The Fine-Tuning Argument Against the Multiverse
A striking phenomenon uncovered by contemporary physics is that the values of the fundamental parameters of nature, such as the universal gravitational constant, the strengths of the weak and strong nuclear forces, the cosmological constant, and the like, appear to fall within extremely narrow life-permitting windows. Contemporary philosophers and physicists often describe this phenomenon as the universe’s being “fine-tuned” for the existence of life. In many cases, the degree to which these parameters appear fine-tuned is rather extreme. The degree of fine-tuning required for the cosmological constant to have a life-permitting value has been estimated, for example, to be 1 part in 10^120! It strains credulity to believe that such parameters fell within their life-permitting windows solely by coincidence. The phenomenon of fine-tuning cries out for explanation.
One frequently proposed explanation is that our universe is part of a vast multiverse. The thought is that if there is a large enough multiverse, and enough random variation within the values of the parameters between universes, it is all but inevitable that some life-permitting universes will arise. Our universe, according to this hypothesis, simply turned out to be one of the lucky, life-permitting ones. And of course, it is no surprise (so the explanation goes) that we would find ourselves observing a life-permitting universe, as we could not have observed any other kind.
Since this explanation allegedly accounts for the phenomenon of fine-tuning, fine-tuning is in turn taken to provide evidence that we do in fact live in a multiverse. In our paper “The Fine-Tuning Argument Against the Multiverse,” however, my coauthor, Philip Swenson, and I argue that the opposite is the case. The phenomenon of fine-tuning, we argue, tends to raise the probability that ours is the only universe. In this blog post, I aim both to provide the background knowledge needed to understand our argument, as well as a brief, intuitive sketch of the argument itself.
Fine Tuning as Evidence for Design?
There is, in addition to the multiverse hypothesis, another frequently proposed explanation for why our universe is fine-tuned—namely, that it is the product of design or some other sort of goal-oriented process. It is commonly held, for instance, that a good explanation of the fact that our universe is fine-tuned is that God designed it to be life-permitting. The hypothesis that our universe was designed by God to be life-permitting, however, is but one version of a more general hypothesis that we refer to in our paper as “the teleological hypothesis.” According to this hypothesis, there is some agent, power, or goal-directed aspect of reality that tends to bring about the existence of at least one or more life-supporting universes.
Many philosophers who argue that the fine-tuning of our universe evidentially supports the teleological hypothesis do so by appealing to a widely endorsed principle of evidence known as “the likelihood principle”:
(Likelihood Principle) E counts as evidence favoring a hypothesis H1 over another hypothesis H2 just in the case that E is more likely on the assumption that H1 is true than it is on the assumption that H2 is true.
In this case (so the argument goes), it is at least not that unlikely that our universe would be life-permitting on the assumption that the teleological hypothesis is true. However, given the incredible degree of fine-tuning required for the universe to be so, it is extraordinarily unlikely that our universe would be fine-tuned on the assumption that the teleological hypothesis is false. Thus, by way of the likelihood principle, the fact that the universe is fine-tuned for life counts as evidence favoring the teleological hypothesis over its denial.
The Multiverse Against Design?
While many philosophers are willing to concede that the fine-tuning of our universe does offer some evidence for the teleological hypothesis, many also deny the evidence is that strong. One reason for this denial is the alleged ability of the multiverse hypothesis to account for the same phenomenon just as well. There are at least two reasons why many philosophers (including ourselves) believe this objection to the teleological argument from fine tuning to be misguided.
First, the multiverse hypothesis and the teleological hypothesis are not mutually exclusive. Perhaps there are multiple universes and also some agent, power, or goal-directed aspect of reality that tends to promote conditions favorable for life. And plausibly, a teleological multiverse hypothesis still ends up more strongly predicting that our universe would be fine-tuned than does a non-teleological multiverse hypothesis. As Robin Collins has pointed out, the teleological hypothesis might be needed to explain, for example, why the multiverse is large enough, and admits of enough variation, to be life-sustaining. And as Michael Rota and Thomas Metcalf have observed, it is also arguably more likely on the teleological multiverse hypothesis that a greater proportion of universes would be fine-tuned, thereby making it more likely that our universe is fine-tuned.
Second, and more importantly, those who claim that this universe’s being fine-tuned for life raises the probability of the multiverse hypothesis appear to commit a probabilistic fallacy. This charge was originally introduced into the literature by Ian Hacking, and it has been further developed and defended by Roger White. Imagine for instance (to adapt Hacking’s example) that I roll an ordinary pair of six-sided dice in my office and observe they come up a double-six. “That’s a really improbable roll!”, I think to myself. “That makes it more likely there are lots of other people rolling dice on campus right now.” It is clear that, in this example, I am reasoning poorly. While it is true that there being multiple people on campus rolling dice makes it more likely that some roll or another on campus will come up a double six, whether the one that I just made comes up a double six remains probabilistically independent of how many other rolls are taking place. For me to take the improbable outcome of my roll to count as evidence for the existence of multiple rolls is to commit a probabilistic fallacy. Hacking and White charge those who take the fact that our universe is fine-tuned to count as evidence for the existence of a multiverse with committing the very same fallacy. The mere fact that something improbable happened here is not by itself evidence for there being multiple independent trials.
The literature on both the fine-tuning argument for design and the fine-tuning argument for a multiverse is extensive. There is much more to be said about all of these issues. But for the purpose of our paper, we take for granted both that the fine-tuning of this universe provides some evidence for the teleological hypothesis and also that Hacking and White are correct in their assessment that our universe’s being fine-tuned for life does not by itself support a non-teleological multiverse hypothesis.
Indeed, we intend to extend the Hacking-White charge by arguing that if both of these things are the case, then not only does the fact that our universe is fine-tuned fail to support a multiverse hypothesis, it provides evidence for the single-universe hypothesis instead! While the argument for this result in our paper is couched in formal terms by way of premises about relationships between conditional probabilities, the intuitive idea behind it is relatively simple.
Why Fine-Tuning Counts Against the Multiverse Hypothesis
There are four logically possible hypotheses to consider: (1) the non-teleological multiverse hypothesis, (2) the non-teleological single-universe hypothesis, (3) the teleological single-universe hypothesis, and (4) the teleological multiverse hypothesis.
First, we note (per Hacking and White) that it is no more likely that this universe is fine-tuned for life given the non-teleological multiverse hypothesis than it is given the non-teleological single-universe hypothesis. So (per the likelihood principle) the fact that this universe is fine-tuned for life counts no more in favor of the non-teleological multiverse hypothesis than it does the non-teleological single universe hypothesis. It is much more likely, however, that this universe would be fine-tuned for life given the teleological single-universe hypothesis than it is given either the non-teleological single-universe hypothesis or the non-teleological multiverse hypothesis. So (again, per the likelihood principle) the fact that this universe is fine-tuned for life favors the teleological single-universe hypothesis over its non-teleological rivals. But how does the teleological single-universe hypothesis stack up against the teleological multiverse hypothesis?
Suppose there are multiple universes. It may be that God (or whatever other agent, power, or goal-directed aspect of reality is responsible for there being life-permitting universes) is interested in ensuring that every universe there is permits life. But there is no guarantee of that. If we look around our own universe, we see that it contains lots of places that are not only devoid of life, but hostile to it. So even if God was keen to ensure that some portions of our universe contained life, God was not all that concerned to make sure that our universe was packed full of life. Perhaps God’s attitude toward the multiverse is similar. Maybe it is really important to God that some universes contain life but not so important that all do. Maybe God even has goals that require some universes in the multiverse to be hostile to life. Perhaps God is interested, for example, in there being all manner of different kinds of universes, not merely life-permitting ones. If God’s goals might be like this, then perhaps it is not all that likely that this universe would be fine-tuned for life given the teleological multiverse hypothesis.
But now suppose that the single-universe hypothesis is true. Remember that it is baked into the teleological hypothesis that God (or whatever other agent, power, or goal-oriented aspect of reality there may be) is keen on ensuring that there are at least some life-permitting universes. And if the single-universe hypothesis is true, this universe affords God’s one and only chance of there being such a universe. So it is highly likely on the teleological single-universe hypothesis that this universe would be fine-tuned for life—more likely than it is on the teleological multiverse hypothesis. So (once again, per the likelihood principle) the fact that this universe is fine-tuned favors the teleological single-universe hypothesis over the teleological multiverse hypothesis.
If we put all these results together the situation is as follows: The fact that this universe is fine-tuned for life fails to support the non-teleological multiverse hypothesis over its single-universe rivals. That same fact does favor the teleological single-universe hypothesis, however, over each of its multiverse rivals. So all things considered, the fact that this universe is fine-tuned for life tends to favor the single-universe hypothesis over the multiverse hypothesis.
We may be grateful that the phenomenon of fine-tuning has driven interest in the multiverse hypothesis, if for no other reasons than how that hypothesis sparks the imagination and makes for fruitful philosophical and cosmological speculation. But if the argument of our paper is sound, the fact that our universe is fine-tuned for life counts against the multiverse hypothesis actually being true.
Kenneth Boyce
Kenneth Boyce is an associate professor of philosophy at the University of Missouri. He specializes in metaphysics, epistemology, and philosophy of religion. He also has an area of competence in the philosophy of science.
I agree that the multiverse alternative is unnecessarily cumbersome and that what appears to be some fine-tuning in our universe does not provide evidence of a multiverse.
It is not surprising that while trying to explain a universe in which life exists, there are theoretical constructs that must accommodate that fact. Constructs that did not allow for that would be deficient. The universe is what it is with life in it. No matter what our universe is, this universe would have been improbable because the possibilities are endless. It is fine-tuned only in the sense that it is just one of many alternatives. Any universe that has something unusual in it would be fine-tuned or special in that sense. The idea of fine-tuning itself is less potent than some theorists believe.
A universe with life in it is no more special than a universe with evolved, but lifeless intelligent robots in it. We could imagine intelligent activity evolving without biological life. If it is the uniqueness of intelligence itself that needs explaining, then evolution within the context of the vastness of space-time very likely is all we need. There is no need to introduce a design or a designer. What has shown progress in theory building in the past are empirical and physically based systems. When we go beyond those, it is difficult to evaluate endless possible theories, and such efforts have never led to progress unless they were somehow rooted in empirical or experimental findings.
As for the “Likelihood Principle,” the teleological hypothesis is only more likely if it is conceded that the explanation of life likely requires the construct of purposeful design. This begs the question. The teleological hypothesis has no greater access to the Likelihood Principle than other hypotheses.
We can ask about purpose and the meaning of life, but the only progress we have made over the millennia has been achieved when we reduce these things to something we can measure and predict. Neuroscience is still in its infancy. I believe that it is where our hope will be rewarded. This requires reductionism. There is confirmatory and disconfirmatory reductionism. Despite various problems with reductionism, it is the only way forward. To posit a designer to explain the universe is to ignore what can be produced on its own by means of evolution throughout the vastness of space-time. When considering the designer hypothesis, it is a good idea to remind ourselves about Hume’s discussion in his “Dialogues Concerning Natural Religion” along with many other sources.
It strains credulity to believe that the parameters of reality fell within the God-permitting window solely by coincidence. The phenomenon of fine-tuning of the logic of reality to be such that there can be a disembodied personal entity which can timelessly act to create universes, with the desires exactly as it has to create embodied life, even though the God itself has no body, cries out for explanation.
The fine tuning argument is a joke because you just piled all the questions about why reality is the way it is into your assumptions about a God would want, but you didn’t bother to ask why it’s at all likely that such an entity would be self-existent in exactly the way it is, with the disposition it has, or want to make this.
Two comments are in order here. First, this article is not in accord with the findings of modern physics. Constants such as the velocity of light, Newton’s constant of gravitation or the mass of the electron have been undermined by the findings of string theory and the research of Barrow and Webb. The variable constant which varies in time and place is the one congruent with modern research. The mathematical basis of string theory postulates as many as six more dimensions of space. Second, depending on the variation of constants in the time and space we call earth history, species naturally would develop depending on the environmental factors then in operation. There is no need for a divine fine tuning.
String theory? Seriously? Let’s wait until it can be tested.
It has survived the test of a half century and is one of the latest verified theories in modern physics. Absolute constants were in vogue in 1965. Much research has been done since then.
Moreover, even if there are absolute constants, the conclusions in your essay do not and cannot follow.
I think you have mistaken my “essay” for someone else’s writing. As for string theory, I am relying on a number of sources, including an article in “Scientific American,” by Brandon Z. Foster, March 25th, 2020. The title is: “Will String Theory Finally Be Put to the Experimental Test?” The answer is: not yet. That is why I have suggested that we wait.
Doesn’t this whole argument fall apart if you add the words “as we know it” between the words “life” and “permitting”? Because if the so called fine tuning only makes “life as we know it” possible, but other forms of life that we couldn’t possibly know because they’re flourishing in other universes that aren’t fine-tuned are possible there, then our universe isn’t really fine-tuned at all. It’s only fine-tuned for life as we know it, and that life arose here in response to actual, albeit arbitrary, conditions.
Burke, of course, I was referring to Kenneth’s essay. But I’m amazed that you receive your information form the Scientific American rather than serious academic journals in physics. Webb and Flambaum, using highly sensitive scientific equipment at the University of South Wales, determined based upon a total of 28 quasar absorption lines that alpha- one of the so-called constants- will increase in six parts every one million years. Thus, the spatial variation of alpha is real. No doubt, as A.J. Ayer observed all testing must be retested and no theory explains all the facts in its domain. However, this essay assumes something which is not true: namely, that constants are ontologically real. This is far from the case. Inconstant constants are a real possibility.
Thanks for your comment, Edward. I appreciate and respect your knowledge about the latest findings related to string theory; however, it is an obvious informal fallacy to suggest that something published in “Scientific American” is suspect because it is not being published in an academic journal. The author is a professional in the field who is also familiar with the relevant literature. I agree that the original essay is flawed in a number of ways, but your concept of “inconstant constants” …looks contradictory to me.
Burke, if you enjoy the Scientific American, that is your choice. As for your remark that inconstant constants are contradictory remember Hagel’s theory that the law of non-contradiction is technically true, but trivial. The reason is that it eliminates the dimension of time. Alpha might be constant for a million years and then change slightly given the added dimensions of space postulated by the mathematics supporting string theory.
So rolling an ordinary pair of six-sided dice and observing them coming up as a double-six is not favoring any of the two hypothesis – a single- or a multiple-universe – one over the other.
How about an ordinary quantum mechanical system/Hilbert space H with the basis {|(1|1)>;|(1|2)>;|(1|3)>;…;|(6|6)>}, where <(i|j)|(k|m)>=0 for i≠k OR j≠m and <(i|j)|(k|m)>=1/36 for i=k AND j=m for any i,j,k,l{1,2,3,4,5,6}?
If we would find or measure the same quantum state to be in different possible states in at the same time, then wouldn’t that be an evidence for the existence of a multiverse?
Call it a “branching-verse” or a “tree-verse” as trees are usually growing by branching.
Thank you, Kenneth, for a thought-provoking article. I have a few responses for your consideration.
No selection effect is considered in the dice-rolling illustration, which seems to undermine its relevance.
We can imagine other illustrations that do contemplate a selection effect. For example, say that you are a sleeping genie at a circus booth, and that you will only be awakened if a circus-goer walks into the booth and correctly guesses a random number between 1 and 1,000,000. If you awake, you would rightly conclude that (a) many circus-goers tried to guess the random number and only the most recent one succeeded, is much more likely than (b) only one circus-goer tried to guess the random number and just happened to be right.
The article talks about the likelihood of “this universe” being fine-tuned several times. This seems to ignore the selection effect too. If we change “this universe” to “whatever universe we happen to observe” the argument breaks down.
The articles claims that fine-tuning decreases the probability of multiverse hypotheses in general, and the non-teleological multiverse hypothesis in particular. If anything it increases* them, as discussed below:
The four hypotheses you lay out are below:
Teleo Non-teleo
Single [A] [C]
Multi [B] [D]
For the sake of example, imagine the probabilities of the four hypotheses, before considering fine-tuning, are as shown below. (The sum of the four probabilities presumably must be 1.)
Teleo Non-teleo
Single 0.25 0.25
Multi 0.25 0.25
So far in this example, the probability of a multiverse is 0.50.
Now introduce fine-tuning. This significantly reduces the likelihood of [C]. Since the sum of the four probabilities must still be 1, the sum of the other three probabilities must increase. For the moment, let’s assume they increase equally*
Teleo Non-teleo
Single 0.33 0.01
Multi 0.33 0.33
Now the probability of a multiverse ([B] + [D]) has /increased/ from 0.50 to 0.66, and the probability of a non-teleological multiverse has /increased/ from 0.25 to 033. They did /not/ decrease.
Granted, fine-tuning also increases the probability of the two teleological hypotheses, which I think is the end goal of any fine-tuning argument. The naturalist should admit this, although he may argue their posterior probabilities are still low. My point is simply that the article incorrectly claims that fine-tuning decreases the probability of a non-teleological multiverse (or multiverses in general).
** Of course, one could argue that all of the increase should be allocated to [A], but I don’t see why. In any case, as long as some portion of the increase goes to /either/ [B] or [D], fine-tuning will be probability-increasing with respect to multiverse hypotheses.
Similarly, one could argue that all of the increase should be allocated to [A] and [B], but again I don’t see why. In any case, as long as some portion of the increase goes to [D], fine-tuning will be probability-increasing with respect to the non-teleological multiverse hypothesis.
I don’t understand how we can say the teleological-single-universe hypothesis is more probable than the non-teleological-multiverse hypothesis. You’re correct that it would be irrational to say some improbable dice throw is explained by the fact that other people are also throwing dice, but that’s not what the multiverse is hypothesizing. Generally it’s hypothesizing that there is some physical mechanism that is generating universes with different constants, constraints, etc. An analogy would more-so be like some kind of lottery ball machine that has 1000 white balls and 1 orange ball. If it generates a ball every second, you’re eventually going to get the orange ball.