There is a YouTube channel, called Inspiring Philosophy (henceforth IP), which is about philosophical apologetics. It has about 45k subscribers, and the videos have high visual production values. One video in particular caught my attention, as it was about the laws of logic.
Despite the relatively large audience and good production values, IP makes some pretty baffling mistakes, and a lot of them are very easy to spell out. I will try to explain the main ones here.
IP’s lack of understanding about the issues involved contributes to a confusion about what is being claimed by his imagined ‘opponents’, and what he is trying to say in reply to them. This fundamental confusion is at the heart of the entire video.
In the very opening section, IP asks two general questions:
“Can we trust the laws of logic? Is logic safe from criticism, or is it just another man made construct built on sand?”
These questions are actually quite vague. What does it mean to ‘trust‘ the laws of logic? Does it just mean ‘Are the laws of logic true?’
More importantly, what exactly does he mean by ‘the laws of logic’? He never specifies what he takes the ‘laws of logic’ to actually be. Commonly in discussions like this, they are taken to be the law of excluded middle, the law of non-contradiction, and the law of identity. They are part of what is known as ‘classical logic‘, which we can think of as a group of logical systems which all share a number of principles, including those laws. We must assume that this is what IP means. Let’s refer to these three laws as the ‘classical laws of logic’
The historical development of logic shows that, in one sense, classical logic is not safe from criticism. Just like mathematics, logic has evolved over time, and it has gone through various changes (see this, and this). In particular, there are logical systems which do not include the classical laws of logic; there are systems of logic which have contradictions in, or which have exceptions to excluded middle, or where identity is treated very differently. So, suppose that IP is asking: ‘are there logical systems which do not include those particular logical laws?’ The answer is: ‘yes, there are non-classical logics‘.
Surely though IP thinks he is asking a more interesting question than this. He wants to ask whether some other non-classical logical system should be regarded as the right one. This is a much more interesting question, and much more difficult to answer. I assume that IP wants to say that the classical laws of logic are the right ones, and all the other non-classical alternatives are not right. That would be a coherent position for him to take: he is defending classical logic against rival non-classical logics.
However, this is not what IP actually articulates throughout the video. The video starts off with a claim which seems to be the target that IP wants to argue against. He says
“Many argue that the laws of logic are not true”.
Here we see the fundamental confusion right at the heart of the video. There are two distinct issues IP never distinguishes between a kind of local challenge to classical logic, and a global challenge to all logic:
Local) “Many argue that classical logic is not the right logic”
Global) “Many argue that there is no right logic at all”
While the first option is clearly something many people do argue, it is not quite clear whether the second option even makes sense. Are there really such people who argue that there is no such thing as logic? Who are these people? IP doesn’t ever say.
One of the main problems in what follows is that IP switches back and forth between the local and global challenge, as if he is unaware of the distinction.
2. The argument
In the first half of the video, IP offers what he calls a “simple argument” to use as a foil to respond to. He does not say where he got this argument from, but I suspect that he got it from here.
The argument goes like this:
- Assume that the laws of logic are true
- All propositions are either true or false
- The proposition “This proposition is false” is neither true nor false
- There exists at least one proposition that is neither true nor false
- It is not the case that all propositions are either true or false
- It both is and is not the case that all propositions are either true nor false
- Therefore, the laws of logic are not true
We need to ignore the fact that the first premise is an example of a command, and is not expressing a proposition. We also need to ignore that the argument is not formally valid; strictly speaking, the conclusion does not formally follow from the premises. You have to assume that by ‘the laws of logic’ we mean to include the law of bivalence. If you want an argument to be formally valid, you cannot keep these sorts of assumptions implicit.
Basically, what is going on with this argument is a challenge to classical logic, or really any logic which has the semantic principle of ‘bivalence’. So it is an example of a local challenge. The principle of bivalence is expressed in premise 2, and it says that each proposition has exactly one of the following two truth values: ‘true’ or ‘false’. This principle is the target of the argument.
The liar’s paradox is notoriously difficult to give a satisfying account of within the constraints of classical logic. Therefore, some people say that the only way to account for it is to give up some aspect of classical logic. Thus, considerations of the liar’s paradox provide some reason for people who argue that classical logic needs to be rejected. In this case, the idea implicit in premise 3 is that the liar’s paradox requires bivalence to be false. They say that the Liar Proposition, i.e. “This proposition is false”, is itself neither true nor false. If they are right about this, then classical logic must be wrong. This is because classical logic says that all propositions are either true or false, but there is a proposition which is neither (i.e. the Liar Proposition).
To defend classical logic against this charge, we would expect IP to argue that the liar’s paradox is not solved by treating the Liar Proposition as neither true nor false, but that it can be solved without giving up any of the assumptions of classical logic. This would undermine the reason given here for thinking that bivalence had an exception.
However, at this point IP starts to show just what a poor grasp he has of what this argument is supposed to be showing, and what he needs to do to defend classical logic against it.
He says that “there are several problems with this argument”, but he criticises premise 2. Now, this is odd, because premise 2 is just an expression of bivalence, which is part of classical logic. If he is defending classical logic, then he should be defending premise 2; yet, he is about to offer a reason to doubt it.
IP says that the problem with premise 2 is that not all propositions are either true or false; some are neither true nor false. His example is the following:
“Easter is the best holiday”.
His reasons for thinking that “Easter is the best holiday” is neither true nor false are strange. He says that that proposition “Cannot be proven true or false” and that it is “just an expression of opinion”. “So,” he continues, “you can have propositions that are neither true nor false. Nothing in either logic or language denies this”.
Now, just hold on a minute. Let’s grant IP’s claim that the proposition “Easter is the best holiday” merely expresses an opinion. This is ambiguous between two different things. On one hand, saying that it merely expresses an opinion might mean that it is just shorthand for:
“My opinion is that Easter is the best holiday”
If that is what IP means, then surely “Easter is the best holiday” can be true. After all, I have opinions, and sometimes they are true. In particular, the proposition “My opinion is that Easter is the best holiday” is true just so long as I really do prefer Easter to all other holidays. It would be false if I happened to prefer Halloween to Easter, etc. What is supposed to be the problem here? If such propositions are expressions of opinion in this sense, that doesn’t mean that they are not true or false.
On the other hand, “Easter is the best holiday” might not be shorthand for “My opinion is that Easter is the best holiday”. It might be taken to be something like: “Yey! Easter!” If that is what IP means, then it doesn’t have a truth-value, but then it isn’t really a proposition at all.
So, it seems like either “Easter is the best holiday” is a proposition with a truth-value, or it lacks a truth-value precisely because it isn’t a proposition. Either way round, it doesn’t seem to be any reason to doubt bivalence.
He also says that it cannot be proven. But if “Easter is the best holiday” is just taken as a proposition, then it can be proven in the same way as any other proposition:
- If p, then “Easter is the best holiday”.
- Therefore, “Easter is the best holiday”.
Why IP thinks we cannot enter “Easter is the best holiday” into a proof like this is a mystery.
IP concludes that the argument doesn’t work, on the basis that propositions like “Easter is the best holiday” are neither true nor false. As we have just seen, his reasons for thinking that this sort of proposition is neither true nor false are pretty unconvincing. But let’s just grant them for the sake of the argument.
He doesn’t seem to realise that if “Easter is the best holiday” is neither true nor false, then he is effectively conceding exactly the thing that the argument was supposed to be showing, i.e. that there are exceptions to classical logic. If his own example were genuinely an example of a proposition that lacked a truth value, this would be enough to undermine classical logic. So, he isn’t showing something about the argument that is wrong; he is just giving another (albeit more flawed) instance of a counterexample to classical logic.
At around 2:20, IP moves on to talk about Kurt Gödel:
“The argument itself is based on Gödel’s theorems, which many think shows logic doesn’t work”.
I think what IP has in mind is that there is another type of challenge to classical logic, this time coming from Gödel’s incompleteness theorems. He gives a statement about what the incompleteness theorems show, but it crucially mistakes (and overstates) their true significance. This leaves IP drawing all the wrong consequences.
IP says that Gödel’s incompleteness theorems show that:
“No consistent system of axioms whose theorems can be listed by an ‘effective procedure’ is capable of proving all truth”
This statement stands out a bit in the video, and it sounds like IP has got it from somewhere, but he never gives any citations for this quote, so we have to guess. My first guess was Wikipedia, and I was right. What is revealing about the quote is what he leaves off. Here is how it shows on Wikipedia:
The quote in full (with the bit he missed off in italics) is:
“No consistent system of axioms whose theorems can be listed by an ‘effective procedure’ is capable of proving all truths about the arithmetic of the natural numbers“.
There is a very big difference between showing that no consistent system of axioms can prove all truth, and showing that they cannot prove all truths about the arithmetic of the natural numbers. I don’t know if he didn’t think the extra bit he left off wasn’t important, or if he did it on purpose to jazz up his point, but either way leaving it off completely changes the significance of Gödel’s incompleteness theorems.
The thing is that (when we look at it properly) Gödel’s incompleteness theorems do not pose a direct local challenge to classical logic. What they show is compatible with non-contradiction, excluded middle and the law of identity all being true (along with all the other principles of classical logic).
What the theorems show is that any system of logic that is powerful enough to express all the arithmetic propositions cannot prove all of them.
So, the result applies to a certain type of logic, called ‘mathematical logic’. This logic is built up out of first-order logic, which is itself a very basic type of classical logic (one that respects all the principles IP presumably wants to defend). If you add the right axioms to this logic, then it becomes capable of expressing things like 1+1=2, etc. Once it is able to do that, we call it mathematical logic. Gödel’s incompleteness theorems apply specifically to mathematical logic.
And because this mathematical logic itself respects the classical principles (it is a type of classical logic), this means that Gödel is just telling us something about the limits of a certain type of classical logic (classical logic that is capable of expressing arithmetic). It is pointing out a limitation in mathematical logic. That is not itself a straightforwardly a reason to think that classical logic is not the correct logic, or that the ‘laws of logic’ are not true.
Except… it might be.
The strange thing about Gödel’s proof is that it shows that arithmetic, and any more complex bit of mathematics, cannot be modelled in classical logic without having ‘blind spots’, where there is something which is true but not provable in that logic. Yet, we might just think that we obviously can prove everything in arithmetic; we might just find the limits of proof in mathematical logic to be an unacceptable consequence. Well, if you did think this, then you could use this as a reason to think that there must be contradictions.
This is because the actual theorems can be thought of as ‘either-or’ statements. They can be thought of as saying ‘either mathematical logic is consistent but has blind-spots, or it has no blind-spots but it has some contradictions in it’ – Gödel is telling us that mathematical logic is either incomplete or inconsistent – either there is something that is true but not provable, or the law of non-contradiction is false.
If you thought that the price (of denying non-contradiction) was worth it so that you didn’t have any of these weird blind-spots in your proof-theory, then you might be willing to accept the inconsistent option. Most people find contradictions more troubling than blind-spots though, and so don’t go that route. But, that is probably the most direct sort of attack you could make from Gödel against classical logic.
If you were feeling charitable, you might think that this is the sort of challenge that IP had in mind. But he dropped off the bit of the quote from Wikipedia which specifically says that Gödel’s theorems are about mathematical logic, not all logic (or even all of classical logic). I find it hard to believe that he didn’t read the end of the sentence he quoted, so either he didn’t understand that the bit he left off is crucial to understand the theorems, or he is deliberately overstating their importance. Either way, it is not great.
Now, if you know a little bit about Gödel, then you might know that in addition to his incompleteness theorems, he is also well known for his completeness theorem. This showed that the basic (classical) first-order logic is actually complete, meaning that it definitely doesn’t have any of those weird blind-spots that the extended mathematical logic has. So without the extra axioms added to first-order logic, it is capable of proving all its own truths.
And this is where we see why leaving off that bit from the Wikipedia quote was so telling. The way IP tells it, the significance of Gödel’s incompleteness theorems is that logic ‘cannot prove all truths’, which sounds like a very profound, almost mystical insight into what people can know and what they can’t. But, in reality, Gödel’s incompleteness theorems only show that some types of logic cannot prove all of their own truths. Admittedly, it is a very important class of logical systems, as it is the ones that model mathematical logic, but it is not as widespread as IP makes out. And Gödel’s completeness theorem actually proves that there are other types of logic for which this is not the case. There are also many other famous completeness theorems in logic (such as Kripke’s celebrated completeness theorem for the modal logic S5, which wouldn’t be possible if IP was right about what Gödel’s incompleteness theorems said!).
IP summarises what he thinks Gödel showed us as follows:
“All Gödel did was show that we are limited in having a total proof of something, but even without Gödel that is intuitively obvious. Many things will only be 99% probably true. But absolute certainty will always be beyond our reach”.
In reality, the significance of Gödel’s incompleteness theorems is not at all intuitive. Almost nobody expected mathematical logic to be limited in the way he showed it was. IP seems to think that Gödel just used maths to show that we can never really know anything for certain. This is demonstrably a bad interpretation of Gödel, and IP clearly has no idea what Gödel really showed us.
On the other hand, I agree that there is no particularly compelling reason to give up classical logic due to Gödel’s incompleteness theorems. I don’t find the idea of accepting contradictions just to get around incompleteness of arithmetic to be persuasive. It’s just a pity that IP wasn’t able to explain what Gödel said, how that was relevant to classical logic, and how it doesn’t mean we should reject classical logic. It’s more a case of a stopped clock accidentally showing the right time.
4. G Spencer-Brown
In the next main bit (around 3:10), IP brings up a different philosopher (or mathematician, depending on how you look at it), G Spencer-Brown, and the section he takes up is from Spencer-Brown’s book, Laws of Form. Now, this is a very strange book on logic, and not within the mainstream work on logic that philosophers usually debate. That is not to say that it is not of any value, but just to be aware that it is already a weird reference. The bit of that book that IP seems to have read is merely the preface, so it is quite easy to check for yourself (just pages ix – xii).
Anyway, IP is going back to the 3rd premise of the argument, which is the idea that the Liar Proposition is neither true nor false. He seems to be saying that Spencer-Brown advocates a solution to the problem which avoids having to postulate that the proposition is neither true nor false. This is presumably done in order to rescue the ‘laws of logic’ from the attack, and to defend classical logic.
So, the thing about the liar proposition, i.e. “This proposition is false”, is that if you assume it has a truth-value (true or false), then it sort of switches that truth-value on you. To see that, assume it is true. That would mean that what it says is the case. But what it says is that it is false. So if it is true, then it is false. The same thing happens if we assume it is false. So, we might say that any input value gets transformed into its opposite output value; true goes to false, false goes to true.
And this feature, or something similar to it, is also seen in the following mathematical example that Spencer-Brown brings up in the preface to Laws of Form. So consider the following equation:
X = -1/X
If you try to solve the equation by assuming that X = 1 (i.e. if we substitute X for 1), then we get:
1 = -1/1
However, -1 divided by 1 equals -1 (because any number divided by 1 equals itself), so: -1/1 = -1. But that means that:
1 = -1/1 = -1
The ‘input’ of 1 gets turned into the ‘output’ of -1. If we try to solve the equation by assuming that X = -1, then we get the converse result (because any number divided by itself equals 1):
-1 = -1/-1 = 1
So the assumption of X = 1, results in an output of -1. And the assumption of X = -1 results in the output of 1. This is a bit like what is going on with the liar proposition if we think of 1 being like ‘true’, and -1 being like ‘false’. In both cases, the input value gets switched to the alternative value.
IP says that the ‘solution’ to this problem is to use an ‘imaginary number‘ i, which is √-1. What he means is that if we assume that X = i, then we get the following solution to the equation:
i = -1/i
Because i is the square root of -1, it is already -1/i. So:
i = -1/i = i
Unlike when we assumed X was 1 or -1, where the output got switched, if we assume the input is i, then the output doesn’t get switched. Ok, got it.
The first thing to note here is that this sort of consideration is what motivated mathematicians to consider changing how they thought about mathematics. And not without some resistance. Descartes apparently used the term ‘imaginary’ as a derogatory term. Nevertheless, mathematicians were convinced that introducing imaginary numbers into their understanding of mathematics, despite being unintuitive to some extent, was warranted due to the utility that doing so brought about. What Spencer-Brown is pointing to is a reason for re-conceiving traditional mathematics.
How does this relate to the liar proposition? Unfortunately for IP, it doesn’t relate in the way he wants it to. Also, he says almost nothing about how this is supposed to relate to the liar’s paradox. He says something, it is not helpful. What he says is:
“The only problem is that we cannot epistemically understand the mathematical usage of i. And thus Gödel was proven right and not the absolute skeptic who doubts logic is true”.
Now, IP is obviously wandering off down the wrong path here. Clearly, IP finds imaginary numbers hard to think about, but it is not clear what that has to do with anything. His comment about Gödel betrays his poor grasp of his work as well. Because Spencer-Brown explained how to use i in an equation, that proves that Gödel was right? Hardly.
What is actually going on here, what IP seems unable to get, is that Spencer-Brown is not advocating for classical logic. In fact, he is quite out-there as a thinker, and proposing something quite radical. Let’s look at what Spencer-Brown says about the mathematical example that IP brought up, and how it relates to the liar paradox:
“Of course, as everybody knows, the [mathematical] paradox in this case is resolved by introducing a fourth class of number, called imaginary, so that we can say the roots of the equation above are ±i, where i is a new kind of entity that consists of a square root of minus one.” (Spencer-Brown, Laws of Form, page xi, bold added by me)
Spencer-Brown is saying that the solution to the mathematical puzzle requires the addition of a “new kind of entity” to mathematics. A new kind of number. He then goes on in the next paragraph to explain how this mathematical lesson applies to logic:
“What we do in Chapter 11 is extend the concept to Boolean algebras, which means that a valid argument may contain not just three classes of statement, but four: true false, meaningless and imaginary.” (ibid)
So Spencer-Brown is playing around with a type of logic which has four truth-values, not two like classical logic has. This makes it a very exotic type of non-classical logic! IP doesn’t mention this passage, which clearly shows Spencer-Brown freely speculating on a type of logic which is very different from classical logic.
So, what we have here is an example of someone saying that the right way to solve the liars paradox is to modify classical logic in some fundamental way. IP seems to think that this example makes the point he wants to make, but if anything it points in the opposite direction completely. Far from showing that the laws of classical logic cannot be questioned, it is an example of someone questioning the laws of classical logic.
So far we have seen that IP has no real idea what the skeptical challenge to logic really consists in. He knows that sometimes people talk about reasons to doubt things like non-contradiction or the law of excluded middle, and he seems to take this to be a very radical attack on logic itself. However, we saw that he presented an argument that attempted to attack the claim that the laws of logic are true, and he hopelessly misunderstood it. It was showing that if the Liar Proposition is neither true nor false, then classical logic isn’t correct. In response, he proposed that “Easter is the best holiday” was neither true nor false, which is itself very poorly argued for, but even if it were correct would be another reason to reject classical logic. He then utterly failed to grasp Gödel, and may have deliberately misstated the theorem’s significance. Lastly, he looked at a passage from Spencer-Brown, but failed to see that if it was correct, it would be a reason to prefer a four-valued logic over the classical two-valued logic.
There is still another half of his video to go, and I will try to get round to debunking the claims made in that half as well when I get a chance.