# I have trouble understanding this fallacy: "If A, then B. Therefore if not-B, then not-A."

About "If A, then B. Therefore, if not-B, then not-A":

From what I understand the conclusion is wrong, because it is not said that A is a sufficient condition for B, (and there may be other conditions required for B, so if they are not present B won't be the case, even if A is the case.).

But I have trouble finding a real life example to this and I'm not sure if it is the consept that I don't understand or it is just the way we express ourselves in natural language that causes the confusion.

Here is an example I could think of, in natural language: "In order to have internet (B), the router must be on (A). So, if there is no internet (not-B), that means the router is not on (not-A)" (Which is not true because there may be a problem with the providers, for example.)

So turning the router on is a necessary condition for having internet but can you then say: "If you turn the router on (A), there will be internet (B)." Saying that is just is not true. So is it just that we don't have a good way to express sufficient conditions (in comparison to necessary conditions) or is it that I just don't understanding some fundamental concept (or both)?

• Comments are not for extended discussion; this conversation has been moved to chat. Feb 26, 2019 at 11:13
• You know "A->B". Then you assume "Not-B" is true. If A were true, you would have both "B" and "Not-B", so necessarily, A is false. Feb 27, 2019 at 9:38
• Its not a fallacy lol. Nov 18, 2019 at 11:05
• Maybe you are confusing that if A->B, and B is true, then we cannot necessarily conclude A. This has one of those formal names I mis-remember just now. Feb 28, 2020 at 15:16
• @puppetsock - The formal name for that fallacy (knowing A->B and B, and incorrectly claiming that proves A) is affirming the consequent. The OP might also have confused their own non-fallacious statemenent with the fallacy of denying the antecedent where someone knows A->B and ~A, and incorrectly claims that proves ~B. Feb 28, 2020 at 17:01

There are multiple good answers here, including references to modus tollens and the contrapositive (both of which are correct).

What helps me understand this concept is a more intuitive/layman's perspective.

Say that if it's raining, you ALWAYS bring an umbrella outside. (Assume that you have a perfect memory). Therefore we have:

Raining -> Umbrella

However, if you have an umbrella, does it mean that it's raining? Well, no. You might be bringing an umbrella for a trip, using it as a parasol in the scorching summer sun, or even simply returning it to a friend.

But if you don't have an umbrella, is it raining? No, and this is supported by modus tollens/contrapositive. That is to say:

Not(Umbrella) -> Not(Raining)

And that makes sense to us. If it were raining, you would've brought your umbrella. We said that you ALWAYS bring an umbrella if it's raining. So if you don't have it, it can't possibly be raining.

Let's apply it to your example:

"In order to have internet (B), the router must be on (A). So, if there is no internet (not-B), that means the router is not on (not-A)"

This is actually not formulated properly; A and B have been reversed. This is probably why it's confusing. (Intuitively, it's in line with what you've been saying: if we don't have Internet, it's not necessarily because of the router.)

So let's word it differently:

If we have an Internet connection (A), the router must be on (B).

InternetConnected -> RouterOn

A -> B

But if the router isn't on, then we can't possibly connect to the Internet. (How would we even do that?) We know this is the case, because we said that if we're connected to the Internet, the router MUST be on.

Not(RouterOn) -> Not(InternetConnected)

Not(B) -> Not(A)

Hope this helps! :)

Edit: My "rain" example was criticized for clarity, since the real-life relationship isn't as straightforward. (Fair enough.) We can apply the same thing to dogs, though, and it works the same way but exploits the IS-A/inheritance relationship you're probably familiar with if you have a CS background.

Corgi => Dog

All corgis are dogs. But are all dogs corgis? No! However, we DO know that if it's not a dog, it's clearly not a Corgi (since a Corgi is a type of dog).

Not(Dog) => Not(Corgi)

Also, to echo the other users: this is not a fallacy. You can find more about logical fallacies here.

‘If not B then not A’ is the contrapositive of ‘If A then B’ and is logically valid.

By saying ‘if A then B’, the author is saying that whenever A happens, B will definitely happen. Hence, if B does not happen, it is clear that A did not happen (if not B then not A).

You are correct: The statement does not show a sufficient condition!

Consider this statement: "If (A) it rains, then (B) the street will be wet."

Assuming the street is always wet after it rained (because it is not covered by a roof or something), the contraposition would hold true!

Contraposition: "If the street is not wet, it did not rain"

I think what is giving you troubles is that the street in this example can be wet without rain (someone sprayed water onto it), therefore the street is not always wet just because of rain. That's called an inverse and differentiation can sometimes be hard.

• That's a strange way of contradicting yourself. A ==> B strictly forces !B ==> !A . That's what the OP was asking... unless you're just pointing out that he contradicted his own question. Feb 26, 2019 at 15:33
• @CarlWitthoft I was saying that the statement is not complete. The street ALWAYS has to be wet when it rained. The internet (in OPs example) works ONLY because of that router. In real life, conversations are not really always fully logical statements. The example he used, shows that well, I think. Feb 27, 2019 at 17:56

Your confusion might stem from the fact that implications in propositional logic don't always correspond intuitively to cause and effect in the real world. I'll get back to your example later; for now, let's focus on a more intuitive implication:

If [a: there is fire], then [b: there is smoke].

This is a natural process in which the cause (fire) always produces the effect (smoke), and that's a situation that can be described with an implication. (In the real world, there are substances that burn without smoke, so let's constrain ourselves to wood fires.) In this situation, you probably agree that the contrapositive is true as well:

If [not(b): there is no smoke], then [not(a): there is no fire].

This makes intuitive sense - we're not observing the effect that is always produced by the cause, so the cause can't be occurring.

We can formally prove why the contrapositive is equivalent to the original statement. Here is the truth table that defines implications (T for true and F for false):

``````| a | b | a -> b |
+---+---+--------+
| F | F | T      |
| F | T | T      |
| T | F | F      |
| T | T | T      |
``````

So an implication is only false if the left hand side is true and the right hand side is false. Let's interpret this via the example. Our implication is intended to be a description of the real world, so all the true rows should correspond to situations that can happen in the real world, and all the false rows should correspond to situations that can not:

1. It can happen that there is no fire and no smoke.
2. It can happen that there is no fire, but there is smoke (which might be produced by something else than fire).
3. It can not happen that there is fire but no smoke.
4. It can happen that there is both fire and smoke.

Now, let's add the contrapositive to the truth table:

``````| a | b | a -> b | not(b) | not(a) | not(b) -> not(a) |
+---+---+--------+--------+--------+------------------+
| F | F | T      | T      | T      | T                |
| F | T | T      | F      | T      | T                |
| T | F | F      | T      | F      | F                |
| T | T | T      | F      | F      | T                |
``````

The `not(b) -> not(a)` is identical to the `a -> b` column, so the two statements are logically equivalent.

In your example, you seem to think that the router is the cause that produces the effect of internet. However, unlike in the fire and smoke example, internet isn't a guaranteed effect of the router alone, and therefore there is not an implication from router to internet. Perhaps surprisingly, there is an implication, but it goes the other way. The router is just one of several necessary ingredients for having internet, and in an implication, the necessary part goes on the right hand side - so even though the router is one of the technical "causes" of the internet, the implication looks like this (assuming that the router is your only means of connecting to the internet):

If [a: you have internet], then [b: the router is on].

This unintuitive order is easier to understand if you stop thinking about an implication as the left hand side being a physical cause of the right hand side, and just think of it as the statement "it is not the case that the left hand side is true while the right hand side is false". Let's confirm that this is a correct description of the real world (the only thing that shouldn't happen is that the left hand side is true and the right hand side is false):

1. It can happen that you don't have internet and the router is off.
2. It can happen that you don't have internet, but the router is on (so you've got a tech problem somewhere else, for example with your computer's network card or with your connection to your internet service provider, or your ISP might have problems - this is why the router being on is not sufficient).
3. It can not happen that you've got internet but the router is off (again, this assumes that the router is your only means of connecting to the internet).
4. It can happen that you've got internet and the router is on.

In this example, the contrapositive is actually the more intuitive one:

If [not(b): the router is off], then [not(a): you don't have internet].

(My source for everything in this answer is the first chapter of "Discrete Mathematics and Its Applications" by Kenneth H. Rosen, but I've added a few Wikipedia links.)

No fallacy

Router necessary for internet

Can be restated

Internet sufficient for router

Let's restate the second more elaborately

Internet (found to be) working is sufficient (evidence) that the router (has to be) working

As others have said, there is no fallacy in a contrapositive argument, i.e. "if A then B. Not B, therefore not A". But your example is faulty:

Here is an example I could think of, in natural language: "In order to have internet (B), the router must be on (A). So, if there is no internet (not-B), that means the router is not on (not-A)" (Which is not true because there may be a problem with the providers, for example.)

What you have written is "if B then A. Not B, therefore not A"--which is of course invalid reasoning. A better way of putting an example of the contrapositive into natural language would be "If we have an internet [connection], then the router is on. The router is off [not on], therefore we do not have an internet [connection]." Which is a valid argument of the contrapositive form.

"In order to have internet, the router must be on" corresponds to the implication "if there is internet, the router is on." The contrapositive is "if the router is not on, there is no internet."

The router being on is a necessary condition for internet, while having internet is a sufficient condition for the router being on. (It can be tricky translating statements to implications, especially when there are modals like "must"! The translation reduces the statement to its barest essence with a material implication, and the translation is usually lossy.)

Your translation from natural English to formal logic is where you go wrong.

"If-A then B" precisely means that A is a sufficient condition for B. - Ie if we have case A we can know that we also have case B.

"In order to have internet (B), the router must be on (A)." describes a necessary condition. This can be written as "Only if A then B" or "If B then A". The contrapositive would be "If Not-A then Not-B"; Ie. "If the router is not on then we do not have internet."

• This is the crux of the matter. "if A then B" does not "require other conditions for B". It simply states that if A holds, then B must also hold. Feb 26, 2019 at 19:09

As others have said, this is not a fallacy.

Visuals can make this easier without the need to follow potentially confusing logical progressions, or the need to know any logical terms. Here's a Venn diagram:

• "If A then B" means that A is a subset of B.
• Anything outside of B ("not-B") is clearly "not-A"
• For good measure, if you also want to consider "not-A", this means "everything outside of A", which could be B, but could also mean "not-B".

Extending this to the router & internet example in the question:

• The internet is only available to you if your router is on
• It is very possible to have no internet availability, even when your router is on
• The internet is certainly not available to you if your router/modem is not on

It's not a fallacy, but it's only guaranteed to hold in Aristotelian logic, where we have the law of the excluded middle, which states that every proposition is either true or false.

Example: Anyone who is a Christian believes that Jesus is their savior.

A is the proposition that my dog is a Christian.

B is the proposition that my dog has a belief in Jesus as her savior.

Now B is false, because my dog has no belief in Jesus as her savior. If you're working in Aristotelian logic, then it's a perfectly valid logical conclusion to say that my dog is not a Christian: if B is false, then A must be false.

But it's at least as reasonable to apply non-Aristotelian logic here. We can say that Christian and non-Christian are silly categories to apply to a dog. Then we say that neither A nor not-A is true. The truth-value of A is undefined. The fact that B is false tells us that A is not true, but it doesn't tell us that not-A is true.

Most mathematicians spend most or all of their time working within Aristotelian logic, but that doesn't mean that non-Aristotelian logic is invalid or not accepted by the mathematical community. Similarly, some people play basketball, some play tennis, and some play both. There is no controversy over which set of rules is the right set of rules, and everyone understands implicitly which set of rules is agreed upon when they're playing.

In most cases in everyday life, Aristotelian logic just isn't a good model of how we reason about the world. If a friend asserts that jazz is more fun than pop music, I would probably answer, "Well, yes and no..."

Yes, you are confused due to the use of some natural language which are by nature all ambiguous/not precise (but that makes them very powerful!). Ancient philosophers were strongly interested on this problem and that leads to formal logic.

For contrapositive the real life example could be:

if it's raining then the ground is wet.

And its contrapositive formulæ :

if the ground is not wet then it is not raining.

Sorry for @eurieka's answer but his example is not so good due to the fact that the real life relation in between rain and umbrella use is not so direct (I never use umbrella raining or not). While it is more easy to agree on rain and wet ground.

In formal logic you can't repudiate a contrapositive, if you think the contrapositive is not correct then the initial proposition is also wrong.

In formal logic, you can't state something like : router on implies internet accessible, and then after add another condition about provider state. If it is the case, then you need to reformulate your problem taking into account everything.

"If A, then B. Therefore, if not-B, then not-A"

This is not a fallacy because what it is stating is equivalent to this:

"If there is A, then there must be a B--every time. Therefore, if there is no B, there could not have been an A, because had there been an A the rule is that there must be a B. So in any case where there is no B, there must have been no A."

You probably have mistranslated the statement :

"In order to have internet (B), the router must be on (A). So, if there is no internet (not-B), that means the router is not on (not-A)"

It is not a fallacy, but rather a mistranslation.

If you have something that comes before `therefore`, it is already a premise.

If A then B

Is already supposed to be the case. It does not say that there are other premises that contribute to the conditional.

It is analogous to: `If it is raining then there are clouds`.

And we can infer from it that if there are no clouds then there is not raining.

So, there is no fallacy here.

But remember this : Deductive reasoning is about the logical form of your argument.

If we suppose (just suppose) that the conditional `if A then B` is true, then it follows that the conditional `If not-B then not-A` is true.

Suppose that the conditional `if A then B` lacks `C`, which is also needed for B to take place.

Well, that's not deductive problem (not a non-sequitur), but it is a problem related either to the truth or reliability of the premise, or simply a mistranslation as I suggested.

If I say that `If you worked then you did make money` , then it is the same as if I say `if you did not make money, then you did not work`.

But whether work is sufficient to make money or not, that's another subject. Deductive arguments are only about the truth of the conclusion, given the premises are true.

Now back to your example :

"In order to have internet (B), the router must be on (A). So, if there is no internet (not-B), that means the router is not on (not-A)"

In this case `in order to have internet (B) the router must be on (A)` should not be translated to `if A then B` , but instead `if B then A`.

If there is internet -B- then the router is ON -A-.

That is the correct conditional here :

• If there is internet then the router is On
• therefore, if the router is not-On then there is no-Internet.

So, your premise was mistranslated .

Consider the clouds example, to see what I mean : Even though clouds can cause rain (with other factors), it does not mean that clouds should be the antecedent in a conditional, so `if there are clouds then it is raining` is a false premise.

But it means that `if it rains then there are clouds`, so the cause should be a consequent, not an antecedent. Think about it.

• Are you sure? Feb 26, 2019 at 13:09
• Clouds do not have to be necessarily right about one's head, it is possible for the clouds to be in a distant area, and wind does the rest. So, it rains therefore there are clouds (either here or somewhere else) Feb 26, 2019 at 13:32
• True, but then that complicates the logic. E.g. no clouds -> no rain, do we mean clouds that you can see? Because the clouds can be elsewhere and causing rain. But if you mean clouds anywhere (whether visible or not), there is always a cloud somewhere (and indeed, probably always rain somewhere). Also, I'm not a meteorologist, but it seems to me we can have a situation where the cloud has just finished evaporating, and the last drops of rain are still falling. Feb 26, 2019 at 13:42
• No, the conditional means "there exists a cloud" whether we see it or not, even if not right above us. There is no rain without clouds. This premise is inductive in nature: so there is an implied "probably". Feb 26, 2019 at 15:05

"If A, then B. Therefore, if not-B, then not-A":

A == Duck

B == Swim.

If A, then B == If it's a Duck then it can swim.

if not-B, then not-A == If this little guy can't swim, then it cannot be a duck.

Why? Because a duck can swim! how can it be a duck but cannot swim?

Isn't this logic easy to see?

• Thanks. I didn't have any reference but was only typing this answer out hoping to help make the logic a little easier. Thanks for the pointer though. Apr 29, 2020 at 23:04

The way I read this, it could either be true or not true, depending on presuppositions.

For example, If the world is covered exhaustively by [(A -A), (B -B)] and exclusionary,
then "If A, then B. Therefore, if not-B, then not-A" would be true.

However, if the world is only partially covered by [(A -A), (B -B)],
(for rocks can be not-ducks isasmuch as swimming hippos are not-ducks)
then "If A, then B. Therefore, if not-B, then not-A" can be false (contains a fallacy).

Consider this example for situating in exhaustive world of either/or:

If you don't like the OCC blueprint (A), you are either corrupt and on the take or you're not educated or intelligent enough to digest the OCC's proposal (B).
What bucket should I put Governor Brainard in?

(If A, then B. Since not-B, not-A;
Since I am either educated enough or not corrupt, I must like OCC.
But I don't like OCC, so must I be either corrupt or uneducated?)

I guess you'd have to make that decision yourself.

Was there a third bucket?

I wouldn't offer up an opinion.

Which do you think applies to me?

You'd have to make a determination on how you'd want to vote.

Is there a third category you'd like to develop for people like me that don't agree with you?

Not at this time.

I would encourage you to understand that some people who don't agree simply have read, have studied it, are economic experts or have been on this committee, leading this committee currently, and simply have a different vision than you do, and I would ask you to please be respectful, particularly in the press, about how you talk about those who disagree with you.
Because rule-making is a collaborative public process and each of us whether we're a Congressperson or just a member of the public has the right to disagree. That doesn't mean we don't understand or that we're somehow on the take under the current system. It just means we have a different vision. ~Katie Porter

Illustration of principle of exhaustion using relation
A = -(-A)
prepared = not (unprepared)

In my opinion, I am not unprepared for what you ask about; ..

"This is why Mrs. LADD-FRANKLIN proposes to call them respectively the principle of exclusion and the principle of exhaustion, inasmuch as, according to the first, two contradictory terms are exclusive (the one of the other); and, according to the second, they are exhaustive (of the universe of discourse)."

Another way of looking at this is with a diagram.

In your example, “If A, then B” is consistent with A and B both being true, or A being false and B being anything — but not with A being true and B being false.  You could draw that as:

``````    A:║ true │ false
B    ║      │
══════╬══════╪══════
true  ║  ✔️   │  ✔️
──────╫──────┼──────
false ║ ✘✘✘  │  ✔️
``````

But that's exactly the same diagram you'd get for “If not-B then not-A.”  So the two are equivalent.

(As per other answers, this assumes classical two-valued logic, of course.  This is basically a Karnaugh map, though that's mainly used for simplifying boolean expressions.  You could alternatively use a truth table, as per other answers, but I find this diagram easier to follow.)

I think that you need to express the problem as so: If A->B implies that B will happen ALWAYS when A has occurred. Then, if B has not occurred, you are sure that A has not occurred.

• but will B only happen if A? Mar 1, 2019 at 19:37

Maybe you can say it is a fallacy, if you consider it this way:

1. If anyone burns this piece of normal paper, then it will become ash. Now it is not ash, therefore no one burned it. This might be obvious and somewhat harder to refute.

2. If anyone kicks this water bottle standing upright on the ground, it will topple and land sideway. It is not appearing as sideway, so nobody kicked it. And many people make judgment this way. But the thing is, it should be "it probably will (or may) topple and land sideway", because there are cases, perhaps one in 100,000 times that it will land upright, as people have shown on YouTube. So in this case if you apply the direct "not B, therefore not A", it will be a fallacy.