The Fabric of Real Things

[-]Bundle_Gerbe13y280

I am confused by these posts. On one hand, Eliezer argues for an account of causality in terms of probability, which as we know are subjective degrees of belief. So we should be able to read off whether X thinks A causes B from looking at conditional probabilities in X's map.

But on the other hand, he suggests (not completely sure this is his view from the article) that the universe is actually made of cause and effect. I would think that the former argument instead suggests causality is "subjectively objective". Just as with probability, causality is fundamentally an epistemic relation between me and the universe, despite the fact that there can be widespread agreement on whether A causes B. Of course, I can't avoid cancer by deciding "smoking doesn't cause cancer", just as I can't win the lottery by deciding that my probability of winning it is .9.

For instance, how would an omniscient agent decide if A causes B according Eliezer's account of Pearl? I don't think they would be able to, except maybe in cases where they could count frequencies as a substitute for using probabilities.

[-]MBlume13y120

OK, let's say you're looking down at a full printout of a block universe. Every physical fact for all times specified. Then let's say you do Solomonoff induction on that printout -- find the shortest program that will print it out. Then for every physical fact in your printout, you can find the nearest register in your program it was printed out of. And then you can imagine causal surgery -- what happens to your program if cosmic rays change that register at that moment in the run. That gives you a way to construe counterfactuals, from which you can get causality.

ETA: There's still some degrees of freedom in how this gets construed though. Like, what if the printout I'm compressing has all its info time-reversed -- it starts out with details about what we'd call the future, then the present, then the past. Then I'd imagine that the shortest program that'd print that out would process everything forward, store it in an accumulator, then run a reversal on that accumulator to print it out, the problem being that the registers printed out from might be downstream from where the value was. It seems like you need some extra magic to be sure of what you mean by "pretend this fact here had gone the other way"

[-]thomblake13y100

For instance, how would an omniscient agent decide if A causes B according Eliezer's account of Pearl?

An omniscient agent would have no reason to decide if A causes B, since causality is a tool for predicting the outcomes of interventions, and the omniscient agent already knows what's going to happen. The concept of "causality" is only useful from a perspective of limited knowledge, much like probability. And the concept of an "intervention" only makes sense in a level of abstraction where free will is apparent.

Pearl addresses this in slide 47 of this lecture. Causality disappears if you consider the entire universe as your object of investigation.

1[anonymous]13y

I think Bundle's question is just that, given the above, how can Eliezer also say that the universe is actually made of cause and effect?

1thomblake13y

Does Eliezer say that the universe is actually made of cause and effect? Also, what work is "actually" doing in that sentence?

5Bundle_Gerbe13y

"Actually" isn't intended in any sense except emphasis and to express that Eliezer's view is contrary to my expectations (for instance, "I thought it was a worm, but it was actually a small snake"). Eliezer does seem to be endorsing the statement that "everything is made of causes and effects", but I am unsure of his exact position. The maximalist interpretation of this would be, "in the correct complete theory of everything, I expect that causation will be basic, one of the things to which other laws are reduced. It will not be the case that causation is explained in terms of laws that make no mention of causation". This view I strongly disagree with, not least because I generally think something has gone wrong with one's philosophy if it predicts something about fundamental physics (like Kant's a priori deduction that the universe is Euclidean). I suspect this is not Eliezer's position, though I am unsure because of his "Timeless Physics" post, which I disagree with (as I lean towards four-dimensionalism) but which seems consonant with the above position in that both are consistent with time being non-fundamental. If he means something weaker, though, I don't know what it is.

1thomblake13y

Yes, I think Timeless Physics puts you on the right track, and it should be pretty clear that "causality" doesn't apply so much at the level of comparing possible states of configuration space, aside from perhaps metaphorically to point to which ones are adjacent to which other ones.

1[anonymous]13y

Well, if I may take up Bundle's question on his behalf, Eliezer said, in the article to which these are comments: and Where I take the second quote to imply an endorsement on Eliezer's part of the claim 'everything is made of causes and effects'. I don't know, I was quoting (without making that clear) Bundle's phrase.

1thomblake13y

That was from a koan, and a good one at that. How would one perform an experiment to determine whether the universe operates on causes and effects? It suggests there might be something wrong with the conception that the universe is "made of" causes and effects.

2[anonymous]13y

Maybe, though given the introduction to the article, I think the koan is about the question 'what counts as meaningful?', not 'is the universe made of causal relations'. But, perhaps more interesting than settling the question of whether or not Eliezer thinks the universe is made of causal relations, we can ask "Is the thesis that the universe is made of causal relations inconsistant with Eliezer's (and your) views on the objectivity of causal relations?' Given your responses, my dialectical instincts are telling me you think the answer to the above is 'Yes, they are inconsistant, and it is false that the universe is made of causal relations'. Is that so?

1thomblake13y

Yes. I thought I actually made that explicit. At least, it's not "made of" causal relations any more than it's "made of" probability.

2[anonymous]13y

Thanks. I think Eliezer is endorsing the 'causal relations are fundamental' reading, and that this apparently conflicts with the idea that causality is the tool of a limited observer. I think he's likely to see these as reconcilable in some way. That, at any rate, is my prediction.

-1Eugine_Nier13y

My prediction is that he never noticed this problem before.

3chaosmosis13y

My prediction is that he will not respond to this line of criticism within the next three days.

3Tyrrell_McAllister13y

Indeed. Eliezer motivates causal graphs by pointing to limitations in the observer: It's just not feasible to gather all of the necessary frequencies to do pure Bayesian conditionalization. But that's saying something about the observer, not about the object that the observer is talking about. We're only talking about causal networks because of our own limitations as epistemic agents. It seems a weird leap to say that reality itself is made out of this stuff (causal networks) that we wouldn't even be thinking about if it weren't for our own imperfections.

3Ronny Fernandez13y

It seems to me that this is the primary thing that we should be working on. If probability is subjective, and causality reduces to probability, then isn't causality subjective, i.e., a function of background knowledge?

-1Dr_Manhattan13y

This seems not in the least contentious, if you're talking about the map of causality.

2Eugine_Nier13y

The question is whether causality exists in the territory at all.

-5folkTheory13y

3Eugine_Nier13y

Thanks for succinctly articulating what was bothering me about this post. Can't upvote this enough.

1MBlume13y

This question seems decision-theory complete. If you can reify causal graphs in situations where you're in no state of uncertainty, then you should be able to reify them to questions like "what is the output of this computation here" and you can properly specify a wins-at-Newcomb's-problem decision theory.

-1adam_strandberg13y

An omniscient agent could still describe a causal structure over the universe- it would simply be deterministic (which is a special case of a probabilistic causal structure). For instance, consider a being that knew all the worldlines of all particles in the universe. It could deduce a causal structure by re-describing these worldlines as a particular solution to a local differential equation. The key difference between causal vs. acausal descriptions is whether or not they are local.

[-]Eliezer Yudkowsky13y120

Mainstream status:

The introduction to causality is intended to be bog-standard. All departures from mainstream academia are errors and should be flagged accordingly.

I can't particularly recall hearing anyone suggest that a universe is a connected fabric of causes and effects in the Pearlian sense of causality. Since Special Relativity there have been many suggestions that a universe is a connected fabric of 'events', or points in spacetime.

I can't particularly recall reading that you can only meaningfully talk about things you can find by tracing causal links (this theory will be developed further in upcoming posts).

[-]aaronsw13y280

Wolfram 2002 argues that spacetime may actually be a discrete causal network and writes:

The idea that space might be defined by some sort of causal network of discrete elementary quantum events arose in various forms in work by Carl von Weizsäcker (ur-theory), John Wheeler (pregeometry), David Finkelstein (spacetime code), David Bohm (topochronology) and Roger Penrose (spin networks; see page 1055).

Later, in 10.9, he discusses using graphical causal models to fit observed data using Bayes' rule. I don't know if he ever connects the two points, though.

5private_messaging13y

Few words: CPT symmetry. The causality is just how we model things at macroscopic scale where the time symmetry is broken thermodynamically. At the bottom level, it's relations that have to be true, and to say that value of one variable in a relation causes value of other variable in a relation is either a minor language misuse or a gross misunderstanding arising from such misuse.

8lukeprog13y

Once again speaking as the author of Eliezer's Sequences and Mainstream Academia... Like Eliezer, I haven't seen anyone suggesting that a universe is a connected fabric of causes and effects in the Pearlian sense. Nor have I seen anyone suggest that one can only meaningfully talk about things that one can find by tracing causal links. If some young philosopher wants to publish some great, attention-grabbing papers in Nous early in their career, I suggest they write up the philosophy-paper versions of the core ideas in The Useful Idea of Truth and The Fabric of Real Things (citing Eliezer, of course).

[-]Mitchell_Porter13y120

What would it even mean to say that the universe is a fabric of cause and effect "in the Pearlian sense of causality"? Pearl originated a new method of causal analysis, not a new concept of causation.

3Pfft13y

I guess it means causation as manipulability, as opposed to e.g. causation as contrafactual?

-2Eugine_Nier13y

How is causation as manipulability "Pearlian"? Pearl's whole point is that it's possible to determine causality without manipulating.

4pragmatist13y

In so far as Pearl has a theory of causation, it is a manipulability theory. Sure, Pearl provides a method for drawing causal conclusions from data about correlations. But what he means by causation is cashed out in terms of how interventions on one variable affect another variable, not just in terms of correlations. Correlations are indicators of causation for Pearl, but they don't constitute causation. The structural equations Pearl uses to represent causal relationships are meant to capture not just de facto mathematical relations between variables, they support counterfactuals. They tell us what would happen to the variable on the left hand side of the equation if the variables on the right hand side were manipulated. See section 6 of the SEP article for connections between the manipulability theory of causation and Pearl's formalism.

-1Pfft13y

But first you need to know what causality is. While the general idea that "causality is about manipulations" was around before Pearl, he certainly popularized it, and his causal bayesian networks and the do-calculus made it something that could be studied mathematically.

-10Bruno_Coelho13y

5[anonymous]13y

When you make this claim about the universe's ontology, it isn't clear whether you mean the is of identity or predication. Are you saying that the things constituting the universe are parts of a connected fabric of causes and effects or that the universe is reducible to causes and effects. The first claim is a standard part of conventional materialism (Pearl's recent contributions notwithstanding). The second claim is one that's set out in Alexander Bird's recent book Nature's Metaphysics: Laws and Properties. This is a somewhat radical program in that it proposes reducing structural properties to dispositional. Are you or aren't you proposing that (for example) squareness can be reduced to cause and effect.

3Eliezer Yudkowsky13y

Quote? SEP link? As mentioned, I've seen "fabric of events" in SR discussion, no one actually saying "fabric of cause and effect", certainly no Pearl.

[-]Eliezer Yudkowsky13y100

Koan 3:

Does the idea that everything is made of causes and effects meaningfully constrain experience? Can you coherently say how reality might look, if our universe did not have the kind of structure that appears in a causal model?

[-]Richard_Kennaway13y120

Physicists tell us that reality is, at the lower levels, described by differential equations. Those describing the temporal evolution of systems are generally hyperbolic differential equations. That is, they have microscopic causal structure: the properties at a point (x,t+dt) depend on the properties only in a small, finite neighbourhood of x at time t. This is what allows larger scale events to be described more abstractly in terms of causal graphs. (A known problem here is the time-symmetry of all these equations. I don't have a solution to that either.)

Some have speculated about the very finest scale being discrete, with fundamental laws of temporal evolution such as those of cellular automata or graph rewriting. That would also provide the fundamental causal structure from which causal structure on the macroscopic level emerges.

A universe whose fundamental laws were all elliptic differential equations -- those in which a change in boundary conditions in one place changes the solution everywhere at once -- would not support causal reasoning. Everything depends on everything else and none of its emergent phenomena would be describable by sparse causal graphs. I'm not sure you could have agents within such a universe, for it to look like anything to them, or any notion of time, but we can imagine such universes from outside.

8Schpwuette13y

A causally disconnected event would not affect our experience, so no, it does not constrain our experience. But it wasn't meant to: it was meant to constrain our maps, not our experience. To stop us from believing meaningless statements. Every causally separated group of events would essentially be its own reality.

5The_Duck13y

No. For any set of observations O1, O2, ..., O_N, you can construct an overprecise causal model that links some central cause to nodes O1, O2, ..., O_N. [That is, our causal model is: "You observed O1 because God willed it. Then you observed O2 because God willed it..."] Thus causal models can explain any set of observations. Constraining experience requires in addition some sort of Occam's Razor that prefers simpler causal models.

1The Dao of Bayes13y

I'd agree with the assertion "Unknowable relationships are equivalent to no relationship". In other words, cause-and-effect is only meaningful if we have some ability to gather information as to the nature of the relationships. Occam's Razor is one such method, but I wouldn't assume that it's specifically required - we could just as easily be born with divine insight as to the cause-effect relationships.

4asparisi13y

To answer this, I need to answer another question: What would the universe look like if it did not have the structure of causal models? If I know that then I can know whether or not that universe looks different than ours. So we're talking about a universe where there is no A --> B causal connections. Where you can't accurately say something like "When I eat, it causes me to be less hungry," or even say anything, since "saying" something requires some sort of causal ability: the ability to push air in intensional patterns. And that universe... really doesn't look like ours. I might want to talk about a universe where there are correlations but not causations, but then how do I explain the correlations? Are they just brute facts? That seems unlikely. But even in that event, I would expect rules to break down in chaotic ways: one day, food wouldn't make me feel nourished. And it wouldn't be because I ate the wrong foods or something like that, it'd just be because the correlation broke down. And that really doesn't look like our universe either. I may be wet because it was raining, but I don't get wet merely because some 'correlation' between 'sunny_day" and "not_wet" broke down.

1ChristianKl13y

Actually we aren't. We are talking about an universe where there are effects that happen without cause. That doesn't mean that effects that have causes aren't allowed.

3asparisi13y

Hm. That is conceptually possible, I suppose. Although I can't really imagine how that sort of demarcation criteria would work.

3CCC13y

Yes, to a limited extent I think I can. A universe without a completely causal structure would have to have a number of effects that have no cause. (This need not apply to all effects in this universe, merely to a subset of effects). The Steady-state universe, a now abandoned alternative to the Big-bang model, may be an example; in the steady-state universe, if I remember correctly, it was theorised that small bits of matter (hydrogen molecules) appeared out of nowhere at random times, at a very low rate. This is an example of an effect without a cause. Alternatively; consider a universe where a man suddenly appears out of nowhere in the middle of a public area. He claims to be a time traveller, from fifty years in the future, coming back in time to prevent some terrible disaster. Though his story is unbelievable, the disaster he predicts is prevented. Fifty years pass, and the time traveller is not sent back to the past (since the disaster was, after all, prevented). This does not cause a paradox (the laws of this universe are that matter can appear at any time, in any arrangement, including a person with false memories of time time travel from a non-existant future) Such a potential universe thus has an effect (the man appears) without a cause.

2selylindi13y

The examples don't quite work. In that steady-state universe, the appearing matter subsequently interacts and so is causally connected to the rest of the universe. You just have to trace the connections backwards, and eventually you reach a stopping point. Similarly with the false time traveler: he causally affected the world, so he's clearly part of a chain of causes and effects. It's a separate question to ask whether everything has a causal in-connection and a causal out-connection. Both your examples, and epiphenomal theories of mind, are meaningfully about unidirectional causal links.

1CCC13y

Hmmm. I had been trying to consider the case where some events have no causes. Now, you're pointing out - as I understand your argument, and please correct me if I am wrong - that the same event can be a cause, and therefore that it is a part of an interconnected causal universe. This is an interesting argument. If I notice the existance of something, then that is a causal relationship; it exists, and this causes me to notice it by some means. Therefore, anything that I observe shares a causal relationship with the fact of its observation; if that is enough to consider it part of a causal universe, then I can only observe a causal universe. I cannot imagine anything that I cannot, by definition, observe in any manner (including by any deduction) constraining my experience. On the other hand, I can (at least in certain imagined universes) observe effects without causes; I can observe the time traveller appearing for absolutely no reason.

0The Dao of Bayes13y

I had assumed the question was "a universe where an event can have out-connection but not in-connection" (and the consciousness one was about having an in-connection but no out-connection). In a universe with NO connections, you wouldn't get patterns, much less emergent patterns - no atoms, no stars, no planets, no life, not even internet! Multiple segregated causal networks would be multiple realities (I think the simplest reference would be to think about two different books or virtual worlds, although there's actually still some connection there) Individual events which have out-connections but not in-connections, by contrast, seems like a fairly interesting question to explore. Why DO we assume that all evens have in-connections? Why not just assume the weather is because Zeus is mad, and Zeus' emotions are beyond the ken of mortal minds?

-3Eugine_Nier13y

Not all, the past doesn't have in-connections and the future doesn't have out-connections.

1The Dao of Bayes13y

Can you explain what you mean by that? I tend to model past events as having even-further-past causes. I definitely model future events as having even-farther-future ramifications. So they'd seem to generally have both in and out.

-3Eugine_Nier13y

I was referring to the past (or future) as a whole.

2Ronny Fernandez13y

At first thought. It seems that if it could be falsified, then it would fail the criteria of containing all and only those hypotheses which could in principle be falsified. Kind of like a meta-reference problem; if it does constrain experience, then there are hypotheses which are not interpretable as causal graphs that constrain experience (no matter how unlikely). This is so because the sentence says "all and only those hypothesis that can be interpreted as causal graphs are falsifiable", and for it to be falsified, means verifying that there is at least one hypothesis which cannot be interpreted as a causal graph which is falsifiable. Short answer, not if we got it right this time. (term clarification) All and only hypotheses that constrain experience are falsifiable and verifiable, for there exists a portion of experience space which if observed falsifies them, and the rest verifies them (probabilistically).

2AlexMennen13y

Asserting that a particular causal model is an accurate description of the universe certainly constrains experience to fit the causal model. Asserting that it is possible to accurately describe the universe with a causal model is tautologically true because a fully connected model fits every possible probability distribution. Asserting that it is possible to accurately describe the universe with a simple causal structure constrains experience to match a simple causal model. If you know that the universe fundamentally works by running some causal model, asserting that the universe runs this particular causal model predicts that you will never find a more accurate causal model for describing the universe. I'm not sure how asserting that the universe fundamentally works by running some causal model would constrain experience.

2selylindi13y

Several possible examples come to mind for a universe with no cause or effect. First is a universe with only one thing in it, so that there's nothing for it to be causally connected to. Second is a universe with multiple things in it that could in principle interact but due to the set-up of the universe never actually do interact. For example, a universe of rigid particles in a void where they would interact if they struck, but the distances between all particles are too great for that to occur in the lifetime of the particles. Third, a universe in which its entities do interact, but nothing ever changes, so there are no nontrivial correlations. Perhaps count a universe of mutually repelling particles in a void, arranged in an unchanging crystalline structure. Fourth, more loosely, is a universe in which its entities do interact and change, but the arrangement of all the things is such that only minimal correlations arise. Perhaps a universe analogous to a closed system containing a gas at thermodynamic equilibrium. Prescinding from those, the idea that everything in our universe is made of causes and effects constrains my expectations in that there should multiple things (check) that actually interact (check) and change (check) and have nontrivial correlations (check). Other than the continuation of such things, I can't readily think of any sense in which the idea constrains my expectations for future experiences.

0MaoShan13y

To clarify you are pointing out that without cause and effect, stars would not even form, or even baryons for that matter. A gas at thermodynamic equilibrium probably would not be conscious, so from there we can go straight for the Anthropic Principle. Otherwise the question has no real solution--like asking what a peacock would look like if eyes didn't exist. (Handoflixue already killed this horse, though--sorry for beating it.)

4CronoDAS13y

Any time a theory of cosmology implies that you're a Boltzmann brain, run. ;)

2Vaniver13y

My first answer is that the idea of 'interventions' comes from causality- if I don't like being wet, and it's raining, I can deploy an umbrella, and that will make me not get wet(ter) from the rain. The idea that everything is made of causes and effects meaningfully constrains actions, because an agent that understands causes can manipulate reality better than an agent which doesn't understand causes. But that's just a further elaboration of what reality looks like now. It's difficult for me to imagine a world without causality- and I don't think that's just because I (like most humans) am a deeply causal thinker. Causal models are just math, and so they can be imagined just like multiple spatial geometries can be imagined, regardless of what the universe's real shape is. I think you would need to have a universe in which no causal models could be justified by the data, which would be an incredibly strange place. And, actually, by the definition of universe you use in the post, a universe without a causal structure would just be a single node, forever alone.

1wizzwizz45y

If our universe did not have the kind of structure that appears in a causal model, then our reasoning would not function properly. Induction would fall completely flat, and I don't think our brains would work right. We probably wouldn't exist in such a world, but if we're taking into account anthropic effects… well, I'm not even sure a human could survive long enough for a single conscious thought, since their state at time t+1 wouldn't follow from their state at time t. After skimming others' replies, I've realised that I've answered a different question. To answer this one: no, but we've got no way of reasoning about or utilising anything else, so it doesn't matter. That there are no uncaused causes of real things (a weaker claim) does meaningfully constrain experience, though.

1[anonymous]13y

If the universe were not governed by cause and effect, then we could not consistently improve our predictions of the future by observing the past, even by a little bit.

1jimrandomh13y

No. I literally assign prior probability zero to the statement that the universe is not made out of causes and effects, because there is causal structure in Turing machines and in all Turing-complete models of computation which could make up the Solomonoff prior. Causal structure is a very broad thing - it's just a sparse graph of interacting entities with a lattice ordering. I can imagine a universe in which the local ordering I observe doesn't go as far forward or back as I thought, and the true everything-is-causes-and-effects structure is pushed one layer back to something completely hidden from me. I can imagine a universe in which I've falsely inferred an ordering which isn't there, and getting confused by cycles in a graph that I thought was causal. But a universe with no causality at the lowest layer - I think causality is inherent in too many things, and that after subtracting those things there's not enough option space left to make a universe out of.

1The Dao of Bayes13y

I think this is one of the few instances where p(0) is actually appropriate! :)

0ChristianKl13y

If you can't imagine an universe that's not made out of causes and effects than talking about such an universe is meaningless. To be able to assign a probability zero to the statement that the universe is not made out of causes and effects you have to be able to imagine a universe that's not made out of causes and effects. Meaningless statements can't be true or false. Speaking about their probability makes no sense.

2Eugine_Nier13y

Before Einstein nobody could imagine without universal time either. In other words, if you can't imagine a universe with property X, that's a fact about you not about the universe.

-1Eugine_Nier13y

In other words, causality is the invisible pink unicorn.

2jimrandomh13y

I don't understand this reply at all, except as an indication that I didn't communicate these concepts as well as I'd hoped.

-1Eugine_Nier13y

The text I quoted in the grandparent seems to be saying that even if the universe doesn't contain causality, we can always postulate an external causality structure even if most of it can't be observed.

1Manfred13y

How about mild violations of causality?

0jimrandomh13y

That isn't a violation of causality, it's just a universe with time that goes infinitely far forward and back.

2Manfred13y

Time appears to stretch backward infinitely, because local causality still looks nice. But actually it only stretches backward to now. If time was actually a straight line rather than a cycle, then one could intervene in this system and it would only change the future. But if you intervened in a cyclic system, you would find that you had changed what looks like the past. In fact, self-consistency constraints would kick in and logically prevent you from some sorts of intervention that would make perfect sense if time stretched off to infinity - it's like using quantum mechanics instead of classical mechanics, where classically it seems like you should be able to have any amount of energy you want. More philosophically, we've run into Aristotle's distinction between self-cause and infinite regress. And also why Bayesian networks are typically defined as directed acyclic graphs, thus putting cyclic graphs outside of the hypothesis-space.

3chaosmosis13y

I think I understand your distinction between infinite regress and cyclical causation, but I don't understand why that implies a cyclical universe violates causality. To rephrase jimrandomh: that isn't a violation of causality, it's just a universe with cyclical time.

1Manfred13y

Well, one can still think of it causally - you can still draw a graph with arrows, at least. But it's atypical causality. Typical causality is like kicking a ball. The ball sits still until you kick it, and you can kick it however you like and it will roll away. But once you have loops, it's like if the ball had to go through a portal to the past and kick itself. As soon as you try to kick the ball, the ball you would have kicked has already gone back to the past and hit itself in a way consistent with the motion of your foot, so it will feel quite unlike kicking the first ball. And in fact it is physically impossible to move your foot in a way inconsistent with the ball being the cause of its own motion, even though trying to kick the ball restricts the possibilities - or rather, being able to try to kick the ball tells you that the possibilities were already restricted... In typical causality, the ball has a reason for moving the way it does - you can trace the motion backwards to some acceptable starting point, like "I kicked it as hard as I could toward the fence." When you add cycles, tracing the chain of arrows back does not need to end at anything you find remotely satisfactory or even unique - "the ball moved because it hit itself because it moved because it hit itself..."

0rkyeun13y

This is a problem with your personal intuitions as a medium-sized multicellular century-lived mammalian tetrapod. No event in this chain is left uncaused, and there are no causes which lack effects in this model. Causality is satisfied. If you are not, that's your problem. Hell, the energy is even conserved. It runs in a spatial as well as a temporal circle, what with the ball hitting itself and skidding to a stop exactly where it was sitting to wait for the next hit. On the other hand, in such a universe quantum mechanics does not apply, because worldlines cannot split, which also removes any possibility of entropy. ALL interactions are 100% efficient.

0Ronny Fernandez10y

Does EY give his own answer to this elsewhere?

0hairyfigment10y

Try to guess what he would say before reading it. You can also click on one of the tags above to read, say, the sequence on epistemology.

0moshez13y

Meditation 3: [Hardest of the meditations, for me.] Let us observe the difference between [post-utopian]-->[colonial alienation] and a connected thing (say [I see you picked Ace]-->[You see you picked Ace] from a deck of cards): In the first case, there is no way to settle an argument about whether Ellie is post-utopian or not. We would predict that it would cause arguments between people that are not settled. Anything connected to the causal web is more likely to lead to settlable arguments, at least among people behaving more-or-less rationally. It is not a perfect test, but it does suggest that I expect to see different things from connected networks and unconnected networks, like people changing their minds.

0Benquo13y

There are three alternatives I can think of: 1) The universe has a joint probability distribution such that events are not independent, but so few events are independent that limiting the description to a causal one does not materially reduce the complexity of the "true" model. In this case, There would only be the joint look-up table. 2) All events are independent. 3) The universe is not well-described by the laws of probability. The first case would make prediction pretty much impossible (since you're never dealing with the exact same set of variables twice). "Same river" and all that. No prediction more specific than "the universe persists in some way" would be true very often on average, since there are so many possible outcomes, and no reason to believe that the next joint probability for the event you care about is similar to the last one. In the second, you could predict roughly the proportion of events based on the proportion of past events, but there would be no discernible pattern, or result of action. In the third, I can't tell you what you're likely to observe, for obvious reasons.

0folkTheory13y

It would be completely random, with all events being equally likely at every point in time. It would have no history, since the past has no effect on the present or future

0Dolores198413y

You could plug a baby's nervous system into the output of a radium decay random number generator. It'd probably disagree (disregarding how crazy it would be) that its observations were best described by causal graphs.

0DuncanS13y

I think it is a mistake to tie the question of what reality is to the particulars of the physics of our actual universe. These questions are about what it is to have an external reality, and the answers to them should be the same whether the question is asked by us in our current universe, or by some other hapless inhabitants of a universe bearing a distinct resemblance to Minecraft. I can imagine types of existence which don't include cause and effect - geometrical patterns are an example - there are relationships, but they are not cause and effect relationships - they are purely spatial relations. I can imagine living in a universe where part of its structure was purely such spatial relationships, and not a matter of cause and effect.

0Bundle_Gerbe13y

Imagine a universe that is made only of ideal billiard balls eternally bouncing around on a frictionless, pocketless billiard table. Essentially the same thing as selylindi's idea of a gas in thermodynamic equilibrium. Imagine yourself observing this universe as a timeless observer, or to aid the imagination, that it's "time" dimension is correlated to our space dimension, so we see the system as an infinite frozen solid, 11 by 6 by infinity, with the balls represented by solid streaks inside that go in straight lines except where they bounce off each other or the boundary of the solid. Now, this system internally has a timelike dimension, except without increasing or decreasing entropy. And the physics of the system are completely reversible, so we have no basis of saying which way is the "future" and which way is the "past" in this system. We can equally well say a collision at one time is "caused" by the positions of the balls one second in the "past" or one second in the "future". There is no basis for choosing a direction of causality between two events. In our universe, our time is microscopically reversible but macroscopically irreversible, because of the fact that the universe is proceeding from a low entropy state (we call that direction the "past") to a high entropy state (the "future"). I am curious, can anyone coherently describe a universe with nothing similar to irreversible time, but with a useful notion of causation? Or with something like irreversible time, but no causation whatsoever? I have tried (for much more than five minutes!) and not succeeded , but I am still far from sure that it's impossible to do. It might be too much to ask to imagine actually being in a universe without causation or time, but perhaps we can think of how such a universe could look from the outside.

0bsterrett13y

What is the difference between constraining experience and constraining expectations? Is there one?

0CronoDAS13y

I think that in a universe without a cause and effect structure, we'd tend to see more chaos than we do. Casual models would tend to fail to correspond to reality; all we could do is attempt to reason over joint probability distributions, which, as was explained, is kind of hopeless.

0dankane13y

Well, my experiences could not be readily predicted by a simple Bayesian net. Thus I would expect the joint probability distribution on all things to be really complicated unless there were some other kind of pattern to it. Heck, maybe my experiences correspond to repeated independent samples from a giant multidimensional Gaussian or something (rotated of course so that the natural observable variables have non-trivial covariance).

0[anonymous]13y

I'm having a hard time making sense of the question. I'm sort of trying to imagine that there are no causal relations, and then opening my eyes to see what's different. Except 'seeing' and 'looking' don't make sense any more. What could it mean for something to 'appear'? I can't get a grip on even the idea of 'experience'. At which point I conclude that either I am not imaginative enough, or the exercise is inconceivable.

0philh13y

(Written before reading other comments.) Assuming our universe is causal, then no you cannot. Any universe you can imagine has laws of physics which can be modelled by the interactions of the neurons in your brain, and are therefore causal. There might be some things that can only be experienced in an acausal universe, but it is not possible for me to imagine them.

0earthwormchuck16313y

In a universe without causal structure, I would expect an intelligent agent that uses an internal causal model of the universe to never work. Of course you can't really have an intelligent agent with an internal causal model in a universe with no causal structure, so this might seem like a vacuous claim. But it still has the consequence that P(intelligence is possible|causal universe)>P(intelligence|acausal universe).

0wedrifid13y

That seems plausible but I must admit that I don't know enough details about possible "acausal universes" to be particularly confident.

1Vaniver13y

If intelligence is seen as optimization power- in the sense of agents that constrain possible futures for their benefit- then it seems clear that the rewards to intelligence are 0 or negative in acausal universes, and so they should be less likely than in universes where they have positive rewards.

0thomblake13y

This one is great - it really does have the contradictory nature of traditional koans. Just try to picture what you would do to test whether the universe was causal, and how we would adjust our models based on such a test.

0Scottbert13y

I suppose I would say that reality would look as if things happened with no observable pattern related to the things that happen before them, but looking at things requires a long causal chain between photons being emitted and signals in my brain. Supposing I happened to somehow flash into existence for an instant in a noncausal world, or that causality suddenly failed, I would not expect to be able to experience anything past that point since my experiences depend on so many causal processes.

3byrnema13y

An acausal world would be a world where there is experience (some kind of 'reality') but without the possibility of contradiction. It is like a dream, where what you experience is a story and that story can be written over, and can shift, at any moment, without contradiction. So at one moment you see a cup, and then the next moment, you didn't see a cup. It's not that the cup changed or that you were mistaken, it's just that at first your experience was "seeing a cup" and then at the next moment your experience was, "I wasn't there to see that cup". The experiences both happened: You saw it. And you weren't there and didn't see it. It's a string of happenings connected by the word 'and' : I saw a cup and I didn't see a cup and I was there and I wasn't hungry and the grape hurt. This isn't an original idea, I realize now. I've read of a place like this in at least a couple science fiction stories.

0RomanDavis13y

I would expect not exist in a way that suggests causality, e.i. being born and then expecting death, rather than the other way around. This is hard for me to imagine because I didn't really evolve for that world. It's possible that our universe doesn't work that way at the smallest level, but it seems might suspicious that random events lead to a largest world that operates very deterministically. Still, it is possible that this is just the manifestation of probabilistic laws at the smallest level. It's definitely paying rent so far,(for those who do the experiments) so that's we're going with, and there hasn't been a good argument or experiment against it yet. Infintesmal "violations" of causal laws as manifestations of probabilistic laws don't seem to effect me very much. Large ones that would pay rent haven't happened on the level that would pay rent on an evolutionary or personal level, and, as I understand it (which is not terribly well) these probably won't happen unless the universe ran from the big bang to heat death a couple hundred times. I can make models in my head where the universe (on my scale) is really chaotic, but looks deterministic because of a conspiracy by matrix gods or whatever, but that seems to violate Occam's Razor, for what that's worth when matrix gods control your life.

[-]thomblake13y70

"A universe is a connected fabric of causes and effects."

But if you're talking about the entire universe, then there aren't any causes and effects, just as there isn't any time. Causality only shows up when you look at a slice of the system, and whether A causes B or B causes A or they are actually independent can vary based on which slice you choose.

[-]loup-vaillant13y60

You've shown no great understanding of philosophy--or even a slight understanding.

To you. I for one have been sufficiently impressed that I'm willing to pay 5 karma to tell you.

For instance, Lukeprog is the only person I know who rejected religion for the right reasons: asking the questions in good faith, then looking up and thinking about the answers. Have you noticed the number of citations at the bottom of many of his main posts? It could have been only to show off, but I bet it would have been noticed here.

The evidence I have say Luke knows more abo... (read more)

[-]Eliezer Yudkowsky13y60

Koan 2:

"Does your rule there forbid epiphenomenalist theories of consciousness - that consciousness is caused by neurons, but doesn't affect those neurons in turn? The classic argument for epiphenomenal consciousness has always been that we can imagine a universe in which all the atoms are in the same place and people behave exactly the same way, but there's nobody home - no awareness, no consciousness, inside the brain. The usual effect of the brain generating consciousness is missing, but consciousness doesn't cause anything else in turn - it's just... (read more)

[-]jimrandomh13y100

Does your rule there forbid epiphenomenalist theories of consciousness - that consciousness is caused by neurons, but doesn't affect those neurons in turn?

A forest does not cause trees, and trees do not cause a forest; they are related, but their relation is of a different kind.

However, if you're willing to abuse definitions a little bit, you can pretend that relation is causal; if you define a forest in terms of things like tree-density, then the forest is "caused by" trees, but it's a causal dead-end which affects nothing in turn. The same holds for neurons and consciousness. However, it would be silly to talk about a bunch of trees in the same place which were not a forest, and similarly silly to talk about a working, talking brain without consciousness.

2ChristianKl13y

The debate isn't about whether "consciousness is caused by neurons" is true but whether the specific arguments that Eliezer made in this thread forbid that "consciousness is caused by neurons" and "consciousness doesn't affect neurons" can both be true. I don't see at all how the argument you are making here has something to do with the rule "For a statement to be comparable to your universe, so that it can be true or alternatively false, it must talk about stuff you can find in relation to yourself by tracing out causal links."

1Eugine_Nier13y

I assume the people arguing that "consciousness is caused by neurons" mean something similar to "the forest is caused by trees" and Eliezer is simply straw-manning/misinterpreting it.

1RichardChappell13y

Nope. Epiphenomenalism is motivated by the thought that you could (conceivably, in a world with different laws from ours) have the same bundles of neurons without any consciousness. You couldn't conceivably have the same bundles of trees not be a forest.

-2Eugine_Nier13y

Good point, thanks.

9CCC13y

My reply (before reading other replies) is that no, such theories are not forbidden. There is a causal link with the rest of the universe; neurons are causing consciousness. Consciousness, in turn, does not need to cause anything else to be part of the universe; it is part of the universe merely by being caused.

8Benquo13y

If consciousness is epiphenomenal, then what generated your question? I assume you are positing non-conscious people who ask all the same questions about consciousness that we do. I don't see a reason to rule out neuron-generated epiphenomena a priori - for example, the most elegant explanation of the phenomena might predict some entity that cannot be measured directly, and never affects anything measurable after its generation - but if they existed, we wouldn't notice and wouldn't ask questions about them. Therefore, whatever real thing it is you're asking about cannot be an epiphenomenon. If true consciousness is an epiphenomenon, then your question is not motivated by experiencing true consciousness. If your question is motivated by experiencing true consciousness, then true consciousness is not an epiphenomenon.

2DaFranker13y

I like your koan responses. They seem like the sort of thinking and arguments that could actually be useful in a discussion, too.

0Benquo13y

Thanks. It started as a tiny set-piece speech about how consciousness couldn't be an epiphenomenon, which is explicitly what the koan isn't asking about. I had to rewrite several times before it seemed like the sort of thing that might actually engage the questioner. I'm still not sure I actually answered the koan, though...

0DaFranker13y

I think your answer is very much what the koan was meant to generate. Simply saying 'No, I don't a priori eliminate epiphenomenal theories' seems like it'd miss the point entirely. You tackle the source of the concern or question, and conclude with a very good "If A, B, if ¬A, C" statement that easily follows from your arguments. More importantly, your answer seems to completely reduce the problem and dissolve the question.

4Emile13y

So if Universe A features epiphenomenal consciousness, and Universe B doesn't, and that we consider the statement that we are in universe A and not universe B, then looking back at the rule: These "causal links" you trace out are part of the map, not part of the territory - you have to be able to deduce their existence. And in this case (unlike the spaceship scenario), there is no way anybody can deduce the existence of the neuron -> consciousness link, since by definition nothing can be observed about the consciousness. => so, firmly in the "meaningless" camp.

3earthwormchuck16313y

You can account for a theory where neurons cause consciousness, and where consciousness has no further effects, by drawing a causal graph like (universe)->(consciousness) where each bracketed bit is short-hand for a possibly large number of nodes, and likewise the -> is short for a possibly large number of arrows, and then you can certainly trace forward along causal links from "you" to "consciousness", so it's meaningful. And indeed for the same reason that "the ship doesn't disappear when it crosses the horizon" is meaningful. We reject epiphenomenal theories of consiousness because the causal model without the (consciousness) subgraph is more simpler than the one with it, and we have no evidence for the latter to overcome this prior improbability. This is of course the exact same reason why we accept that the ship still exists after it crosses the horizon.

2moshez13y

[Cheating, since I already read some Zombie sequences, but have not read any replies in this thread] The consciousness causes you to speak of consciousness, which is the result of neurons in your brain firing your jaw muscles (and other muscles, and so on). If it was epiphenomenal enough that none would talk about it, we wouldn't have this question in the first place. [Has consciousness] --> [Writes books/blogs on consciousness] Causually connects consciousness to the universe.

2Richard_Kennaway13y

The rule does not rule out epiphenomenalism. It only rules out entities without causal connection in any direction with the material world, and epiphenomenalism says that there is a connection, albeit only in one direction. Epiphenomenalism is ruled out for other reasons. Epiphenomena can be considered in terms of causal graphs. Suppose you have two causal graphs proposed to describe some phenomenon. In one, there is a variable called X, with many arrows in and out. The second graph is identical to the first, except that X is renamed to a new variable Y, then X is added back in with a causal arrow from Y to X, and no other arrows impinging on X. Y is stipulated to be unobservable and un-intervenable on, and X to be observable but un-intervenable on in both graphs. Given these limitations on observations and interventions, is there any experiment that can distinguish between the two graphs? I think there is not. If not, what does it mean to say that there is such a Y, and that it is Y, not X, that is producing the apparent causal effects of X? It's almost as if at the meta-level, it is Y, not X, that is the epiphenomenon.

2Vaniver13y

Drill down on "outside the universe," as it seems to conflict with the definition of universe we're using here. Presumably, we mean something which has access to the nodes in the connected fabric of causes and effects, but is not an input to any of them- much like the proposed consciousness. If consciousness is caused by neurons, it's in the fabric, and our omniscient observer will be able to see whether or not consciousness is a node in the fabric or not. We'll avoid the infinite regress by enforcing that all observers are inside the universe / the connected causal fabric, but they are differentiated by where they are in the fabric. From inside the fabric, we can only comment confidently on the nodes we have access to, and implicitly the nodes which are ancestors of those nodes. Agents like us are both causal ancestors and descendents of each other- we see clear feedback loops that convince everyone of the reality of other agents (barring those confused by solipsism, of course). There could be silent observers (of the consciousness variety or otherwise) that are descendents but not ancestors of the nodes we have access to- and because they're not ancestors of the nodes we have access to, we can't comment confidently on them! But by Occam's Razor, it's generally better to not put such nodes into your model of the universe, since they don't give you any additional predictive or manipulative power, but still cost cognitive resources to lug around. (It seems to me that a principle that only relies only on the structure of the causal graph which excludes silent observers will also exclude the spaceship that has passed the horizon- and so either we need a principle from outside the model, like "my intuition says spaceships don't disappear but silent observers aren't real," or to put them in the same bin.)

1wizzwizz45y

The only problem with epiphenomenalist theories of consciousness is that the thing we call consciousness does have an effect on our actions (proof: answer the question "how do you know you're conscious?" out loud), so the thing they call "consciousness" must be something different. However, these rules don't say that this thing is impossible or meaningless; by definition, it's caused by neurons, and so (if it actually happens) would be causally linked to reality, and hence meet the criteria for being "real".

1moridinamael13y

Impossible, no; meaningless, yes. Consciousness in reality causes us to emit sentences about an awareness of subjective experience, so there is a coupling between neurons and the conscious algorithm being run by those neurons. If consciousness were merely a passive awareness with no causal influence upon anything else in reality, it would not be causally entangled with the universe. This is as close to "meaningless" as I can imagine.

1[anonymous]13y

A one way causal connection cannot be observed. If neurons cause consciousness, but consciousness does absolutely not affect anything else, then there is literally no way to observe consciousness, and so you must in addition to the description of the causal universe include the fact that there are things that are affected by the universe's existence but does not affect the universe, and furthermore that consciousness is one such thing. A strictly simpler theory is that there is a causal universe, without all that epiphenomea cruft. Also, the viewpoint of epiphenomenal consciousness is very much triggering my pattern maching of 'human specific inbuilt stupidity'.

-1Eugine_Nier13y

How's this different from the case photon leaving my future light cone?

1[anonymous]13y

it is different because the mathematical model of the universe makes a mention of the photon, but not of the consciousness. In a theoretical future version of python, the photon example would be: import universe universe.Photon().leave(universe.observers["Eugine_Nier"].light_cone) While the conssciousness example would be: import universe import epiConsc epiCons.Consciousness(universe.observers["Eugine_Nier"]).observe() On a side note,remember that there are no fundamental photons nor light cones. There is only whatever QFT is the true one; the rest are mathematical consequences.

1The_Duck13y

As I read it, the rule does not forbid epiphenomena. Epiphenomena are caused by ordinary things that we can observe, so they are connected to us by causal links, even though we can never discover these links, since epiphenomena do not themselves cause anything we can observe. This seems like a coherent way the universe could be, but isn't. The machine simulating the universe could constantly scan it for blue objects and populate a list of all blue objects--but without using that list for anything further. This would be an epiphenomal blue-ness tag.

1beoShaffer13y

I would say that it isn't ruled out a prior because a causal chain does run from the observable (the neurons) to consciousness.

0Paul_G13y

"We can imagine any number of universes, that does not always lead to a good argument. In this case, the main issue with the argument is that while we can imagine that universe, it doesn't look like ours. There's no talk of consciousness, there's no self-reflection. Those are things in reality clearly caused by a link between our thoughts and our brains, one that goes in both directions. Imagining a world in which people act exactly like people do now, but without a consciousness, strays so clearly outside the bounds of Occam's Razor that there doesn't seem to be any point in thinking about it. Adding in a mysterious 'zombie master' to make the zombies act as though they had consciousness... Well at this point, we're not talking about anything remotely resembling reality. This entire thought experiment in no way gives us any truths about reality whatsoever. It is completely meaningless."

0Dolores198413y

It does not. Epiphenomenal consciousness could be real for the same reason that the spaceship vanishing over the event horizon. It's Occam's Razor that knocks down that one.

0CronoDAS13y

A strict reading of the rule suggests that it doesn't; consciousness is caused by neurons, and we observe neurons.

0hairyfigment13y

As with the psychic cousin, this particular rule doesn't seem to forbid it outright (at least not in the obvious way). But a node thrown in with only one link to the rest of the universe still seems suspicious. It smells of an ad hoc hypothesis even before we ask if it means "consciousness" in the usual sense. (For this one I only saw Emile's comment.)

0dankane13y

With the above solution, then yes epiphenomenalist theories of consciousness are meaningful. They clearly describe networks in which consciousness connected to the rest of reality that you experience. On the other hand, I think that this is because the above statement fails at its intended goal rather than that these theories are actually meaningful. In particular I think one should append to the above: "Furthermore, it must not be possible to remove this stuff (and perhaps some other stuff) from your causal network without affecting things that you can experience."

0AlexMennen13y

For any precise definition of consciousness that fits epiphenomenalism, it would be meaningful. However, to my knowledge, no such precise definition has been put forward.

0GDC313y

It doesn't rule it out. Unless you're directly observing those epiphenominal nodes, Occam's razor heavily decreases the likelihood of such models though, because they make the same predictions with more nodes.

0philh13y

(Written without reading other comments.) The theory does not forbid a node from having no causal effect on anything else, which is what I think an epiphenomenal consciousness would be. But such a node could not be measured. A logical positivist would say this makes the theory meaningless, and I'm inclined to agree. Conceivably, if the universe runs on a computer, there could be a node which cannot be measured from inside the universe but could be measured from outside. Perhaps debug info is turned on, and any time neurons fire in patterns that match some regex, the word "consciousness" is printed to stderr. I would not say this is meaningless; if it is true, it gives us a way of communicating with whoever is watching stderr (but no guarantee that such a being exists or would talk back). But it is not suitable as a theory of consciousness, unless you want to stipulate that neurons write about consciousness for reasons that are completely uncorrelated to consciousness.

0The Dao of Bayes13y

Draw a Venn Diagram, with "things that are causes" and "things that are effects" (get-caused-by-other things). The intersection in the middle is "reality" Consciousness, in this model, is "effect, but not cause" - it doesn't change any of our anticipations about the rest of the universe, and is therefor meaningless (but not intrinsically impossible) Psionics, in this model, is being asserted as "cause, without effect" - either it actually is effected by the state of reality, and thus correlates with reality (in which case it is part of "reality"), or else it is not correlated to reality and thus cannot produce useful information on the state of said reality.

0The Dao of Bayes13y

Note that "cause, without effect" does include the possibility of "a particle and anti-particle spontaneously enter the universe AND have a measurable effect on said universe before vanishing". I can't think of any reason to conclude that this is impossible, and it would certainly be meaningful. The point still stands that such events would by necessity be random processes, not ones that correlate with the world - you can't gain actual information on what card someone drew this way.

0Scottbert13y

Well, my first thought was that it doesn't rule epiphenomenal consciousness out. It's strange that people would still talk about consciousness without it, but you can posit that people are just programmed to talk about consciousness for some reason (it's at least conceivable). Then I looked at the next guy's answer (asparisi) and thought he had a point: Does our theory of causal links allow for causes to have probabilistic effects? (It's different to say that 'human brains sometimes cause consciousness than to say 'human brains can cause ANYTHING, like a blue goblin appearing in front of you or the universe being destroyed and replaced with another one') I'm not sure. If it does, then epiphenomena-that-sometimes-don't-happen are okay. If not, they aren't. THEN I thought, but if consciousness is an epiphenomenon then what's strange is that people talk about it at all -- by definition, we cannot be aware of an epiphenomenon. But there could be another cause for the discussion of something we can't interact with in any way. After all, people's talk about gods is not caused by gods. There are other reasons to rule epiphenomena out, but a world with P-zombies is at least conceivable, even if it requires a lot of unlikely assumptions.

0RomanDavis13y

I'm not sure. And am not sure how you would you do an experiment to check. My rules aren't data typed into a computer program on which the universe runs, they're descriptions of the universe as experienced through my senses and processed through my mind be things like "inference" and colored by things like the "expectation of beauty", and "Occam's Razor." The reason I don't believe in the epiphenomenal theory of consciousness is because of the evidence against it, starting with my awareness, the existence of all this talk about awareness, and ending with fuzzier sort of thinking like, "Animals seem awake and aware and aware that they're aware." Oh, that and saying that consciousness doesn't cause anything you can sense seem a violation of Occam's Razor, while consciousness not effecting anything, ever, even in principle, seems to be a rejection of causality itself.

0ThrustVectoring13y

It seems like a meaningful statement. "Consciousness" is a label for the cause of people talking about consciousness. The epiphenomenal consciousness model replaces that with nothing. It's a meaningful arrangement of the causal net. It seems just rather unlikely, given that most people talk about consciousness - there's probably something behind it.

-1asparisi13y

Hm. Well, it does prevent the thought experiment, at minimum. If we take the rule as true, then it is incoherent to talk about a universe where the neurons in your brain are doing all the same stuff but don't give rise to consciousness. We can talk about a different structure that might be composed of neurons, but not the same structure. Whether you believe in epiphenomena or not, we take consciousness to be caused by neurons in our universe. So another universe with the same neuronal structure that followed this rule would make the thought experiment fail. So if I have the right sort of structure of neurons and this rule, then I have consciousness. The 'epiphenomena' stops doing any work: the universe looks the same whether or not I posit the epiphenomena. So the epiphenomena isn't explaining consciousness anymore. So while the rule doesn't forbid epiphenomenal theories of consciousness per se, it does mean that if those epiphenomena are supposed to not be a causal result of the neurons in any conceivable universe, then they can't be a causal result of the neurons in this universe.

[-]Ronny Fernandez13y50

I have a plausibly equivalent (or at least implies Ey's) candidate for the fabric of real things, i.e., the space of hypotheses which could in principle be true, i.e., the space of beliefs which have sense:

A Hypothesis has nonzero probability, iff it's computable or semi computable.

It's rather obviously inspired by Solomonoff abduction, and is a sound principle for any being attempting to approximate the universal prior.

-2Eugine_Nier13y

What if the universe permits hyper-computation?

0Ronny Fernandez13y

Hmm, it depends on whether or not you can give finite complete descriptions of those algorithms, if so, I don't see the problem with just tagging them on. If you can give finite descriptions of the algorithm, then its komologorov complexity will be finite, and the prior: 2^-k(h) will still give nonzero probabilities to hyper environments. If there are no such finite complete descriptions, then I gotta go back to the drawing board, cause the universe could totally allow hyper computations. On a side note, where should I go to read more about hyper-computation?

[-]Psy-Kosh13y50

Koan 4: How well do mathematical truths fit into this rule of defining what sort of things can be meaningful?

1dankane13y

This seems especially difficult noting that although we can claim that things are caused by certain mathematical truths, it doesn't really make sense to include them in our Bayesian net unless we could say, for example, how anything else would be different if 2+2=3.

3Peterdjones13y

What sort of things?

3dankane13y

Well I know that when I drop something the distance it falls after time t is roughly 1/2 g t^2 where g~10 m/s^2. When I drop something off of a 20m high building, I can reasonably claim that the fact that it takes roughly 2s to reach the ground is a consequence of the above, and of the mathematical truth that 1/2 10 2^2 = 20.

4Peterdjones13y

That's a physcial truth, and a local one at that. The mathematical expression of a physcial fact is not a "mathematocal truth" because, pace Tegmark, most mathematical truths don't model physical facts. What casues objects to fall is gravity. Maths does not cause it, any more than words do.

2dankane13y

I agree that mathematical truths do not have effects on their own. But when combined with mathematical formulations of laws of reality they do have observable consequences. The timing of a falling projectile above is a consequence of both a mathematical formulation of the law of gravity and a purely mathematical arithmetical fact. If you somehow want to describe the universe without mathematics, good luck.

0endoself13y

An event can have more than one cause. My uncertainty about the value of some variable in an equation is related to my uncertainty about the outcome of an experiment in exactly the way that makes Pearlean methods tell me that both the value of t in the equation and the physical truth that g ≈ 10 m/s^2 are causes of the amount of time that the object takes to fall. This is just a fact about my state of uncertainty that falls directly out of the math.

4Peterdjones13y

"Falls out of the math" doens't mean "caused by math" any more than "expressed in math" means "caused by math".

0endoself13y

Sorry, I was unclear. I meant that the causal structure where the equations of physics cause the outcome of the experiment falls out of the Pearlean causal math, not that the outcome of the experiment falls out of the physical math (though the latter is of course also true).

1Peterdjones13y

I think that still has the same problem. The (edit:) math is the map, causes are in the territtory.

-1endoself13y

I'm not sure what you mean. Pearlean causality, as I understand it, is about maps. You put in a subjective probability distribution and a few assumptions and a causal structure comes out.

1Peterdjones13y

If it's only about maps, its doubtful whether it deseres to be called causality. ETA: Indeed, EY seems to take the view that correlation is in the map, and causation in the territory:- "More generally, for me to expect your beliefs to correlate with reality, I have to either think that reality is the cause of your beliefs, expect your beliefs to alter reality, or believe that some third factor is influencing both of them. This is the more general argument that "To draw an accurate map of a city, you have to open the blinds and look out the window and draw lines on paper corresponding to what you see; sitting in your living-room with the blinds closed, making stuff up, isn't going to work." Correlation requires causal interaction; and expecting beliefs to be true means expecting the map to correlate with the territory. "

0endoself13y

I'm not sure what Eliezer believes here; the way he talks about causality is why I added the "as I understand it" to the grandparent. I think that even if it is just in the map we should still call it causality. It's a useful concept and it is similar to the concept of causality we had before we understood these insights. It's expected that when you come to understand something better it won't turn out behave exactly the way you thought it did.

0Peterdjones13y

I don't see any insight that causality is on the map not the territory. I think EY's overall point is that supernatural claims can be assesed by the same episetmology as scientific ones.

-1Slackson13y

Most mathematics has isomorphism to typographical or computational rules. I'm pretty sure these can be encoded into a causal diagram which connects with the real world.

0Shmi13y

Do you mean that some of those strings are useful in modeling the "real world"? Say, by providing a way to discover further causal diagrams?

3Slackson13y

Not necessarily. I've realized I'm more confused than I thought I was.

-3jslocum13y

Mathematics is a mental construct created to reliably manipulate abstract concepts. You can describe mathematical statements as elements of the mental models of intelligent beings. A mathematical statement can be considered "true" if, when an intelligent beings use the statement in their reasoning, their predictive power increases. Thus, " '4+4=8' is true" implies statements like "jslocum's model of arithmetic predicts that '4+4=8', which causes him to correctly predict that if he adds four carrots to his basket of four potatoes, he'll have eight vegetables in his basket" I'm no sure that "use the statement in their reasoning" and "their predictive power increases" are well formed concepts, though, so this might need some refining.

-3Bruno_Coelho13y

I find mathematics most about future physics laws who will be discovered. Math without empirical confirmation is more difficult to link, but normally is a matter of time to find a application.

-1TimS13y

Hrm? You think some future experimental results will say something meaningful about whether mathematics should accept the axiom of choice? Even if the universe is inconsistent with ZFC, why does that imply studying ZFC based mathematics should stop?

[-]pragmatist13y40

But while deep philosophical dilemmas such as these are never settled by philosophers, they are sometimes settled by people working on a related practical problem which happens to intersect the dilemma. There are a lot of people who think I'm being too harsh on philosophers when I express skepticism about mainstream philosophy; but in this case, at least, history clearly bears out the point.

It's worth mentioning that the first people to rigorously apply Bayes net methods to causal discovery (as far as I know; correct me if I'm wrong) were philosophers. ... (read more)

[-]Eliezer Yudkowsky13y40

Koan 1:

"You say that a universe is a connected fabric of causes and effects. Well, that's a very Western viewpoint - that it's all about mechanistic, deterministic stuff. I agree that anything else is outside the realm of science, but it can still be real, you know. My cousin is psychic - if you draw a card from his deck of cards, he can tell you the name of your card before he looks at it. There's no mechanism for it - it's not a causal thing that scientists could study - he just does it. Same thing when I commune on a deep level with the entire un

... (read more)

[-]CCC13y140

My reply, before reading other replies, is that the question is wrongly posed. The described phenomena can be part of a causal universe, because they are causal processes.

Consider the psychic cousin. I draw a card from his deck - it is (say) the three of clubs. Let us further assume that he correctly guesses the card that I have drawn, and does so in 99% of trials. In such circumstances, his statement of the card that I have drawn ('you have drawn the three of clubs!') is caused by my drawing the three of clubs. The mechanism of that causality may not be known, but it is there.

Similarly for the example of "communing on a deep level with the entire universe in order to realize that your partner truly loves you" - there's a causal link, there, and the cause is "your partner truly loves you". (Personally, I'd prefer to check the conclusion by some other means in order to prevent observer bias effects, but this becomes tricky for this example).

Purely spiritual phenomena, therefore, are firmly inside the realm of causal processes, even if they are not yet fully understood.

[-]MixedNuts13y160

"Your partner truly loves you" could also be a purely spiritual phenomenon outside the realm of science. This implies that it has no observable consequences; a partner who truly loves you is no less and no more likely to dump you than one who doesn't. This is an unusual idea of true love.

6CCC13y

It could be. However, the consequences of the questioner's "communing with the universe" are observable; I can observe whether you claim that your partner truly loves you or not afterwards. Since this is an observable consequence, I therefore conclude that if it is possible to commune with the universe in such a way, and if the results of such communing are correlated at all to the state "your partner truly loves you", then that state has consequences (i.e. whether or not you say that it is true after communing with the universe) and thus can be part of a causal universe.

0Ender13y

And since it has observable consequences, you can do science to it! Yay!

1Yosarian213y

Right. If there was no causal link between the card drawn and the psychic predicting the card, then the psychic wouldn't get the right answer. There has to be a causal link, or else the psychic answer would have nothing to do with the card that you picked, and then it wouldn't be any better then random.

[-]Vaniver13y130

Causal diagrams don't require understanding of the links- if the content of the card influences your cousin's claim, then we would draw an arc from the card to the claim in our causal diagram, even if we don't know how he's influenced by the card.

(Privately, I would suspect the speaker's belief is not rigorous, and would not put much weight by it, but why challenge them on a factual matter when we're discussing causal networks?)

It's not clear to me that you've sufficiently expanded your belief that your partner truly loves you, or what it means to commune with the universe, or what separates physical and non-physical processes. If we define as real things which are part of reality, and acknowledge that your belief in your partner's true love exists in reality, why are we not content to find real causes for that belief?

(Of course, there are many status reasons to turn to spiritual causes. But the instrumental benefits are probably outweighed by the epistemic detriments- or, at least, I should not privately believe their claim, even if I consider it worthy of public endorsement.)

[-]Richard_Kennaway13y110

"Magic" just means "I don't know how it works." (Pratchett, somewhere in the Tiffany Aching novels.)

If the world is exhibiting some regularity, you can always ask how that happens. "No mechanism" is itself a prediction: that none will be found. Where did that prediction come from?

9asparisi13y

Taboo the word real for a moment. Also any related words, like 'actual.' What do you mean when you say that it can be real? You say your cousin can tell you the name of a card before he looks at it, after a random draw of the deck. But he does it. It's an act on his part. This isn't a mere word-issue either: you don't think the same thing will happen if I try to predict a card you draw from a deck. So you are talking about a direct link between his statement and the card. Unless you think he is doing it by chance (so that there is just a correlation, and an extremely improbable one at that) you think that there is a causal link between the card drawn and the prediction. Saying you don't know what it is, is not the same thing as saying that it is not there. Likewise, when you say that you commune with the universe, you are stating some act, "communing" and some result "finding out that your partner loves you." You don't expect this to fail the next time you do it, or you wouldn't perform the act to that effect. (You might perform it to "see if it works this time" but that would be another matter.) So I don't think you really believe what you think you believe. Why would you "commune with the universe" if it did not cause effects such as "realizing your partner truly loves you"?

6jimrandomh13y

Mechanisms can be well hidden; if you deny that there is mechanism, you're just cheating yourself out of the chance to figure out what the mechanism was. When you "commune with the universe", you're actually putting your brain into a meditative state where it queries large volumes of remembered weak evidence that you can't recall by normal means. You can use that state for other things, too (but you'll fool yourself if you don't understand its limitations). When your cousin does card tricks, he's explicitly presenting you with a hard puzzle - figure out just how he's cheating. Answers are in books, but custom forbids giving spoilers. Figure it out, and you can fool others the same way.

5TheOtherDave13y

Probably I would start with "You say about your cousin that it's not something that scientists could study. That's an interesting claim. Suppose I tried to study it... say, suppose I asked him to demonstrate this ability a hundred times, and wrote down for each demonstration what card I drew, and what card he named. That's one way to start to do science, to observe the phenomenon and record my observations. Would something prevent me from doing that? Would the ability itself stop working? Something else?" Their answer would give me more of a sense of what I was dealing with. It doesn't engage with the reality question directly, but how I engage with the reality question will depend a lot on their answer to this question.

1Shmi13y

Suppose that the Great Psychicator who imbues all psychics with their amazing powers hates Science and revokes their abilities the moment one decides to systematically study them, so all such experiments lead to a null result. This model "explains" the null results found by the James Randi Educational Foundation. This might look far-fetched, but recall that "Nature" already behaves like it, say in the double-slit experiment. Replace "psychic abilities" with "interference pattern produced by an electron psychically detecting the other slit" and "systematic study" with "adding detector to one of the slits". The moment you start this "systematic study" the electron's "psychic abilities" to sense the presence of the other slit disappear without a trace. How would you proceed?

9TheOtherDave13y

(nods) This is basically (modulo snark) the position of several people I know who believe in such things, and at least one LW contributor. Faced with that response, I usually pursue a track along the lines of "Ah, I see. That makes sense as far as it goes, but it gets tricky, because it's also true that we frequently perceive patterns that aren't justified by the data at all, but without systematic study it's hard to tell. For example, (examples)." Followed by a longish discussion to get clear on the idea that some perceived patterns are indeed hallucinatory in this sense, even though your cousin's psychic power isn't necessarily one of those patterns. This sometimes fails... nobody I know actually claims that all perceived patterns are non-hallucinatory, but some people I know reject the path from "not all perceived patterns are non-hallucinatory" to "some perceived patterns are hallucinatory." Which I generally interpret as refusing to have the conversation at all because they don't like where it's headed. a preference I usually respect out of politeness. If it succeeds, I move on to "OK. So when I see a pattern that goes away when I begin systematic study, there are two possible theories: either the phenomenon is evasive as you describe, or my brain is perceiving patterns that aren't there in the first place. How might I go about telling the difference, so I could be sure which was which? For example, if I wake up in the middle of the night frightened that there's an intruder in my house, what could I do to figure out whether there is one or not?" Which moves pretty quickly to the realization that an intruder in my house that systematically evades detection becomes increasingly implausible the more failed tests I perform, and at some point the theory that there simply isn't such an intruder becomes more plausible. I generally consider this a good place to stop with most people. Lather, rinse, repeat. They have one track that supports "I believe X no matter h

3Shmi13y

But things keep disappearing from my house at random! Surely it's an evidence for an invisible intruder, not only for my memory going bad! And this never happens in the office, so it can't be my memory! Therefore intruder!

4TheOtherDave13y

I'm not really interested in role-playing out a whole conversation. If you insist that your invisible intruder, like your cousin's psychic ability, is real and evasive, I look for a different example. If you insist that everything you ever think about, however idly, is real and evasive, I tap out and recommend you seek professional help.

-3Shmi13y

I thought LW was that professional help...

5satt13y

Hmm, my pay slips must be getting lost in the post.

0ChristianKl13y

This assume's that the person you are talking to didn't perform any tests that provide them evidence for their belief. If you are facing someone who got his ideas from reading books, that might work. If you are facing someone who does have reference experiences for his belief, things get a bit different. You are basically telling them that they intruder that they found in their house is a hallucination. The observer could go and study his cousin systematically. The cousin does 1000 trials and no trial shows any evidence that his cousin isn't psychic. If the observer believes "the more experiments fail to provide evidence for X, the less I should believe X", the huge quantity of experiements dictate to himself that he should believe that his cousin is psychic. The idea that the cousin uses a trick is supposed to become increasingly implausible the more failed tests the observer performs. Some experiements are obviously systematically flawed. Doing more of those experiments shouldn't lead you to increase your belief. The debate is more more about which experiements are systematically flawed than it's about "I believe X no matter how many experiments fail to provide evidence for it," vs "the more experiments fail to provide evidence for X, the less I should believe X".

0TheOtherDave13y

It doesn't assume this, it infers it about a particular person from the evidence provided by shminux above. The interlocutor shminux is describing rejects the idea that experimental results can be definitive on this question, which is different from the position you describe here. (Anyone who starts out asserting the former, then switches to the latter in mid-stream, is no longer asserting a coherent position at all and requires altogether different techniques for engaging with them.) I'm not quite sure what you mean by "the debate". Is there only one? That surprises me; it certainly seems to me that some people adopt the stance shminux described, to which I responded. All that aside, I certainly agree with you that my response to someone taking the stance you describe here (embracing experimentalism as it applies to psychic phenomena in theory, but implementing experiments in a problematic way) should differ from my response to someone taking the stance shminux describes above (rejecting experimentalism as it applies to psychic phenomena).

-2Eugine_Nier13y

That depends on what you mean by "experiment". If you mean doing a proper replicable controlled experiment than there is no experimental evidence. If you mean any evidence based on observation than there is experimental evidence. In other words, there is evidence for the intruder, just not scientific evidence in the sense of this post.

0TheOtherDave13y

I don't in fact mean, by "experiment", any evidence based on observation. I agree that there is evidence for (and against) the intruder, and did not say otherwise, although in general I don't endorse using "evidence" in this sense without tagging it in some way (e.g., "Bayesian evidence"), since the alternative is reliably confusing.

6chaosmosis13y

1. How does the Great Psychicator distinguish between Science and normal events? Maybe we can trick it. 2. This model is more complicated and so it's less probable.

0Shmi13y

What would you propose? True. But so is Quantum Mechanics vs Classical Mechanics. (Un)Fortunately, CM is not enough to explain what we see (e.g. the interference pattern). To make a parallel with the issue at hand, the null-psychic model alone (without the complicated machinery of cognitive biases) does not explain the many non-scientifically tested claims of psychic powers.

3chaosmosis13y

A model where people are delusional about psychic powers or making claims about them they don't believe is less complicated than a model with a Great Psychicator because we already know that people are delusional in lots of ways similar to that. RE: tricking it. The problem is that we can't observe one of the groups involved here, or we invalidate the experiment. But it seems likely to me that all conditions that could prohibit it from working when science is involved would also prohibit the claims about it working from having any basis in reality. In other words, either it can be tricked, or those people making the claims must concede that they don't really have any basis to assert that it works. Any conditions which allow it to prohibit science should also preclude anyone from being justified in making claims about the reasons why it works. An infinitely complex Great Psychicator would be smart enough to fool our experiments though, I think.

0Shmi13y

Actually, all this entity would have to do is to foil any attempts to go meta on a certain class of phenomena, like psychic powers (and plenty of others, say, homeopathy). In a simulated universe this could be as "simple" as detecting that a certain computation is likely to discover its simulated nature and disallow this computation by altering the inputs. A sensible security safeguard, really. This can be done by terminating and rolling back any computation that is dangerously close to getting out of control and restarting it from a safe point with the inputs adjusted. Nothing infinite is required. In fact, there is very little extra complexity beyond sounding an alarm when dangerous things happen. Of course, anything can be "explained" in the framework of a simulation.

2thomblake13y

But mixing "certain computation" and "discover" like that is mixing syntax and semantics - in order to watch out for that occurrence, you'd have to be aware of all possible semantics for a certain computation, to know if it counts as a "discovery".

0Shmi13y

Not at all. You set up a trigger such as "50% of all AI researchers believe in the simulation argument", then trace back their reasons for believing so and restart from a safe point with less dangerous inputs.

2thomblake13y

If your simulation has beliefs as a primitive, then you can set up that sort of trigger - but then it's not a universe anything like ours. If your simulation is simulating things like particles or atoms, then you don't have direct access to whether they've arranged themselves into a "belief" unless you keep track of every possible way that an arrangement of atoms can be interpreted as a "belief".

0Shmi13y

Sure, if you run your computation unstructured at the level of quarks and leptons, then you cannot tell what happens in the minds of simulated humans. This would be silly, and no one does any non-trivial bit of programming this way. There are always multi-level structures, like modules. classes, interfaces... Some of these can be created on the fly as needed (admittedly, this is a tricky part, though by no means impossible). So after a time you end up with a module that represents, say, a human, with sub-modules representing beliefs and interfaces representing communication with other humans, etc. And now you are well equipped to set up an alert.

0chaosmosis13y

If the Great Psychicator uses triggers on a level of reality less precise than the atomic or subatomic ones, then I believe its triggers could not possibly be precise enough to A. prevent science from discovering psychic powers and simultaneously B. allow normal people not doing science access to its psychic powers. If there's a flaw in its model of the universe, we can exploit that and use the flaw do to science (this would probably involve some VERY complex work arounds, but the universe is self consistent so it seems possible in theory). So the relevant question is whether or not its model of the universe is better than ours, which is why I concede that a sufficiently complex Great Psychicator would be able to trick us.

-2Eugine_Nier13y

No, it just needs to be better at optimizing than we are.

-2chaosmosis13y

I don't know exactly what you mean by "optimizing", but if your main point is that it's an issue of comparative advantage then I agree. Or, if your point is that it's not sufficient for humans to have a better model of reality in the abstract, we'd also need to be able to apply that model in such a way as to trick the GP and that might not be possible depending on the nature of the GP's intervention, I can agree with that as well.

0chaosmosis13y

Yeah, thanks. This is part of what I was trying to get at. And I'm further contending that if the semantics for every possible scientific experiment were invalidated, then in doing so the hypothetical Great Psychicator would also have to invalidate the semantics for any legitimate claims that psychic powers worked. The two categories overlap perfectly. This isn't intended to be an argument from definition, I hope that is also clear.

4Benquo13y

I am interested in the idea that what your cousin does happens without a mechanism. Does this mean that the process does not involve your cousin at all? That it happens whether or not your cousin is around? If so, why do you say that it's your cousin who does it? I could ask the same kind of questions about your communing with the universe, but let's take these things one at a time.

4jesch13y

If your cousin reliably names my card, it is causally correlated to my drawing the card. I never specified the nature of the casual relationship-whether it's mirrors, marked cards, or the cards whispering in his brain, you told me that my action and his statement are connected. If 'purely spiritual phenomena' have effects, they're causal processes, connected at the very least to your state of mind- if they don't have effects I'm not sure what we're arguing about.

3Scottbert13y

Okay, hypothetical mystic dude. You said: "IF you draw a card... [then] he can tell you the name of your card". Sounds causal to me! Otherwise he could tell me the name of my card whether I draw it or not! Also, you commune with the universe TO realize that your partner loves you. If you don't believe the results of your divination are caused by your partner's love why are you doing it? In short, you may believe you believe that these are 'non-causal processes', but on the level that determines your behavior, you believe they are causal processes. I suspect this is because either labeling these things non-scientific is important to you for some reason (love of mystery, or perhaps it's what your peer group says to believe) or you don't understand what the words 'non-causal process' mean and it's just a password.

2moshez13y

Replying without reading any of the other answers. Apologies in advance for redundancy: Meditation 1: The psychic cousin is indeed connected to the network of things. Let's assume that it works, for simplicity, on decks of two cards: an Ace and a King. Probabilities: Moshe picked Ace/Cousin says Moshe picked Ace -- 0.4 Moshe picked King/Cousin says Moshe picked Ace -- 0.1 Moshe picked Ace/Cousin says Moshe picked King -- 0.1 Moshe picked King/Cousin says Moshe picked Ace -- 0.4 The True Love/Communing is more complicated: Does True Love have any discernible effect? If we assume True Love, say, changes the probability of having a fight (or some property of the fight -- for example, a fight without reconciliation inside of 24 hours), then we should have a diagram: True Love --> Communing says True Love | | \/ No fight and resulting joint probability distributions. Since "fighting" is something observable (by a trained psychologist, say, who puts them in the "Love Lab" http://www.gottman.com/49847/The-Love-Lab.html) we have connectedness.

2earthwormchuck16313y

I believe that your cousin can, under the right circumstances, reliably guess which card you picked. There are all sorts of card tricks that let one do exactly that if the setup is right. But I confidently predict that his guess is not causally separated from the choice of card. To turn this into a concrete empirical prediction: Suppose that I have your cousin repeat his trick several times (say 100) for me, and each time I eliminate any causal interaction between the the deck and your cousin that I can think of. Then I predict with probability ~1 that, within several rounds, your cousin will not be able to guess which card I draw. If this prediction turns out to be wrong, and your cousin can still reliably guess which card I drew even when we are in lead boxes on opposite sides of the planet, then I will gladly discard my model of reality, since it will have predicted a true state of affairs to be impossible. Even this wouldn't be enough to prove your cousin's power is acausal, since there are conceivable causal influences that I couldn't eliminate easily.For example, maybe we live in a computer simulation that runs standard physics all the time, except for a high-level override which configures your cousin's visual cortex to a see a picture of every card that anyone draws from a deck. But I would certainly require evidence at least that strong before even considering the possibility that your cousin can guess cards acausally. A scientist is just a person who learns about things by looking at them (ie causally interacting with them), plus some social conventions to make up for human fallibility. Do you claim that the social conventions of science stop one from learning about this thing by looking at it, or that one cannot learn about this thing by looking at it at all. If the former, then again I confidently predict that you are wrong. If the latter, then I can only ask how you learned about the thing, if not by looking at it. You mean to say that your confide

2ChristianKl13y

Both cases are quite different. In the first case we have an observation. The cousin can know the name of a card without looking at the card. The observer has no explanation about how the cousin knows. The observer therefore reasons says that cousin has a property P that causes him to know the card. The moment you call P "being psychic" you add a lot of connotations that have nothing to do with what the observer knows about the process. That added baggage is meaningless. Case two is about metaphors. The observer engages into an acitivity A to get goal B. Goal B is about having a change in his own belief system that the oberver now believes C. C being that the partner loves her. So the person assumes that there's a clear causal relationship between A and B. Some form of meditation could feel like the observer engages in A and allow him to reach B. There are two struggles: Firstly, the semantics of naming A. Naming A can plausibly help with doing A successfully. Members of this community will probably more comfortable with naming A differently. Secondly there the issue of whether a belief change in my own mind can be "real". That category of claims didn't get yet discussed in this sequence, but I think it makes sense to speak of belief changes as real. ---------------------------------------- There another issue about this koan. If you ask a buddhist whether they belief that the universe is a connected fabric of causes and effects, they'll answer: "Of course I believe in karma". You don't need to believe in a reductionist worldview to say that the world is made up of cause and effect.

1wizzwizz45y

Written in a somewhat fake wisdom manner, without reading other replies: If you chose a different card, would he not say a different card's name? Therefore, the card he says is causally linked to your choice of card. And would you not hear different words if he said something different? Therefore, your observations are causally linked to the card he says. Just because we do not know how something works – nay, even if it is Unknowable™ – that doesn't make it causally unshackled from reality.

1Dolores198413y

1: If your cousin can demonstrate that ability using somebody else's deck, under experimental conditions that I specify and he is not aware of ahead of time, I will give him a thousand dollars. 2: In the counter-factual case where he accomplishes this, that does not mean that his ability is outside the realm of science (well, probably it means the experiment was flawed, but we'll assume otherwise). There have been a wide range of inexplicable phenomena which are now understood by science. If your cousin's psychic powers are real, then science can study them, and break down the black box to find out what's inside. There are certainly causal arrows there, in any case. If there weren't, we wouldn't know about it. 3: If your strongest evidence that your partner loves you is psychic intuition, you should definitely get a prenup.

1hairyfigment13y

CCC and others have given most of my answer. As a practical matter, I would ask: 'Do you need to commune with the entire universe in order to realize your partner loves you? If you have no good evidence taken from your relationship, from the interactions between you, that seems like a giant warning sign.' Note too that the part about the psychic cousin (if we falsely assume that it has a clear meaning which the character wants to state openly) seems meant to describe a graph with a "spiritual" node that connects to the cousin's card-naming behavior and to nothing else that we can observe. Which does not violate this particular rule, but does seem needlessly complicated.

1GDC313y

Mechanism is beside the point. Mechanism is just causal nodes in between. Having no mechanism just means there is a direct connection.

1beoShaffer13y

There may or may not be intervening steps, but if your cousins predictions are accurate we can create causal models that either have predictions->card drawn or card drawn->predictions.

1AlexMennen13y

Or predictions <-- lurking variable --> card drawn.

0AnthonyC13y

Or Great Psychiator->Card drawn ^ Great Psychiator->predictions Which might look the same to use here within the universe but is far less likely as a prior.

0Paul_G13y

As someone with some experience dealing with this, having learned how difficult it is to fix I would reply something like "You are wrong. If you want to learn WHY you're wrong, tell me and we can work on this together. Otherwise, I'm going to go now." Playing the game a bit: "Okay, bear with me a moment, this is going to sound a little odd. I'm not sure what you mean by "outside the realm of causal processes". Does that mean it happens on its own, with no outside influence at all? Nothing causes it, it just... Happens? Even if it's a 'magic' skill, shouldn't he be the one to activate it? I mean, worst case scenario, it's caused by someone drawing a card from a deck. It doesn't happen completely independently of reality, it's CAUSED by something. If your cousin is the only one with this power, I'm sure he could be studied by the scientists and they could figure out what lights up in his brain as he does it. A minor note, I was once a card magician, and there are very specific ways to either force people to choose the card you want, or to figure out what card they've chosen. I can show you a few, if you want. Next, 'communing with the entire universe' is a pretty arrogant thing to say, isn't it? I never got any communication, anyway. Question for you - how would it feel to look deeply inwards, ask 'the universe' questions, and receive answers from your own mind? Would it feel much different from what you feel now? Usually it's better to assume that confusing or 'unexplained' things are happening in your mind, not in reality. You FEEL like the universe has told you that he loves you, but that would look exactly the same as if it was just your unconscious mind telling you. How often have people said that they were deeply, permanently in love, but then it didn't work out? Do you really think you're that much better than everyone else?"

0somervta13y

In reality, I would deny the statement about what can be don by his cousin. But for the purpose of this exercise, I would say that if there was truly no causal mechanism, you would have no basis for saying that it was true, because you would have no evidence for it. In fact, you wouldn't even be aware of it. If your cousin can name my card, he does so through a causal mechanism of some kind.

0CronoDAS13y

In real life? "You're a loony." ::walks away:: I suspect that's not the answer you're looking for, though. ;)

0philh13y

(Written without reading other comments.) Let us stipulate that your cousin is indeed psychic. This is not a purely spiritual phenomenon. Then there is certainly correlation between me drawing QH and between him saying "queen of hearts". Probably no direct causal link - I'd guess it has intermediate nodes related to patterns of neurons in his brain - but the graph certainly has a path between the two nodes. If something does not correlate with anything, then it is not real in the sense that it does not determine your experiences.

0RomanDavis13y

Well first of all, we're not perfect philosophers of perfect emptiness. We get our beliefs from somewhere. So it's true that all sorts of things are true that we have no evidence of. For instance, it's very, very likely there's life outside our solar system, but I don't have any evidence of it, so I act as if it's not true because in my model of the universe, it's very unlikely that that life will affect me during my natural lifetime. I would even go far as to say that there may be matter beyond the horizon of the matter that expanded after the big bang, or that we're all running on an alien matrix, or that God is real but he's just hiding, and I act as if it's false. Not because they're untrue, or unlikely to be true, as I have no way to tell. But because I am very, very unlikely to ever, ever get evidence about any of those things, and they probably will never, and probably could never (especially in the near future) affect me. Not so much a "Nuh uh," as a "So what?" You know your partner loves you based on evidence. If you have no evidence (from past experience or otherwise), then you are very likely wrong. Love operates according to mechanisms, and we understand some of those mechanisms. Similarly, just because you don't understand the mechanism by which your psychic cousin works, doesn't mean there isn't one. He could be getting unbelievably lucky, or he could be playing a trick, or there could be things we don't know yet that really truly give him psychic powers. You don't know what the mechanism is, but you haven't really investigated either, have you? Even if you never find out what the mechanism is, how much evidence is that that there is no mechanism? Lastly, I'm not sure, "no mechanism" even makes sense. What does it mean for something to have no mechanism? What does a thing that doesn't have a mechanism look like? How would you tell? So, from the top: A Priori, Making Beliefs Pay Rent, No One Knows What Science Doesn't Know, What is Evidence?, Fragi

2Blueberry13y

If there's no evidence of it (circumstantial evidence included), what makes you think it's very likely?

1MugaSofer13y

I assume they meant direct evidence. It's truth or falseness do not affect us, we must extrapolate from our knowledge of the structure of the universe.

[-]sfwc13y30

On the other hand, while "post-utopian" is linked to "colonial alienation" and vice versa, these two elements don't connect to the rest of the causal fabric - so that must not be a universe.

If I really want to, there's an easy way for me to sidestep this. I just postulate something called "post-consciousness" which is caused both by colonial alienation and by particular arrangements of neurons in the brains of particular people (in a similar way to that in which epiphenomenalists would say consciousness is caused). Presto! ... (read more)

0adam_strandberg13y

I think it makes more sense to say that this test rules out ideas that can't actually be tested as hypotheses. An idea can only be tested by observation once it is formulated as a causal network. Once it's formulated as a testable hypothesis, you can simply discard this epiphenomenal example by Solomonoff induction.

[-]Manfred13y20

I like this post a lot, even though it's short (or, er, bipartite). It covers the stuff it needs to.

3RHollerith13y

In fact, I'd prefer for it to be 2 posts: one before "Causal Diagrams and Causal Models" in the sequence, one after.

4Bugmaster13y

Agreed. The paragraph instructing me to drop what I'm doing and go read some other post, felt quite jarring.

[-]PhilGoetz13y10

and yet the people who first delineated and formalized a critical hint about the nature of reality, the people who first discovered what sort of things seem to be real,were trying to solve a completely different-sounding question.

You should give some credit to the logical positivists, because you seem to be trying to re-invent logical positivism. It would be nice to explain how what you're getting into differs from logical positivism.

It would also be nice to do a critical analysis of the refutation of logical positivism. Students today are taught that... (read more)

[-]abramdemski13y00

I feel that it would be meaningful to claim that there are actually other universes which don't causally interact with ours, although (like the claim about photons disappearing when they are sufficiently far away from us) it would not be testable. This seems to be a counterexample to the causal connection requirement. [A good scientific theory would posit a common cause for these universes, but that is beside the point: it seems conceivable that they have no common cause.]

As such, I'm tempted to claim that what is real is just what actually is. There isn't an internal criteria for determining that, such as causal connectedness.

(This does not force any particular position about what is meaningful.)

[-]folkTheory13y00

Is there some reason this article hasn't been promoted to the frontpage?

[-]DuncanS13y00

It's meaningful and false, rather than meaningless, to say that on March 22nd, 2003, the particles in the center of the Sun spontaneously arranged themselves into a short-lived chocolate cake. This statement's truth or falsity has no consequences we'll ever be able to test experientally. Nonetheless, it legitimately describes a way reality could be, but isn't; the atoms in our universe could've been arranged like that on March 22nd 2003, but they weren't.

I actually think this a confusing statement. From a thermodynamic perspective, it's not impossible ... (read more)

2chaosmosis13y

More generally, Eliezer takes an explicit and very detail oriented approach to some aspects of some topics, but takes a very informal and relaxed approach to other topics. It seems to me that he pays attention to detail more when he agrees with the arguments he is outlining. This has the effect of making the arguments he opposes seem artificially weaker. This isn't directly relevant to the chocolate cake issue, but your comment reminded me of this.

1DuncanS13y

I agree - I think this is because Eliezer's intent is to explain what he believes to be right, rather than to worry too much about the arguments of those he doesn't agree with. An approach I entirely agree with - my experience is that debate is remarkably ineffective as a means of reaching new knowledge, whilst teaching the particular viewpoint you hold is normally much more enlightening to the listener, whether they agree with the viewpoint or not.

[-]drnickbone13y-10

It's true, but not testable, to say that a spaceship going over the cosmological horizon of an expanding universe does not suddenly blink out of existence

This example has been used by Eliezer before in the Sequences, and it is a bit problematic under the latest physical theories. Assuming cosmological horizons behave like black hole event horizons, the theories say the following:

The spaceship doesn't "blink out of existence"; instead it becomes more and more red-shifted, and never crosses the horizon from our own point of view.
Further, in

... (read more)

3Eliezer Yudkowsky13y

Why on Earth would they? Edit: Also that paper does not say anything about cosmological horizons converting anything into Hawking radiation! It's an entirely different and stranger argument.

6pragmatist13y

In asymptotically flat spacetimes, the event horizon of a black hole is defined (roughly) as the boundary of the region of spacetime that can be seen by an immortal observer (more precisely, it's the boundary of the causal past of future null infinity). If you extend this definition to generic spacetimes, then it applies to the cosmological event horizon in a spacetime with positive cosmological constant. Because of this formal similarity, a number of results in black hole thermodynamics (specifically, the laws of black hole mechanics) can be generalized to the cosmological horizon. Anyway, one response (Susskind's) to the sort of thing drnickbone brings up, in the case of black holes, is black hole complementarity. We seem to have this paradox when something falls into a black hole. On the one hand, we don't want information to disappear beyond the event horizon, so the information must be absorbed into the event horizon itself, theoretically readable off the structure of the horizon (or off Hawking radiation). On the other hand, from the perspective of an infalling observer, nothing special happens as she passes through the horizon. It certainly doesn't seem to her as if the information she carries has been smeared over the event horizon. Susskind's response is essentially that both of these things happen. The information is both reflected and transmitted by the event horizon. But the no-cloning theorem rules out the possibility that the reflected and transmitted information are two separate things. Instead, they are the same event described from different perspectives. From an outside observer's perspective, the information is painted onto the horizon. From an infalling observer's perspective, the information passes right through the horizon. Extending this idea to the cosmological horizon and a spaceship leaving my horizon: From the perspective of the spaceship nothing special has happened. After all, why should they care about my horizon. From my perspective,

3drnickbone13y

An update. I didn't realise this a couple of months ago, but it seems there has been a big controversy brewing recently about black hole complementarity, and whether it is consistent. There was a key paper by Polchinski and three others in August Complementarity or Firewalls?; see also his guest post in Discover. The basic argument is that there is a new black hole paradox: quantum states on the edge of a horizon have to be fully entangled BOTH with Hawking radiation that has already emerged from the black hole AND with neighbouring states that are just inside the black hole. And that is not possible, because there is a "monogamy" of quantum entanglement. Further, complementarity doesn't help, because an observer could in principle collect the entangled radiation that had already emerged from the black hole, distill it, and then bring it into the black hole to meet its duplicate entangled state inside, which would lead to quantum cloning. Oops. Instead Polchinksi et al propose that there is a "firewall" at the black hole event horizon which would destroy the second entanglement, and also destroy any observers going into the black hole. There seems to have been a big fight on the high energy physics archive, with lots of authors drafting papers in an attempt to refute Polchinski et al, then withdrawing them or heavily editing them. Bousso also had a go, claiming that they had misunderstood complementarity, then retracted; his latest version argues that they've found a genuine paradox after all. To be fair, not many of these physicists/cosmologists agree with the firewall solution, probably because it can leads to observers suddenly disappearing into flame without warning (it is possible to reach an event horizon around a very large black hole in otherwise normal space, with no outward-sign that is coming, then smash into the firewall and die). That violates the same sorts of physical intuitions that Eliezer raises in the main article (and which I challenged). It's

0drnickbone13y

P.S. I found the following interesting paragraph in Bousso et al which considers Susskind's complementarity proposal. It seems to make a big difference whether you define an observer first (and then consider his causal horizon), or define a causal patch first (together with its horizon) and then consider where the observer might be: If we consider it in "many worlds" terms, then the wave function over the causal patch contains many different branches, and human observers (indeed our whole Milky Way galaxy) get to exist in lots of those branches. But in only a very few of the branches is our galaxy near the centre of the patch and able to continue to survive for 100s of billions of years; in most of the other branches we are somewhere off-centre, and will be smeared out against the horizon very much sooner. The fact that the Milky Way will continue to exist in some branches does not mean we should expect to survive in our branch. This reminds me somewhat of the discussions on quantum suicide.

-2drnickbone13y

Thanks for this, though in a way, Susskind's interpretation seems to be even weirder than that of Bousso et al. In Susskind's view, we would have to say that every galaxy apart from ours has an end-of-time experience, and gets smeared out on the horizon (or thermalized by de Sitter radiation), still in an average of about 5 billion years. But our own doesn't... so in 100 billion years or so we will be the lucky sole survivors in a universe containing a single remaining galaxy. Yet we are not really "lucky" because every other galaxy is experiencing the same thing from its own galacto-centric viewpoint. And while these individual amazing survivor stories are all consistent, there is no globally consistent story where all the galaxies continue to survive, just moving further and further apart. Strange...

3drnickbone13y

Well horizons might not behave exactly the same (this is all theoretical physics) but there is quite a long chain of papers arguing that Hawking radiation arises from all sorts of causal horizons, and for the same sorts of reasons that motivate it for black hole horizons. See Gibbons-Hawking effect or look up "de Sitter radiation" on Google scholar. Here's just one paper. With regard to your edit, the paper by Bousso et al does in fact discuss physical interpretations of the "end of time" effect, and scrambling into radiation appears to be the authors' preferred interpretation. See Section 5.3 on "causal patch measure" and this paragraph:

-1MugaSofer13y

Because they're both horizons, of course!

-3wedrifid13y

That question sprung to mind too, right before "How could that even make sense?" and a few hundred milliseconds later "What are the precise engineering details required for me to use this effect (and the implied possibility for FTL communication) to go back and sink the first fleet, thereby preventing my own existence?". (My reply is to the version conveyed within the grandparent, not to the strange argument. It is probably best for me to hold off reading said strange argument until is confirmed as reasonable by a sufficiently trusted source. Given the complexity of the subject I may forget to not-believe something I read.)

0drnickbone13y

Well there's nothing in the Gibbons-Hawking effect (all causal horizons have a temperature and emit thermal radiation), or in the referenced paper of Bousso et al to imply FTL travel. Where did you get that idea? The "end-of-time" papers may be wrong but they're not that wrong...

[-]Ronny Fernandez13y-10

I have to ask, how does this metaphysics (cause that's what it is) account for mathematical truths? What causal models do those represent?

My bad:

Someone already asked this more cleverly than I did.

[-]Decius13y-20

3) If causes and effects do not meaningfully constrain experience, the universe could look like literally anything from moment to moment, regardless of what it was a moment ago. If that were the case, it is certain that the universe would enter every possible state given enough 'time'- even the state where you are reading a comment on the internet with a memory of sensory perception of a consistent world.

1chaosmosis13y

Could something else constrain experience, even if causes and effects didn't?

0Decius13y

There is nothing which could have the effect of constraining the universe without causation. It is possible that experience is constrained without a cause for that constraint. What could possibly have the effect of constraining experience in the general absence of causality?

1chaosmosis13y

This is what I was trying to get at.

1Decius13y

However, without causation, the constraints on experience (or the fact that experience was constrained) would have no effect.

[-]mytyde13y-30

Philosophy is analogous to masturbation, as science is to sexual love. How can you manage the latter if you've never learned how to use your most important organ by practicing the former?

Yudkowsky is indeed too "hard" on philosophy, even as he practices it himself.

[This comment is no longer endorsed by its author]Reply

[+]chaosmosis13y-60

LESSWRONG
LW

LESSWRONG
LW

43

The Fabric of Real Things

43

43

vs.