"What Is a Strange Loop and What is it Like To Be One?" by Douglas Hofstadter (2013)

Many years ago, reading about Gödel's incompleteness theorem gave me an image of the nature of a human self or "I". I tried to spell this vision out in my 1979 book "Gödel, Escher, Bach", with mixed success. The book itself did very well, but the ideas that drove it were not so well understood. Nearly three decades later, in my 2007 book "I Am a Strange Loop", I tried to express these same ideas more simply and more directly, and I hope I did a bit better the second time around. In this talk I'll try to get across the crux of these intuitions about the mysterious concept of "I".
Douglas Hofstadter
Indiana University
Professor Hofstadter received his B.S. in mathematics from Stanford University (1965), and his M.S. and Ph.D. in physics from the University of Oregon (1972, 1975). His Pulitzer-prize-winning book Gödel, Escher, Bach: an Eternal Golden Braid (1979) has had considerable impact on people in many disciplines, ranging from philosophy to mathematics to artificial intelligence, to music, and beyond. He has written four other books, numerous articles, and, for a number of years, wrote a column for Scientific American.
Recorded at Strange Loop conference (thestrangeloop...) in St. Louis, MO, Oct 2013.

[@xdcountry]

I’m lucky to have read GEB back in HS - love this guys work/perception on things. Incredible mind.

[@falklumo]

Thanks for bringing this gem back to life. I have a similiar background to Douglas (i.e., PhD in Theoretical Physics) and have gone through similiar ideas, except that I didn't publish a book about ;) If you don't mind, I have a couple annotations to make for Douglas to read ... First, I agree 99% with a few exceptions only. But first, let's agree that 2021 and with the advent of GPT3 and similiar deep learning networks, the problem of perception can be considered settled. We also learned that perception alone doesn't bring us as close to consciousness as we might have hoped for. In the talk here, I hear about five things: - perception - concept - causality - free will - "I" And therefor, I miss the most important thing which rules them all: - prediction model Our brains learn, train and evolve a "prediction model" which is able to run "what if" scenario simulations which eventually are fed-back into our other perception channels. This has obvious overwhelming advantages for survival and is the only explanation why esp. humans can afford such power hungry brains: their "prediction models" are more evolved, far beyond a pure "outer world" simulation, with abilities to abstract. Moreover, our prediction models (must) incorporate a model of self. As this is the crucial component which can ultimately be controlled. And because the model of self within our prediction model would otherwise hit an endless recursion, it *must* contain a model of self with free will, i.e., with an unpredictable will. Our perceived free will is a modelling necessity maximising our chances for survival. It that sense, it is both real and an illusion. Free will is real in the sense that our prediction model of self DOES have free will. Note that our prediction model doesn't necessarily attribute free will to models of other actors, human or animal. Some of them have predictable enough behaviour that we perceive them as deterministic with the illusion of free will only (you don't know but I do what you or cat is going to do). Note that a neural network or brain doesn't need such prediction models. E.g., the autonomous driving network in a Tesla has perception but no such model, just hard-wired response behaviour (reflexes). And if it has such a model, it may not incorporate a model of self. And if it does, it needs not be a self with perceived free will. That's the last stage only. One thing I don't know is if this stage is a necessary or even sufficient condition for consciousness. It may be that consciousness needs one more thing: e.g., a hunt for purpose, whatever, or some strange loop ;) We'll know soon enough because within 20 years, self-driving cars need to incorporate refined prediction models for other actors such as vehicles (incl. their interactions) and themselves (incl. their own behavioural model, e.g., an unconstrained will). Or self-driving cars will not replace human drivers. Such cars would be self-aware for sure. Conscious we'll see ... The programming of such cars would be to make their passengers as happy as possible with causing as little as possible damage overall. In summary, the talk by Douglas misses the term "model". Great talk otherwise.

[@ByronRuth]

He certainly said "model," specifically around 49:01 (possibly before that as well). But in any case, this notion of a model with feedback was at the core of the talk (at least how I perceived it). Every decision we make is backed by a prediction first. We are not _state machines_ at the level of consciousness and therefore there are always multiple possible outcomes.

[@RhizometricReality]

Have you seen google io project LaMDA?

[@HumanBladeG0D]

Is it your idea that this prediction model is a factor of consciousness, a byproduct or in fact yet another strange loop feedbacking into itself? Self referential object with data feedback loop etc.

[@gustavosicaviani7553]

causality derives models cause there are many cases

[@cube2fox]

Note that language models are curiously similar to prediction models. A language model takes text as input and "predicts" the continuation of that text as output. A prediction model, presumably, takes experiences (sensory observations) as input and predicts future experiences as output.

[@nakor555]

Consciousness is a big library where all the concepts are making iterations to be the more efficient for the being they're computing for....

[@HappyMomma412]

I just want corduroy to come back! 🤦🏾‍♀️ It’s so nostalgic. No, but in all seriousness. Still listening, but this is great. 🥰💜

[@user-hy9nh4yk3p]

Corduroy may - rub many people - the wrong way, though. Hee hee. (Fare thee well)

[@nospoko5262]

Self referencing is massive compression now that's nice idea.

[@ElectricChaplain]

"To think is to forget a difference" - Borges

[@faster-than-light-memes]

Crystal clear down to earth. Amazing

[@youtubebane7036]

I think the big bang was the beginning of a strange loop from our perspective. It is also key in finding a theory of everything. There is a geometry involves, a patern or frequency and some other thing I'd like to call mind that allows infinity to be calculated. If you think of each strange loop as an infinite and then give them a temporal identity, as one can only exist at one time, then you can pit infinities into a hiarchy. Identity is only assigned by minds.

[@user-hy9nh4yk3p]

In Raja Yoga meditation - identity precedes mind - almost the - Divine stamp of authenticity - that is then - thought about - via the mind. Identity remains - even when the mind function is dissolved. (This hint from Heartfulness) Fare thee well.

[@DrorF]

This was just a wonderful lecture! Fascinating stuff 👏🙂

[@DisperseControl]

Thanks for sharing. This is great ❤

[@or6060]

how can a desire for chips and a desire for health exist independently of who you are in your entirety? they are consequences of it. they are not "just" your desire for chips or health, you have them because of what has happened in your past. they may be small elements of that, but they are not separate.

[@klausfaller19]

It is the I, we here to understand. The purpose of life is the discovering and loving of the I. The I is in constant change. The secret of a happy life is, to love and embrace these changes of the I, without the interference of the me.

[@carlosluis1970]

no I, no spirit, no soul, just self reference to the whole body as a point of reference. we are machines!

[@tambwetombo781]

Good lecture. I disagree with the point that "I" or a rainbow are somehow "less real" than a physical object. This categorisation into real/fiction or mental/physical is a purely abitrary and illusionary division.

[@brianhershey563]

I understand the abstraction 🙏

[@PREMKUMAR-wz6gq]

So a strange loop is a computational system having a coarse grained perception of itself and other things in its environment?

[@CBruceNL]

I believe a strange loop is a self referencing system that functions in a predictable way to achieve a specific ends. I think of the different base math systems. Base 8 or base 10 math are both useful systems, but they don't work with each other. They are both too closed loops that work within their internal reference system. Humans then are an exceptionally complicated system of these self referencing feedback loops. And I could be super wrong, but there is no better way to get the right answer on the internet

[@heater5979]

What do you mean by a "computational system"? A simple computational system might just produce an output that is the same of its inputs, a = b + c but it has no idea of what it is doing. It does not know it is calculating as sum. It is just doing what it does, which may be my means of transistors, relays or other mechanical means. The whole notion of computation comes from those who built the system and observe its inputs and outputs. That is to say us. In general I have this feeling that the universe does not compute, it does not follow any laws of physics, it just does what it does. The notions of computations, mathematics, the laws of physics are just concepts that exist in human brains.

[@brianhershey563]

It's not compression, it's not ridiculous, it's not simplification, it's efficiency, the master attractor of all energy dissipation. 🙏

[@platykurtic5510]

To summarize, it is not so much that France is conscious like us but that we are conscious like France. But if knowledge is power, France is bacon.

[@anthonychristie7781]

42:40 ..."France is a 'teathing" metropolis".... Teaming? Seething? The teaming mass of neurons "he" affectionately/ruefully "knows" as Doug, went with seething. Good ol' "free" will!

[@lani0]

the France analogy alone is worth the price of admission

[@user-hy9nh4yk3p]

This 'free will ' is subject to continual debate - sometimes, gentle reflection - otherwise, weapons are brandished - also. Enough. Fare thee well.

[@youtubebane7036]

A funneling process? So basically a geometry? What if it's infinity being funneled? Couldn't the value at the jumction where one trip up the strange loop ends at it's own lower point be used in some modified fashion to translate an infinity into a finite value at least for the purpose of computation? I believe we can assign identity to individual infinities to truly define cardinality past numerical infinities by assigning identity to each one base on it's place in an infinite series of strange loops. From my understanding a strange loop doesn't necessarily begin exactly where it ended but it begins in the corresponding place that is by all measures identical. So more of a strange spiral kinda like the fibonacci sequence when viewed as a fractal. In other words after each strangeloop is completed then the next one would be in a hierarchy that can be given an identity. As every Infinity share things with other infinities. As in everything about any Infinity would be shared with any other infinity or could be I should say. In other words there's always going to be at least one more Infinity that shares a given property of an Infiniti. The only thing that another Infinity would not be able to share that would be impossible for it to would be the identity of any said given individual Infinity that is assigned to them or their position in strangeloop hierarchy. And if they are part of the hierarchy that means only one Infinity could exist at one time because it is an Infinity and nothing else that can exist at that time. Or in the same place. So that means each individual Infinity in the strange Loop would have its own identity. And this would still be true even if the beginning of each strangeloop with exactly the same strange Loop. Because it would be very the very Act of having identity that distinguishes that distinguish them from each other and caused the next strangeloop the be one up or one down in the hierarchy. What I need is a type of number system like how they came up with complex numbers to assign to these things I am calling identity. Did I need to translate the act of giving identity into a mathematical operation because that is how we are going to calculate Infinities and give them finite values or flip them back around to 0 or negative Infinity depending on what we're doing. Physics is full of Infinities that are being calculated without people even realizing it. The Heisenberg uncertainty principle is one set because it is the the value of infinity that would be associated with position or velocity when unknown before measurement as in the velocity or position of said object can be infinite or any place. The act of measurement reactive measurement Narrows this value down from an infinite value to a finite. Or from an infinitely continuous playing to a finite point for location. Or you can look at it as an infinite continuous playing of singular points or coordinates. And it doesn't matter if the plane you are working with as in our space-time is actually infant in or not because it is potentially so just because the process of counting never end and once the universe began it has been growing that means it is potentially infinite. I know the uncertainty principle involves probabilities but the probabilities of and certain particle being anywhere anytime are literally infinite but at some places they are far more likely to be but there is no place in existence where the probability is zero of finding said object. So basically during measurement you are taking an infinite plane and giving it finite point and coordinates and then for the more narrowing it down to one of these finite points. And then once you have it stopped at that point so you know where it is you don't know how fast it's going anymore because you stopped it. Anyting in movement doesn't have a position as it is always in transition on some scale. There's an infinite number of speeds and object or anything else can be traveling at even if there is a local speed limit that matches the speed of time. Which you guys called the speed of light. Mathematically there is no limit to the velocity of an object if there isn't anything to stop it from accelerating. Isn't the very mass of an object that stopped it from accelerating past the speed of light. Why does the photon obey light speed? Obviously that is the speed of causality but what forces the lightspeed to obey causality? Is there something about the four dimensions of time-space that causes the photon to slow down or to come into existence? So I believe the act of assigning identity do anything or to have identity or have identity emerge from it has to do with some kind of mental operation as identities are properties that are only given or understood by things of a mental nature as they would be meaningless to anything else so there would be nothing else that can either understand the Identity or give it to said thing. Even if the infinite number of infinities occur in the same place they could not all occur at the same time and vice-versa so this means each Infinity must have its own individual identity. This identity is what the strangeloop is from and it is what separates one Infinity from the next. This number or identity has to be some type of number that can be defined by some type of operation I have of yet to have thought of or heard of from someone else. But it has to do with geometry and a pattern or a frequency or dimension value and a mental operation and these somehow allow an Infiniti either to start over again or be translated into an Infinity of the opposite polarity or the number zero or a finite number. I believe we will find that this depends on the order we perform the operation in and what its polarity is in the first place. An energy skill that is used to assign a value to a virtual photon decreases on one side of its equation as in its length well increases in the magnitude of its magnitude of its power or energy. These values would be infinite if not for this unique property that I think is another strange Loop. One that doesn't involve any type of mental operation maybe but one whose operation that translates one energy to the other arm translate energy to link can also be used to translate another function like a numeric function into another force of mind. I'm sure I'm getting onto something here because all things could be reduced down to something else except for mine. You can't reduce nothing physical below the level of Mind where they showed mind is emergent from that phenomenon but you can reduce every single thing there is down to mental causation just by the fact you can say mind is what causes it by imagining it or thinking of it. Physicist another material list love to say that mind is an emergent property of the universe but they can never ever describe exactly how that is but opposite is that it's very easy to say the universe emerges from mind. And this has to do with data or information this has to do with data or information and the number zero and the number infinity and the concept of conception or thought. Specifically it has to do with the Paradox that absolute nothingness is an impossibility to exist because the information describing what it is would have to exist. Is that information is so far beyond Infinity in its magnitude that since we exist in as well as he that consists of infinity and nothing there is an abundance of leftover energy that must go somewhere in that goes to creating everything. But all this is beside the point because all of that information and stuff comes from inside of my mind because that's the only place anyting like information can be conceived in or perceived in for that matter

[@alexgonzo5508]

Interiorization and encapsulation are cognitive processes that can be used to conceive of an infinite process as a finite value. Interiorization involves taking an external action or process and mentally representing it internally. This allows us to think about and manipulate the process in our minds. For example, consider the infinite decimal expansion of the number 1/3, which is 0.3333... The process of calculating the decimal expansion can be interiorized by understanding it as a repeating pattern. By recognizing the pattern, we can encapsulate the infinite process into a finite representation of 0.3, which captures the essence of the infinite process. Encapsulation involves encapsulating or summarizing a complex or infinite process into a simpler or finite representation. In the context of infinity, encapsulation allows us to capture the essential characteristics of an infinite process in a more manageable form. For instance, consider the process of computing a limit in mathematics. The limit may involve an infinite number of steps or calculations. However, through encapsulation, we can summarize the behavior of the process by stating that the limit converges to a specific finite value. This encapsulation allows us to work with and reason about the infinite process in a more practical and meaningful way.

[@jamescareyyatesIII]

The problem with calling the self an illusion is the implication that there could be a self which isn't an illusion. Can you concieve of a self which is authentic and not an illusion? If not, saying the self is illusion is true but trivial.

[@christopherscragg7018]

When was this talk given?

[@StrangeLoopConf]

This was recorded in 2013 as a keynote at the Strange Loop conference

[@borntobemild-]

Can someone get some of Douglas Hofstadters work over to the people programming GPT-3?

[@xieyuheng]

我们有「因果」这个概念，正是因为我们经常观察到在时间序列中，自己的行为能经常引发固定的结果。 所以「因果」的概念产生于「我」这个概念，反过来用「因果」来解析「我」的结构就有点肤浅了。

[@asdf8asdf8asdf8asdf]

All concepts, including 'causality', originate from an "I". Using the I-created concept of causality to go back explore the structure of "I" is, in fact, brilliant reverse-engineering.

[@ergenekonualkslayanliberal1077]

“因果关系”的概念是如何起源于“我”的概念的。 以及我们如何分析“我”？

[@AsharRM]

Yes it's quite a strange loop

[@deplant5998]

This sentence no verb.

[@blenderpanzi]

I see "I" as an abstraction.

[@deplant5998]

This ‘l’ is not an ‘i’ ... it is a lowercase ‘L’

[@yoananda9]

Fascinating, but yes ... the consciousness IS a quantum computer. That explains so much I don't understand why it's still considered "quantum woowoo".

[@trillionman2105]

Doesn't explain that much honestly

[@deplant5998]

There is no “I” in REALiTY

[@carlosluis1970]

exactly!

[@mindseye123]

This intro is massively self referential lol!

[@Miletus]

Human persons are artificially generally intelligent, autonomous, biological machines. A human being becomes a person by creating (programing) a virtual self, not by harbouring a ghost in a biological machine. I am my virtual self, an artificial strange loop, not a ghost or a spirit. Yes, I am an abstraction and I am real - just as real as my physical body - just as real as any valid abstraction such as the number of me (i.e., 1). To exist (i.e., to be real} is to be a member of the universal class, Ʋ, such that Ʋ = { x | x=x }. Class membership is binary; there are not degrees of being real. Virtuality, artificiality, and physicality are all aspects of reality, not degrees of reality. I am an abstraction, I am a real object of consciousness, and I am conscious. Indeed, I am conscious of my virtual self, an artificial strange loop. Therefore, I am a person.

[@adamsawyer1763]

If "the real causality" all takes place at the level of atoms, cells organs etc then why does evolution need to "force us to really believe" in these "airy fairy ideas" like 'I' and our free-will? This whole attempt to reconcile our conscious experience with our current mathemarical models of the physical universe feels like an argument ad absurdum which points more to the physics being wrong in some way than our conscious experience being somehow illusory. Its a valiant effort on behalf of the reductionist physicalist point of view but one that is ultimately self defeating.

[@hyphenpointhyphen]

All just mappings

[@robchr]

So being conscious is anything that can be anthropomorphized ;-). Concepts like France are nothing but human constructs. I think of consciousness as organisms with executive function with memory and an awareness of self.

[@trudyandgeorge]

i think you've missed the point. Referring to France in that way was a heuristic designed to upload the idea of one's self rapidly. You said "Concepts like France are nothing but human constructs", absolutely, and concepts like "myself" are nothing more than evolutionary constructs (equally as "real").

[@andybaldman]

France is more than a human construct. You want an organism with executive function, memory, and an awareness of self? Pick any country that has ever sought to conquer another country (e.g., WWII). A country can be both an organism and a human construct. Also, the concept of anthropomorphization itself is faulty, because it is anthropocentric. It assumes man is the only thing that can be conscious, so how could anything else have the same agency, executive function, and intelligence as man? However that's where the error lies. If consciousness can exist in forms other than man, then the entire idea of anthropomprhization falls apart, because it isn't about projecting consciousness onto other forms. Man becomes just one of multiple forms in which consciousness can exist, and consciousness itself becomes the reference, not man.