Reply to Naturalism Next on Psychophysical Harmony and Whether I Can Read

I am, in fact, literate

0 Introduction

CW: some pettiness.

The folks over at Naturalism Next—who generally write good and interesting things—have written quite a lengthy reply to my reply to them on the subject of psychophysical harmony. Their reply is filled with accusations that I didn’t read their piece carefully and misunderstood it in ways so egregious that they indicate a deep failing on my behalf—their section where they explain why I’m wrong is simply labeled misunderstandings. They claim, for example:

the approach he took epitomizes precisely the opposite approach to philosophy than that outlined above. Any careful reading of our piece juxtaposed with [his] demonstrates that he did not make even a first attempt to charitably interpret the arguments we made or carefully read what he wrote. He frequently misconstrues our arguments, skips over entire sections of our post, and makes points that were already addressed without mentioning our responses.

I am not going to address everything that they said in their post for two reasons. First, it is a 15,000-word piece, and I have other things to do with my life. Second, many of the issues they raise are rather technical and relate to issues I haven’t thought about in-depth since writing the post. As such, it would be quite time-consuming to refamiliarize myself with the issues surrounding, for example, the highly technical views about the content of belief and Putnam’s twin earth case. I will also not address their criticisms of phenomenal conservatism, according to which intuitions confer justification, for that is a much broader methodological view that I’ve already addressed at some length in other articles and would only muddy the waters in responding to this piece.

Reading the first draft, I was somewhat worried. They come out of the gates swinging, accusing me of egregious misunderstandings and failing to carefully read their post, all the while being extremely adamant and abrasive about my own views. This is technical stuff, so it’s possible that I did, in fact, egregiously misunderstand. But when I carefully combed through the things I allegedly missed, almost all of them are wrong, and in fact rely on misunderstandings either of my post or what it intended to do. To my surprise, I discovered that not only was my piece not littered with misreadings and skipped chunks of texts, it actually holds up pretty well.

1 Concerns

Under the broad header of misunderstandings, their first section is labeled “concerns.” In their original post, Naturalism Next provides a series of views that can explain psychophysical harmony. They note that, on account of this, to most naturalists, psychophysical harmony won’t be a problem that motivates belief in theism because the theistic explanation of psychophysical harmony will be worse than one of the naturalistic explanations, even if none of those views are very plausible. I argue that each of these views are implausible. The authors express considerable outrage about my failure to quote them on this point about the broader dialectical context, saying:

it was inappropriate … to not even mention or try to reconstruct that point we were making, and to offer responses to this section as if they engaged with our point when they did not. A naive reader would make the wrong assumptions about our claims in that section if they only read [his] piece, because he omitted any discussion of our dialectical aims, and what he did say implies that our arguments were different than the ones we made.

This claims strikes me as particularly baffling. For one, my post was not exclusively a reply to them. My post was a defense of the psychophysical harmony argument from pretty much all the objections that have ever been made to it. I cited them as a helpful source for explaining many of the objections and addressed the ones they raised. As such, it’s perfectly reasonable for me not to address everything they said in the article about the psychophysical harmony argument. If I’m making a cumulative case for some view, and a different piece argues that many other views can explain some piece of data, it’s reasonable to quote them without quoting their explanation of the dialectical context.

Second, it is reasonable not to quote every point one disputes in an article that one is disagreeing with. My piece, which was part 1 of a series, was already about 7,500 words and involved disputing things said by about a dozen different authors. When one quotes a book, for instance, even to disagree, they do not need to summarize all things said in the book for fear of misleading “a naive reader”—one so naive apparently that they’re unaware that when one quotes a source they are not quoting everything they’ve ever said.

Next, I’m accused of three more errors. One of these is a genuine error, and the other two are not and rely, in fact, on misreadings of what I said. Were I Naturalism Next, I’d take the opportunity to make a snarky comment about the important things philosophy taught me being the importance of charitably reconstructing an opposing view.

I claimed that the various views they describe as avoiding the problem of psychophysical harmony all entail that zombies are inconceivable and Mary doesn’t learn anything. It’s true that they all entail that zombies are inconceivable, however, they don’t all entail Mary doesn’t learn anything. Here, I was wrong. I was correct, however, about zombies and inverted qualia.

I next claimed that error theory is implausible because it implies that excruciating agony isn’t bad. They claim that these kinds of responses “trade on the ambiguity between stance-independent badness and badness.” I’ve addressed this claim in other articles; it seems like it is not merely that excruciating agony is stance dependently bad, meaning we have a negative judgment of it, but it is stance independently bad in that it is bad even if people have a positive judgment towards it. Here I did not respond in depth to NN’s objection to error theory because I have done that elsewhere and it would have muddied the waters too much. But if someone argues that a view that I think is very implausible on other grounds avoids some philosophical problem, it’s reasonable to state that.

[He] goes on to claim that since theism entails that error theory is false, theism predicts that pain leads to aversion behavior. This misses the point. It is not being disputed that on theism there are normative facts, and that theism predicts, insofar as there are sentient creatures whose consciousness and dispositions are governed by psychophysical laws, that such laws will be normatively good. The point is that the psychophysical harmony argument proceeds with the assumption that normative harmony (our behavioral states line up with our psychological states in a normatively valuable way) is a datum in need of explanation. Yet, that this datum obtains is precisely what the error theorist denies.  For the datum of normative harmony at least, there is nothing to be explained for the error theorist. Nonetheless, we anticipated that there may be a way to state the psychophysical harmony argument that is consistent with error theory, and what we say about this looks quite similar to what Adelstein said (see footnote 1 in the concerns section). Why would Adelstein bring up this objection but fail to state that we were aware of it?

Their claim that “this misses the point” misses the point. On naturalism combined with error theory, one has no reason to think that pleasure and pain would pair with physical states in ways such that damage is correlated with pain and beneficial things are correlated with pleasure. Their claim is that if error theory is true then the same would be true on theism; that if pain isn’t bad and pleasure isn’t good then there’s no reason to think that pain, for example, would pair with harmful physical states. However, if error theory might be false, then one should update on the fact that pain pairs with harmful bodily stimuli and conclude that both theism and error theory are true.

Theism+moral realism explains why pain pairs with desctructive bodily stimuli. None of the other views do. So if one thinks theism is remotely plausible they should update in favor of theism+moral realism. Nothing that NN says remotely undermines this point.

The point they raise about how one could construct the psychophysical harmony argument without needing to deny error theory is raised in footnote 2, not footnote 1 (see, I can read carefully). Second, it is very different from the probability point I raised—I’ll include a quote stating it in a footnote¹. Third, what I said earlier is obviously applicable—in replying to a piece, one need not quote every point made in the piece, especially if the point is only somewhat related to the point one intends to make and not refuted at all. I cited their piece as a useful list of objections—nothing more.

2 Normative Harmony

The next section criticizes my section defending the normative harmony argument. Montesinos’s argument is roughly the following (he says more words, so if you want the full experience, read his piece): we come to form our beliefs about pain through a conceptual framework in the real world. As such, our beliefs about pain are derived from what pain is like in the real world which leads to mistaken judgments about what pain would be like in scenarios very different from the real world.

My response was to point out that, while it’s true that we might miss the mark somewhat, proposing we miss it by as much as would be required for normative harmony is extremely implausible. It implies that if disembodied minds were tortured, that wouldn’t be bad, for the badness of pain resides in its functional role. Montesinos bites the bullet and says I didn’t deal with his debunking. But I think some beliefs are sufficiently implausible that pointing out the implications of them is enough to discredit a view, even if you can tell a just-so story of how one would come to be wrong about it. If a person thinks that torturing babies isn’t bad then you don’t have to address their error theory—that alone is enough to give one a reason to reject the view. This holds even with disembodied babies.

3 Semantic Harmony, Causation and Explanation

I won’t address this section in detail. Let me just admit to an error I made: I said the psychophysical harmony was about reports not judgments. However, in Crummett and Cutter’s original paper, they talk about reports and judgments. That’s my bad.

4 Understated evidence

Montesinos and Benjamin claim “BB provides two responses to this section. His first point implies that he did not read this section carefully, the second entails that he did not read it at all.” Now, while it’s reasonable to address one part of a series without reading the whole thing, it’s ironic that their first claim establishes that they did not read the further posts in the series, either carefully or at all.

Their criticism of psychophysical harmony claims that it understates the evidence—that while the world is harmonious in many respects, it is disharmonious in a sufficient number of respects that make it not strongly predicted on theism. In response, I claim that the theist will have to have some explanation for the many blemishes and imperfections of the world and this explanation will also explain the limited instances of psychophysical harmony. Montesinos and Benjamin say “We cited a wide array of different kinds of disharmony and most of it is not even prima facie explained by any theodicy we know of. To illustrate, let’s use just one example of disharmony we cited: that our perceptual beliefs about the richness of color perception in the periphery are false.”

Before I address this charge, I want to note that even if there aren’t theodicies that explain the data, this would mean that I am wrong, not that I “did not read this section carefully.” As it happens, I don’t really think any of the theodicies succeed, but I wouldn’t describe any person who raises a theodicy as having not read the things their critics say.

As it so happens, I’ve listed some theodicies that are fairly general and explain all of the blemishes and imperfections in the world. The soul-binding theodicy claims that for us to remain the same through the beatific vision requires a fairly random assortment of experiences and, as a result, evils are neither expected nor unexpected. Skeptical theism, which is not technically a theodicy but is a defense against the problem of evil nonetheless, says we’re not in a position to ascertain the odds of various evils on theism. Finally, the free will theodicy given by Crummett says that evils in the world are the result of immoral acts taken by demons and programmers of the simulation.

My next response involves noting that, though our laws are not perfectly harmonious, they are in the top .00001% most harmonious laws. As a result, the full evidence favors theism. The Naturalism Next team takes considerable umbrage at my failure to quote their reply to his argument. It is true that I did not explicitly address their counterargument. Was this because I didn’t read it? No, it was because it was already addressed by the things I said.

Their argument is that on theism, we should expect one of the most harmonious sets of laws, while on naturalism we should expect any of the laws. As a result, laws that are in the middle in harmoniousness favor naturalism! Now, dear reader, can you think of something I said that would interact with this point? Their claim, which hinges on the supposed fact that on theism one would expect extremely harmonious laws is addressed by my first point that theodicies will explain why that isn’t so!

It’s true, I could have pointed this out in the article. But it was already about 7,500 words and would take almost an hour to read out loud. In such a case, it’s worth being concise and not pointing out when people make points that your arguments address. Again, I did not claim to be addressing everything said in their piece; I merely claimed that it was a helpful catalogue of many objections to psychophysical harmony that I planned to respond to.

They finally note that this is a reason to favor views other than theism or naturalism which have significant imperfections. I agree that this data point by itself should raise the credence of them. They would perhaps be ruled out on other grounds though, like prior probability.

5 The Revenge Objection

The revenge problem claims that the same factors that make psychophysical harmony unlikely also make theism unlikely because God is psychophysically harmonious. To be clear, I think this could be right and is a reason to think theism is less likely, but I’m not clear on whether this works.

One reply given is that theism is simple because it follows from a simple property like omnipotence or perfection. This can be used to explain God’s psychophysical harmony because a perfect being will necessarily be harmonious. Benjamin replies by suggesting that this response is unworkable on account of omnipotence or perfection requiring harmony. Something is omnipotent because of their power, not the other way around. Therefore, omnipotence assumes rather than explains psychophysical harmony.

In response, I argued that a property of parts can be explained in virtue of the whole. For instance, an apple pie may constitutively depend on its atoms, but the reason the atoms are as they are may depend on the apple pie (they are where they are because someone wanted to bake an apple pie). Benjamin expresses some confusion about the exact point that I was making and I don’t think my original piece was sufficiently clear. The idea is that the explanation for various parts that constitute a whole can depend on the whole itself. This will be especially true if there is such thing as top down causation, as dualists often suppose. If this is so then the theist has a good explanation of why God has these properties that would be implausible if brutely posited—they follow from perfection. Benjamin says “If we accept that an apple pie is dependent on its constituent atoms, then I have no idea what it is to further add that the atoms, in turn, constitutively depend on the pie as a whole.” It means that the facts about them are best explained in reference to facts about the apple pie. If you want to know why they are as they are, a helpful explanation will come by referencing the apple pie.

Next, Benjamin addresses the claim that theism isn’t vulnerable to the challenge because it posits simple laws of the form “that which God wills, is.” He claims in his original piece “God's laws could be stated as "God can will anything" or it can be stated as "God can will X, & Y & Z and so on" for some infinite, or astronomically large set of states of affairs.” This is what I was addressing when I said “The approach where you just list all the things that can be done given by TT is clearly unworkable—it implies full omnipotence is no more complicated than omnipotence minus being able to perform one random fact.” Benjamin notes that I haven’t given a good reason why that implication is false. It’s true that I don’t give any additional reason why it’s false. But I submit that any view that implies an omnipotent being is no more likely than a being that is almost omnipotent but can’t do five random things is false. This implication strikes me as very implausible and I think views having very implausible implications gives one a reason to reject them. If you ask most ordinary people whether they think God is more likely than a being that can do everything except, for example, create squirrels, they will report that he is.

Unfortunately, Benjamin has missed the second part of my reply when he says:

However, as it turns out, in fact the point I made doesn’t turn on God’s laws being just as probable, or not greatly more probable than any given epistemically possible distribution of laws. God’s laws can be a million times more probable than any given distribution of laws, but, once again, there are just so many, likely infinitely many, epistemically possible, disharmonious ways the psycho-divine laws conceivably could have been (e.g, God tries to will a frog into existence, but he causes a set of dominoes to fall over instead etc. etc.) that (by parity of reasoning for the improbability of psycho-physical harmony) so long as God’s psycho-divine laws being harmonious is not maximally or close to maximally more a priori probable than other conceivable distributions, that God's laws are harmonious should be a very surprising fact and astronomically improbable indeed. 

Here I was not claiming that merely by virtue of claiming that there are a lot of different psychophysical laws one has to assign them equal probability. Rather, I was giving a reason why simply listing all the things a being can do is not a good way to evaluate the simplicity of its power. I agree that theism has a pretty low prior because there are so many ways reality can be, though in a later piece in the series I argue it’s not vanishingly low.

Next, Benjamin says:

BB also claims that since psychophysical harmony is just so shocking and improbable on naturalism, and we shouldn't be too confident about the prospects of the revenge objection, psychophysical harmony is still evidence for theism. What this misses is that the point of the revenge objection is not to dispute that we should update in favor of the theistic hypothesis when taking into account the data of psychophysical harmony.

My claim is about what to do under conditions of uncertainty. If there is some possibility that explains an otherwise miraculous fact, and it might have a very low probability or might have a pretty high probability, based on its explanatory power you should think it’s more likely that it has a high probability. To see this, suppose that there are two possibilities: first that a royal flush is a very special arrangement of a deck of cards, second that a royal flush is just a random arrangement of a deck of cards. One is evaluating two hypotheses: the first claims that a deck is rigged and the second claims it isn’t.

Given that the deck is rigged+royal flush is special hypothesis explains this otherwise implausible event, if one is initially unsure of how to calculate the prior probability of the hypothesis they should then think that the royal flush being special hypothesis is more likely. Similarly, if you’re not sure what the prior probability of theism is, and then you get good evidence for theism, you should update in favor of views that give it a higher prior.

Finally, Benjamin says:

Finally, BB asserts that since there are at least somewhat plausible views on intrinsic probability on which theism would be simple, we should give some credence to such views, and factor that into our probabilistic calculations. As tempting as a response like this may be, it doesn't work. The reason this doesn't work is that the view we use to calculate intrinsic probability is the very resource that informs how we determine the prior probability of the hypothesis under discussion in the first place, which, as stressed, is a crucial step in Bayesian updating. I've nowhere claimed that my view on assigning the priors is maximally plausible, but we have to assign priors somehow otherwise our posterior probability will be inscrutable and thus whether we should adjust our beliefs in favor of theism given psychophysical harmony will be indeterminate.

This is not true! To see this, let’s distinguish first-order and second-order methods of assigning prior probabilities. A first-order method describes the ideal way to assign priors—it would give a formula for deciding upon them. A second-order method would tell you how to assign priors given uncertainty about the first-order methods. In this way, a second-order method is analogous to a theory of how to act under moral uncertainty.

I submit that one should, to assign priors, take the various views that they have some credence in about assigning priors, and multiply the prior those theories give to some hypothesis by the odds they give to that hypothesis being true. For example, if there are two ways of deciding on prior probabilities, and I’m split 50/50 between them, and one says that God is certain to exist and the other says God is certain not to exist, I should take the average and start God with a 50% prior. Similarly, in this case, if one is not sure exactly how to take prior probabilities, because there are some views on which God is very intrinsically probable, they should give it a non-trivial second-order prior.

6 Conclusion

Well folks, that’s a wrap. I probably will respond to what NN says about intuitions at some point, and I might address the more technical sections if doing so strikes my fancy at some point. I think that the psychophysical harmony argument holds up quite well and that, contrary to their assertions that my “response … is a failure,” and that I skipped over entire sections, I made a few small errors, but nothing deserving the quite sharply worded criticism.

1. “To see how, consider the parallel move made in discussions around the problem of evil: in response to atheists, the theist asserts that if naturalism is true then ‘good’ and ‘evil’ have no meaning or there are no true moral statements, and so the atheist cannot make claims about the evidential weight of ‘evil’. One good response the atheist should make to this claim is to point out that the data in question can be stated neutrally as to avoid any metaethical commitment just by referencing facts about pain, suffering, the lack of wish fulfillment, etc. If theism is true, because God is all-good, these facts are unexpected, but not if naturalism is true. Similarly, in the case of the psychophysical harmony argument, the advocate might be able to state the data not as being that pain and pleasure serve their proper normative roles—a fact that antirealists will deny—but instead that pain and pleasure are correlated with certain behaviors. Then, they can say that on theism, because God is all-good, we should expect these correlations, while on naturalism, we have no reason to.”