The Imaginary Magical Super-Robot Theory of Truth

I want to refine what philosophers call the "correspondence theory of truth," articulating it as, what I call, the "imaginary magical super-robot theory of truth".  Sometimes the correspondence theory of truth is regarded as untrue(!) or false(!) or meaningless, but I maintain that this stems from a misunderstanding, or more accurately an ambiguity - or more accurately still, a few ambiguities.  There are ambiguities about what we mean when we speak of the correspondence theory of truth, and there are ambiguities about what we mean when we say that something is "meaningless."

Is the word "unicorn" meaningless?  No.  Nor is the word "dragon."  Unicorns and dragons don't exist, but that doesn't mean that these words are meaningless.  Similarly, "caloric" isn't meaningless, and it isn't meaningless to say "The atomic weight of mercury is 16."  That is a meaningful, yet false statement.  For a statement to be false is not at all equivalent to saying that it is meaningless.  (Indeed, there are some people who would insist that for a statement to be meaningful, it must be falsifiable.  I'm not sure that I would go that far, but it's much more sensible than claiming that any false statement is meaningless.  I would go so far as to say that if you can demonstrate that a sentence is false, you have simultaneously proven that it is not meaningless.)

Similarly, the phrases "unicorn horn" and "dragon breath" are not meaningless.  They are fairly well-defined and well-understood properties of beings that don't exist.  You may imagine unicorn horns differently than I do, but that doesn't render the phrase "unicorn horn" any more meaningless than the phrase "19th century painting."  When I utter the phrase "19th century painting," you may picture a Whistler while I'm picturing a Gaugin, which are indeed very different, but this doesn't render the phrase meaningless, and with a bit more discussion we will understand each other perfectly.  Similarly, if I say, "cat's paw," one person may picture the end of a cat's leg, while another imagines a tool for pulling out nails, and a third thinks of an expendable agent.  But this simply means that phrase "cat's paw" has more than one meaning - i.e., it is very meaningful, not meaningless.

The point here is that we can make meaningful utterances about imaginary beings and events.  Of course, everyone who has ever read fiction knows this.  It's only blockheaded wannabe philosophers who have trouble with the concept.

If you have followed me so far, then the next step in my argument is that just as we can imagine a dragon or a unicorn, we can also imagine a robot, which I will call the imaginary magical super-robot for reasons that will become clear.  We can imagine the imaginary robot a few different ways.  The imaginary magical super-robot contains either a certain finite (though very large) set of propositions about the world, or, more likely, some kind of function, or finite set of functions, for producing propositions about the world, which it stores in memory banks.  Each proposition is the first part of an (at least) two-part data structure.  For each of these propositions, the robot would then have an algorithm, a test or set of tests, to determine whether that proposition corresponds to a reality that is external to the robot's memory banks.  If the proposition does indeed correspond to an external reality, the robot stores a secondary piece of information in the second part of the data structure: the value "true".  If not, the robot stores the value "false".  Perhaps we can imagine that these would be the only possible results, or we can imagine more possible results: not just "true," and "false," but also "meaningless," "partly true," "I don't know," "still working on it," etc..  (I, for one, would at the very least like the robot to be able to employ, in addition to "true" and "false," categories such as "valid," and "invalid," "sound," and "unsound."  Perhaps this would necessitate data-structures that contain more than two parts, and indeed, we can imagine this getting pretty complicated, pretty quickly.  But you get the basic idea.)  In each case, the robot would compare the actual reality with its own representation of reality, and see if they match.  In fact, we can define "the truth" as "the set of propositions that an imaginary magical super-robot would determine to be true, according to its algorithm."  Notice that this is not one theory of truth, but many, many theories of truth - one for each algorithm that we can consider.

(Is it bothering you that the word "true" is in the definition?  Fine.  We can put it this way.  The robot sorts propositions about the world into two (or more) sets: set A, and set B (and set C, D, etc., if necessary).  We define truth as "every proposition the imaginary magical super-robot sorts into set A, according to its algorithm."  Satisfied?)

Now, maybe one day such a super-robot will actually exist, in reality.  Maybe it already exists.  Maybe an alien civilization built one, in some unknown corner of the universe.  But then again, maybe not.  Far more likely not.  Maybe humans will never build such a super-robot.  Maybe humans just aren't smart enough to build such a thing, and never will be.  Maybe no civilization anywhere in the universe will ever build such a robot.  Maybe.

Is it impossible that a super-robot could ever exist which would be capable of determining any truth from any falsehood?  Maybe.  I don't know.  You tell me.  If you have an argument that such super-robots are impossible, I'm all ears.  And if you have a knock-down, perfectly tight, sound, logical argument to demonstrate this, once and for all, I would be very interested to hear it, and I think that would be useful and vitally important information.  I would consider such an argument valid, and I would even go so far as to call it "true" (heh, heh).

(Here: if you're interested in producing a proof that a super-robot that can determine truth from falsehood is impossible, I'll get you started.  You might want to begin by thinking about Gödel's Incompleteness Theorem, and how a Turing machine is analogous to a finite set of axioms and an algorithm for deriving theora.  Of course, even if you manage to figure that out, that's hardly helpful, since the goal here is not to prove that a robot's list of truths will be incomplete, but rather to show that it can't distinguish any truth from any falsehood at all.   But it's a start.)

But even then - even if you were capable of convincing me that it is impossible for such truth-determining super-robots to exist in reality, because they would break the laws of physics or something, I would still maintain that the concept of "truth" as defined through this theory of truth is not meaningless, because even if these robots cannot exist in reality, they can still exist in my imagination.  I can still imagine these robots, even if you have proven that they can never exist in reality, and therefore the attributes of them, including the putative algorithm that governs them, is no more meaningless than the phrase "unicorn horn".  That is why the word "magical" appears in the phrase "imaginary magical super-robot theory of truth."  Perhaps such robots could actually exist in reality - in which case we would simply have a super-robot theory of truth.  But even if they couldn't exist in reality, that doesn't render the concept meaningless, and so the imaginary magical super-robot theory of truth is still meaningful.

It strikes me that many people try to reject the correspondence theory of truth by arguing that human beings are not imaginary magical super-robots and don't think about the world the way an imaginary magical super-robot would.  Well, yes, obviously.  Human beings are not imaginary magical super-robots.  Duh.  But that really has no bearing on the validity of the imaginary magical super-robot theory of truth.  It's completely beside the point.  

Human beings have biases.  That's true(!).  To me, that doesn't imply that humans will never be able to identify any truth about anything, but maybe it does, to you.  But either way, that really has absolutely nothing to do with the imaginary magical super-robot theories of truth.

So for instance, when Richard Rorty tries to argue against the correspondence theory of truth, in "The Mirror of Nature" and elsewhere, or when Hegel argued against "sense certainty" at the beginning of The Phenomenology of Spirit, especially as interpreted by Robert Brandom, who renders Hegel into a kind of post-Pragmatist, these seem to me to be arguments about the kind of consciousness of which beings like humans are capable.  They do not seem to be arguments about the possibility of truth-correspondence per se, categorically including non-human minds.  The advent of artificial intelligence research has fundamentally and permanently transformed the bounds of the conversation, not so much because of any new facts that have been discovered, but because this gives us a new vocabulary of speculation about the universe of possible minds.  (This seems especially fortuitous given Hegel's conclusions about "speculative science," a phrase that he uses in a much different, yet compatible way.)  It does mean that the argument of The Phenomenology of Spirit begins with a less-than-airtight immanent criticism.  Though the phenomenology remains valid, the imaginary magical super-robot theories of truth have not been eliminated as the beginning of a parallel trajectory for argumentation over the meaning of truth.

It is especially ironic (and not in the way he'd like) that a pragmatist like Rorty's argument against correspondence theories depends on the truth of human nature and the human condition, utterly ignoring minds in other possible worlds.  After all, without reference to truth, there is nothing separating us from other possible worlds.  If it is useful and clarifying to think about imaginary magical super-robots - and it is - then, from a Pragmatist perspective, they are just as real as protons.

Now, some people might find it questionable that an imaginary magical super-robot theory of truth is a theory of truth that bases itself on something that is admittedly completely imaginary.  They may think that this seems like a shaky foundation for an ontology.  But to my mind, this is not a bug, but a feature.  If a theory of truth depended on something to be true, this would be logically circular. It may or may not be true that a super-robot capable of sorting propositions into sets is possible in real life, but whether it is true or not really does not matter for the argument to work.  It may be true (if, say, aliens have developed such a robot) but even if it is true, this does not matter.   The truth of the situation is completely orthogonal to the strength of the argument.  The strength of the argument does not depend on the truth of the situation in any way.  And this is exactly the kind of argument we should be looking for as the basis for a theory of truth.  The meaning of truth should arise out of this imaginary situation, rather than the situation depending upon anything that is true.

Are there other theories of truth?  Probably.  But, precisely because the foundation of this kind of argument is purely imaginary, and does not itself depend on truth, imaginary magical super-robot theories of truth may be the least bad type of argument of which we are capable.  The strength of the argument is precisely that it is purely speculative.

That said, this is only the beginning of a theory of truth - the framework in which a theory of truth may be constructed, not the substance of the definition, which would be the robot's algorithm for sorting propositions itself.  A perfectly reasonable criticism at this point would be that if we can imagine a robot that sorts propositions one way, we can just as easily imagine a robot that sorts propositions inversely - that is, for any robot X, which sorts propositions into sets A and B, we can just as easily imagine a robot Y that sorts everything that X would sort into A into B and everything that X would sort into B into A.  Thus the meaning of truth that would arise from Y's algorithm would be the exact opposite of the meaning of truth that would arise from X's algorithm.  Thus it could be argued that we haven't solved anything - we've merely shifted the problem into a new form.  Fair enough, but establishing a clearer framework in which to consider the problem - one which avoids the pitfalls and foibles of the specificity of human experience - is, in itself, is an important step forward.

It might be further argued that any super-robot, since it is necessarily the product of human invention, will necessarily recapitulate all of the biases and other characteristics of human consciousness.  With this, I flatly disagree.  First of all, there's no reason to assume that this imaginary super-robot was invented by humans - it may well have been invented by aliens, or something else, as previously discussed.  Second of all, and more importantly, even if the super-robot were built by humans, the idea that it would recapitulate the characteristics of humanity is simply a false assumption.  After all, calculators are made by humans, yet calculators are not conscious the way humans are, do not think the way humans think, and do not make the same kinds of errors that humans make.  And there's no logical reason to think that they would.

But, to be fair, imaginary magical super-robot theories of truth may open their own can of worms.  Consider the following: let's imagine that there's a super-robot capable of successfully determining truth from falsehood.  Now consider the proposition: "A super-robot capable of successfully determining truth from falsehood exists."  We ask the robot whether this proposition is true.  How does it respond?  Well... it's a little ambiguous, isn't it?  I mean, this robot does exist - in our imagination.  But it doesn't exist in reality.  So how would it respond?  Would it think that it is real?  Well, clearly, it depends on how it is programmed - that is, what its algorithm would be.  Would we want to program an imaginary magical super-robot to know that it is merely imaginary?  Well, since its very function is to separate truth from falsehood, the answer is yes.  Thus, when we ask the imaginary magical super-robot whether it exists, the answer would most likely be "no."  

Which brings up some other questions.  Although a real super-robot might be able to determine truth from falsehood better than humans can, an imaginary super-robot is presumably limited to human imagination.  True enough.  But even though humans may never be able to build a super-robot capable of doing what this imaginary magical super-robot can do, we can build machines of greater and greater power, which could be thought of as very low-resolution simulations or emulations of such a super-robot.  And these can be thought of not only as valuable in their own right, but also as ways of training our imagination.  But this is a feedback loop, and a potentially unhelpful one.  The simulations of the super-robot, instantiated in the real world, would presumably model themselves on what our admittedly limited imagination of what that super-robot would be; and at the same time, our imagination would equally be trained on these simulations.  This presents no guarantee that we would get any closer to the imaginary magical super-robot.

That said, I don't think the above is a death-blow to the imaginary magical super-robot theory of truth.  After all, what the above argument ignores is the key to the way that any super-robot itself would zero in on the truth - namely, error correction.  Presumably the super-robot would be better than humans at getting to the truth, but certainly we can expect that it would be at least as good as humans, and therefore that it would employ something like the scientific method as human beings currently understand it.  Working day and night, faster than any human being can work, who knows of what a machine that uses the scientific method to determine the truth might one day be capable?