How to Build an Artificial Human

June 19, 2021 entirelyuseless Science Artificial Intelligence, Body, Causes, Deepmind, Emotion, Ems, Ex Machina, Future, Google, GPT-3, Learning, Life, Lifelogging, Nick Bostrom, Orthogonality Thesis, Prediction, Predictive Models, Predictive Processing, Reinforcement Learning, Robin Hanson, Self, Self-Reference, Senses, Technology, Thought, Universality, Universals, Values

I was going to use “Artificial Intelligence” in the title here but realized after thinking about it that the idea is really more specific than that.

I came up with the idea here while thinking more about the problem I raised in an earlier post about a serious obstacle to creating an AI. As I said there:

Current AI systems are not universal, and clearly have no ability whatsoever to become universal, without first undergoing deep changes in those systems, changes that would have to be initiated by human beings. What is missing?
The problem is the training data. The process of evolution produced the general ability to learn by using the world itself as the training data. In contrast, our AI systems take a very small subset of the world (like a large set of Go games or a large set of internet text), and train a learning system on that subset. Why take a subset? Because the world is too large to fit into a computer, especially if that computer is a small part of the world.
This suggests that going from the current situation to “artificial but real” intelligence is not merely a question of making things better and better little by little. There is a more fundamental problem that would have to be overcome, and it won’t be overcome simply by larger training sets, by faster computing, and things of this kind. This does not mean that the problem is impossible, but it may turn out to be much more difficult than people expected. For example, if there is no direct solution, people might try to create Robin Hanson’s “ems”, where one would more or less copy the learning achieved by natural selection. Or even if that is not done directly, a better understanding of what it means to “know how to learn,” might lead to a solution, although probably one that would not depend on training a model on massive amounts of data.

Proposed Predictive Model

Perhaps I was mistaken in saying that “larger training sets” would not be enough, at any rate enough to get past this basic obstacle. Perhaps it is enough to choose the subset correctly… namely by choosing the subset of the world that we know to contain general intelligence. Instead of training our predictive model on millions of Go games or millions of words, we will train it on millions of human lives.

This project will be extremely expensive. We might need to hire 10 million people to rigorously lifelog for the next 10 years. This has to be done with as much detail as possible; in particular we would want them recording constant audio and visual streams, as well as much else as possible. If we pay our crew an annual salary of $75,000 for this, this will come to $7.5 trillion; there will be some small additions for equipment and maintenance, but all of this will be very small compared to the salary costs.

Presumably in order to actually build such a large model, various scaling issues would come up and need to be solved. And in principle nothing prevents these from being very hard to solve, or even impossible in practice. But since we do not know that this would happen, let us skip over this and pretend that we have succeeded in building the model. Once this is done, our model should be able to fairly easily take a point in a person’s life and give a fairly sensible continuation over at least a short period of time, just as GPT-3 can give fairly sensible continuations to portions of text.

It may be that this is enough to get past the obstacle described above, and once this is done, it might be enough to build a general intelligence using other known principles, perhaps with some research and refinement that could be done during the years in which our crew would be building their records.

Required Elements

Live learning. In the post discussing the obstacle, I noted that there are two kinds of learning, the type that comes from evolution, and the type that happens during life. Our model represents the type that comes from evolution; unlike GPT-3, which cannot learn anything new, we need our AI to remember what has actually happened during its life and to be able to use this to acquire knowledge about its particular situation. This is not difficult in theory but you would need to think carefully about how this should interact with the general model; you do not want to simply add its particular experiences as another individual example (not that such an addition to an already trained model is simple anyway.)

Causal model. Our AI needs not just a general predictive model of the world, but specifically a causal one; not just the general idea that “when you see A, you will soon see B,” but the idea that “when there is an A — which may or may not be seen — it will make a B, which you may or may not see.” This is needed for many reasons, but in particular, without such a causal model, long term predictions or planning will be impossible. If you take a model like GPT-3 and force it to continue producing text indefinitely, it will either repeat itself or eventually go completely off topic. The same thing would happen to our human life model — if we simply used the model without any causal structure, and forced it to guess what would happen indefinitely far into the future, it would eventually produce senseless predictions.

In the paper Making Sense of Raw Input, published by Google Deepmind, there is a discussion of an implementation of this sort of model, although trained on an extremely easy environment (compared to our task, which would be train it on human lives).

The Apperception Engine attempts to discern the nomological structure that underlies the raw sensory input. In our experiments, we found the induced theory to be very accurate as a predictive model, no matter how many time steps into the future we predict. For example, in Seek Whence (Section 5.1), the theory induced in Fig. 5a allows us to predict all future time steps of the series, and the accuracy of the predictions does not decay with time.
In Sokoban (Section 5.2), the learned dynamics are not just 100% correct on all test trajectories, but they are provably 100% correct. These laws apply to all Sokoban worlds, no matter how large, and no matter how many objects. Our system is, to the best of our knowledge, the first that is able to go from raw video of non-trivial games to an explicit first-order nomological model that is provably correct.
In the noisy sequences experiments (Section 5.3), the induced theory is an accurate predictive model. In Fig. 19, for example, the induced theory allows us to predict all future time steps of the series, and does not degenerate as we go further into the future.
(6.1.2 Accuracy)

Note that this does not have the problem of quick divergence from reality as you predict into the distant future. It will also improve our AI’s live learning:

A system that can learn an accurate dynamics model from a handful of examples is extremely useful for model-based reinforcement learning. Standard model-free algorithms require millions of episodes before they can reach human performance on a range of tasks [31]. Algorithms that learn an implicit model are able to solve the same tasks in thousands of episodes [82]. But a system that learns an accurate dynamics model from a handful of examples should be able to apply that model to plan, anticipating problems in imagination rather than experiencing them in reality [83], thus opening the door to extremely sample efficient model-based reinforcement learning. We anticipate a system that can learn the dynamics of an ATARI game from a handful of trajectories,¹⁹ and then apply that model to plan, thus playing at reasonable human level on its very first attempt.
(6.1.3. Data efficiency)

“We anticipate”, as in Google has not yet built such a thing, but that they expect to be able to build it.

Scaling a causal model to work on our human life dataset will probably require some of the most difficult new research of this entire proposal.

Body. In order to engage in live learning, our AI needs to exist in the world in some way. And for the predictive model to do it any good, the world that it exists in needs to be a roughly human world. So there are two possibilities: either we simulate a human world in which it will possess a simulated human body, or we give it a robotic human-like body that will exist physically in the human world.

In relationship to our proposal, these are not very different, but the former is probably more difficult, since we would have to simulate pretty much the entire world, and the more distant our simulation is from the actual world, the less helpful its predictive model would turn out to be.

Sensation. Our AI will need to receive input from the world through something like “senses.” These will need to correspond reasonably well with the data as provided in the model; e.g. since we expect to have audio and visual recording, our AI will need sight and hearing.

Predictive Processing. Our AI will need to function this way in order to acquire self-knowledge and free will, without which we would not consider it to possess general intelligence, however good it might be at particular tasks. In particular, at every point in time it will have predictions, based on the general human-life predictive model and on its causal model of the world, about what will happen in the near future. These predictions need to function in such a way that when it makes a relevant prediction, e.g. when it predicts that it will raise its arm, it will actually raise its arm.

(We might not want this to happen 100% of the time — if such a prediction is very far from the predictive model, we might want the predictive model to take precedence over this power over itself, much as happens with human beings.)

Thought and Internal Sensation. Our AI needs to be able to notice that when it predicts it will raise its arm, it succeeds, and it needs to learn that in these cases its prediction is the cause of raising the arm. Only in this way will its live learning produce a causal model of the world which actually has self knowledge: “When I decide to raise my arm, it happens.” This will also tell it the distinction between itself and the rest of the world; if it predicts the sun will change direction, this does not happen. In order for all this to happen, the AI needs to be able to see its own predictions, not just what happens; the predictions themselves have to become a kind of input, similar to sight and hearing.

What was this again?

If we don’t run into any new fundamental obstacle along the way (I mentioned a few points where this might happen), the above procedure might be able to actually build an artificial general intelligence at a rough cost of $10 trillion (rounded up to account for hardware, research, and so on) and a time period of 10-20 years. But I would call your attention to a couple of things:

First, this is basically an artificial human, even to the extent that the easiest implementation likely requires giving it a robotic human body. It is not more general than that, and there is little reason to believe that our AI would be much more intelligent than a normal human, or that we could easily make it more intelligent. It would be fairly easy to give it quick mental access to other things, like mathematical calculation or internet searches, but this would not be much faster than a human being with a calculator or internet access. Like with GPT-N, one factor that would tend to limit its intelligence is that its predictive model is based on the level of intelligence found in human beings; there is no reason it would predict it would behave more intelligently, and so no reason why it would.

Second, it is extremely unlikely than anyone will implement this research program anytime soon. Why? Because you don’t get anything out of it except an artificial human. We have easier and less expensive ways to make humans, and $10 trillion is around the most any country has ever spent on anything, and never deliberately on one single project. Nonetheless, if no better way to make an AI is found, one can expect that eventually something like this will be implemented; perhaps by China in the 22nd century.

Third, note that “values” did not come up in this discussion. I mentioned this in one of the earlier posts on predictive processing:

The idea of the “desert landscape” seems to be that this account appears to do away with the idea of the good, and the idea of desire. The brain predicts what it is going to do, and those predictions cause it to do those things. This all seems purely intellectual: it seems that there is no purpose or goal or good involved.
The correct response to this, I think, is connected to what I have said elsewhere about desire and good. I noted there that we recognize our desires as desires for particular things by noticing that when we have certain feelings, we tend to do certain things. If we did not do those things, we would never conclude that those feelings are desires for doing those things. Note that someone could raise a similar objection here: if this is true, then are not desire and good mere words? We feel certain feelings, and do certain things, and that is all there is to be said. Where is good or purpose here?
The truth here is that good and being are convertible. The objection (to my definition and to Clark’s account) is not a reasonable objection at all: it would be a reasonable objection only if we expected good to be something different from being, in which case it would of course be nothing at all.

There was no need for an explicit discussion of values because they are an indirect consequence. What would our AI care about? It would care roughly speaking about the same things we care about, because it would predict (and act on the prediction) that it would live a life similar to a human life. There is definitely no specific reason to think it would be interested in taking over the world, although this cannot be excluded absolutely, since this is an interest that humans sometimes have. Note also that Nick Bostrom was wrong: I have just made a proposal that might actually succeed in making a human-like AI, but there is no similar proposal that would make an intelligent paperclip maximizer.

This is not to say that we should not expect any bad behavior at all from such a being; the behavior of the AI in the film Ex Machina is a plausible fictional representation of what could go wrong. Since what it is “trying” to do is to get predictive accuracy, and its predictions are based on actual human lives, it will “feel bad” about the lack of accuracy that results from the fact that it is not actually human, and it may act on those feelings.

Kavka’s Toxin

January 5, 2019January 6, 2019 entirelyuseless Ethics, Uncategorized Belief, Decision Theory, Decisions, Future, Gregory Kavka, Kavka's Toxin, Prediction

Gregory Kavka discusses a thought experiment:

You are feeling extremely lucky. You have just been approached by an eccentric billionaire who has offered you the following deal. He places before you a vial of toxin that, if you drink it, will make you painfully ill for a day, but will not threaten your life or have any lasting effects. (Your spouse, a crack biochemist, confirms the properties of the toxin.) The billionaire will pay you one million dollars tomorrow morning if, at midnight tonight, you intend to drink the toxin tomorrow afternoon. He emphasizes that you need not drink the toxin to receive the money; in fact, the money will already be in your bank account hours before the time for drinking it arrives, if you succeed. (This is confirmed by your daughter, a lawyer, after she examines the legal and financial documents that the billionaire has signed.) All you have to do is sign the agreement and then intend at midnight tonight to drink the stuff tomorrow afternoon. You are perfectly free to change your mind after receiving the money and not drink the toxin. (The presence or absence of the intention is to be determined by the latest ‘mind-reading’ brain scanner and computing device designed by the great Doctor X. As a cognitive scientist, materialist, and former student of Doctor X, you have no doubt that the machine will correctly detect the presence or absence of the relevant intention.)

Confronted with this offer, you gleefully sign the contract, thinking ‘what an easy way to become a millionaire’. Not long afterwards, however, you begin to worry. You had been thinking that you could avoid drinking the toxin and just pocket the million. But you realize that if you are thinking in those terms when midnight rolls around, you will not be intending to drink the toxin tomorrow. So maybe you will actually have to drink the stuff to collect the money. It will not be pleasant, but it is sure worth a day of suffering to become a millionaire.

However, as occurs to you immediately, it cannot really be necessary to drink the toxin to pocket the money. That money will either be or not be in your bank account by 10 a.m. tomorrow, you will know then whether it is there or not, and your drinking or not drinking the toxin hours later cannot affect the completed financial transaction. So instead of planning to drink the toxin, you decide to intend today to drink it and then change your mind after midnight. But if that is your plan, then it is obvious that you do not intend to drink the toxin. (At most you intend to intend to drink it.) For having such an intention is incompatible with planning to change your mind tomorrow morning.

The discussion goes on from here for some time, but the resolution of the puzzle is easier than Kavka realizes. There is only a problem because it is implicitly assumed that the belief that you will or will not drink the toxin is something different from the intention to drink it. But in the case of voluntary actions, these are one and the same. The reason you cannot intend to drink the toxin without thinking that you will end up drinking it is simply that the intention to drink the toxin is the belief that you will end up drinking it. If the brain scanner works correctly, it registers that you intend to drink the toxin if you in fact think you will end up drinking it, and it registers that you do not intend this if you in fact think you will not drink it.

Is there a problem on the practical level? That is, is it possible for someone to get the million, or is it impossible because everyone in such a situation would expect that they would reconsider tomorrow morning, and therefore they will not believe that they will end up drinking it?

Possibly, and for some people. It is entirely possible in some situations that beliefs about what you will in fact do, apparently simply based on facts, entirely prevent certain decisions and intentions. Thus if someone has tried dozens of times in the past to give up smoking, and consistently failed, it will become more and more difficult to intend to give up smoking, and may very well become impossible.

However, Kavka gives a theoretical argument that this should be impossible in the case of his thought experiment:

Thus, we can explain your difficulty in earning a fortune: you cannot intend to act as you have no reason to act, at least when you have substantial reason not to act. And you have (or will have when the time comes) no reason to drink the toxin, and a very good reason not to, for it will make you quite sick for a day.

Again, it may well be that this reasoning would cause an individual to fail to obtain the million. But it is not necessary for this to happen. For the person does have a reason to intend to drink the toxin in the first place: namely, in order to obtain the million. And tomorrow morning their decision, i.e. their belief that they will drink the toxin, will be an efficient cause of them actually drinking the toxin, unless they reconsider. Thus if a person expects to reconsider, they may well fail to obtain the million. But someone wanting to obtain the million will also therefore plan not to reconsider. And tomorrow morning their belief that they will not reconsider will be an efficient cause of them not reconsidering, unless they reconsider their plan not to reconsider. And so on.

Thus, someone can only obtain the million if they plan to drink the toxin, they plan not to reconsider this plan, and so on. And someone with this plan can obtain the million. And maybe they will end up drinking the toxin and maybe they won’t; but the evening before, they believe that they factually will drink it. If they don’t, they fail to obtain the million. And they may well in fact drink it, simply by carrying out the original plan: going about their day without thinking about it, and simply drinking that afternoon, without any additional consideration of reasons to drink or not drink.

There is also a way to obtain the million and avoid drinking, but it cannot happen on purpose. This can happen only in one way: namely, by being lucky. You plan on every level not to reconsider, and expect this to happen, but luckily you end up being mistaken, and you do reconsider, despite expecting not to. In this case you both obtain the million, and avoid the drink.

Perfectly Random

December 30, 2018 entirelyuseless Logic, Mathematics, Metaphysics, Probability, Science, Uncategorized David Hume, Definitions, Form, Future, Induction, Matter, Natural Laws, Numbers, Randomness, Time

Suppose you have a string of random binary digits such as the following:

00111100010101001100011011001100110110010010100111

This string is 50 digits long, and was the result of a single attempt using the linked generator.

However, something seems distinctly non-random about it: there are exactly 25 zeros and exactly 25 ones. Naturally, this will not always happen, but most of the time the proportion of zeros will be fairly close to half. And evidently this is necessary, since if the proportion was usually much different from half, then the selection could not have been random in the first place.

There are other things about this string that are definitely not random. It contains only zeros and ones, and no other digits, much less items like letters from the alphabet, or items like ‘%’ and ‘$’.

Why do we have these apparently non-random characteristics? Both sorts of characteristics, the approximate and typical proportion, and the more rigid characteristics, are necessary consequences of the way we obtained or defined this number.

It is easy to see that such characteristics are inevitable. Suppose someone wants to choose something random without any non-random characteristics. Let’s suppose they want to avoid the first sort of characteristic, which is perhaps the “easier” task. They can certainly make the proportion of zeros approximately 75% or anything else that they please. But this will still be a non-random characteristic.

They try again. Suppose they succeed in preventing the series of digits from converging to any specific probability. If they do, there is one and only one way to do this. Much as in our discussion of the mathematical laws of nature, the only way to accomplish this will be to go back and forth between longer and longer strings of zeros and ones. But this is an extremely non-random characteristic. So they may have succeeded in avoiding one particular type of non-randomness, but only at the cost of adding something else very non-random.

Again, consider the second kind of characteristic. Here things are even clearer: the only way to avoid the second kind of characteristic is not to attempt any task in the first place. The only way to win is not to play. Once we have said “your task is to do such and such,” we have already specified some non-random characteristics of the second kind; to avoid such characteristics is to avoid the task completely.

“Completely random,” in fact, is an incoherent idea. No such thing can exist anywhere, in the same way that “formless matter” cannot actually exist, but all matter is formed in one way or another.

The same thing applies to David Hume’s supposed problem of induction. I ended that post with the remark that for his argument to work, he must be “absolutely certain that the future will resemble the past in no way.” But this of course is impossible in the first place; the past and the future are both defined as periods of time, and so there is some resemblance in their very definition, in the same way that any material thing must have some form in its definition, and any “random” thing must have something non-random in its definition.

Violations of Bell’s Inequality: Drawing Conclusions

August 17, 2018 entirelyuseless Metaphysics, Science, Uncategorized Absolute Time, Action at a Distance, Albert Einstein, Aristotle, Bell Inequality, David Hume, Determinism, Erwin Schrödinger, Future, Hugh Everett, Induction, Intelligibility, Isaac Newton, John Bell, Many Worlds, Mark Alford, Marvin Edwards, Ptolemaic System, Quantum Mechanics, Relativity, Vagueness

In the post on violations of Bell’s inequality, represented there by Mark Alford’s twin analogy, I pointed out that things did not seem to go very well for Einstein’s hope for physics, I did not draw any specific conclusions. Here I will consider the likely consequences, first by looking at the relationship of the experiments to Einstein’s position on causality and determinism, and second on their relationship to Einstein’s position on locality and action at a distance.

Einstein on Determinism

Einstein hoped for “facts” instead of probabilities. Everything should be utterly fixed by the laws, much like the position recently argued by Marvin Edwards in the comments here.

On the face of it, violations of Bell’s inequality rule this out, represented by the argument that if the twins had pre-existing determinate plans, it would be impossible for them to give the same answer less than 1/3 of the time when they are asked different questions. Bell however pointed out that it is possible to formulate a deterministic theory which would give similar probabilities at the cost of positing action at a distance (quoted here):

Moreover, a hidden variable interpretation of elementary quantum theory has been explicitly constructed. That particular interpretation has indeed a grossly non-local structure. This is characteristic, according to the result to be proved here, of any such theory which reproduces exactly the quantum mechanical predictions.

Nonetheless, I have set aside action at a distance to be discussed separately, and I would argue that we should accept the above surface appearance: the outcomes of quantum mechanical experiments are actually indeterministic. These probabilities represent something in the world, not merely something in our knowledge.

Why? In the first place, note that “reproduces exactly the quantum mechanical predictions” can be understood in two ways. A deterministic theory of that kind would say that because the details are unknown to us, we cannot know what is going to happen. But the details are there, and they in fact determine what is going to happen. There is still a difference on the object level between a world where the present fixes the future to a single possibility, and one in which the future is left open, as Aristotle supposed.

Of course there is no definitive proof here that we are actually in the situation with the open future, although the need for action at a distance in the alternative theory suggests that we are. Even apart from this, however, the general phenomena of quantum mechanics directly suggest that this is the situation. Even apart from violations of Bell’s inequality, quantum mechanics in general already looked exactly as we should have expected a world with an indeterminate future to look.

If this is the case, then Einstein was mistaken on this point, at least to this extent. But what about the deterministic aspect, which I mentioned at the end of this post, and which Schrödinger describes:

At all events it is an imagined entity that images the blurring of all variables at every moment just as clearly and faithfully as does the classical model its sharp numerical values. Its equation of motion too, the law of its time variation, so long as the system is left undisturbed, lags not one iota, in clarity and determinacy, behind the equations of motion of the classical model.

The answer is that this is deterministic not because the future, as we know it, is deterministic, but because it describes all of the possibilities at once. Thus in the case of the cat it includes both the cat living and the cat dying, which are two possible outcomes. It is “deterministic” only because once you have stated all of the alternatives, there is nothing left to say.

Why did Einstein want a deterministic theory? He openly admits that he does not have a convincing argument for it. It seems likely, however, that the fundamental motivation is the conviction that reality is intelligible. And an indeterministic world seems significantly less intelligible than a deterministic one. But this desire can in fact be satisfied by this second kind of “determinism”; thus Schrödinger calls it “one perfectly clear concept.”

In this respect, Einstein’s intuition was not mistaken. It is possible to give an intelligible account of the world, even a “deterministic” one, in this sense.

Einstein on Locality

Einstein also wanted to avoid “spooky action at a distance.” Admitting that the future is indeterminate, however, is not enough to avoid this conclusion. In Mark Alford’s twin analogy, it is not only pre-determined plans that fail, but also plans that involve randomness. Thus it first appears that the violations of Bell’s inequality absolutely require action at a distance.

If we follow my suggestion here, however, and consequently adopt Hugh Everett’s interpretation of quantum mechanics, then saying that there are multiple future possibilities implies the existence of multiple timelines. And if there are multiple timelines, violations of Bell’s inequality no longer necessarily imply action at a distance.

Why not? Consider the twin experiment with the assumption of indeterminacy and multiple timelines. Suppose that from the very beginning, there are two copies of each twin. The first copy of the first twin has the plan of responding to the three questions with “yes/yes/yes.” Likewise, the first copy of the second twin has the plan of responding to the three questions with, “yes/yes/yes.” In contrast, the second copy of each twin has the plan of responding with “no/no/no.”

Now we have four twins but the experimenter only sees two. So which ones does he see? There is nothing impossible about the following “rule”: if the twins are asked different questions, the experimenter sees the first copy of one of the twins, and the second copy of the other twin. Meanwhile, if the twins are asked the same question, the experimenter sees either the first copy of each twin, or the second copy of each twin. It is easy to see that if this is the case, the experimenter will see the twins agree, when they are asked the same question, and will see them disagree when they are asked different questions (thus agreeing less than 1/3 of the time in that situation.)

“Wait,” you will say. “If multiple timelines is just a way of describing a situation with indeterminism, and indeterminism is not enough to avoid action at a distance, how is it possible for multiple timelines to give a way out?”

From the beginning, the apparent “impossibility” of the outcome was a statistical impossibility, not a logical impossibility. Naturally this had to be the case, since if it were a logical impossibility, we could not have coherently described the actual outcomes. Thus we might imagine that David Hume would give this answer:

The twins are responding randomly to each question. By pure chance, they happened to agree the times they were asked the same question, and by pure chance they violated Bell’s inequality when they were asked different questions.

Since this was all a matter of pure chance, of course, if you do the experiment again tomorrow, it will turn out that all of the answers are random and they will agree and disagree 50% of the time on all questions.

And this answer is logically possible, but false. This account does not explain the correlation, but simply ignores it. In a similar way, the reason why indeterministic theories without action at a distance, but described as having a single timeline, cannot explain the results is that in order to explain the correlation, the outcomes of both sides need to be selected together, so to speak. But “without action at a distance” in this context simply means that they are not selected together. This makes the outcome statistically impossible.

In our multiple timelines version, in contrast, our “rule” above in effect selected the outcomes together. In other words, the guideline we gave regarding which pairs of twins the experimenter would meet, had the same effect as action at a distance.

How is all this an explanation? The point is that the particular way that timelines spread out when they come into contact with other things, in the version with multiple timelines, exactly corresponds to action at a distance, in the version without them. An indeterministic theory represented as having a single timeline and no action at a distance could be directly translated into a version with multiple timelines; but if we did that, this particular multiple timeline version would not have the rule that produces the correct outcomes. And on the other hand, if we start with the multiple timeline version that does have the rule, and translate it into a single timeline account, it will have action at a distance.

What does all this say about Einstein’s opinion about locality? Was he right, or was he wrong?

We might simply say that he was wrong, insofar as the actual situation can in fact be described as including action at a distance, even if it is not necessary to describe it in this way, since we can describe it with multiple timelines and without action at a distance. But to the degree that this suggests that Einstein made two mistakes, one about determinism and one about action at a distance, I think this is wrong. There was only one mistake, and it was the one about determinism. The fact is that as soon you speak of indeterminism at all, it becomes possible to speak of the world as having multiple timelines. So the question at that point is whether this is the “natural” description of the situation, where the natural description more or less means the best way to understand things, in which case the possibility of “action at a distance” is not an additional mistake on Einstein’s part, but rather it is an artifact of describing the situation as though there were only a single timeline.

You might say that there cannot be a better or worse way to understand things if two accounts are objectively equivalent. But this is wrong. Thus for example in general relativity it is probably possible to give an account where the earth has no daily rotation, and the universe is spinning around it every 24 hours. And this account is objectively equivalent to the usual account where the earth is spinning; exactly the same situation is being described, and nothing different is being asserted. And yet this account is weird in many ways, and makes it very hard to understand the universe. The far better and “natural” description is that the earth is spinning. Note, however, that this is an overall result; just looking out the window, you might have thought that saying that the universe is spinning is more natural. (Notice, however, that an even more natural account would be that neither the earth nor the universe is moving; it is only later in the day that you begin to figure out that one of them is moving.)

In a similar way, a single timeline account is originally more natural in the way a Ptolemaic account is more natural when you look out the window. But I would argue that in a similar way, the multiple timeline account, without action at a distance, is ultimately the more natural one. The basic reason for this is that there is no Newtonian Absolute Time. The consequence is that if we speak of “future possibilities,” they cannot be future possibilities for the entire universe at once. They will be fairly localized future possibilities: e.g. there might be more than one possible text for the ending to this blog post, which has not yet been written, and those possibilities are originally possibilities for what happens here in this room, not for the rest of the universe. These future alternatives will naturally result in future possibilities for other parts of the world, but this will happen “slowly,” so to speak (namely if one wishes to speak of the speed of light as slow!) This fits well with the idea of multiple timelines, since there will have to be some process where these multiple timelines come into contact with the rest of the world, much as with our “rule” in the twin experiment. On the other hand, it does not fit so well with a single timeline account of future possibilities, since one is forced (by the terms of the account) to imagine that when a choice among possibilities is made, it is made for the entire universe at once, which appears to require Newton’s Absolute Time.

This suggests that Einstein was basically right about action at a distance, and wrong about determinism. But the intuition that motivated him to embrace both positions, namely that the universe should be intelligible, was sound.

Open Past

August 15, 2018August 13, 2018 entirelyuseless Metaphysics, Science, Uncategorized Aristotle, Determinism, Double-Slit Experiment, Erwin Schrödinger, Future, Lucretius, Quantum Mechanics

Suppose that Aristotle was right, and the future is open. What would things be like in detail?

There are many ways things could go, so for concreteness let’s assume that (in some local area) there are approximately 100 possibilities for the next second, and approximately 100 x 100, or 10,000 possibilities for the next two seconds.

Then the question arises: do some of the two-second outcomes have overlapping paths? In other words, suppose we take the first option in the first second. Are all of the outcomes we can reach different from all of the outcomes we could reach if we took the second option in the first second?

It is at least plausible that some overlapping paths can exist. For example, something might swerve to the left in the first second, and then to the right in the second second, ending up just where it would have been if it had swerved to the right in the first second and to the left in the second. Let’s suppose it turns out this way. Thus we have situation A and time A, and situation B and time B, with a first and second path, both of which lead from situation A at time A, to situation B at time B.

When we get to situation B, what does the world look like? In particular, if someone is in situation B and says, “let’s look at the world and figure out what just happened,” what does it look like? Consider three different accounts:

It looks like situation B except also that it looks like we took the first path
It looks like situation B except also that it looks like we took the second path
It looks like situation B, and we can’t tell which path was taken

The problem is evident. These are three different situations. If things currently look different, the situation is different. So these cannot possibly all be descriptions of situation B. And in particular, only the third is a reasonable description of the situation we should expect. We have set up the situation so that there is no difference in our current situation, whether the first or second path was taken. So of course in situation B it will be impossible to know which path was taken.

But what does that look like, exactly? “We don’t know” is not a description of a situation, but a description of our state of knowledge. What is it about situation B that makes it impossible to tell which path was taken? What happens if you describe the situation as exactly as possible, and then explain why that “exact” description still does not determine which path was taken?

Consider again Schrödinger’s confusion about his cat. The reason why the notion of “bluriness” came up at all was not merely that the wave equation seems to describe something blurred, but also because the actual results of experiments suggest that something blurred took place. Thus for example in double-slit experiments, interference patterns suggest that something is going through both slits at once, while if detectors are added to determine what, if anything, is going through the slits, one seems to find that only one slit is used at a time, and the interference pattern goes away.

This fits the above description of situation A and situation B almost perfectly. In the double slit experiment, there are two paths that could be taken to arrive at the same outcome. But that “same outcome” is not one in which it looks like the first path was taken, nor one in which it looks like the second path was taken, but one in which the outcome’s relationship to the path appears to be confused. And on the other hand, if we can tell which path was taken, as we can when we add detectors, there is no such confusion, because the outcomes no longer overlap; the outcome where the first detector registers is not the same as an outcome where the second detector registers.

In this sense, quantum theory is simply the situation where Aristotle was right about the indeterminacy of the future, with the minor addition that it turned out to be possible to get to the same future by more than route.

Note, however, that this implies the worrisome outcome that I suggested in that post. Just as the future is indeterminate, so is the past. Just as the present has many possible future outcomes, there are many past paths that could have resulted in the present.

Open Future

August 13, 2018August 11, 2018 entirelyuseless Logic, Metaphysics, Science, Uncategorized Aristotle, Determinism, Eternalism, Future, Modal Realism, Necessity, Paradoxes, Presentism, Time, Time Travel

Let’s return for a moment to the question at the end of this post. I asked, “What happens if the future is indeterminate? Would not the eternalist position necessarily differ from the presentist one, in that case?”

Why necessarily different? The argument in that post was that eternalism and presentism are different descriptions of the same thing, and that we see the sameness by noting the sameness of relations between the elements of the description. But if the future is open, as Aristotle supposed, it is hard to see how we can maintain this. Aristotle says that the present is open to either having the sea battle tomorrow or not having it. With an eternalist view, the sea battle is “already there” or it is not. So in Aristotle’s view, the present has an open relationship to both possibilities. But the eternalist view seems to be truly open only to the possibility that will actually happen. We no longer have the same set of relationships.

Notice the problem. When I attempted to equate eternalism and presentism, I implicitly assumed that determinism is true. There were only three states of the universe, beginning, middle, and end. If determinism is false, things are different. There might be beginning, middle, and two potential ends. Perhaps there is a sea battle in one of the potential ends, and no sea battle in the other.

This suggests a solution to our conundrum, however. Even the presentist description in that post was inconsistent with an open future. If there is only one possible end, the future is not open, even if we insist that the unique possible end “currently doesn’t exist.” The problem then was not eternalism as such, but the fact that we started out with a determinist description of the universe. This strongly suggests that if my argument about eternalism and presentism was correct, we should be able to formulate eternalist and presentist descriptions of an open future which will be equivalent. But both will need to be different from the fixed “beginning-middle-end” described in that post.

We can simply take Aristotle’s account as the account of presentism with an open future. How can we give an eternalist account of the same thing? The basic requirement will be that the relationship between the present and the future needs to be the same in both accounts. Now in Aristotle’s account, the present has the same relationship to two different possibilities: both of them are equally possible. So to get a corresponding eternalist account, we need the present to be equally related to two futures that correspond to the two possiblities in the presentist account. I do not say “two possible futures,” but “two futures,” precisely because the account is eternalist.

The careful reader will already understand the account from the above, but let us be more explicit. The eternalist account that corresponds to the presentist account with an open future has multiple timelines, all of which “exist”, in the eternalist sense. The reader will no doubt be familiar with the idea of multiple timelines, at least from time travel fiction. In a similar way, the eternalist reworking of Aristotle’s position is that there is a timeline where the sea battle takes place, and another timeline where the sea battle does not take place. In this view, both of them “actually” happen. But even in this view, an observer in the middle location will have to say, “I do not, and cannot, know whether the sea battle will take place or not,” just as in Aristotle’s view. For the observer cannot traverse both timelines at once. From his point of view, he will take only one, but since his relationship to the two possibilities (or actualities) is the same, it is indeterminate which one it will be.

Even if one cannot prove my account of equivalence to be wrong, the reader may worry. Time travel fiction frequently seems incoherent, and this suggests that any view with multiple timelines may also be incoherent. But this potential incoherence supports the equivalence, rather than subtracting from it. For as we noted in the post on Aristotle, there is a definite appearance of incoherence in his position. It is not even clear how his view is logically possible. So it would not be surprising, but quite natural, if views which are intended to be equivalent to his position are also not clearly coherent. Nonetheless, the multiple timelines description does have some logical advantage over Aristotle’s position, in the sense that “the sea battle will take place in timeline A” does not even appear to contradict “the sea battle will not take place in timeline B.”

Aristotle on Future Contingents

August 11, 2018August 10, 2018 entirelyuseless Logic, Metaphysics, Science, Uncategorized Aristotle, CPT Symmetry, Deliberation, Determinism, Excluded Middle, Free Will, Future, Necessity, Non-Contradiction, Past, Prediction, Propositions, Quantum Mechanics, Time, Truth

In Chapter 9 of On Interpretation, Aristotle argues that at least some statements about the future need to be exempted from the principle of Excluded Middle:

In the case of that which is or which has taken place, propositions, whether positive or negative, must be true or false. Again, in the case of a pair of contradictories, either when the subject is universal and the propositions are of a universal character, or when it is individual, as has been said,’ one of the two must be true and the other false; whereas when the subject is universal, but the propositions are not of a universal character, there is no such necessity. We have discussed this type also in a previous chapter.

When the subject, however, is individual, and that which is predicated of it relates to the future, the case is altered. For if all propositions whether positive or negative are either true or false, then any given predicate must either belong to the subject or not, so that if one man affirms that an event of a given character will take place and another denies it, it is plain that the statement of the one will correspond with reality and that of the other will not. For the predicate cannot both belong and not belong to the subject at one and the same time with regard to the future.

Thus, if it is true to say that a thing is white, it must necessarily be white; if the reverse proposition is true, it will of necessity not be white. Again, if it is white, the proposition stating that it is white was true; if it is not white, the proposition to the opposite effect was true. And if it is not white, the man who states that it is making a false statement; and if the man who states that it is white is making a false statement, it follows that it is not white. It may therefore be argued that it is necessary that affirmations or denials must be either true or false.

Now if this be so, nothing is or takes place fortuitously, either in the present or in the future, and there are no real alternatives; everything takes place of necessity and is fixed. For either he that affirms that it will take place or he that denies this is in correspondence with fact, whereas if things did not take place of necessity, an event might just as easily not happen as happen; for the meaning of the word ‘fortuitous’ with regard to present or future events is that reality is so constituted that it may issue in either of two opposite directions. Again, if a thing is white now, it was true before to say that it would be white, so that of anything that has taken place it was always true to say ‘it is’ or ‘it will be’. But if it was always true to say that a thing is or will be, it is not possible that it should not be or not be about to be, and when a thing cannot not come to be, it is impossible that it should not come to be, and when it is impossible that it should not come to be, it must come to be. All, then, that is about to be must of necessity take place. It results from this that nothing is uncertain or fortuitous, for if it were fortuitous it would not be necessary.

The argument here is that if it is already true, for example, that I will eat breakfast tomorrow, then I will necessarily eat breakfast tomorrow, and there is no option about this and no ability of anything to prevent it. Aristotle is here taking it for granted that some things about the future are uncertain, and is using this as a reductio against the position that such claims can be already true. He goes on to give additional reasons for the same thing:

Again, to say that neither the affirmation nor the denial is true, maintaining, let us say, that an event neither will take place nor will not take place, is to take up a position impossible to defend. In the first place, though facts should prove the one proposition false, the opposite would still be untrue. Secondly, if it was true to say that a thing was both white and large, both these qualities must necessarily belong to it; and if they will belong to it the next day, they must necessarily belong to it the next day. But if an event is neither to take place nor not to take place the next day, the element of chance will be eliminated. For example, it would be necessary that a sea-fight should neither take place nor fail to take place on the next day.

These awkward results and others of the same kind follow, if it is an irrefragable law that of every pair of contradictory propositions, whether they have regard to universals and are stated as universally applicable, or whether they have regard to individuals, one must be true and the other false, and that there are no real alternatives, but that all that is or takes place is the outcome of necessity. There would be no need to deliberate or to take trouble, on the supposition that if we should adopt a certain course, a certain result would follow, while, if we did not, the result would not follow. For a man may predict an event ten thousand years beforehand, and another may predict the reverse; that which was truly predicted at the moment in the past will of necessity take place in the fullness of time.

Further, it makes no difference whether people have or have not actually made the contradictory statements. For it is manifest that the circumstances are not influenced by the fact of an affirmation or denial on the part of anyone. For events will not take place or fail to take place because it was stated that they would or would not take place, nor is this any more the case if the prediction dates back ten thousand years or any other space of time. Wherefore, if through all time the nature of things was so constituted that a prediction about an event was true, then through all time it was necessary that that should find fulfillment; and with regard to all events, circumstances have always been such that their occurrence is a matter of necessity. For that of which someone has said truly that it will be, cannot fail to take place; and of that which takes place, it was always true to say that it would be.

Yet this view leads to an impossible conclusion; for we see that both deliberation and action are causative with regard to the future, and that, to speak more generally, in those things which are not continuously actual there is potentiality in either direction. Such things may either be or not be; events also therefore may either take place or not take place. There are many obvious instances of this. It is possible that this coat may be cut in half, and yet it may not be cut in half, but wear out first. In the same way, it is possible that it should not be cut in half; unless this were so, it would not be possible that it should wear out first. So it is therefore with all other events which possess this kind of potentiality. It is therefore plain that it is not of necessity that everything is or takes place; but in some instances there are real alternatives, in which case the affirmation is no more true and no more false than the denial; while some exhibit a predisposition and general tendency in one direction or the other, and yet can issue in the opposite direction by exception.

Now that which is must needs be when it is, and that which is not must needs not be when it is not. Yet it cannot be said without qualification that all existence and non-existence is the outcome of necessity. For there is a difference between saying that that which is, when it is, must needs be, and simply saying that all that is must needs be, and similarly in the case of that which is not. In the case, also, of two contradictory propositions this holds good. Everything must either be or not be, whether in the present or in the future, but it is not always possible to distinguish and state determinately which of these alternatives must necessarily come about.

Let me illustrate. A sea-fight must either take place to-morrow or not, but it is not necessary that it should take place to-morrow, neither is it necessary that it should not take place, yet it is necessary that it either should or should not take place to-morrow. Since propositions correspond with facts, it is evident that when in future events there is a real alternative, and a potentiality in contrary directions, the corresponding affirmation and denial have the same character.

This is the case with regard to that which is not always existent or not always nonexistent. One of the two propositions in such instances must be true and the other false, but we cannot say determinately that this or that is false, but must leave the alternative undecided. One may indeed be more likely to be true than the other, but it cannot be either actually true or actually false. It is therefore plain that it is not necessary that of an affirmation and a denial one should be true and the other false. For in the case of that which exists potentially, but not actually, the rule which applies to that which exists actually does not hold good. The case is rather as we have indicated.

Basically, then, there are two arguments. First there is the argument that if statements about the future are already true, the future is necessary. If a sea battle will take place tomorrow, it will necessarily take place. Second, there is the argument that this excludes deliberation. If a sea battle will take place tomorrow, then it will necessarily take place, and no place remains for deliberation and decision about whether to fight the sea battle. Whether you decide to fight or not, it will necessarily take place.

Unfortunately for Aristotle, both arguments fail. Consider the first argument about necessity. Aristotle’s example is that “if it is true to say that a thing is white, it must necessarily be white.” But this is hypothetical necessity, not absolute necessity. A thing must be white if it is true that is white, but that does not mean that “it must be white, period.” Thus for example I have a handkerchief, and it happens to be white. If it is true that it is white, then it must be white. But it would be false to simply say, “My handkerchief is necessarily white.” Since I can dye it other colors, obviously it is not simply necessary for it to be white.

In a similar way, of course it is true that if a sea battle will take place, it will take place. It does not follow at all that “it will necessarily take place, period.”

Again, consider the second argument, that deliberation would be unnecessary. Aristotle makes the point that deliberation is causative with respect to the future. But gravity is also causative with respect to the future, as for example when gravity causes a cup to fall from a desk. It does not follow either that the cup must be able not to fall, nor that gravity is unnecessary. In a similar way, a sea battle takes place because certain people deliberated and decided to fight. If it was already true that it was going to take place, then it also already true that they were going to decide to fight. It does not follow that their decision was unnecessary.

Consider the application to gravity. It is already true that if the cup is knocked from the desk, it will fall. It does not follow that gravity will not cause the fall: in fact, it is true precisely because gravity will cause the fall. In a similar way, if it true that the battle will take place, it is true because the decision will be made.

This earlier discussion about determinism is relevant to this point. Asserting that there is a definite outcome that our deliberations will arrive at, in each case, goes against our experience in no way. The feeling of “free will,” in any case, has a different explanation, whether or not determinism is true.

On the other hand, there is also no proof that there is such a determinate outcome, even if in some cases there are things that would suggest it. What happens if in fact there is nothing ensuring one outcome rather than another?

Here we could make a third argument on Aristotle’s behalf, although he did not make it himself. If the present is truly open to alternative outcomes, then it seems that nothing exists that could make it be true that “a sea battle will take place,” and false that “a sea battle will not take place.” Presumably if a statement is true, there must be something in reality which is the cause of the statement’s truth. Now there does not seem to be anything in reality, in this scenario, which could be a cause of truth. Therefore it does not seem that either alternative could be true, and Aristotle would seem to be right.

I will not attempt to refute this argument at this point, but I will raise two difficulties. First of all, it is not clear that his claim is even coherent. Aristotle says that “either there will be a sea battle or there will not be,” is true, but that “there will be a sea battle” is not true, and “there will not be a sea battle” is not true. This does not seem to be logically consistent, and it is not clear that we can even understand what is being said. I will not push this objection too hard, however, lest I be accused of throwing stones from a glass house.

Second, the argument that there is nothing in reality that could cause the truth of a statement might apply to the past as well as to the future. There is a tree outside my window right now. What was in that place exactly 100 million years ago to this moment? It is not obvious that there is anything in the present world which could be the cause of the truth of any statement about this. One might object that the past is far more determinate than the future. There are plenty of things in the present world that might be the cause of the truth of the statement, “World War II actually happened.” It is hard to see how you could possibly have arrived at the present world without it, and this “necessity” of World War II in order to arrive at the present world could be the cause of truth. The problem is that there is still no proof that this is universal. Once things are far enough in past, like 100 million years, perhaps minor details become indeterminate. Will Aristotle really want to conclude that some statements about the past are neither true nor false?

I will more or less leave things here without resolving them in this post, although I will give a hint (without proof at this time) regarding the truth of the matter. It turns out that quantum mechanics can be interpreted in two ways. In one way, it is a deterministic theory, and in this way it is basically time reversible. The present fully determines the past, but it equally fully determines the future. Interpreted in another way, it is an indeterministic theory which leaves the future uncertain. But understood in this way, it also leaves the past uncertain.

Necessary Connection

July 22, 2018 entirelyuseless Metaphysics, Uncategorized Causes, Color, David Hume, Epistemology, Essence, Experience, Explanation, Future, Induction, Kantian Error, Knowledge, Necessity, Senses, Subjectivity

In Chapter 7 of his Enquiry Concerning Human Understanding, David Hume says about the idea of “necessary connection”:

We have looked at every possible source for an idea of power or necessary connection, and have found nothing. However hard we look at an isolated physical episode, it seems, we can never discover discover anything but one event following another; we never find any force or power by which the cause operates, or any connection between it and its supposed effect. The same holds for the influence of mind on body: the mind wills, and then the body moves, and we observe both events; but we don’t observe– and can’t even conceive– the tie that binds the volition to the motion, i.e. the energy by which the mind causes the body to move. And the power of the will over its own faculties and ideas– i.e. over the mind, as distinct from the body– is no more comprehensible. Summing up, then: throughout the whole of nature there seems not to be a single instance of connection that is conceivable by us. All events seem to be entirely loose and separate. One event follows another, but we never can observe any tie between them. They seem associated, but never connected. And as we can have no idea of anything that never appeared as an impression to our outward sense or inward feeling, we are forced to conclude that we have no idea of ‘connection’ or ‘power’ at all, and that those words– as used in philosophical reasonings or in common life– have absolutely no meaning.

This is not Hume’s final word on the matter, as we will see below, so this has to be taken with a grain of salt, even as a representation of his opinion. Nonetheless, consider this caricature of what he just said:

We have looked at every possible source for an idea of mduvvqi or pdnfhvdkdddd, and have found nothing. However hard we look at an isolated physical episode, it seems, we can never discover anything but events that can be described by perfectly ordinary words; we never find any mduvvqi involved, nor any pdnfhvkdddd.

We could take this to be making the point that “mduvvqi” and “pdnfhvdkdddd” are not words. Other than that, however, the paragraph itself is meaningless, precisely because those “words” are meaningless. It certainly does not make any deep (or shallow for that matter) metaphysical or physical point, nor any special point about the human mind. But Hume’s text is different, and the difference in question is a warning sign of Kantian confusion. If those words had “absolutely no meaning,” in fact, there would be no difference between Hume’s passage and our caricature. Those words are not meaningless, but meaningful, and Hume is even analyzing their meaning in order to supposedly determine that the words are meaningless.

Hume’s analysis in fact proceeds more or less in the following way. We know what it means to say that something is necessary, and it is not the same as saying that the thing always happens. Every human being we have ever seen was less than 20 feet tall. But is it necessary that human beings be less than 20 feet tall? This is a different question, and while we can easily experience someone’s being less than 20 feet tall, it is very difficult to see how we could possibly experience the necessity of this fact, if it is necessary. Hume concludes: we cannot possibly experience the necessity of it. Therefore we can have no idea of such necessity.

But Hume has just contradicted himself: it was precisely by understanding the concept of necessity that he was able to see the difficulty in the idea of experiencing necessity.

Nonetheless, as I said, this is not his final conclusion. A little later he gives a more nuanced account:

The source of this idea of a necessary connection among events seems to be a number of similar instances of the regular pairing of events of these two kinds; and the idea cannot be prompted by any one of these instances on its own, however comprehensively we examine it. But what can a number of instances contain that is different from any single instance that is supposed to be exactly like them? Only that when the mind experiences many similar instances, it acquires a habit of expectation: the repetition of the pattern affects it in such a way that when it observes an event of one of the two kinds it expects an event of the other kind to follow. So the feeling or impression from which we derive our idea of power or necessary connection is a feeling of connection in the mind– a feeling that accompanies the imagination’s habitual move from observing one event to expecting another of the kind that usually follows it. That’s all there is to it. Study the topic from all angles; you will never find any other origin for that idea.

Before we say more, we should concede that this is far more sensible than the claim that the idea of necessity “has absolutely no meaning.” Hume is now conceding that it does have meaning, but claiming that the meaning is about us, not about the thing. When we see someone knock a glass off a table, we perhaps feel a certainty that it will fall and hit the floor. Experiencing that feeling of certainty, he says, is the source of the idea of “necessity.” This is not an unreasonable hypothesis.

However, Hume is also implicitly making a metaphysical argument here which is somewhat less sensible. Our feelings of certainty and uncertainty are subjective qualities of our minds, he suggests, not objective features of the things. Therefore necessity as an objective feature does not and cannot exist. This is not unrelated to his mistaken claim that we cannot know that the future will be similar to the past, even with probability.

What is the correct account here? In fact we already know, from the beginning of the conversation, that “necessary” and “possible” are meaningful words. We also know that in fact we use them to describe objective features of the world. But which features? Attempting to answer this question is where Hume’s approach is pretty sensible. Hume is not mistaken that all of our knowledge is from experience, and ultimately from the senses. He seems to identify experience with sense experience too simplistically, but he is not mistaken that all experience is at least somewhat similar to sense experience; feeling sure that two and two make four is not utterly unlike seeing something red. We want to say that there is something in common there, “something it is like,” to experience one or the other. But if this is the case, it would be reasonable to extend what we said about the senses to intellectual experiences. “The way red looks” cannot, as such, be an objective feature of a thing, but a thing can be objectively red, in such a way that “being red,” together with the nature of the senses, explains why a thing looks red. In a similar way, certainty and uncertainty, insofar as they are ways we experience the world, cannot be objective features of the world as such. Nonetheless, something can be objectively necessary or uncertain, in such a way that “being necessary” or otherwise, together with the nature of our minds, explains why it seems certain or uncertain to us.

There will be a similarity, however. The true nature of red might be quite strange in comparison to the experience of seeing red, as for example it might consist of surface reflectance properties. In a similar way, the true nature of necessity, once it is explained, might be quite strange to us compared to the experience of being certain or uncertain. But that it can be explained is quite certain itself, since the opposite claim would fall into Hume’s original absurdity. There are no hidden essences.

Truth and Expectation III

July 14, 2018 entirelyuseless Logic, Metaphysics, Uncategorized Belief, Bob Seidensticker, Future, Mormonism, Prediction, Richard Rorty, Scott Sumner, Truth

Consider what I said at the end of the last post on this topic. When our Mormon protagonist insists that his religion is true, he improves the accuracy of his expectations about the world. His expectations would actually be less accurate he decided that his religious beliefs are false.

Now we said in the first post on truth and expectation that in part we seem to determine the meaning of a statement by the expectations that it implies. If this is the case, then why should we not say that the Mormon’s beliefs are definitely true? In fact, in the post linked above about truth in religion, I suggested that people frequently do mean something like this when they say that their own religion is true. Nonetheless, it is easy to see that truth in this particular sense does not imply that each and every claim in the religion, understood as a claim about the world, is true.

But why not, if one’s expectations become more accurate, just as with statements that are clearly true? As I noted in the earlier post, to say that “that man is pretty tall” is a statement about the man. It is not a statement about myself, nor about my expectations, even if the meaning is partly determined by these things. So ultimately the truth or falsehood of the claim about the man is going to be determined by facts about the man, even if they need to be understood as facts about the man in relation to me and my expectations.

Consider again Scott Sumner’s anti-realism. Scott claims that we cannot distinguish between “our perception of reality, and actual reality.” As I said there, this is right in the sense that we cannot consistently hold the opinion, “This is my opinion about reality, but my opinion is false: reality is actually different.” But we can recognize the distinct meanings in “my perception of reality” and “actual reality.” Scott’s failure to recognize this distinction leads him to suggest on occasion that our beliefs about certain matters are just beliefs about what people in the future will believe. For example, he says in this comment:

You misunderstood Rorty. He is not recommending that you try to trick you colleagues into believing something that is not true. Rather he is merely describing what society regards as true. And who can deny that society tends to regard things as true, when they believe them to be true. No you might say “but what society believes to be true is not always really true.” But Rorty would say that statement means nothing, or else it means that you predict that a future society will have a different view of what is true.

Most people, without even realizing it, assume that there is some sort of “God-like” view of what is “really true” which is separate from what we believe is true, and or will believe in the future to be true. Rorty is an atheist. He believes that what society’s experts believe is the best we can do, the closest we can come to describing reality. Rorty would suggest to Hayek “if you want to convince other economists, use persuasive arguments.” I think that is very reasonable advice. It is what I try to do on this blog.

This is much like Bob Seidensticker’s claim that moral beliefs are beliefs about what society in the future will believe to be moral. In both cases, there is an unacceptable circularity. If our belief is about what people in the future will believe, what are the future people’s beliefs about? There is only one credible explanation here: people’s beliefs are about what they say their beliefs are about, namely the very things they are talking about. Moral beliefs are about whether actions are good or bad, and beliefs describing the world are about the world, not about the people who hold the beliefs, present or future.

This does not imply that there is “some sort of ‘God-like’ view of what is ‘really true’.” It just implies that our beliefs are distinct from other things in the world. Sumner is suggesting that a situation where everyone permanently holds a false belief, forever, is inconceivable. But this is quite conceivable, and we can easily see how it could happen. Just now I counted the cups in my cupboard and there are exactly 14 (if anyone is surprised by that value, most of those are plastic.) It is entirely conceivable, however, that I miscounted. And since I also have some cups that are not currently in the cupboard, if I did, I will probably never get it right, since I will just assume there was some other assortment. And since there is presumably no way for the public to discover the truth about this, society will be permanently deluded about the number of cups in my cupboard at 11:16 AM on July 14, 2018.

What would it mean, then, if it was not “actually true” that there were 14 cups in my cupboard? It would be determined by facts about the cups and the cupboard, not by facts about me, about society, about my expectations or society’s expectations. It would not be actually true if there were, in fact, only 13 cups there, even if no one would ever know this.

This all remains related to expectations, however. I don’t think I miscounted, so I think that if I go again and count them, I will get 14 again. If I did miscount, there is a good chance that counting again would result in a different number. Now I don’t intend to bother, so this expectation is counterfactual. Nonetheless, there are at least conceivable counterfactual situations where I would discover that I was mistaken. In a similar way, to say that the Mormon holds false religious beliefs implies that there are at least counterfactual situations where the falsehood would be uncovered by a violation of his expectations: e.g. if he had been alive at the time and had followed Joseph Smith around all the time to see whether there were any golden plates and so on.

Nonetheless, one cannot ultimately separate the truth of a statement from facts about the thing the statement is about. A statement is true if things are the way it says they are, “the way” being correctly understood here. If this were not the case, the statement would not be about the things in the first place.

Skeptical Scenarios

June 3, 2018 entirelyuseless Metaphysics, Religion, Science, Uncategorized Acceptance, Alexander Pruss, Amoris Laetitia, Antichrist, Brain-in-Vat, Details, Essence, Form, Future, Jehovah's Witnesses, Knowledge, Medjugorje, Patience, Physics, Pope Francis, Prediction, Reductionism, Relation, Ron Conte, Skeptical Scenarios, Skepticism, Truth, Universe

I promised to return to some of the issues discussed here. The current post addresses the implications of the sort of skeptical scenario considered by Alexander Pruss in the associated discussion. Consider his original comparison of physical theories and skeptical scenarios:

The ordinary sentence “There are four chairs in my office” is true (in its ordinary context). Furthermore, its being true tells us very little about fundamental ontology. Fundamental physical reality could be made out of a single field, a handful of fields, particles in three-dimensional space, particles in ten-dimensional space, a single vector in a Hilbert space, etc., and yet the sentence could be true.

An interesting consequence: Even if in fact physical reality is made out of particles in three-dimensional space, we should not analyze the sentence to mean that there are four disjoint pluralities of particles each arranged chairwise in my office. For if that were what the sentence meant, it would tell us about which of the fundamental physical ontologies is correct. Rather, the sentence is true because of a certain arrangement of particles (or fields or whatever).

If there is such a broad range of fundamental ontologies that “There are four chairs in my office” is compatible with, it seems that the sentence should also be compatible with various sceptical scenarios, such as that I am a brain in a vat being fed data from a computer simulation. In that case, the chair sentence would be true due to facts about the computer simulation, in much the way that “There are four chairs in this Minecraft house” is true. It would be very difficult to be open to a wide variety of fundamental physics stories about the chair sentence without being open to the sentence being true in virtue of facts about a computer simulation.

If we consider this in light of our analysis of form, it is not difficult to see that Pruss is correct both about the ordinary chair sentence being consistent with a large variety of physical theories, and about the implication that it is consistent with most situations that would normally be considered “skeptical.” The reason is that to say that something is a chair is to say something about its relationships with the world, but it is not to say everything about its relationships. It speaks in particular about various relationships with the human world. And there is nothing to prevent these relationships from co-existing with any number of other kinds of relationships between its parts, its causes, and so on.

Pruss is right to insist that in order for the ordinary sentence to be true, the corresponding forms must be present. But as an anti-reductionist, his position implies hidden essences, and this is a mistake. Indeed, under the correct understanding of form, our everyday knowledge of things is sufficient to ensure that the forms are present: regardless of which physical theories turn out to be true, and even if some such skeptical scenario turns out to be true.

Why are these situations called “skeptical” in the first place? This is presumably because they seem to call into question whether or not we possess any knowledge of things. And in this respect, they fail in two ways, they partially fail in a third, and they succeed in one way.

First, they fail insofar as they attempt to call into question, e.g. whether there are chairs in my room right now, or whether I have two hands. These things are true and would be true even in the “skeptical” situations.

Second, they fail even insofar as they claim, e.g. that I do not know whether I am a brain in a vat. In the straightforward sense, I do know this, because the claim is opposed to the other things (e.g. about the chairs and my hands) that I know to be true.

Third, they partially fail even insofar as they claim, e.g. that I do not know whether I am a brain in a vat in a metaphysical sense. Roughly speaking, I do know that I am not, not by deducing the fact with any kind of necessity, but simply because the metaphysical claim is completely ungrounded. In other words, I do not know this infallibly, but it is extremely likely. We could compare this with predictions about the future. Thus for example Ron Conte attempts to predict the future:

First, an overview of the tribulation:
A. The first part of the tribulation occurs for this generation, beginning within the next few years, and ending in 2040 A.D.
B. Then there will be a brief period of peace and holiness on earth, lasting about 25 years.
C. The next few hundred years will see a gradual but unstoppable increase in sinfulness and suffering in the world. The Church will remain holy, and Her teaching will remain pure. But many of Her members will fall into sin, due to the influence of the sinful world.
D. The second part of the tribulation occurs in the early 25th century (about 2430 to 2437). The Antichrist reigns for less than 7 years during this time.
E. Jesus Christ returns to earth, ending the tribulation.

Now, some predictions for the near future. These are not listed in chronological order.

* The Warning, Consolation, and Miracle — predicted at Garabandal and Medjugorje — will occur prior to the start of the tribulation, sometime within the next several years (2018 to 2023).
* The Church will experience a severe schism. First, a conservative schism will occur, under Pope Francis; next, a liberal schism will occur, under his conservative successor.
* The conservative schism will be triggered by certain events: Amoris Laetitia (as we already know, so, not a prediction), and the approval of women deacons, and controversial teachings on salvation theology.
* After a short time, Pope Francis will resign from office.
* His very conservative successor will reign for a few years, and then die a martyr, during World War 3.
* The successor to Pope Francis will take the papal name Pius XIII.

Even ignoring the religious speculation, we can “know” that this account is false, simply because it is inordinately detailed. Ron Conte no doubt has reasons for his beliefs, much as the Jehovah’s Witnesses did. But just as we saw in that case, his reasons will also in all likelihood turn out to be completely disproportionate to the detail of the claims they seek to establish.

In a similar way, a skeptical scenario can be seen as painting a detailed picture of a larger context of our world, one outside our current knowledge. There is nothing impossible about such a larger context; in fact, there surely is one. But the claim about brains and vats is very detailed: if one takes it seriously, it is more detailed than Ron Conte’s predictions, which could also be taken as a statement about a larger temporal context to our situation. The brain-in-vat scenario implies that our entire world depends on another world which has things similar to brains and similar to vats, along presumably with things analogous to human beings that made the vats, and so on. And since the whole point of the scenario is that it is utterly invented, not that it is accepted by anyone, while Conte’s account is accepted at least by him, there is not even a supposed basis for thinking that things are actually this way. Thus we can say, not infallibly but with a great deal of certainty, that we are not brains in vats, just as we can say, not infallibly but with a great deal of certainty, that there will not be any “Antichrist” between 2430 and 2437.

There is nonetheless one way in which the consideration of skeptical scenarios does succeed in calling our knowledge into question. Consider them insofar as they propose a larger context to our world, as discussed above. As I said, there is nothing impossible about a larger context, and there surely is one. Here we speak of a larger metaphysical context, but we can compare this with the idea of a larger physical context.

Our knowledge of our physical context is essentially local, given the concrete ways that we come to know the world. I know a lot about the room I am in, a significant amount about the places I usually visit or have visited in the past, and some but much less about places I haven’t visited. And speaking of an even larger physical context, I know things about the solar system, but much less about the wider physical universe. And if we consider what lies outside the visible universe, I might well guess that there are more stars and galaxies and so on, but nothing more. There is not much more detail even to this as a guess: and if there is an even larger physical context, it is possible that there are places that do not have stars and galaxies at all, but other things. In other words, universal knowledge is universal, but also vague, while specific knowledge is more specific, but also more localized: it is precisely because it is local that it was possible to acquire more specific knowledge.

In a similar way, more specific metaphysical knowledge is necessarily of a more local metaphysical character: both physical and metaphysical knowledge is acquired by us through the relationships things have with us, and in both cases “with us” implies locality. We can know that the brain-in-vat scenario is mistaken, but that should not give us hope that we can find out what is true instead: even if we did find some specific larger metaphysical context to our situation, there would be still larger contexts of which we would remain unaware. Just as you will never know the things that are too distant from you physically, you will also never know the things that are too distant from you metaphysically.

I previously advocated patience as a way to avoid excessively detailed claims. There is nothing wrong with this, but here we see that it is not enough: we also need to accept our actual situation. Rebellion against our situation, in the form of painting a detailed picture of a larger context of which we can have no significant knowledge, will profit us nothing: it will just be painting a picture as false as the brain-in-vat scenario, and as false as Ron Conte’s predictions.

	entirelyuseless on Richard Dawkins and the Simpli…
	Philip on Richard Dawkins and the Simpli…
	theusic on A New Beginning
	Shannon on Aristotle and the Ends of the…
	entirelyuseless on Zombies and Ignorance of the F…