still my personality. I still choose, even though the universe can predict all of my choices, because the thing I do the choosing with is part of the process.
And this seems very like the argument about intelligence and instinct. If I respond in a certain way to an event, is it because I am intelligent and "thinking" about my response, or is it instinctual and coded into my meat to respond this way? How would I tell the difference?
Same with free will, how would I tell the difference between a choice I freely made and one I didn't?
This is completely contradicted by the fact I could talk about that experience, which does imply some control from the observer to the physical world. Which makes the whole thing paradoxical. The only way I can square it is with my religious beliefs.
Can one? One of the most disturbing short stories I've ever read is "Love Is the Plan, the Plan Is Death" by Alice Sheldon (writing as James Tiptree). It is the first-person narrative of an unusually self-aware member of a non-technological intelligent species trying to make a life different from "the plan", the species' instincts, in the face of an oncoming slow-motion multigenerational catastrophe. His efforts end up, all for contextually rational and agentive reasons, reiterating his species instinctual lifecycle.
My takeaway from that, from other readings, and from self reflection, is that we are puppets that may or may not become aware of our strings; but if we cut them, we die.
But the intelligent work is in the connections, exchange, and integration of information.
I think that security for human groups should be thought of in this context. Many strong communication links or maybe a holistic network are required for preventing sub-colonies from becoming "other".
Or, at least, I could. It seems I lost the knack of controlling saccades. When I was younger I was able to suppress saccades and induce nystagmus-like eyeballs movement.
Ah, I still can do it after some tries. Amplitude of induced nystagmus seems to have diminished though. And it's more uncomfortable than it was.
Intelligence isn't well defined. According to our current understanding of physics, everything is cause and effect, which throws free will out the window, so what do we even mean by intelligence? Is it just when chains of cause and effect become so complex we can't understand them any more? And if so, why does consciousness even exist at all?
The only consistent theory of consciousness I'm aware of is panpsychism, which seems very unsatisfying.
I guess the other option is to divorce consciousness from physical master entirely, but then we have kind of opened ourselves up to almost any kind of woo.
It really is a hard problem.
Integrated Information Theory seems interesting, at least, but far from flawless. Like panpsychism though, I don't think it's falsifiable.
To me, it seems like a bit of a trick - take a philosophical question, propose an answer, stuff some mathematics into the answer, maybe even a few minor empirical predictions, make people think those added bits make the answer more likely to be correct and more respectable than competing answers which don’t do that - I don’t think that actually works though
> Like panpsychism though, I don't think it's falsifiable.
Well, I don’t think any theory in this area is falsifiable. Either you give up demanding falsifiability or embrace agnosticism (we don’t know and we maybe never will)
Maybe give up on falsifiability, since presenting it as an absolute must-have is self-defeating (the idea that falsifiability is a must-have is itself unfalsifiable)
We don’t know whether the apparent indeterminism of QM ultimately reduces to determinism or not. It depends on which interpretation of QM you prefer, and none of them has any strong empirical support
> which throws free will out the window,
As well as assuming deterministic QM, you also assume incompatibilism. Reject incompatibilism and you can still have free will even if physical reality turns out to be 100% deterministic
> so what do we even mean by intelligence?
People disagree greatly as to what “intelligence” means, but this is the first time I’ve seen anyone suggest that the issue turns on free will. Usually people present “definition of intelligence” and “does free will exist?” as questions having a significant degree of orthogonality
> Is it just when chains of cause and effect become so complex we can't understand them any more?
Mediaeval philosophers commonly viewed intelligence or intellect as a power, the power to engage in conceptual thought. A “power” is a casual concept - it is the possibility of causing something - but possibility, not necessity. It is orthogonal to the question of free will, so long as our position on free will enables us to meaningfully speak of “things we could have done but never actually did”, “thoughts we could have thought but never actually thought”…
It was our dog’s birthday the other day. She’s a smart dog, but she’s never going to understand the concept of a birthday… she enjoyed some special treats but she’ll never understand why she got them on that occasion. She has intelligence, but not the specifically conceptual sort of intelligence mediaeval philosophers were talking about. And free will doesn’t enter into that
> And if so, why does consciousness even exist at all?
I think intelligence and consciousness are distinct issues. Consciousness is having subjective experiences; intelligence is having thought processes of a certain kind. One entity might have consciousness without intelligence, another might have intelligence without consciousness.
> The only consistent theory of consciousness I'm aware of is panpsychism, which seems very unsatisfying.
I don’t think the competitors to panpsychism are necessarily inconsistent. Of course, if you reject panpsychism, you need a non-trivial criterion to decide what is conscious - and people will ask you to justify that criterion - and maybe all you can say is that it is axiomatic - and I get why proposing axioms can feel unsatisfying, but it isn’t strictly speaking inconsistent, assuming your axioms are consistent with each other. By the Münchausen trilemma, every quest for justification must end in either infinite regress, circularity, or axioms - and if axioms feel unsatisfying, infinite regress and circularity are just as unsatisfying - so maybe we are just doomed to feel unsatisfied.
> I guess the other option is to divorce consciousness from physical master entirely, but then we have kind of opened ourselves up to almost any kind of woo.
I don’t know what you mean by “divorce… entirely”-do dualists (of whatever kind) do that? Do idealists? From my own idealist viewpoint, I reject the claim that idealism necessarily entails “almost any kind of woo”, and I think the thought that it does relies on misunderstanding or misrepresenting idealism, or else confusing certain versions of it which maybe do do that with other versions which don’t
NYT 2023 https://archive.ph/DpMMr
https://www.sas.upenn.edu/~cavitch/pdf-library/Nagel_Bat.pdf
Do we really have to resort to total bullshit claims like this in order to make people care about this subject? Knowing what is and is not intelligent has fuck-all to do with not destroying the environment we as a species rely on for survival. The people and forces at work care not one whit. You could prove that chickens can write symphonies and it would make no difference.
> Which brings us to the most striking idea — that some types of electrical oscillations could mediate an experience of self.
> In 2021, Hanson found that similar electrical activity — spontaneous low-frequency oscillations — is evident across many different organisms, from E. coli to humans. She concluded that across a diverse range of creatures, the oscillations may have a shared function: constructing a single organismal whole from many parts.
That's fascinating stuff which deserves the title (and the long context). 2021 is pretty damn new as far as theories of mind go.
> this is very literally about the first steps in how brains evolved
Well, no; they even highlighted this section:
> “Intelligence, according to some, is a biological function that evolved not with humans or brains but way back in some form to the earliest organisms, a fundamental biological function like respiration.”
It would definitely connect to how brains evolved, but the article and the main idea are a lot broader than that.
> It's akin to theories of abiogenesis like the one involving clay crystals "reproducing"
Well no; unless the clay crystals are producing interesting electrical oscillations.
> it's steep to say this is about minds, which only works if we accept the various parts of the article that urge us to stop being so egotistical and anthropocentric as to expect that a mind should be capable of doing something clever.
It's not clever to live for 2,000 years? And, are only clever people worthy of the label of having a mind? ... Because I know an awful lot of people with no interest in doing anything particularly clever.
And yes, your comment is anthropocentric... Definitively so. Whether something has a "mind" depends on the definition used, and if you define that as "what humans call clever", then yeah you're being anthropocentric.
(I meant very loosely akin, because it's another example of some complex phenomenon emerging from matter.)
It seems to me that a key characteristic of consciousness is it's informational character. It is representational of a cognitive state, it's interpretive, introspective. It seems to me that for any system to be conscious it must at a minimum be generating and interpreting representational structures. I'm sure that's nowhere even close to being a sufficient criterion, but I don't see how it can't be a necessary one.
I think that's key to subjectivity, because different systems can have radically different ways to represent and interpret even the same phenomenon they have representations of. The details of the representational structure, and it's network of associations with other representational structures in the system, are intrinsically tied to details of the system processing and interpreting the representation.
Hanson, Alison. "Spontaneous electrical low-frequency oscillations: a possible role in Hydra and all living systems." Philosophical Transactions of the Royal Society B 376.1820 (2021): 20190763.
Discussion of some of it here with M. Levin: https://www.youtube.com/watch?v=bZPhqwrSC24
If it’s not:
https://www.sciencedirect.com/science/article/abs/pii/S01676...
If everything is pulsing with electricity and they are syncing to something global …
From the paper:
Examples of biological systems that have been modeled using PCOs include cardiac pacemakers [2], crickets that chirp in unison [3], and rhythmic flashing of fireflies [4]
I'd say you probably don't know enough to say that with such certainty, and should probably avoid jumping to conclusions lest you be Dunning-Krugered.
I would then encourage you to ask ChatGPT something like: "How are SELFOS (Spontaneous electrical low-frequency oscillations) different to PCOs (pulse-coupled oscillators)".
Tl;dr - SELFOs and PCOs aren't the same.
I'll absolutely take that.