In two earlier posts now, I have expended energy on trying to clarify exactly what is wrong with the "simulation argument"—not because I think it is particularly likely to persuade anyone, nor because it would have any consequences if it did, but because it is a kind of epitome of faulty reasoning. If we can manage to pinpoint exactly where it goes astray, therefore, we can use what we have learned to defeat any number of other sophistries. More specifically, it will enable us to think our way out of that species of theological argument that seeks to work upon our sense of wonder at the inherent "implausibility" of our particular universe.
Having made the earlier arguments, I don't intend to revisit them fully here. I have already made the case to the best of my ability that the core of the argument's fallacy is to be found in the "principle of indifference." I will, however, lay out these arguments in abbreviated form once again, so that we don't need to turn back to those earlier attempts.
To recapitulate, the simulation argument wishes to persuade us that we are almost certainly living inside a computer simulation. How do we know this? Because future generations very likely will possess technology advanced enough to create a perfectly realistic simulation of reality, including lifelike imitations of the past, and the possible forms of this artificial past are infinite. So, we have before us an infinite number of hypotheses in which our experienced reality is nothing but a computer simulation. But we have only one hypothesis that we could describe as the naïve realist view (which most of us hold day-to-day in practice)—namely, that the objective world exists outside of us and matches our conception of it.
Very well, we think. We don't disagree. It is indeed possible that we are living in a computer simulation. All of modern philosophy from Berkeley, Hume, and Kant on agrees that we have little if any way of knowing the state of external reality (the "thing-in-itself") and whether or not it matches our conception of it. Any experience has already passed through our conceptual apparatus before becoming intelligible to us, and therefore whatever is taking place outside our minds—and which may or may not have made the impression that led to our perception of it—will be forever unknowable. We cannot say anything about its attributes, or even whether or not it exists independently of us.
Confronted with this kind of skeptical reasoning, most of us throw up our hands and conclude that this is either a problem without solution, or else it is really no problem at all. We can, with Kant, pursue a transcendental solution that is really not a solution (since it bites the bullet by accepting the skeptical premise); along similar lines, we can invoke the techniques of the analytic philosophers and the logical positivists to say that our experience of an object is simply what we mean by the concept of its existence, and that to think of it existing or not existing outside our minds, or any other reference to the realm of the numinous, is simply an inconceivable bit of nonsense. Or more simply, we say that we cannot, by definition, know what exists beyond our minds, and therefore need not concern ourselves with it.
The "simulation argument" wishes to push us further, however, than a mere acknowledgment of this ultimate unknowability. Remember, it argues that we are almost certainly living in a computer simulation. This is because, in order to determine the probability of our living in the naïve realist world, as opposed to the simulated one, we need to set one as a ratio of the other. As Hume once explained his concept of probability, we determine the relative likelihood of various outcomes by conjuring images in our minds of each possibility—and the outcome with the greater force, vividness, and number of examples on its side wins the day.
Recall that the argument has already claimed that there are an infinite number of hypotheses in which our world is a computer simulation. Whereas there is only one possible outcome in which the world outside our minds really exists, and looks and behaves in exact resemblance to our conceptions of it (i.e. the naïve realist position). Thus, if we are trying to set a ratio between them, in order to determine the probability of our living in a realist world, we find that it can only be something like 1/infinity; which is to say zero. (In a more technical formulation, as the number of possible computer simulation hypotheses is endlessly multiplied, our probability function for the realist position (f(x)=1/x) approaches zero asymptotically.)
Okay, so what's wrong with that line of reasoning? Let us not dwell on the fact that precisely analogous arguments could be used to prove virtually any other skeptical hypothesis. There are an infinite number of possible worlds in which we may be merely brains hovering in vats (they could be vats of innumerable different shapes and sizes, after all); an infinite number of possible worlds in which we exist as merely figments in the minds of other creatures or of a supreme Being; and so forth. Thus, each of these skeptical hypotheses is "certain" by precisely the same logic as the computer simulation one—and yet they are also mutually exclusive of one another.
When we have innumerable arguments that are equally "certain," yet which by definition cannot all be true, we begin to suspect that we have perhaps deployed a faulty sense of the word "certain" in reaching our conclusions. This consideration, however, can only tell us that the simulation argument is false; it does not precisely inform us why it is wrong, or how we know it is wrong. To get at this, as I have argued in the earlier posts, we must invoke probability theory; specifically the principle of indifference.
In order to illustrate this, however, I think it will be helpful to once again point out the resemblance between the simulation argument and many versions of the cosmological argument and/or argument from design found in modern religious apologetics.
These lines of reasoning, such as we find for instance in the "fine-tuning" argument, wish to impress us with the conviction that there is something inherently implausible about the world in which we find ourselves, given that (as modern physics avers) it operates according to a set of physical laws and constants that are extraordinarily fine-grained and precise (being calculated out to many decimal points), and if any change in this mechanism were to be effected (if any constant were altered even at the seventeenth decimal point, let's say), the entire system would collapse.
To put the "fine-tuning" argument into simulation-ese: we might say that we are faced here with a number of possible hypotheses. In one hypothesis—that of scientific naturalism—we simply "happen" to live in the one universe governed by this incredibly precise set of cosmological constants (we're just lucky that way). In the other possible scenarios, any one of these constants could have been ever-so slightly different, and the entire universe would have collapsed as a result. Trying to establish our probability ratio between the lucky world and the not-so-lucky non-existent worlds, therefore, we discover that once again, it is 1/infinity. Thus, the odds of our living in the fortunate, fine-tuned, and really-existent world are approaching zero.
Yet, we plainly do exist, and plainly do exist in such a fine-tuned universe that allows for our existence. The fact that we enjoy such a near-impossible outcome would therefore (on this reasoning) seem to be a miracle. And the fact of this miraculous occurrence points very strongly (the argument runs) to the intervention in our universe of something other than mere chance. In other words, it points to the existence of some sort of governing intelligence or providential arrangement that has worked to our good: that is (and this part is sometimes left for us to infer), it suggests the existence of a Deity.
The above reasoning might strike some people, for an instant at least, as compelling and irrefutable. We cannot get around the power of that now-familiar ratio: 1/infinity; nor fail to experience a sense of bafflement as we watch it inexorably approximate to zero.
Before being carried away from all our settled convictions, however, we should pause to consider the fact that we could construct a precisely similar probability function for literally every other event that has ever taken place.
Take the fact that I just sneezed while writing that last paragraph. I might have sneezed at literally any other nanosecond in the infinitely-divisible fabric of time. Thus, if I had been trying to predict the likelihood of my sneezing at exactly the moment I did, I would have had to set the value of sneezing at that moment to 1 and the value of sneezing at any other moment at infinity. Once again, our 1/x function will approximate zero. Thus, the fact of my sneezing at the particular moment I did is "impossible"; and the fact that I did, nonetheless, is therefore nothing short of miraculous.
Plainly, something has gone haywire in our conceptions of probability here. If every event that has ever taken place is actually impossible, then none of our other probabilistic reasoning has ever been correct, and we probably shouldn't be permitted to use the tools of this science in order to make any arguments about the structure of reality in the first place (thus trapping ourselves in a logical circle).
So where exactly did we go wrong? What is incorrect about the kind of probabilistic reasoning deployed in the simulation argument and the "fine-tuning" argument above?
To start with, we are faced in at least the latter of the two cases with something like the "anthropic principle." In other words, we could only engage in this kind of reasoning within a universe that exists; we could only live and think in a world that allowed for our existence; so the existence of our universe in accordance with some sort of physical laws and cosmological constants has to be taken as a given. The thought of non-existent worlds in which we might have been born is not available to us, and therefore cannot be used to multiply alternative hypotheses.
More fundamentally, however, we are also running up against the principle of indifference. The important thing to remember is that any given event will have more than one possible outcome. In this sense, each individual outcome is "unlikely," but it is also not particularly surprising when it happens, because some such outcome had to happen, and this particular one was just as likely as the others, and—in accordance with the indifference principle—it might as well have been this one as another.
To use the familiar example, suppose I were to roll a hundred-sided die. The odds are fairly low (1/100) that the number 32 would come up. But if I had not predicted such a throw in advance, and the throw had already happened, then the fact that I lived in such a world in which a hundred-sided die had been cast and 32 had come up, would not strike me as in any way remarkable. This is due to the indifference principle: I had no reason to prefer the number 32 to any of the other 99 options.
Of course, one can employ sophistries to try to provoke a false sense of wonder at this outcome. After having rolled the die and found that 32 came up, we could retroactively categorize our various potential outcomes in ways that make this seem remarkable. We could create a set of 99 possible outcomes, which we label "Not-32," all of which share only a single trait in common: namely, that they are not 32. Then we could compare the size of these two sets. We could say, "My goodness! How astonishing! We had two possible outcomes, here. According to one outcome, we would roll a 32. According to the other outcome, we would roll a member of the set of 'Not-32.' Our odds of rolling the former were 1%. Our odds of rolling the latter were 99%. How perfectly extraordinary, then, that we managed to roll that 1 in a 100 chance!"
Of course, that reasoning is absurd. We are plainly just toying with concepts after the fact in order to elide the fact that, before the throw was a fait accompli, we were perfectly indifferent to whether the outcome was 32 and any of the other 99 options. Now, if we had established our two-fold division before the throw happened, labeling a 32 option and an 'Not-32' set (as is done with chance predictions), and then the die really had landed on 32, that would strike us as rather remarkable. It would be, at the very least, an extraordinary coincidence. But this would only be the case because (in that hypothetical) we were not in fact indifferent to the various possible outcomes before we rolled the die.
Both the person making the simulation argument and the apologist using the "fine-tuning" reasoning are in the position of the person labeling the 'Not-32' set after the fact. To take the simulation argument, we have here a realist hypothesis (which we already acknowledged at the top—and which all modern philosophy would admit—is merely a hypothesis, and is not certain, and cannot even be tested or falsified in any way), as well as an endless variety of other possible accounts of reality. In truth, the realist hypothesis is indifferent. It is just as likely to be the case as any of the other non-realist hypotheses, and we have no reason to prefer any of them to it.
So too, with regard to the fine-tuning argument, we also have here multiple possible hypotheses. I would not say, in this case, that they are actually indifferent, due to the anthropic principle discussion above (the "non-existent" universe options were never available to us, let us recall, so we cannot count them as equivalent possibilities). Suppose we ignore this critical point, however, and we say merely that one hypothesis is the scientific naturalist one, and the others are the infinite number of possible worlds with slightly different cosmological constants that would have collapsed upon themselves.
There is no reason to place the realist or scientific naturalist hypotheses in one category, and all the others in a different set, now that—whichever one provides the accurate account—we are living in the state of the fait accompli. It should therefore not in any way surprise us or strike us as remarkable if the naïve realist account should happen to be correct, any more than it should if one of the computer simulation accounts turned out to be correct. Likewise, it should not be regarded as remarkable that we live in a universe with the precise cosmological constants that we do in fact have (however precise those may be), since we had to live in some such a universe, with some such constants, and it might as well be these as any others. (Indeed, if designer there were, they chose some strangely arbitrary constants, rather than pleasant round numbers.)
John Maynard Keynes, in his Treatise on Probability, explicates the point:
Uninstructed common sense seems to be specially unreliable in dealing with what are termed ‘remarkable occurrences.’ Unless a ‘remarkable occurrence’ is simply one which produces on us a particular psychological effect, that of surprise, we can only define it as an event which before its occurrence is very improbable on the available evidence. But it will often occur—whenever, in fact, our data leave open the possibility of a large number of alternatives and show no preference for any of them—that every possibility is exceedingly improbable à priori. It follows, therefore, that what actually occurs does not derive any peculiar significance merely from the fact of its being ‘remarkable’ in the above sense. Something further is required before we can build with success.
Or, to quote from a passage that I have cited before in these discussions, I find that a character in John Updike's novel Roger's Version has already pinpointed the difficulty in entirely nontechnical language. When confronted with a version of the cosmological "fine-tuning" argument, deployed by an evangelical graduate student, the Roger of the book's title offers the following casual refutation:
I do worry a bit about this concept of probability. In a sense, every set of circumstances is highly improbable. It is highly improbable, for instance, that a particular spermatozoon out of the millions my father ejaculated that day [...] would make its way to my mother's egg and achieve my particular combination of genes [...] but some such combination [...] was likely, and mine as probable as any other. No?
So much I have argued or hinted at before in the earlier posts, of course. We are going over familiar ground.
The reason why I wished to go through it all again, however, is that it recently struck me, on reading Hume, that there is yet another foundational problem with the species of reasoning of which both the simulation and the fine-tuning arguments partake. This is the fact that, even if all of their reasoning were sound, and the arguments were to be admitted, and if none of the refutations offered above were valid, then they would still manage to establish nothing positive whatsoever, in terms of propositional content. And an argument that establishes nothing convinces us of nothing, and therefore cannot be said to be persuasive of anything.
Suppose, that is to say, that we accept the simulation argument as valid. We have been convinced that we are certainly not living in a real and tangible world of external objects, but are instead living in a computer simulation. We then ask what we mean by a "computer simulation," in this sense, and what can be said about it with any degree of certainty.
We can say, for one thing, that this "computer simulation" created a fully convincing replica of a hypothetical objective reality. We can say, to whoever designed it, "job well done!" It perfectly maps onto our expectations for what a world would be like, if it really did exist outside of us.
What can we say of the designer of the simulation, beyond this? Do we know who they were? Were they humans from the future? Aliens? Machines? The Singularity? Do we know what their intentions might have been? Whether they had an intelligence like ours? Whether they are in fact a "they" at all, or merely an arbitrary or natural force, a happenstance of matter? Whether the simulation was in fact designed at all, or merely existed from the beginning of time? When we try to answer these questions, we cannot search—by definition—outside the subjective experience of the simulation itself. The simulation is all we have about our eyes; and it is offering us no clues as to its creators.
Very soon, we discover that we cannot possibly answer any of these questions, and therefore our saying that "we are living in a computer simulation" had no propositional content, and that what we really meant by the words "computer simulation" was simply the perceptual experience of reality, transmitted through the apparatus of the human mind—the existence of which no one ever did dispute or ever could dispute in the first place. So we are right where we started.
Suppose now we turn to the fine-tuning argument, and we are this time persuaded of its truth and honesty. We say, "wow! It IS remarkable [deploying our "uninstructed common sense," as Keynes would call it] that we happen to exist in this one universe, with this one very precise set of cosmological constants. How extraordinary! It must have required a miracle—some sort of intervention from outside the system—to make it possible in the first place. We can therefore say that the universe must have been created by some sort of force, entity, or Being who exists outside of it.
Having gone this far, and granted this much (which, of course, I am not willing to do—but, let us ignore this for the sake of argument), we then must ask: "very well, what do we know—based on the evidences of their creation—about this force, entity, or Being who made the universe?"
Well, we can say that that force, entity, or Being must have been a kind of force, entity, or Being capable of creating this particular universe. And... and...
Nothing else. We can go no further. There is nothing more to be known about it. All the evidence we have of it is enclosed within this one universe and this one perceptual apparatus available to us.
Of course, we could go on to speculate. We could say that this Being has personality and quasi-human traits. We could ascribe any number of attributes to it. But we would have no particular reason to back up any of these hypotheses, since all we know of it is that it is "that which created the universe."
Does this mean, admitting that this force is unknowable, that therefore one conception of it is as valid as another? No, because it is possible to form or argue for none of these conceptions, and to say merely that what we are talking about is "that which created the universe." And by the principle of logical parsimony, we are prohibited from saying anything further about it. We have no reason to say that it is a God or a personality or an intelligence or anything else. Any specific theory as to what it might be is multiplying entities needlessly.
Hume explains the point more fully, in refutation of various arguments from design, in Section XI of his Enquiry Concerning Human Understanding. Of course, the ever-wily and sly Hume is careful not to place these arguments into his own mouth. Rather, he invents a dialogue with a "friend who loves sceptical paradoxes," in which he himself is presented as the more conventional adherent to pious doctrines. Nonetheless, Hume puts the best arguments into the mouth of the friend, who puts these arguments in turn—by a kind of double transference—into the mouth of a hypothetical version of the historic Epicurus. In this voice, he argues:
When we infer any particular cause from an effect, we must proportion the one to the other, and can never be allowed to ascribe to the cause any qualities, but what are exactly sufficient to produce the effect. A body of ten ounces raised in any scale may serve as a proof, that the counterbalancing weight exceeds ten ounces; but can never afford a reason that it exceeds a hundred. If the cause, assigned for any effect, be not sufficient to produce it, we must either reject that cause, or add to it such qualities as will give it a just proportion to the effect. But if we ascribe to it farther qualities, or affirm it capable of producing other effects, we can only indulge the licence of conjecture, and arbitrarily suppose the existence of qualities and energies, without reason or authority.
Of course, these are unsatisfyingly narrow bounds in which to confine our positive assertions about the world. We would wish to know something further about metaphysical realities, and to ascribe something to them, without simply admitting that they are forever—and by definition—beyond our ken. But this discontent with the "limits of human understanding," as Hume called them—this desire to know more—does not in itself give us any reason to think that there is more, or that we have any means of finding it.
The best we can do is to adopt Hume's own approach: refute the pretensions of metaphysical arguments that wish to go further, when we encounter them (this jousting with the simulation argument is a humble attempt along these lines) and—the rest of the time—to concern ourselves with matters that lie closer to home; trusting to the presumption that we all of us, regardless of the speculations we entertain, will continue to live each day, and continue to behave, precisely as we would whether we were living inside a computer simulation or not.
No comments:
Post a Comment