What is truth?

[This may become part of a book.]

What is truth? There are complicated ways of explaining “truth” that get all tangled up in questions about reality and perception, but we’re going to use a very simple one: truth is what makes the future less surprising.

No matter what you think you are and no matter what “reality” may be, the experience that you have to deal with (like every other human being) is of being thrown into a surrounding that does things independently of your thoughts. Shit happens, and you have to deal with it. The first step to dealing with it is to be able to predict it.

So, for example, if somebody says to you “It’s raining outside,” the meaning of that claim is a bundle of implied predictions, including “If you go outside without a hat, hood, or umbrella your head will get wet.” You test the truth of that claim by checking if those predictions are true. You don’t have to know what water “really is”, or for that matter what “reality” is. (We’ll get to what “reality” is later; it’s not actually very complicated when you start from here.)

All truth claims can be unpacked as predictions. Sometimes they’re predictions about obvious, directly observable events in our immediate environment (“Rain will make your head wet.”). Sometimes they’re predictions about events we can’t observe directly but which have consequences we can observe (“Electricity is a flow of electrons,” or “Genetic information is carried by DNA.”) Sometimes they’re predictions about the distribution of outcomes in repeated tests (“A flipped coin will fall heads-up half the time and tails-up half the time.)

Sometimes truth claims are predictions about states of mind in other people – but even these are tested by observable consequences. If I say “Cathy likes chocolate,” for example, you could check that claim by offering Cathy some and seeing if she smiles.

If anyone makes a “truth” claim at you that you can’t unpack into testable predictions, be careful. It may be that you don’t understand the claim but it’s still true – if you don’t know what the properties of an electron are, for example, you’re not going to get much meaning out of the truth claim “Electricity is a flow of electrons”.

But it may also be that the claim is meaningless. A classic example is the sentence “Green ideas sleep furiously.” How would you tell if this is true? What consequences could you check? You can only assign a meaning to this sentence if you can answer these questions.

A more concise way of putting it is that every truth claim corresponds to a set of experiments, not necessarily in the formal sense with test tubes and lab coats but in the informal way that we might stick a hand out a window to see if water falls on it.

And now we can say what “reality” is; it’s wherever the experiments happen. It’s whatever observables are accessible to us.

We can also say what “theory” is. A theory is just a machine for generating predictions. We judge the theory’s “truth” by whether those predictions are correct. And, remember, we make predictions because we need to cope with the shit that happens. A theory is a survival adaptation: we are theory-builders because we are prediction-needers because we are goal-seekers because we are survival machines.

All other distinctions, even those as basic as the one between “me” and “everything else”, are consequences of our need to theory-build so we can generate predictions so we can cope with the shit that happens.

Often, this theory-building has happened out of sight of our conscious minds. Some theories (like the one that there’s a “me”, and an “everything else”) are formed so early in life that by the time we learn to speak we’ve forgotten the process. Others, like the theory that reality consists of “objects” separated by “space”, are wired deep into the evolved structure of our nervous systems.

[Having got this far, we can start talking about mental models and perception and the map-territory distinction and ontology. The point is *not to try to do those first*; that way lies only confusion. And, BTW, good for you you if you recognized this as Peirce crossed with Heidegger with a dash of Husserl.]

121 comments

  1. I may be dense (and should probably preface every statement emitted at 5:00 a.m. with that phrase), but what about past-truths? “The Declaration of Independence was signed July 4, 1776.” I think we can agree that that is a true statement, but what predictions does it imply? That people will agree with you when you say it?
    I suppose we have deduced this truth by working backwards from the consequences, saying “well, these people were writing about it in this time period, and we have this document dated July 4, 1776.”
    But those aren’t really testable as consequences of that truth. Must we then discard all history as unprovable?

  2. “The Declaration of Independence was signed July 4, 1776.” I think we can agree that that is a true statement, but what predictions does it imply?

    It implies that if you date the Declaration of Independence it will give a year close to 1776. Yes it wont give an exact date, thats because our methods to check the veracity is limited, in this case. There might be similar limitations for statements about the future too. Some predictions made by General Relativity have not been verified yet because they require technology that is currently beyond us. This does not imply anything about the correctness of General Relativity.

  3. Ben: It predicts that there will be certain artefacts in the world that fits the statement. There are lots of experiments you can do on these that could falsify the statement. (However, I don’t agree that you could prove it in the strict sense.)

  4. That’s all good for science hypothesis, but what about pure mathematics. In mathematics you can for example have euclidean and non-euclidean geometry, differing in whether one axiom is true or false. Well… unless you modify the meaning of prediction from “checking with reality” (“preforming experiment”) to “making formal proof (using allowed methods)”.

  5. ESR: You might want to be more explicit about what inferences are valid. For example, Cathy smiling when you give her chocolate can imply lots of things other than her liking chocolate; perhaps she likes *you* and smiles because she thinks you giving her chocolate is a sign that you like her, even though she hates chocolate. The same goes for vastly less likely situations: sticking your hand out the window and feeling water fall doesn’t necessarily mean it rains, it could be someone spraying water just above your window. I realise you get this, but the text doesn’t make the point. It’s an important point because it tightly limits absolute truth (and reality) to things we *directly* observe; the rest are inferences and we can not be completely sure about their truth.

    What about probability claims on non-recurring events? Strictly, you can say that a probability claim is true by definition (i.e. it was derived according to a process whose results we’ve agreed to call “probability”), but you can’t perform experiments to find out whether the meaning is true. I guess what I want to understand is: How does your thoughts on absolute truth relate to making inferences (which is what you need to do to deal with shit that happens)?

    Typo: s/for example, for example,/for example,/

  6. “And, remember, we make predictions because we need to cope with the shit that happens. A theory is a survival adaptation: we are theory-builders because we are prediction-needers because we are goal-seekers because we are survival machines.”

    As a hack TV sports journalist might say, “please talk about” the difference between theory-building involving human behavior and the purely physical phenomena you mention above. Since much of the important shit we need to cope with is the result of how other humans will behave (individually or collectively), how do we go about making such predictions?

  7. You have an extra “for example” in the following paragraph;
    “Sometimes truth claims are predictions about states of mind in other people – but even these are tested by observable consequences. If I say “Cathy likes chocolate,” for example, for example, you could check that claim by offering Cathy some and seeing if she smiles.”

  8. Why prediction and not simply identification? To start with prediction means, like it or not, you are bringing in a whole chain of knowledge that precedes prediction, including an implicit view of reality (i.e. that it is knowable and consistent). Before you can predict you must observe. Entities act, they act according to their identity. Truth is identifying entities and their attributes in a manner consistent with reality, that is as you indicate observable. Predictions and tests are crucial here, it’s how you can go on to integrate and discover principles and ultimately build a smart phone or fly to the moon.

    Epistemology and questions on are a very interesting topic, but if you back into metaphysics by starting there you are either a) implicitly bringing in a metaphysics or b) likely to end up anywhere examples of this latter abound throughout the history of philosophy.

    You seem to be saying that the purpose of consciousness is the identification of the things around you for the purpose of preserving and furthering your own life. If so I certainly agree.

  9. @Ben:

    “The Declaration of Independence was signed July 4, 1776.” I think we can agree that that is a true statement, but what predictions does it imply? That people will agree with you when you say it?

    Ah, but in this example, we’re in luck, for, we can go to the Rotunda and retrieve the actual Declaration of Independence, We can see the actual date on the document. Furthermore, we can look at other evidence markers, such as the kind of paper used, the handwriting style in use, the kind of ink used. We could probably also use some sort of radio-dating method on the paper or the ink. Furthermore, we can look at other historical records that will corroborate the evidence we see in the Declaration itself.

    Other historical truths may not be as easy to validate, but historians definitely work from evidence in this way. So the statement “the Declaration…was signed on July 4, 1776,” is most certainly testable and verifiable.

    @Carl/esr:

    For example, Cathy smiling when you give her chocolate can imply lots of things other than her liking chocolate; perhaps she likes *you* and smiles because she thinks you giving her chocolate is a sign that you like her, even though she hates chocolate.

    esr’s example isn’t a very good one — or perhaps it’s an excellent springboard for debate ;) Here there be dragons; the dragons of ‘psychology’ and ‘soft science.’ Even if Cathy doesn’t smile, it doesn’t mean she doesn’t like chocolate. She could just think you’re kind of creepy and is therefore is afraid to accept the gift of chocolate. Or, it could be that she loves chocolate, but she’s on a diet and angry and frustrated that you would even offer her chocolate at a time like this. There are lots and lots of reasons why Cathy may or may not smile when offered the chocolate; we can only know the truth of Cathy’s emotion by continued observation of Cathy. Even asking Cathy herself may not produce the truth of the matter, as people lie about their emotions all the time.

    It’s an important point because it tightly limits absolute truth (and reality) to things we *directly* observe; the rest are inferences and we can not be completely sure about their truth.

    Exactly, but complex problems like the Cathy/chocolate one we can break down into a series of a smaller, supporting truths, as in my text above regarding Carl’s Declaration of Independence example. With continued observation of Cathy, we can determine Cathy’s true mental and emotional state with regard to chocolate in context with other things we can learn about Cathy. Sure, this is a ‘soft science,’ meaning we can’t know for absolute certain, but we can make a determination within a range of statistical probability.

    Our inferences can’t always be proven 100% correct; even in the Declaration example: radio dating isn’t going to pindown the exact date, and we have to rely on corroboration from other documents that could conceivably be lying. Maybe there is a vast conspiracy to conceal the fact that the true date of the signing is July 5, for example. Not likely, but possible. And since it’s not really likely, we simply apply Occam’s Razor and say that since most evidence seems to point to the date being July 4, and there doesn’t seem to be much point in lying about it, the simplest explanation is that the date is, indeed, July 4.

    That being said, if you live your life needing to prove every specific “truth,” it could be that you’re obsessive-compulsive or just need to get a life. ;)

    ESR says: Morgan’s account of historical truth and the way we can analyze the Cathy/chocolate problem into smaller ones with observable consequences is exactly correct.

  10. Can you unpack this a little more? What about statements testable in theory but not practice, such as the current temperature on the far side of Jupiter? What about the truths of mathematics? Lastly, of course, what experiment corresponds to the truth claim, “every truth claim corresponds to a set of experiments”? Is this a different category of claim?

    Re-reading what I’ve just written, I guess that you want to divide statements into analytical and synthetic. Presumably then the statements of mathematics plus “every truth claim corresponds to a set of experiments” are simply true by definition. Is that about right?

  11. I was going to leave a big reply, but then I realized that would be uncharitable because this is a very short essay and can’t possibly deal with all of the objections I might raise. Also, it seems that many of the objections I was going to raise have already been raised by others. You have imported the problems of logical behaviorism as Carl points out. You also can’t account for a notion of historical truth as Ben points out because starting with the notion of predicting the future by definition ignores the past. And as Jakub and Nancy point out, it’s impossible to account for certain varieties of a priori truth such as mathematics and logic when starting with an empirical account of truth (I realize that I have oversimplified math and logic but you’re smart so you get the drift).

    In any case I’ll be interested in seeing how you deal with these problems in the future. The history of philosophy is full of the wreckage of epistemologists trying to account for all of this stuff, but hope springs eternal. Good luck!

  12. This sounds kind of like an attempt to summarize “The Simple Truth”; I’m guessing this because you just recommended another of Yudkowsky’s works, and the emphasis on a straightforward empirical view of truth looks like the same thing.

    It does provide a nice illustration of why Yudkowsky chose to use fiction; a naive view of truth–beliefs should imply expectations–is pretty obvious, but is vulnerable to nitpicking and “what about pure mathematics” and so on. Putting it in narrative form does a very good job of making obvious things obvious; it’s harder to argue against fiction.

  13. Just one thought: the question “what is truth” makes a big assumption, namely that the meaning of “truth” is something we find rather than something we define. You are entitled to make the claim that “truth” is something that effectively predicts the future. However, other people may very well define truth to mean something entirely different, for example, truth is what the Bible says, or truth is an accurate description of the putative objective reality, or truth is what makes me feel good about me. Pick your poison.

    The problem is, as always in philosophy, that words don’t have very precise meanings. You claim that “truth” is such an such, but the fact is that your definition of truth does not correspond well with what most people mean when they say truth. (I’d say the most common definition would be “an accurate description of objective reality”, which of course begs the question.) It is generally accepted that words mean what the majority of people think they mean — this is after all the basis of human communication.

    That is not to say that there is no utility to your philosophical description, no doubt there is, I found it interesting to think about. However, I am not sure you have the right to define what “truth” is for other people. Perhaps using arbitrary names for your concepts might reduce their controversy. I think it is worth pointing out that jargon is useful because it can be defined more precisely within the community of expertise, and is not subject to the insults of the vagueness of human language.

    By using a word in common usage and redefining it, you are almost hijacking it to grab all the semantic net the word has and applying it to your new definition. To give a prescient example, it has become common for the Democratic party in the USA to claim that the Supreme Court overturning a law is “judicial activism.” They do this to hijack the negative associations of the expression, to neuter the term when applied to themselves.

    In that context I’d ask, why define “truth” why not define kwijibo, and have at it?

  14. Where does math fit into this?

    Mathematical statements are considered to be “true” or “false” given a prior set of assumptions. Consider Euclidean geometric truisms such as “the sum of the angles of a triangle is 180 degrees”, which break down in non-Euclidean geometries. Whole new frontiers of mathematics have been opened up by changing or refusing to make some of the fundamental assumptions underpinning existing theory.

    Ultimately, though, even mathematics has its feet firmly grounded in reality: we devised systems of measurement and counting in order to observe and manage objects in the real world. The power of mathematics comes from its isomorphism to reality; a mathematical statement like “2+2=4” could be taken as a predictive claim of the form: “If I had 2 apples and someone gave me 2 more, I would have four apples.” More complicatedly, theoretical physics predicted the existence of black holes with math, but it wasn’t until we devised ways of observing actual black holes that these mathematical statements became meaningful; i.e., we knew what to look for and could begin searching for observed phenomena that matched our expectations.

  15. The problem is, as always in philosophy, that words don’t have very precise meanings.

    I’d restate that to say that many words have many meanings, and different people define the same word differently; even the same person will define a word differently for different contexts. (I got horrible English grades for about a month after moving to a different school district that used different definitions for the words describing grammar, before I caught on to the changes.) It is therefore crucial when using a word that has many meanings to keep straight which meaning is used.

  16. # Jeff Read Says:
    > The power of mathematics comes from its
    > isomorphism to reality;

    Of course it is worth pointing out that mathematics is isomorphic to reality because it was designed to be isomorphic to reality — it isn’t like it was just a coincidence. This is of particular interest when you consider your point about the theoretical discovery of black holes before the observational discovery of black holes. Mathematics is designed to map to the realities of our day to day life on earth (though as it becomes more sophisticated, that is less so.) Isn’t it surprising that the mathematics isomorphic to our little rock are also, apparently, isomorphic to the mathematics of a location and time a million light years away from here? That seems to me to be a non obvious result, and one that makes me think.

  17. The Monster Says:
    > I’d restate that to say that many words have many meanings,

    What you are saying is correct, but it is also different than what I am saying. Words are like circles on the domain of semantics. Some words have multiple circles centered in different locations, (for example “I love you” and “I love ice cream” have their centers in different locations.) But I am referring to the radius not the center point. For example, in a recent discussion, Morgan and I disagreed over the meaning of the word “game.” I said a game was an amusement, or a competition with arbitrary goals and arbitrary rules. He pointed out that some Casino games were not very amusing. I’d say that, for example, professional poker is not a game, some people would say it is. Depends on how large the radius of your circle is.

  18. The problem is, as always in philosophy, that words don’t have very precise meanings. You claim that “truth” is such an such, but the fact is that your definition of truth does not correspond well with what most people mean when they say truth. (I’d say the most common definition would be “an accurate description of objective reality”, which of course begs the question.) It is generally accepted that words mean what the majority of people think they mean — this is after all the basis of human communication.

    I just read some debates on news.yc involving a Derrida[0] fanboy. My patience for this kind of thing is thin.

    Yes, there are many definitions of truth. This is an attempt to isolate the definition which works reliably for the vast majority of people. And lots of people use this definition every day, though they may not know it. If you’re a kid and your mother asks you “Did you clean your room” and you say “Yes”, she is not going to consult her bible, nor is she going to rely on the answer that works best for her. If she’s doing her job right the first thing she’s going to do is test the implicit prediction that when she goes upstairs she will find a clean room.

    [0] I personally think Derrida was a genius but not in the way most people understand him: he has all the earmarks of a troll who did it for the lulz, and devised a particularly virulent and effective strategy for trolling philosophers. :)

  19. >If anyone makes a “truth” claim at you that you can’t unpack into testable predictions, be careful.

    How do you see origins science with regards to this?

  20. # Jeff Read Says:
    > I just read some debates on news.yc involving
    > a Derrida[0] fanboy.

    Never heard of him.

    > This is an attempt to isolate the definition which
    > works reliably for the vast majority of people.

    Which is, as I said, something of great utility. However, what I am saying is “get your own word.” By using a very common word, and giving it a significantly different meaning, you are making a confusing mess where the implications of the old and the new get mixed up together.

    > If you’re a kid and your mother asks you
    > “Did you clean your room” [etc…]

    This fits in quite well with what I think is the most common idea of the word truth, namely that it is an accurate description of a supposed objective reality. Mommy certainly believes Junior has a room, for example. I’ll grant you that if Mommy took Eric’s meaning, she would have largely the same result, but I doubt very much that that is what she thinks “truth” is.

    Which is not to say that Eric’s definition is bad. If I have a map of Chicago from the 1950s, I might be able to find my way about. But the map does not represent what I mean by “Chicago.” Similar perhaps, but not the same. And, with the wrong map, I might very well end up in some bad neighborhoods.

  21. >Where does math fit into this?

    A mathematical truth claim is a prediction about the existence of a proof, that is a reduction of the claimed theorem to agreed-on axioms by means of agreed-on formal operations. Such proofs are often thought to be timeless and absolute (and I used to believe this myself) but they’re not, because all proof-seekers are fallible. In particular, an exhibited proof may be erroneous.

  22. >Presumably then the statements of mathematics plus “every truth claim corresponds to a set of experiments” are simply true by definition. Is that about right?

    Actually, you can unpack “every truth claim corresponds to a set of experiments” into an assertion that all other predications of “truth” eventually become circular or ungrounded. But it’s probably simpler to regard that as a definition.

  23. >This sounds kind of like an attempt to summarize “The Simple Truth”

    Well spotted. It’s actually part of an email conversation I’m having with Eliezer in which I’m proposing a nonfiction explanation that is much simpler and more effective than one he had written, which you can’t see as it’s part of an unpublished ms.

    I should add that I’ve been using this analysis for about as long as Eliezer has been alive :-)

  24. Never heard of him.

    Jacques Derrida, a 20th century French philosopher credited with the establishment of the technique of deconstruction. Postmodern types — the sort of people most often seen objection to rigorous definition of terms — are fond of him.

  25. >Just one thought: the question “what is truth” makes a big assumption, namely that the meaning of “truth” is something we find rather than something we define.

    Jeff Read did a pretty good job of answering this objection. I’ll add that some definitions of “truth” work (in that they help you cope with the shit that happens) and some don’t. If you are content to play games with your map, you can argue endlessly over the definition of “truth” with no consequences. If you want to cope with the shit that happens in the territory, you’re going to end up with a definition equivalent to this one.

    The equivalence may be disguised by language and logic errors. For example, you appear to believe that “objective reality” is a sort of irreducible prior concept that has to appear in any account of “truth”. This is wrong, though it’s one of the least bad ways to screw up. When you talk about “objective reality”, you’re confusing the largest generator of mutually coherent predictions in your mind with your phenomenal field.

    (I’d unpack that last sentence, but I think you’re smart enough to do it yourself, and you’ll learn more about the issue if you do.)

  26. >When you check a prediction, is all you can get another prediction?

    No. Eventually you get a sense datum you can check, a direct observable.

  27. I’m surprised no one has mentioned Logical Positivism or Verificationism. This post (to my admittedly casual ears (eyes?) ) sounds like a summary of those philosophies.

    (Unless, of course, this is so blindingly obvious that it didn’t need mentioning, or I completely misunderstand what you’re saying)

    Logical Positivism (from what I could glean from “Language, Truth, and Logic”) separates all propositions as either a) verifiable via experiment (where “experiment” is fairly loosely defined, as esr points out) or b) tautologies (all mathematical statements fall into this category) or c) complete nonsense, neither true nor false.

    I did find it a bit weird that the logical positivists regard the entirety of mathematics as one great big system of tautologies, on a par with “All bachelors are unmarried”. You can know these sorts of truths a priori, because they don’t refer to anything in “experience”. Their truth is derived from the meaning of the words/symbols. But I guess it does make a certain amount of sense. The fact that some tautologies can be surprising and unexpected is simply attributed to a limited imagination on our part.

    1. >I’m surprised no one has mentioned Logical Positivism or Verificationism. This post (to my admittedly casual ears (eyes?) ) sounds like a summary of those philosophies.

      Yes, it would have that effect. Those two theories differ mainly in edge cases that my simple exposition deliberately avoids. But if you want to associate it to a school with a history, “instrumentalism” would be better.

      A significant difference from all three of these schools is the adaptationist part of the account. Logical positivism got itself in trouble because its theorists wanted a pure correspondence theory of truth, disdaining motivation. But there’s a class of objections about “theory-ladeness” (by Hilary Putnam and others) that cannot be answered within that frame. To answer Putnam you need to reinstate the idea of direct observables, and you need a principled reason to do that.

      It was actually that mad old Nazi bastard Heidegger who solved this problem, but his language was so difficult that most analytical philosophers haven’t realized it was a solution yet.

  28. It’s nice to see someone who really understands the scientific method the way even few scientists these days do. (And the Harry Potter fan fic you linked to is *spectacular*. I think I’m only on Chapter 6 or so, but I keep giggling hysterically at it in fits of happiness.)

    I was so sure that when I went to college, I’d finally meet a group of smart people to talk to that understood this. Then grad school. Then silicon valley. But no. Very few people understand it at all. Science is just a prop to use in their culture war and they wield it badly. Or to squeeze money from the taxpayers.

    (And I feel quite validated finding out that no, reducing homocistine levels does nothing for your heart. I can’t believe they convinced people it would based on *nothing* other than a correlation.)

  29. @esr:

    Actually, you can unpack “every truth claim corresponds to a set of experiments” into an assertion that all other predications of “truth” eventually become circular or ungrounded.

    That’s a very insightful observation; it’s one I’ve also had, but have failed to put into words so succinctly. I’ll bet you could make some cash as a writer. ;)

  30. But I am referring to the radius not the center point.

    That isn’t awfully problematic, however. The existence of “pink” doesn’t invalidate the concepts “red” and “white”; it helps to clarify that they define a continuous rather than discrete phenomenon.

    Statements that apply to most “games” but not to all would still be truthful, but only when understood in their proper context. When people attack the notion of objective truth, it’s often by dropping context or the artificial construction of “anti-concepts” — words that represent sets of existents that have no conceptual common denominator to warrant assigning them a concept.

  31. JessicaBoxer: That is not to say that there is no utility to your philosophical description, no doubt there is, I found it interesting to think about. However, I am not sure you have the right to define what “truth” is for other people. Perhaps using arbitrary names for your concepts might reduce their controversy.

    Jeff Read’s reply was right on the money. Non-empirical ways of knowing sound pretty on paper, but when the chips are down and the kneecaps are busted, you’re going to want the good stuff. People who use the word “truth” to refer to something other than “the thing that lets you know some guy just broke your kneecaps” should get a different word. Yes, I know they were there first.

    The problem with these other ways-of-knowing is that they’re ludicrous to apply in toto; they only seem to apply when the believer wants a get-out-of-jail-free card to avoid believing unpleasant. We’re all empiricists; some of us just punch fewer holes in our empiricism.

  32. It looks more like a definition of knowledge (as opposed to fallacy) than a definition of truth (as opposed to untruth).
    So you are basically hijacking the term truth to emphasize that it’s not the territory but the map is that counts.

  33. esr: I’m proposing a nonfiction explanation that is much simpler and more effective than [“The Simple Truth”]

    I wish you the best of luck, but I have doubts on how effective such a presentation could be. Judging by the feedback in this thread, a fictional treatment of the subject is clearly an emergent phenomenon with special causal powers; the fictional elements seem to act as a bung keeping the magic in the bucket.

    Then again, I’m assuming that someone with a wacky view of truth would read “The Simple Truth” and be enlightened, because it seems to address the standard objections in a way that makes them appear obviously foolish. But, then, I already agreed when I read it, and it didn’t include a heap of comments below arguing against a naive view of truth.

    1. >I wish you the best of luck, but I have doubts on how effective such a presentation could be.

      Er, you read my intentions wrong. Eliezer worked up a nonfiction definition that this is better than. His was almost unbelievable roundabout, haring off into Tarski schemata and coherentism and a lot of other unnecessary elaboration.

      I agree that the fictional version is better propaganda. But it’s better propaganda only because it’s subtly dishonest – it’s using a rhetorical strategem to cozen people into not noticing the edge cases and formulating objections. Sometimes, when you’re trying to accustom people to a new way of thinking, it’s OK to do this and backfill later. But that’s not good enough for what I was trying to do.

  34. Why do we value truth? Only for the practical use of it – being able to influence the environment by making predictions about it?

    Well the scientific revolution began in astronomy – finding truths about the movements celestial bodies of which Galilei and co. could have absolutely no hope of influencing. It was more like curiosity – or as in the case of Kepler, simply an aesthetical attraction to beautiful Platonic systems. Kepler’s problem with the Ptolemaian system was that it it made celestial bodies revolve not around the center of Earth but slightly off it. And such an assymetry he as a good Neo-Platonist just could not stomach – he needed another theory that would make things revolve in a beautiful spherical around the center of something. To the Neo-Platonist, god just does not make ugly, assymetrical stuff and that was actually their main argument until Copernicus, who was the first empiricist i.e. the first actual scientist amongst them. To the modern mind it is really astounding how Galilei didn’t give a damn about empirical evidence. He wrote tide works in 24 hour cycles because of the revolving of the Earth. Some folks who actually saw a see told him that tide actually works in 12 hour cycles. Galilei’s answer was pretty much “whatever, probably just the sea was very shallow where you were”.

    It means that curiosity and the seeking of aesthetical beauty in theories should not be undervalued. It is a great irony in the history of science that Galilei’s opponents, the geocentric Jesuit astronomers were methodologically speaking better scientists because they based their theories on actual observations, althogh very crude ones (“If I am on a moving wagon, I can feel I am moving. How comes the Earth moves and we cannot feel it?” As dumb as it may sound in 2010 this attitude was actually closer to an actual scientific approach than all that fantasy about the perfect spherical paths of heavenly bodies which, by a strange irony of history, actually proved to be closer to and a precursor of the actual truth.)

    1. >It means that curiosity and the seeking of aesthetical beauty in theories should not be undervalued.

      You’re right, of course. These are the normal operating noises of our adaptive machinery. When they’re absent, the meta-strategy of truth seeking is not being improved as rapidly as it should be.

  35. ESR,

    what do you think about that stuff that there is a distinct and different “natural” and “human” realms of truth?

    http://en.wikipedia.org/wiki/Wilhelm_Dilthey#The_Distinction_between_the_Natural_Sciences_and_the_Human_Sciences

    http://en.wikipedia.org/wiki/Lifeworld (I mean the stuff by Husserl)

    As Scruton have put it, we make predictions about natural events which are based on causes, but we make decisions based on reasons regarding on our life – hence they are about two very different things. I find it very convincing. If you see a ball rolling down a street you will explain it by cause which is rooted in the past of the ball – but if you see a man walking down the street towards the library, you will explain it by reasons and decisions which are rooted in his future: he wants to get there.

    Also, what do you think about Murray Rothbards criticism of empiricism with regard to the human/social sciences: he said if you repeat and experiment with different people then you have lots of confounding factors, while if you repeat it with the same people you have the problem that they have learned from the previous ones.

    1. >what do you think about that stuff that there is a distinct and different “natural” and “human” realms of truth?

      Two words: special pleading.

      >Also, what do you think about Murray Rothbards criticism of empiricism with regard to the human/social sciences:

      Methodological problems with experimental design are a different thing from fundamental problems in confirmation theory.

  36. A mathematical truth claim is a prediction about the existence of a proof, that is a reduction of the claimed theorem to agreed-on axioms by means of agreed-on formal operations.

    I assume given your background that you know Gödel’s first incompleteness theorem. Are you abandoning the principle of the excluded middle here and classifying Gödel’s sentence as meaningless?

    (For the non-mathematicians here: Gödel’s sentence’s is the proposition “this proposition has no proof”. If a formal system is consistent, Gödel’s sentence must be true but unprovable.)

    1. >I assume given your background that you know Gödel’s first incompleteness theorem. Are you abandoning the principle of the excluded middle here and classifying Gödel’s sentence as meaningless?

      I used to be a metamathematician, and could write a detailed analysis of this particular pathological case, but I’m just not going to go down that rathole. I’ll just hope you know what is meant by “constructivism” and leave it at that.

  37. Panvan> “The Declaration of Independence was signed July 4, 1776.” I think we can agree that that is a true statement, but what predictions does it imply?

    Let me preface this with: I’ve done no study in this area other than my own thinking, so I’m sure Eric and others will stomp all over me on this one and I’ll deserve it :^).

    In my internal model of the world, I’ve always separated “Truth” from “Fact”. Fact is what is and immutable (a “datum” to use Eric’s words, but I feel that word is inappropriately limited in scope some how). “Truth” is what people take away from fact, and is there for (at least partly) subjective in a POV way.

    So, the solution to your problem is that “The Declaration of Independence was signed July 4, 1776.” is a statement of fact (or at least would be by adding “in Philidelphia, PA”. The truth is coincident to that fact if you are in the same frame of reference (lets say, on the correct side of the international date line as a completely contrived example, but might be a bit different if you’re on the other side of midnight).

    This distinction allows me to try (and as Jessica Boxer and others will attest, not always succeed ;^) to see things from somebody else’s point of view and try to use that to reverse engineer their own internal model of the world. In doing so, I’m doing as Eric asserts, in that I’m attempting to find the “truth” about how the other person thinks, and so this allows me to make a prediction on how they will react to different but similar stimuli.

    Anyway, that’s my $0.02 (well, not sure it’s worth that much).

  38. @dgreer:

    The statement itself is a testable prediction, which is why I neglected to address the question of “what predictions does it imply?”

    Actually, it’s not a true statement at all. I simply pretended it was to avoid writing a dissertation about the veracity of the statement. According to the Library of Congress, July 4 was the date the Continental Congress approved the document, but the document was actually signed by most of the delegates on August 2.

    Notice my last statement. Now compare the statement to the original: “The Declaration of Independence was signed July 4, 1776.” What’s the main difference? Besides the facts presented in them, that is?

    The difference is that that they don’t make the same statement. The first one says “The Declaration…was signed July 4, 1776,” while the second says “According to the LIbrary oF Congress…the document was signed…on August 2.” Each makes a testable prediction, but the second statement is a somewhat weaker claim; I haven’t made any claim beyond saying that the Library of Congress says….

    In other words, any claim that can be objectively verified is a testable prediction. Stronger claims require stronger proofs; likewise, weaker claims require weaker proofs — In this case, a simple hypertext link is sufficient. ;)

  39. “All truth claims can be unpacked as predictions.” So far, so good (and I generally agree) – but how do we evaluate the veracity of a prediction? Your definition of verification or falsification makes a radical difference in what “truth” you end up defining.

    Does one failed experiment doom a truth to falsehood? Does it take repeatability, and of what kind? Do we vote (and who, and how)? How do you evaluate the veracity and competence of the experimenters? How do you adjudicate disagreements in the evaluation of an experiment? Are there ways to impeach an outcome?

    (A different way to put this: The outcome of an experiment is itself a “truth” in the sense of your proposal.)

    So I’d say that so far, you haven’t really furnished a useful definition of “truth”. Tell me your rules for “checking if those predictions are true,” and we can do business. :)

    What audience are you planning to write your book for?

    Cheers
    — Perry

    1. >Does one failed experiment doom a truth to falsehood? Does it take repeatability, and of what kind? Do we vote (and who, and how)? How do you evaluate the veracity and competence of the experimenters? How do you adjudicate disagreements in the evaluation of an experiment? Are there ways to impeach an outcome?

      These are all excellent questions, and I suspect you know sound answers as well as I do, but they are beyond the scope of the introduction I was writing. The actual goal was to show that (1) theory-building is motivated behavior, and (2) doing confirmation theory before epistemology or ontology is a simpler way to bootstrap your way to a sound naturalistic philosophy than reversing that order as is normally done.

      That second point is one that really irritates me. Academic philosophers persist in getting it wrong, and they’ve been out of excuses for this error for at least fifty years now.

      >What audience are you planning to write your book for?

      General audience. The “book” is only an idea at this point; that text was in the nature of an experiment to see how simple I could make an exposition of some basic analytic philosophy.

  40. >…doing confirmation theory before epistemology or ontology is a simpler way to bootstrap your way to a sound naturalistic philosophy than reversing that order as is normally done.

    This sounds a great deal like saying “developing a testing strategy and testing tools before you start programming is a simpler way to bootstrap your way to sound code rather than reversing that order as is normally done.”

    (Not trying to be too off-topic)

  41. doing confirmation theory before epistemology or ontology is a simpler way to bootstrap your way to a sound naturalistic philosophy than reversing that order as is normally done.

    IOW, we shouldn’t start with definitions or exploring what is considered to be “known,” but instead we should make testable predictions, set out to prove them, and then connect that with definitions, classifications, and “existing knowledge” later? Sort of a “shoot first, ask questions later policy.” Obviously, as you point out, this deserves further exploration, perhaps starting with why this should be so; I think Yudkowsky’s essay “The SImple Truth,” demonstrates this best: the fictional society basically had no decent, epistemology or ontology, yet our intrepid hero was able, through experimentation, to arrive at a realistic conclusion.

    1. >IOW, we shouldn’t start with definitions or exploring what is considered to be “known,” but instead we should make testable predictions, set out to prove them, and then connect that with definitions, classifications, and “existing knowledge” later?

      Yup. Well put.

      >This sounds a great deal like saying “developing a testing strategy and testing tools before you start programming is a simpler way to bootstrap your way to sound code rather than reversing that order as is normally done.”

      Ha. Yes, the analogy is apt and not misleading.

      >I did recognize it as Pragmatism, so I get partial credit.

      Yes. Peirce was like Dijkstra’s famous remark about Algol: a great improvement over most of his successors.

  42. > And, BTW, good for you you if you recognized this as Peirce crossed with Heidegger with a dash of Husserl.

    I did recognize it as Pragmatism, so I get partial credit.

    A nearly-forgotten but very important book that needs to be read widely again is “Evolution and the Founders of Pragmatism” by Philip Wiener (1949).

  43. I’ve always defined “truth” as the accurate perception or conception of reality. The accuracy determination fits with your testing modality. In addition, I’ve always defined “reality” as that which exists regardless of what we think of it. I frequently assert, to the dismay of my children, that “reality remains as it is regardless of what you think about it.”

  44. Actually, I like your first definition of, “Truth is what makes the future less surprising, which is more general than your later claim about predictions. It works even where there is not enough information to predict the future. It is also related to Marc Stiegler’s point in David’s Sling about sometimes all that is necessary in addressing a potential problem is recognizing that it is possible, that way if it does happen you can just get on with dealing with its consequences without also having to fight disbelief that it happened.

  45. I like this article. You effectively start from “I think therefore I am” and build a powerful working definition of reality on top of that. Which is impressive given how little we really know about the nature of reality.

  46. Another practical definition of reality – “Reality is what trips you up when you walk around with your eyes shut.” From one of L Neil Smith’s novels, probably Forge of the Elders, it sounds sort of like Sam the dog.

  47. “I also like: reality is that which refuses to go away when I stop believing in it” – that would be bad news, because it would mean lots of important things in our lives from professional success through love affairs to kicking an addiction or bad habit would not be parts of reality.

  48. > […] The actual goal was to show that (1) theory-building is motivated behavior, and (2) doing confirmation theory before epistemology or ontology is a simpler way to bootstrap your way to a sound naturalistic philosophy than reversing that order as is normally done.

    I quite agree with you there. Science *works* as a pragmatic machinery for developing successful theories. It works whether we have a satisfying epistemology for *why* it works or not. (It is evolutionary knowledge, after all.) I might consider a philosophy-first approach if it won an Occam’s-razor contest with science (whose practice is pretty ad-hoc and messy, after all), but I haven’t found any.

    > That second point is one that really irritates me. Academic philosophers persist in getting it wrong, and they’ve been out of excuses for this error for at least fifty years now.

    An academic philosopher tends to want philosophy to rule over science. This means epistemology must not be contingent on uncontrolled feedback from the environment, or you’ve just demoted your livelihood from “queen of the sciences” to “engineer of reality checks.” (Nothing wrong with that, of course – but if that’s what you want to be, what are you doing in a philosophy department?)

    >>What audience are you planning to write your book for?
    > General audience. The “book” is only an idea at this point; that text was in the nature of an experiment to see how simple I could make an exposition of some basic analytic philosophy.

    Well, there is the “here’s an interesting idea” kind of book, and then there’s the “here is how you might want to order your life” kind. For the former, have fun. For the latter, you’ll have to come up with a reason why *for the reader* a “truth” notion based on validation is superior to one based on consensus or authority. The problem is that for a majority of the populace, this may not objectively be true…

    Cheers
    — perry

  49. “he fictional society basically had no decent, epistemology or ontology, yet our intrepid hero was able, through experimentation, to arrive at a realistic conclusion.”

    I don’t want to take stands myself but for a balanced view I would like to point out a rather powerful critique of empiricism I found (PDF): http://inform.nu/Articles/Vol8/v8p189-210Mende.pdf

    Excerpt:

    “Science is characterised by the existence of deductive explanatory theory.
    Yet inductive empirical methods are inappropriate for creating deductive explanatory theory.
    So inductive empirical methods are unlikely to produce a science.
    The empiricist doctrine restricts researchers to inductive empirical methods.
    So it impoverishes research by inhibiting progression to scientific status. “

    1. >Science is characterised by the existence of deductive explanatory theory.

      That paper is a lot of tut-tutting and impressive verbiage wrapped around nothing much.

      What’s wrong with Mende’s case in a nutshell; he’s got a deduction fetish, he’s mistaken means for ends. Sure, deductive logic is very useful in science. But the hard part is constructing a map that has two properties: (a) it’s a reasonable description of the territory, (b) deduction is possible within it. Construction of that map is necessarily an inductive, empirical process, refuting Mende’s ridiculous claim that “empirical methods are inappropriate for creating explanatory theories”. Once you get past this stage deduction is relatively easy.

      It is true that induction alone won’t do it. There’s an element of imagination, of creative synthesis that’s required. My single favorite example of such is Kekule’s dream of snakes biting their tails that led him to the cyclic structure of benzene. But this isn’t deduction either – rather, it’s a sort of holistic pattern recognition. Mende excludes that middle, selectively misquoting better thinkers than he is to read induction and empiricism out of the game simply because they aren’t the entire process of theory building. He then throws everything else into a bin labeled “deduction”, whether it belongs there or not.

      Honestly, I could write a better critical paper on the philosophy of science than this. At least I’ve actually had a moment of Kekule-like insight and originated a generative theory of something, even if it was only how open-source development works; I’m betting Mende hasn’t. Watching him divagate about scientific discovery is like watching your dried-up virgin aunt theorize about sex.

  50. “For the latter, you’ll have to come up with a reason why *for the reader* a “truth” notion based on validation is superior to one based on consensus or authority. The problem is that for a majority of the populace, this may not objectively be true…”

    I would agree with that. Eric likely would wind up preaching to the choir (other Rationals); but nevertheless such a book may function as a useful primer for young scientists, engineers and businesspeople, where it can do a great deal of good.

    “It is true that induction alone won’t do it. There’s an element of imagination, of creative synthesis that’s required…But this isn’t deduction either – rather, it’s a sort of holistic pattern recognition. ”

    In layman’s terms, either you see it or you don’t. Holistic pattern recognition is exactly it.

  51. We can distinguish between at least two kinds of relevant theories, corresponding roughly to distinction between “the equations” and “the boundary conditions” in typical physics theories. This causes difficulties when applying ESR’s “a more concise way of putting it is that every truth claim corresponds to a set of experiments” formulation to theories of the first type, like Newton’s laws of motion; Lakatos wrote at some length about this.

    We can also distinguish two types of questions to ask about the theories we want to arrive at using inductive reasoning: “what algorithm should we use to search for the best theory (hopefully better than just enumerating all theories in ascending order of length)” and “given that we have found two candidate good theories, which one is better?” Some of the comments here seem to mix those two questions together, e.g., JB talking about creative synthesis and holistic pattern recognition. Appealing to creative synthesis or holism for the search algorithm may be OK if you like, but that’s not so much a philosophical problem as an engineering problem. The problem that philosophers have much to say about is about which of any two theories are best.

    It has long seemed to me that one can resolve the difficulties pointed out by Lakatos by (1) using Occam/MDL ideas to rank complete-including-the-boundary-conditions theories in order of preference and (2) interpreting the value of equations-without-boundary-conditions theories (e.g. Newton’s) in this framework as roughly “if you are allowed to treat Newton’s theory as a predefined library subroutine, the minimal programmatic description of your complete-including-the-boundary-conditions theory will be shorter.” But I got a bad grade for this idea in my philosophy of science class, so who knows?:-)

  52. I’m not sure that mathematical statements are about the same sort of objective, observable reality as statements of material fact…but rather statements of logic? In practice, misapplying mathematics is one of the most efficient ways to go wrong with certainty. As Einstein said, “As far as the laws of mathematics refer to reality, they are not certain; and as far as they are certain, they do not refer to reality.” (A bit of Wittgenstein here, maybe?)

    Or is the argument that this is a bad distinction, and truth is predictive power, full stop?

    1. >Or is the argument that this is a bad distinction, and truth is predictive power, full stop?

      Yes. For discussion of the issue of the applicability of mathematics to perform deductions about empirically-observed systems, see my essay The Utility of Mathematics. Pay particular attention to the part where I point out that the choice of a mathematical formalism to apply to a physical system has to be inductively justified. Such induced theories are subject to predictive check – in fact, that’s the only way they can be checked.

  53. I think it’s best to regard formal truths and empirical truths as separate concepts entirely, joined only by an unfortunate overloading of the word “true”. This overloading is an unfortunate historical accident, but it’s an accident that’s about as likely to go away as conventional current is.

  54. I think it’s best to regard formal truths and empirical truths as separate concepts entirely, joined only by an unfortunate overloading of the word “true”. This overloading is an unfortunate historical accident, but it’s an accident that’s about as likely to go away as conventional current is.

    An accident? It’s just that English doesn’t have two words for “true” meaning “honest, ethical, pure” and and “true” meaning “empirically-observed, correct, etc.” I don’t know if that’s an historical accident or just the nature of the language. Any linguists care to answer that question?

  55. Honestly, I could write a better critical paper on the philosophy of science than this. At least I’ve actually had a moment of Kekule-like insight and originated a generative theory of something, even if it was only how open-source development works; I’m betting Mende hasn’t.

    “The folly of mistaking a paradox for a discovery, a metaphor for a proof, a torrent of verbiage for a spring of capital truths, and oneself for an oracle, is inborn in us.” — Paul Valery

  56. esr:
    the choice of a mathematical formalism to apply to a physical system has to be inductively justified

    An illustrative example well worth the time for anyone how hasn’t ready it already:

    Maciej Cegłowski had a brilliant blog entry a couple of months back detailing how the British Navy found a fool-proof cure for scurvy in the 18th century, then completely lost it in the 19th because they were never able to put the correct theoretical framework to their experimental and engineering know-how.

    Quote:
    Now, I had been taught in school that scurvy had been conquered in 1747, when the Scottish physician James Lind proved in one of the first controlled medical experiments that citrus fruits were an effective cure for the disease. From that point on, we were told, the Royal Navy had required a daily dose of lime juice to be mixed in with sailors’ grog, and scurvy ceased to be a problem on long ocean voyages.

    But here was a Royal Navy surgeon in 1911 apparently ignorant of what caused the disease, or how to cure it. Somehow a highly-trained group of scientists at the start of the 20th century knew less about scurvy than the average sea captain in Napoleonic times. Scott left a base abundantly stocked with fresh meat, fruits, apples, and lime juice, and headed out on the ice for five months with no protection against scurvy, all the while confident he was not at risk. What happened?

  57. Jeff Hawkins in his theory of the brain characterizes the neocortex as a memory-prediction system. Intelligence then flows from this function, corresponding nicely to ESR’s truth and reality definitions.

  58. Truth is what makes the future less surprising.

    I like this. It ties in very nicely with my own personal system of morality, which, like your description of truth, has a few complications and wrinkles wrapped around a very simple core statement. My core statement is that morality derives from long-term critical thinking. In general, the longer the term and the more accurate the thinking, the more moral the person. For example, it’s hard to think about events 200 years from now without thinking at least a little bit about your personal effect on the environment and on those people around you.

    That psychopaths are often immoral is at least partly explained by their broken risk/reward calculations:

    Psychopaths also have a markedly distorted sense of the potential consequences of their actions, not only for others, but also for themselves. They do not, for example, deeply recognize the risk of being caught, disbelieved or injured as a result of their behaviour.

    By this definition of morality, it’s not surprising that criminals often have lower than average IQs. While it doesn’t take a genius to figure out that if you get caught you will go to jail, the Dunning-Kruger effect insures that a lot of criminals have an unrealistically high estimation of their own ability to avoid capture.

    So, for me, it’s a good working definition. But just like people nickel-and-dime you on your definition of “truth,” many people take umbrage at the assertion that they are just being good to avoid negative effects. There’s also the unending discussion about morals vs. ethics. Here, you have to remember that most moral codes actually have the benefit of a lot of critical thinking by multiple individuals over the course of centuries or millenia, so that often, depending on whose definition is used, morals and ethics are the same at the limit.

    1. >My core statement is that morality derives from long-term critical thinking.

      This would be a better definition of ethics. That is, it approximates both folk and academic usage of the word “ethics” better than it does folk and academic usage of the word “morality”.

      Actually, what you’ve done is rediscover a philosophical position called ethical egoism. I’m one of these myself.

  59. Actually, what you’ve done is rediscover a philosophical position called ethical egoism.

    Certainly looks similar in many perspectives. Maybe I missed it, but when I was researching this previously (a few years ago), I didn’t notice anything about “long-term”, especially about post-lifetime long-term. That may be a meaningless distinction (because you can certainly read that into ethical egoism), but there’s nothing in my (admittedly limited) reading about ethics that talks about this. To me, the exceedingly long-term and exceedingly well thought-out positions are what makes morals and ethics the same “at the limit”.

  60. >Where does math fit into this?

    >> A mathematical truth claim is a prediction about the existence of a proof, that is a reduction of the claimed theorem to agreed-on axioms by means of agreed-on formal operations. Such proofs are often thought to be timeless and absolute (and I used to believe this myself) but they’re not, because all proof-seekers are fallible. In particular, an exhibited proof may be erroneous.

    The status of truth of a proof as defined by common arbitrary rules does not change. A proof is static; once committed to paper and viewed through the lens of the defined rules, does not become erroneous. It’s the perceived truth (the truth claim?) of a proof *by the proof-seekers* that is variable.

    To put it another way, a program written from the language definition exactly is correct. If it won’t compile (or if an program disallowed by the language rules compiles), that’s an issue with your toolchain, not the program.

    1. >A proof is static; once committed to paper and viewed through the lens of the defined rules, does not become erroneous.

      Unfortunately, this is not true. It ceases to be true once proofs reach a level of complexity that is above that which a single gifted mathematician can understand. A famous recent case was the characterization theorem for finite groups, a *huge* thing stitched together from contributions by twelve different mathematicians. No individual understands all of it and no individual (neither a human being or a mechanical proof checker) has verified all of it. There could be a fatal error in there that is beyond the effective capacity of that group of mathematicians to find.

      The point here is that for theorems with proofs that large, “proved” becomes semi-definite because the proofs are so difficult to check.

  61. > The point here is that for theorems with proofs that large, “proved” becomes semi-definite because the proofs are so difficult to check.

    Isn’t that just an issue verifying the truth claim, though? Or do things not have meaning if we don’t presently understand them? What if we have better methods for checking large proofs in the future? Temporal relativism just seems weird.

    1. >Isn’t that just an issue verifying the truth claim, though? Or do things not have meaning if we don’t presently understand them? What if we have better methods for checking large proofs in the future? Temporal relativism just seems weird.

      Yeah, I know. Mathematicians hate it. They like to think of mathematical proofs as timeless absolutes. And if we had infinite computational capacity and never made mistakes, they would be.

  62. > What if we have better methods for checking large proofs in the future?

    “Better” in this case does not imply “perfect”, and never will.

    There’s always a non-zero chance of human error, and even if that chance seems small, you can never ever calculate those odds. Mathematicians who believe in perfect proofs and religious people who believe in divine revelation are no different. Both need to admit that there is at least one weak link in any belief they have: themselves.

  63. ESR,

    There is still something that’s bugging me about this theory. What about the descriptive sciences? What about that guy – I forgot his name – who is a member of half a dozen academies but never ever had any sort of a theory, prediction or hypothesis in his whole life: he simply discovered and described about 500 species of ants?

    Of course this sort of stuff is at the root consisting of predictive and testable statements – but of millions of little ones, instead of a few big ones, which makes it too costly to formulate them in a predictive manner. Thus this sort of descriptive scientist does not really make hypotheses or does not put his observations in a predictive form, he basically just writes “here is what I have observed, go check it yourself if you don’t believe it”.

    Isn’t it possible that being exposed to different kinds of sciences tends to have the effect of having different biases about science as such? And the different kinds are different mainly in “a few really big hypothetical questions” vs. “millions of tiny observations without much theory behind them – just describing the way things are”.

    I mean, you were exposed to a lot of physics in your youth and physics is very typically a “few very big hypothetical questions” field, thus you can be biased in the direction that science is all about predictive hypotheses.

    I went to a high school where half of the class consisted of biology freaks who knew the Latin name of every weed and bone. Biology is very typically a “only a few theories, but millions of little facts which simply describe the way things are, without much hypothesising about them” kind of field, and thus my bias is that science is plain simply information in systematic form: the main difference between my biologist classmates and fishermen was that the stuff fishermen knew about fish was less systematic, less categorized. Thus I have a very different attitude to science, my attitude, my bias, based on this exposure to the strong focus on categorizing plants, animals, bones with Latin names into various categories, often a tree structure, is that basically if someone puts anything into a systematic form, making neat categories and sub-categories, I’m tempted to think it may be true, because it sounds much like science – the kind of science I was exposed to.

    1. >There is still something that’s bugging me about this theory. What about the descriptive sciences?

      There are at least two possible answers to this question, one methodological and one social.

      The methodological answer is that a classification (indeed, any sort of predication about the observable world) is a bundle of predictive claims. That is, if you tell me how to recognize two different species of mammal, you also bundle in (for example) a prediction that two individuals of those species won’t breed in the wild. You’ve noted this yourself. The fact that classifications are not normally expressed in terms of the predictions they make doesn’t deprive those predictions of their importance.

      Socially…just because it’s the business of science to make predictively testable claims doesn’t t mean that every scientist gets to be the author of a predictive theory. Some people just collect observations for the same reason some people just dig ditches; it has to be done, and if it’s what you’re best suited for it’s where you’re more likely to end up. There are places in science for obsessive-compulsives as well as for imaginative synthesists.

  64. Fair enough. The reason for me asking this question is that I’m actually not very interested in the natural sciences, much more in the humanities/soft/social sciences kinds, anthropology, history, sociology, philosophy etc.

    In these fields there is always the problem how far can you stretch the definition of science and where do you step into the fields of pure speculation. Which would not be a big problem in itself, but there is a much bigger one: in the natural sciences, you can play it methodologically safe and still talk about something relevant and interesting. Not so in the humanities. Methodologically safe sociology for example is unbearebly boring and irrelevant to anything that’s actually an important human interest, it’s the kind of stuff no one reads without getting paid for it. Actually interesting and relevant social sciences always test the boundaries of science a bit – they are always a bit suspicious of being partially speculation. Max Weber, Mircea Eliade, Eric Voegelin etc.

    Thus it is very crucial for the humanities how science is defined. The definition as generating predictive maps – many of them would fail that definition Evidence-based, verifiable information expressed in a systematic form – that test is much easier to pass.

  65. BTW. Isn’t it interesting that science have always won its biggest victories in the fields where the questions themselves are easy and uncontroversial? The easy and uncontroversial question of “Just what the hell is things made of” brings you everything from quantum physics to molecular biology, the similarly easy and uncontroversial question “just why do celestial bodies move the way they do” brings you everything from Galilei to general relativity and so on. When the questions are harder to formulate, are more controversial etc. we are looking more into philosophy and less into science – and run into the problems I mentioned above.

    1. >When the questions are harder to formulate, are more controversial etc. we are looking more into philosophy and less into science – and run into the problems I mentioned above.

      Beware temporal bias. Questions that seemed philosophical and intractable two centuries ago (like, say, how living and nonliving things are different in nature) are firmly within known science now. Two centuries from now, expect questions we now think of as philosophical to be within science. I think the nature of mind is one that seems particularly likely to crack during the next century.

  66. Eric,
    I enjoyed reading your post and it made me happy to find out that I had reached to the same conclusion as you on what is oft called “theory of truth”. The affinity between Peirce and Heidegger is surprising. They share a great many notions, though employ a different language to talk about them. What Heidegger calls “the hermeneutic circle”, for instance, is already presupposed in Peirce’s account of the process of meaning-making or semeiosis. What Peirce and Heidegger basically share is an anti-Cartesian approach to knowledge and truth, which is also based on a phenomenology (or as Peirce put it, phanersocopy) of ordinary life.
    Affinities can be found with many naturalist philosophies, but myself a libertarian, I’d like to point out a very interesting one.
    Pragmatism stems from the doctrine of spontaneity first exposed by Alexander Bain, a 19-th century philosopher influenced by the Scottish Enlightenment. Hayek (who applied the notion of spontaneity to social orders) himself based his gnoseology on that of the philosophers within the Scottish Enlightenment (like Hume, Smith and Ferguson), and explicitly rejected “continental” socialism on the grounds that is was a projection of the Cartesian metaphor to society.
    So Hayek, Peirce and Heidegger go hand in hand, and it is no surprise that we libertarians find ourselves comfortable with them, since that is perfectly consistent.
    Cheers,
    Pablo

  67. While there is a Human Will, a desire to survive and a accomplish things, there is actually no “Truths or Falsehoods”.
    Even in Physics, we have no idea what an electron is nor do we understand which way it flows whether it be from Positive to Negative or the reverse. We do not understand the concept of time, light, electromagnetism, gravity, or even the human emotions of a sobbing woman.

    You can judge how much we know about ‘Truth’ by how much we know about anything which is nothing…

    1. >Even in Physics, we have no idea what an electron is nor do we understand which way it flows whether it be from Positive to Negative or the reverse.

      You typed your babbling nonsense on a machine that demonstrates that we do in fact understand electrons in the only way that matters – we can make them do things for us.

  68. What about epistimological dependencies such as a priori vs. a posteri? I think the argument provided only accounts for a posteri knowledge since the validation of a given truth is dependent upon the experience of testing the truth out (i.e. Give Cathy chocolate). Does not knowledge contain truths that according to your “theory” need to be tested out?

    1. >What about epistimological dependencies such as a priori vs. a posteri?

      I’m not sure what you mean by this. Explain?

  69. >What about epistimological dependencies such as a priori vs. a posteri?

    >I’m not sure what you mean by this. Explain?
    What I was trying to articulate was that I could know that 2 + 2 = 4 is true without having to prove it by experimentation or experience. I think the example you gave was that a posteri in that I would have to go outside to experience “wetness” before I could validate that the theory of “it is raining outside” to be true. So I’m thinking that you could prove a theory without having to do it by experience. If that makes sense.

  70. >No. Never. “A priori” knowledge is available only in formal systems with no actual content. See my essay The Utility of Mathematics for discussion.

    Will do. So it sounds like your conclusion is that any truth claim that cannot be validated through experience is false and therefore not a truth?

    1. >So it sounds like your conclusion is that any truth claim that cannot be validated through experience is false and therefore not a truth?

      No. Any “truth” claim that cannot be confirmed or disconfirmed by experience is meaningless. It cannot rise to the level of false until you unpack it to a set of predictions which are then disconfirmed.

  71. Are you saying then that is true that “truth is what makes the future less surprising”? In that case, how can we actually use that truth to make the future less surprising? Philosophical discussions sometimes accept some things as ‘true’ and then proceed to develop an argument dialectically (though I don’t think we could call this ‘math’ nor ‘formal’), are these meaningless?

    What about phrases like ‘Venus is hot’ pronounced by a mayan on 1000BC where there was actually no way to check it? Was that meaningless before but true now?

    1. >Philosophical discussions sometimes accept some things as ‘true’ and then proceed to develop an argument dialectically (though I don’t think we could call this ‘math’ nor ‘formal’), are these meaningless?

      No. They’re ways of expressing contingency – truth in a possible universe, which may or may not be this one. The meaning is “If predictively verifiable premise A were true, then we would expect predictively verifiable consequence B to be true”.

      >What about phrases like ‘Venus is hot’ pronounced by a mayan on 1000BC where there was actually no way to check it? Was that meaningless before but true now?

      Some claims which are in principle confirmable are not in practice confirmable. Such claims are not meaningless; they’re like contingent claims. “If we could measure the temperature of Venus…” They’re interesting because they might be confirmable someday.

  72. > The meaning is “If predictively verifiable premise A were true, then we would expect predictively verifiable consequence B to be true”.

    What I meant was that some times in philosophical discourse, some basic premises are taken as apriori knowledge (ie. I think, therefore I am). Even a sentence like “All truth claims can be unpacked as predictions” is basically a definition, or axiom, or something accepted for practical purposes, or how can that sentence be predictively proved?

    > Some claims which are in principle confirmable are not in practice confirmable. (…) “If we could measure the temperature of Venus…”.

    I was thinking in that some phenomena might seem outside the reach of men at a given time (I don’t know if it’s the case, but for culture where planets were regarded as gods, I suppose considering that you could actually go to Venus and measure something could be similar to say ‘If we could see love, it would be pink’, but it might be nonsense), and in that case the meaningfulness of the sentence might change with time.

    Another thing you could mention in the post, is that under that definition, truth is temporal, since the next experiment might disprove something that was considered true until then. That’s basically the definition is what Popper would have given for truth (I think Popper’s truth was something more stable, and for him propositions were not actually true, but instead ‘closer’ to THE truth than others, but is similar), though I think he would differ with the definition of ‘meaningless’ (for what I mentioned on the first paragraph).

    One last point I found confusing was where you said ‘It may be that you don’t understand the claim but it’s still true’. That gives the idea that the claim might be true, even when you are unable to make predictions yet (for ignorance in that case), but it can’t, since it would only be a truth once you tested that the predictions are correct (otherwise truth would exist independently of our experience, but I think that’s not your idea).

    1. >Even a sentence like “All truth claims can be unpacked as predictions” is basically a definition, or axiom, or something accepted for practical purposes, or how can that sentence be predictively proved?

      As you said, it’s a definition. One can’t “prove” it in any a priori sense, only observe that other attempts to define “truth” end up being circular or ungrounded. One ends up asking how else might “truth” be defined and showing that othere definitions fail to the extent they’re not equivalent to the predictive criterion.

      >Another thing you could mention in the post, is that under that definition, truth is temporal, since the next experiment might disprove something that was considered true until then.

      That’s certainly the case. Truth is a semi-definite predicate; you can know when a truth claim has been falsified, but it is usually not possible to know that it will never be falsified.

      >One last point I found confusing was where you said ‘It may be that you don’t understand the claim but it’s still true’.

      Certainly. It might be expressed in a language I don’t speak – Sanskrit, or homological algebra. The relevant question is not whether I can make predictions from it, but whether it is in principle possible for someone to make predictions from it – that is, whether it is possible to construct a confirming or disconfirming experiment.

  73. But then “All truth claims can be unpacked as predictions” is not a truth? And if is not a truth, is it meaningless? Or is there something else in the middle?

    > The relevant question is not whether I can make predictions from it, but whether it is in principle possible for someone to make predictions from it – that is, whether it is possible to construct a confirming or disconfirming experiment.

    Ok, I suppose that when you say truth then you are talking more about a quality, and not saying the sentence is true but the sentence is truth? ‘The moon is made of cheese’ is certainly false, but you say it’s ‘truth’ in the sense that I can make predictions from it, and once checked, see if it’s true or false, right? I think I made the mistake of mixing ‘truth’ with what actually is ‘true’ in your exposition.

    Thanks for all the replies! Didn’t want to argue much (or take too much of your time, I already owe you too much time for using your amazing software through the ages), it just that at first glance it looked as a too scientificist look at what ‘reality’ is, with which I don’t feel specially comfortable, and had to write something ;).

    1. >But then “All truth claims can be unpacked as predictions” is not a truth? And if is not a truth, is it meaningless? Or is there something else in the middle?

      It’s a definition. You’re free to use a different definition of “truth”, but you’ll find that if you try you’ll end up thrashing, fighting your own terminology, and probably ending up with a hidden or not-so-hidden circularity in your system.

      >Ok, I suppose that when you say truth then you are talking more about a quality, and not saying the sentence is true but the sentence is truth?

      I think this question is a pretty good example of fighting your own terminology. I did write “truth is what makes the future less surprising”. True statements make the future less surprising. False statements make it more surprising. If you believe “The moon is made of green cheese” you will be surprised when someone shows you a moon rock. On the other hand, if I say “The moon is made of thwuddle”, this claim is neither true nor false until I give a referent for “thwuddle”.

      >at first glance it looked as a too scientificist look at what ‘reality’ is, with which I don’t feel specially comfortable

      Ah, so that’s your actual problem. Well…as long as you continue looking for a non-“scientificist” account of reality I predict that you will waste your own time and that of anyone who listens to you.

  74. What about the laws of logic? They are purely abstract and you cannot even ask the question, “What is truth?” without using them. How would you ever prove them to be true given your reasoning around empiricism? How would you “prove” the law of non-contradiction given your argument?

  75. “every truth claim corresponds to a set of experiments”

    To what set of experiments does this truth claim correspond?
    Does it not entail that if such a sset of experoments were possible that it would require knowledge of *all* truth claims? If it does not then how does it verify the quoted truth claim by its own standard?

    1. >To what set of experiments does this truth claim correspond?

      That is exactly the interesting question about any truth claim. To answer it, you have to inquire into the meanings of the individual terms within individual truth claims.

  76. > I think this question is a pretty good example of fighting your own terminology.
    I was referring to that the phrase ‘It may be that you don’t understand the claim but it’s still true’, doesn’t really make sense for me, since for it to be _true_ (and not just meaningful) it should have been contrasted with experience, and if I can’t understand it, I can’t contrast it; maybe you could say it’s true once someone somewhere actually checked it, but that would be odd in such a big universe, or once someone tells me he contrasted it… So my conclusion was that by ‘truth’ you were maybe referring to a quality, a synonym of meaningful perhaps, since the phrase does make the world less surprising _once_ you contrast it with experience (and if it’s evaluated to true or to false is not relevant, it just tells you a different thing that will remove some surprise from the world).

    > as long as you continue looking for a non-”scientificist” account of reality I predict that you will waste your own time and that of anyone who listens to you
    I don’t _look_ for such an account, but I’m not willing to accept, given our current understanding of complex phenomena, theories that, based strictly on experience and logical reasoning, exclude other means of experimentation that might open new insights on how approaching ‘truth’. Maybe sometime we can explain things that we now classify as ‘mysticism’ in a scientific way, but we can’t do that now, so accepting that even those things that we can’t really explain follow the same rules as those that we can, seems like limiting the breadth of exploration. And closing the door to things we cannot check by experience as meaningless, removes any relevance on justifications made on moral or personal grounds (until we can ‘scientificize’, sigh, ethics), leading to places I don’t want to visit.

    1. >I don’t _look_ for such an account, but I’m not willing to accept, given our current understanding of complex phenomena, theories that, based strictly on experience and logical reasoning, exclude other means of experimentation

      What “other means of experimentation” are there? If you’re checking hypotheses by experiment you’re doing science; you may not be doing it formally, and your execution may be flawed, but you’re still doing science.

      >Maybe sometime we can explain things that we now classify as ‘mysticism’ in a scientific way, but we can’t do that now, so accepting that even those things that we can’t really explain follow the same rules as those that we can, seems like limiting the breadth of exploration.

      You don’t seem to be distinguishing between (1) allowing for the possibility of different rules for ‘mystical’ phenomena than we presently understand, and (2) proposing that the methods by which we confirm truth claims need to be changed just because the label ‘mysticism’ has been attached to what we’re studying. The former position is reasonable; the latter is just a way to confuse yourself.

  77. It is your truth claim to defend. It is just assumed to be “true” for the entire post. If it cannot account for itself then I must ask: how you know it is true?

    1. >It is your truth claim to defend. It is just assumed to be “true” for the entire post. If it cannot account for itself then I must ask: how you know it is true?

      Because it produces the results we want from a definition of “truth”.

  78. Because it produces the results we want from a definition of “truth”.

    It doesn’t produce the results that I want, namely that it cannot account for itself.

    1. >It doesn’t produce the results that I want, namely that it cannot account for itself.

      Then you lose. No definition of “truth” can do that, any more than you can haul yourself up by your own bootstraps. Definitions that pretend otherwise rely on smuggling in a conclusion in some way.

  79. esr: Because it produces the results we want from a definition of “truth”.

    defectivebit: It doesn’t produce the results that I want, namely that it cannot account for itself.

    esr: No definition of “truth” can do that…

    So the reason you offered for knowing that your definition of “truth” is true is actually false, since no definition can (by your admission.)

    1. >So the reason you offered for knowing that your definition of “truth” is true is actually false, since no definition can

      You’re playing cute games with the map, but I don’t care – I’m only interested in the territory. I’ll leave you to your sterile tail-chasing.

  80. That is fine. I was just trying to discover if you had a valid reason for believing your method was consistent with itself. I simply applied the method back to itself to see if it did what you claimed it did and demonstrated that it cannot.
    Cheers and g’day.

  81. >That is fine. I was just trying to discover if you had a valid reason for believing your method was consistent with itself. I simply applied the method back to itself to see if it did what you claimed it did and demonstrated that it cannot.
    >Cheers and g’day.
    Good example of relative truth. Eric only cares about truth so far as it remains consistent with his presuppositions. I prefer to go the route of what Metallica sang in “Through the Never”, “..pursuit of truth no matter where it lies.”.

    1. >Eric only cares about truth so far as it remains consistent with his presuppositions.

      What nonsense is this? The whole point of starting with confirmation theory rather than ontology is so you don’t have to have presuppositions. Well, other than “I have a phenomenal field.”

  82. > What “other means of experimentation” are there?

    Art has a big part of just doing, experiencing, without seeking rational justification, and in the processes it can find things that ‘feel’ right or better, you cannot define them (or at least every definition is lacking), and there are no explicit / conscious hypothesis under them (and yeah, you cannot logically explain this to someone that has not experienced something similar, that is why is not rational!). For example, stage players talk about finding ‘truth’ (their word, not mine ;)) on the stage, which is not defined as reproducing something realistically, but reaching a given state on stage where emotions are transmitted with full strength to the public. There’s no known recipe to reach it, but when it happens, almost everyone involved feel the impact. Something similar happens on other types of art.

    Of course, in the future, we might find a perfectly scientific explanation, but we don’t have it now, and, at this time, understanding (which slowly gets you better at replicating it) comes not from rational thinking but from experience itself, which helps you get a better grasp of that ‘quality’, even if you cannot verbalize it entirely. It’s might be a different kind of learning, a different kind of knowledge, and as such, might need a different conception of truth. This is as far as it can get from scientific experimentation.

    My last post for an interesting discussion, otherwise this thread will never end ;). Thanks for keeping up with the replies.

  83. >What nonsense is this? The whole point of starting with confirmation theory rather than ontology is so you don’t have to have presuppositions. Well, other than “I have a phenomenal field.”
    But doesn’t ontology need to precede confirmation theory? How can anything be made intelligible without starting with presuppositions? Isn’t that putting the cart before the horse?

    1. >But doesn’t ontology need to precede confirmation theory? How can anything be made intelligible without starting with presuppositions?

      I don’t need to know what ontological kind dogshit is to scrape it off my shoe.

      Ontology is like any other sort of theory; it’s a way to tie observed causal regularities into bundles in order to lower the complexity overhead of reasoning from experience. Experience comes first, but too many philosophers rush to do ontology too early out of bad habit.

  84. Slightly off topic: Can you recommend an introduction to philosophy that isn’t complete nonsense? I’ve attempted to read a few philosophical texts in my life only to conclude that the authors were wasting my time.

  85. What about truth claims involving faith? “I believe God exists because I feel Him in my heart.”

    Is that a prediction? Would you say that the speaker is predicting internal warm fuzzy feelings?

    But of course this is impossible to verify for an outside observer. I suppose the speaker is also predicting future events in the afterlife, but there is no way for living people to verify.

    1. >Is that a prediction?

      No. It’s actually a statement about what the speaker thinks constitutes observational evidence, not in itself a truth claim. The implication is that if the speaker did not have the feelings he points at, he might not believe in the existence of God.

  86. OK but allow me to play devil’s advocate (no pun intended) and point out that ~90% of the human race would parse it as a truth claim.

    What about simply: “God exists.”

    How is that phrase structurally any different from “Rain exists”, one of the examples from your article?

    1. >What about simply: “God exists.”

      In isolation, meaningless because it doesn’t point at any consequences of the existence of God. Normally, however, it is embedded in a discourse in which consequences are drawn from the existence of whatever kind of God the speaker has in mind. Those consequences may unpack to predictions or they may not. When they do not unpack to predictions, they are meaningless. Or theology. Whichever. :-)

Leave a comment

Your email address will not be published. Required fields are marked *