If I Can't See It, It Doesn't Exist

A general argument for deflating our ontology

Those pesky philosophers inflating their ontologies again…

Many people - especially philosophers - have a tendency to inflate their ontologies with all sorts of strange things like numbers, groups, chairs etc. I think this is often misguided, and I try here to show why holding that such entities exist is usually unjustified.

1 - A Counterfactual Argument against Realism

The argument I want to push here basically comes down to the question: What would the world look like if X didn’t exist? I think that in many philosophically contentious cases, the answer will be: Exactly the same. If this is true, then the existence of X does not explain anything, and so we have no reason to believe that X exists. I think the best understanding of this argument will be through taking a few examples:

1.1 - Social Groups

We might think that there is a group which we call “football fans” (or “soccer fans” for the uncivilized apes across the pond). Some will think that this group exists. It is not just that there are people who exist, but some of the people who exist come together to form the group which we call “football fans”, and this group is a separate part of our ontology. Now ask yourself the question: What would the world look like if the group “football fans” didn’t exist? Here I am of course not asking about the people who make up the group, but rather the entity just described, which is the group. The answer: Exactly the same. Nothing would change - or rather nothing would change for us. We can explain every part of everything we ever experience without reference to groups. The behavior of the group “football fans” is explained entirely by the behaviors of each individual person of that group, and the entity that is the group does no extra causal work. There is of course something that would be different, namely that the group would not exist (obviously), but all the people who are group realists now would still be group realists and for the exact same reasons.

It is here important to emphasize that I am not just talking about states of affairs outside your head - importantly your mental states would be exactly the same. If groups themselves cause nothing and explain nothing, then your belief in groups, and whatever considerations you take to support the existence of groups would be there, completely independently of whether groups exist or not. That means that you have a defeater for any evidence you cite in favor of the existence of groups: You would have the exact same evidence and the exact same attitude towards that evidence, regardless of whether groups existed or not. You may of course come up with a theory of groups where their existence does make some sort of difference, but it is not at all clear what difference they even could make which would not be fully explained by their members - though I would be interested in a group-realist showing it to me.

This is pretty damning for group realism, I think. For something to be evidence in favor of a hypothesis, it has to be more likely given the hypothesis than given the negation of the hypothesis. But any possible evidence you could ever, even in principle, point to, would be just as likely given group realism as given group anti-realism, since groups don’t affect anything (or at least affect no experiences in us). This means that it is actually impossible to have evidence in favor of group realism, and so we should just abandon it, since it makes our worldview more complicated than it would otherwise be.

I think (or at least hope) that most people would actually agree with this, if you press them on it: Group-talk is just a useful way of conceptualizing large amounts of people with shared attributes which we are interested in - and there aren’t actually existing abstract entities “out there” called groups. But I think this serves as a useful example to get us into the right mindset when we turn to more contentious examples.

1.2 - Ordinary Objects

Another example (and one which I have previously written on) is ordinary objects. You are probably sitting on a chair right now. Most people would probably think that that chair exists. More specifically there exists a composite object which is a chair, and which is a further object beyond its proper parts. Now consider again what the world would look like if there was no further object, but only the proper parts making up that object. The answer again: Exactly the same. Any property that the chair has is wholly and perfectly explained by the properties of its proper parts. The color is explained by absorption and reflection of specific wavelengths of light, which is explained fully by the atoms that make up the chair; the shape is explained fully by the arrangements of the atoms; the hardness is explained by the strength of the bonds between atoms etc. There is literally nothing - or at least no part of our experience - which is explained by the composite object itself, and so there is no possible evidence to be had in favor of the existence of composite objects.¹

1.3 - Mathematical Platonism

A lot of philosophers also believe that mathematical entities actually exist. One popular argument for this is the Quine-Putnam Indispensability Argument:

1. We ought to have ontological commitment to all and only the entities that are indispensable to our best scientific theories.

2. Mathematical entities are indispensable to our best scientific theories.

3. Therefore, we ought to have ontological commitment to mathematical entities.

It should be clear how the argument I have given provides a straightforward defeater for 1. Why is this? Well, mathematical entities do not interact with the world. This means that the world would look exactly the same, whether numbers exist or not; particles would move around in the exact same ways, and we would describe their movement in the exact same way. So the fact that mathematics is indispensable to our scientific theories does nothing to show that mathematical entities exist - they would also be indispensable to the theories, if they did not exist, and we would notice no difference.

This is of course not the only argument for platonism about mathematical entities, but I think you can make a similar move for each argument. The problem is not the arguments, but the underlying theory: If your theory makes no difference to our evidence then you cannot be justified in believing it, and so a theory where numbers make no difference to the world is just dead on arrival.

1.4 Idealism

This last example will especially help highlight exactly how this inference works, as it will be a case where we should not deflate our ontology. One could be tempted to think that this argument also applies to the physical world: We only have direct awareness of our mental experience, and so the world could look exactly the same to us, without physical objects existing. The thing is, with stuff like groups and mathematics, it is not just that we can explain the world without reference to these things, it is that they do no explanatory work at all. There is no explanandum that is better explained (or explained at all really) by their existing; they are just abstract entities that “sit on top” of other facts, just being there and doing nothing, to put it in spatial metaphors. And so we can “take them off”, and nothing changes. But the physical world does explain a lot of stuff, and it does so simply and coherently. The fact that I see a screen, move my head and see it from another perspective, press a key and change the light-patterns on the screen etc., is explained very simply and coherently by the fact that there is a screen there, and which I am in some way perceiving. If we want to explain this on an idealist picture, we can do so, but we will have to posit a lot more brute facts about my experience, which are much more elegantly explained by the external world. For example the theory might say something like: “S experiences X at time t, and if S has intention P at time t, then S experiences Y at time t+1, and…” Each of these statements will have to be brute, unless there is some underlying explanation (like the physical world) which unifies them. Alternatively the theory is not as specific, and leaves open what we could experience. But in that case the physical-world thesis makes much more specific predictions, and is thus explanatorily superior.

So we can perhaps rephrase the argument in terms of parsimony: If two theories make the exact same predictions, then we should pick the more parsimonious theory. This works for cases 1-3, because adding the new kind of entity does not change our predictions at all, but makes our theory more complicated. But with idealism we either get a really complex theory (by adding all our experiences as brute postulates of the theory) or a theory which is terrible at making predictions (by leaving it open what sort of experiences we might have).

2 - Side Note: Schaefferian/Neo-Aristotelian Meta Ontology

Perhaps some of you, dear readers, have a meta ontology similar to that of Jonathan Schaeffer, where the question of ontology is not about what exists (we allow basically anything to exist), but rather about grounding-relations between the things that exist. Those who hold such a view can still accept my arguments, we just have to rephrase the terminology. Instead of saying that composite objects or groups don’t exist, we simply say that they are mind-dependent or grounded in minds. We can then say that the world would look no different to us if they were mind-independent, and so we can have no evidence that they are mind-independent.

3 - Phenomenal Conservatism

I think that phenomenal conservatism is probably the correct view of epistemology, and so it would be in order to address an objection that might arise from this view. Michael Huemer formulates the view as follows in his 2007 paper “Compassionate Phenomenal Conservatism”:

• If it seems to S that p, then, in the absence of defeaters, S thereby has at least some degree of justification for believing that p.

We can then give an argument like this: It seems to me that chairs and numbers exist, and so I have some degree of justification for believing that they do. I think this fails for the simple reason that it fails to consider the conjunct about defeaters. The argument that I have given above is exactly that: A defeater. Defeaters are generally classified in two types²:

1. Rebutting defeaters - Reasons in favor of holding the negation of P (or Q which is incompatible with P), where P is whatever you think you are justified in believing. For example: You read in a magazine that your friend Larry is dead, but he then walks in the door.

2. Undercutting defeaters - Reasons for thinking that the ground for your belief that P is not a good reason to believe P. For example: You read in a magazine that your friend Larry is dead, but you then read on the front cover that it says “The Onion”.

I think it should be clear that the argument given in this post is an undercutting defeater for our justification. Just like reading “The Onion” on the front cover of a magazine tells us that there is no reason to believe the truth of the claims of that magazine (that is, the claims of the magazine are independent of the truth of those claims) - just like that, realizing that we would have the exact same appearances and intuitions about something regardless of whether it is the case or not, tells us that those intuitions are not a good way to form beliefs. And so phenomenal conservatism doesn’t save us from this argument.

4 - A Couple of Ways Out

I think that the argument I have given is pretty much decisive for people who hold to the general “philosophical orthodoxy”. That is, a general naturalism/physicalism (perhaps paired with physical determinism), or really any view where mental states are identical to or supervene on physical states, and the physical world is causally closed. This is because these views entail that anything which does not cause anything physically cannot in principle affect your experience and mental states, and so cannot in principle affect your evidence. So even if you came to consider a philosophical argument in favor of realism for some non-causal kind, which did not in any way reference contingent physical facts, you would still have no evidence for it, since your mental states and thus your assenting to the argument would still occur in the exact same way, whether the entity in question existed or not.

4.1 - Interactionist Dualism or Something Close Enough

But if one does not hold such a view, for example if one is an interactionist dualist, then this sort of argument will be weaker, though still quite strong, I think. Basically, you could in principle have evidence in favor of something like mathematical entities, but we need to have some plausible story of how we come to have any connection with these entities. I think a somewhat plausible such story (perhaps the most plausible) is that souls (or minds, if that is less scary) somehow directly grasp these abstract facts about things like morality and numbers. Our directly grasping these facts causes us to have intuitions about these things, which give us justification. This is essentially an add-on to phenomenal conservatism (although it could perhaps work under other epistemologies) which removes the defeater from before: Because our souls have some sort of access to these abstract entities, the world would not be exactly the same whether they existed or not - namely we would not have the same intuitions about these things (or it would at least be less likely).³

4.2 - Theism

I think theists have an even better explanation of how we could be justified in believing in these sorts of kinds: God knows of the existence of such kinds infallibly, meaning he knows about them even if they don’t and can’t interact with anything ever. Furthermore, he would want his creatures to be rational truth-seekers or something like that, and therefore he would create creatures with accurate intuitions, not just about mundane matters like how many fingers are in front of you and whether your coffee is hot, but also stuff like mathematics and morality. This model even works under physicalism+determinism: God could just create the initial conditions plus the laws in such a way that his creatures would come to have the right sorts of intuitions (though I don’t think there are many theists who also believes in physicalism+determinism).

One problem with this story is that it is perhaps too flexible. I can tell the story in a way where God wants his creatures to realize that they have no justification for believing all these abstract categories, because the world would look the same regardless of whether they existed or not. And so he created his creatures so they could understand this amazing argument I have given, making them realize that they had no reason to believe in numbers and chairs.

5 - Conclusion

I think this argument can be summed up in the short phrase: If X existing would make no difference to our evidence, then we cannot be justified in believing in X, unless X makes our theory more parsimonious. I think this is clearly true, and the two positions I described as ways out of the argument are essentially just ways of making it so that X existing would make a difference. But I still think that everyone should agree that we are never justified in believing in superfluous entities that do not affect our evidence at all.

1. Or, well, that is not entirely true. If there are no composite objects, then the world could not be gunky, and so that is one consideration which could count in favor of the existence of composite objects.

2. Sometimes the third type of no-reason defeaters are also included, but those are not relevant here, since there clearly is a reason (namely the seeming).

3. Fellow blogger Bentham’s Bulldog has outlined a view like this.