Jason writes about the existence of God, and one of the questions that arose is how one justified induction. Jason writes:
First, on induction: I disagree that I accept the utility of induction by means of faith. This inference seems to rest on a false dilemma, namely that we accept all propositions either by induction or by faith.
Other means exist besides these two. In the case of induction, I simply find that rejecting it leads to a sort of mental paralysis, one in which I can permit myself no thoughts at all about the external world. Worse, whenever I try to think this way, I always fail. Following Kant — I think — the inductive aspect of our reasoning may well be a priori. We’re stuck with it, whether we want it or not.
As such, we might as well make the best of it. I take this to mean putting my inductive inferences into the least contradictory order that I can. It’s that or I go back to paralysis.
This seems more like a pragmatic reason to act as if the world we experience is real than an apriori reason to believe. Apriori epistemic reasons are not the same as reasons we have to accept in order to avoid paralysis. After all, I could consistently say that it is not clear if the world we experience really exists but there are pragmatic reasons to treat it as such anyway. I will give such an argument later. Jason may as well concede that he has no argument to show that the probability that the word really is as or close to how we experience it is close to 1. And that is required in order to vindicate the common sense belief that the world we experience is real.
Veronica Dire writes:
Right. I consider “brain in a jar” or “Descartes demon” scenarios so far fetched to be irrelevant. They are curiosities to while away the time. Thing is, nowadays we know a fair amount about how brains work, how our senses work, how learning works — not everything of course; there is a long, torturous path between what we know and what we would like to know, and we may never travel the full distance. But we know a lot. Epistemology is today an empirical science. And “the thing in itself” is atoms and particles and complex forces, and we know that we can never perceive the full thing. Our abilities to measure are very limited. Our models are imperfect. Our theories go so far, but there must be further, some next layer of turtles we have not yet puzzled out. But to jump from this manifest ignorance to phenomenology, or worse, God, simply is not justified. Such is fanciful and nothing more.
She is just wrong about this. Evil demon or systematic deception scenarios are relevant and cannot be ruled out just because they sound strange. Consider how we would try vindicating such a move. If someone were to ask us why we think that it is unlikely that an evil demon is deceiving us, we would say that it is far fetched, i.e. it lies too far away from things which we have experienced. But this circularly assumes that we have till now not been deceived by evil demons. If an evil demon did in fact deceive us, we would not be aware of such deception. Thus, we cannot claim to have not encountered any evil demons without presupposing the conclusion that we are trying to justify.
Evidentiary Relation and Possible Worlds
The construction of these sceptical scenarios strikes at the heart of fundamental questions of epistemology. Consider the following principle/definition which everyone should accept.
Evidentiary Relation (ER): Some R counts as an epistemic reason for a proposition P if and only if R being true makes P more likely to be true and R and P are distinct.
Cases when R is identical to P, that is self evident propositions are justified differently.
Self Evidence: A proposition P is self evident if and only if its negation is logically impossible.
To illustrate how ER works, I will provide an example. For the purposes of the example, I will set aside the various sceptical scenarios
Suppose I experienced seeing five green books on a shelf. The fact that I experience seeing five green books on the shelf makes it more likely to be the case that there really are five green books on the shelf. It might turn out that I was miscounting the number of green books. Therefore, the fact that I experienced seeing five books is not dispositive. One way to spell this out is to use the possible-worlds framework. In some possible worlds, the shelf has only one green book. Of those possible worlds, the fraction in which I experience seeing five green books is very low or even zero. We can similarly describe probabilities of me experiencing five green books when there are other numbers of green books on the shelf. All counted in, we should expect that in most of the possible worlds where I experience seeing five green books, there are in fact five green books. This proportion goes up once we start excluding possible worlds where I am drunk, sleepy or cognitively impaired in other ways. In addition, there is, for lack of a better term, a “story” we can tell that accounts for how it can be that our experience of seeing five green books stands as evidence for there being five green books. In this case, it happens that there is some plausible causal story that we can tell. Quite plausibly, we can posit that any perceptual capacity must involve some causal mechanism by which information is transferred to the receiving perceptual structure. If there were no causal mechanism, it would seem impossible that the perceptual structure would be able to obtain information about the fact in question.
The only constraint on any account is that the account must entail that we have the experiences we think we have. Even when we relax the constraint against sceptical scenarios, our own subjective experiences still constrain possible accounts. We cannot be deceived into experiencing something we are not experiencing.
Each account is logically connected to our experiences. The confidence we ought to have in any one account just is the fraction of accounts occupied by that account. There are a number of upshots.
1. More specific accounts are less probable. This is to be expected as we would need more evidence to fill in details.
2. Accounts with fewer premises that are not inferred from any other premise are, all else equal, more probable. This is because the probability of a given account is equal to the product of the probabilities of its independent premises.This gives us a nice derivation of Occam’s Razor.
3. Things like gravity, strong force, weak force and electromagnetism are thus (at the minimum) logical objects we posit in order to make the content of our subjective experience more probable. In fact, positing the existence of such logical objects which entail our subjective experiences makes for a good explanation as possible worlds in which we merely happen to have the experiences we have without any uniting explanation are improbable for reasons mentioned in (2)
4. Once you posit the existence of such logical objects, the expectation that future events will look like past ones is almost trivial. The existence of those objects entails that certain future experiences will occur. That we are only highly confident about the existence of such logical objects just means that we will only be highly confident about what our future experiences will be. This solves the old problem of induction
5. It is possible that the new problem of induction could be solved as well. Given that both grue and green have the same phenomenological character before time t, the story that connects the wavelength reflected from the emerald to the perception of green/grue should be the same. However, the details of any such story should make it unlikely that the phenomenology would change at time t. If there is a story that logically links physical property to phenomenology, then there is no reason to posit a change in phenomenology absent some intervening cause. Any story about a grue object would have to posit some such intervening cause in order to spell out what a grue object would be. This could, depending on the explanations available range from being only slightly more complicated than explanations involving green or involve so many more moving parts as to be very nearly precluded. This largely depends on what the most plausible story about grue is.
Sceptical Scenarios and Possible Worlds
Till now, I have mostly bracketed away the possibility of evil demons, brains in vats and other sceptical scenarios. However, we can consider that for just about every sort of possible world in which the various objects posited are real, we can imagine conceptions of justice in which they are systematic deceptions of one sort or another. Even if such scenarios are more complicated than the realist ones, they are not so complicated that the realist option becomes almost certain. As I wrote in Jason’s previous post:
Brain- in- vat is merely a place holder for a whole class of theories in which we are systematically deceived by some mechanism. By eliminating narrative details, we can posit two worlds. W1 describes a possible world (or a class of possible worlds) in which we are not systematically deceived in some serious way. W2 represents a possible world (or a class of worlds) in which we are systematically deceived*. Let us now consider how many more moving parts are necessary for W2. Logically, we do not need to posit anything else in this world other than the content of the deception, the mechanism of deception and the person being deceived. This is at most two or three more moving parts. W2 is thus negligibly more complex than W1. Let me grant that the mechanism of deception must have at least as many moving parts as the content of deception. If I grant this, W2 has just twice as many moving parts as W1. Since it is the case that we are either systematically deceived or we are not, we get two equations
p(W1) + p (W2) = 1 …………………… (1)
p(W2) = p(W1) x p(W1) …………….. (2)
solving for p, p(W1) = 0.62
So while we have more reason to think that we are not systematically deceived than to think that we are, the reasons are not so strong as to warrant the near certainty with which we attribute to the “external world is real” proposition.
So, even if not quite equal in number, it seems that there is still approximately a 0.4 probability that we are deceived. This means that we should be closer to reserving judgment i.e. being indifferent between realism and anti-realism than we are to endorsing realism.
Notice however, that even in the anti-realist scenarios, the deception has to have most of the same properties as the real world. The sorts of rules that relate particular features of the deception to the subjective experience will have to be the same. In fact, the key difference is just the metaphysical status of the objects of apprehension. Illusory food still tastes the same way and has the same effect on illusory body in mass deception world as real food has on real body in realist world. Since this is the case, we can just say that our normal inductive practices tell us things about the world that we observe without committing ourselves to whether this is real or illusory.
Consider a cup of cocoa. The reason why the hypothesis that there is a cup of cocoa in front of me is highly probable is that there are few other stories to tell in which I see the cup, and where when I further see my hands touching the cup I also feel the cup, and the heat of it. When I further pick up the cup and drink the cocoa, I only taste the cocoa when it lands on my tongue. It is not logically necessary that all these distinct sensations coincide. Positing the existence of a single object which has certain properties explains the coincidence of sensations. Consider what the evil demon would have to do to reproduce this. He would have to reproduce in every sense but one, all the properties of a real cup. Out of all the sceptical scenarios, the ones in which there is an illusory cup with illusory cocoa is the most likely scenario. That is, of the various theoretical explanations of the illusory properties, the same sorts of reasons that impel us to think that there is a real cup of cocoa would also incline us to think that there is an illusory cup of cocoa in the illusion case. So, we can know that there is a cup of cocoa. We just don’t know whether it is illusory or real.
To further complicate things, when the illusion is so complete, the difference between illusion cocoa and real cocoa seems merely semantic. After all, there is a sense in which the illusion exists too. After all, if the illusion did not exist, there would be no way in which the illusion could produce experiences in us. Real and illusory, then in these sorts of cases seem to be some obscure metaphysical properties which may not really matter for most purposes.
In addition, suspending judgment about the metaphysical status of the world we observe does not change which courses of action are more rational to pursue.
Consider the following example where we decide whether to attempt to jump off a building
If the observed world is real, U(Jumping) < U(not jumping) given that we care about living and are reluctant to die.
If the observed world is not real, maybe nothing would happen to us if we did try to jump off a building. Perhaps something good will happen. Perhaps something bad will happen. Either way, based on the principle of insufficient reason, I have no expected change in utility even if I do jump. We can thus assign our utilities for jumping and not to be some value k
Given that we adjust our probabilities according to the calculation given above, the expected utility for jumping is
E(U-jump) = 0.4 x k + 0.6 x U(jumping)
E(U-no jump) = 0.4 x k + 0.6 x U(not jumping)
E(U-jump) < E(U- no jump)
In fact, since it seems that the expected utility is always going to be transformed in some monotonic way, the preference ordering is undisturbed and the ranking of various courses of action remains the same. Worries about pragmatic consequences of suspending judgment are in this case unwarranted. Whichever course of action is rational when we know that the external world** is real would be the same even if we suspended judgment about it.
We thus don’t have to assume that the external world is real in order to do science or go about our daily lives. At most, we just have to acknowledge some possibility that the world we observe is real as far as daily practical problems are concerned. On the theoretical end, we just have to be careful and be modest about the claims that we can make. However, as we have seen, this does not invalidate our current scientific knowledge.
*it is irrelevant whether this deception is caused by some agent. All that is relevant is that the mismatch between reality and perception exists.
**There is a sense in which there has to be a world external to us. The question I am concerned with is whether the world that seems to be external to us is real.