If I'm given a well formed formula $\varphi$ that only has the logic symbols $\land,\lor,\neg$. I want to define a formula $\varphi^*$ that is a result of switching every sign $\land$ to $\lor$ and every $\lor$ to $\land$ and every atom to its negation in $\varphi$.

So, I thought defining $\varphi^*$ will be :

Base step: $\varphi$ is an atom then $\varphi^*=\neg{\varphi}$.

Induction step: Here I'm not having a clear thought about defining this so i was thinking of letting $\alpha,\beta\in{WFF}$ that satisfy $\varphi$ .then i gave my self some examples and saw that i need to apply negation to the formulas $\alpha,\beta$ that construct $\varphi$ based on counting $\land$,$\lor$.

My examples:

$\varphi=(p_1\land p_2)\implies \varphi^*=\neg\varphi$.

$\varphi=(p_1\land p_2)\lor p_3 \implies\varphi^*=\neg\neg\varphi$

Is this correct?

  • $\begingroup$ Induction step $\alpha := \varphi_1 \lor \varphi_2$. Then $\alpha^* := \varphi_1^* \land \varphi_2^*$, and so on. $\endgroup$ May 10, 2018 at 11:19
  • $\begingroup$ Sorry but I'm a novice at this stuff ,can you please clarify more to where each symbol belongs to? ,And preferably a full solution to the induction step definition.And is my induction step entirely incorrect ? $\endgroup$ May 10, 2018 at 11:28

2 Answers 2


Terminological note: There isn't really such a thing as an 'inductive definition'. Rather, what you are trying to do is to come up with a recursive definition.

However, once such a recursive definition is in place, you can use structural induction to prove things about it.

For example, using the recursive definition as given by @Taroccoesbrocco, we can now prove by structural induction that $\varphi^* \Leftrightarrow \neg \varphi$ (I wouldn't say $\varphi^* = \neg \varphi$, since within the context of metalogic the '$=$' is used to indicate that the two formulas are syntactically/symbolically the same symbol string, whereas of course you just want to prove logical equivalence, for which we use $\Leftrightarrow$)

OK, so here goes that proof:

Base: $\varphi$ is atomic. Then by definition, $\varphi^* = \neg \varphi$

Step: There are three cases to consider:

Suppose $\varphi = \neg \varphi_1$. By inductive hypothesis we can assume $\varphi_1^* \Leftrightarrow \neg \varphi_1$. But that means that $\varphi^* = (\neg \varphi_1)^* = \neg \varphi_1* \Leftrightarrow \neg \neg \varphi_1 = \neg \varphi$

Suppose $\varphi = \varphi_1 \land \varphi_2$. By inductive hypothesis we can assume $\varphi_1^* \Leftrightarrow \neg \varphi_1$ and $\varphi_2^* \Leftrightarrow \neg \varphi_2$. Hence: $\varphi^* = (\varphi_1 \land \varphi_2)^* = \varphi_1^* \lor \varphi_2^* \Leftrightarrow \neg \varphi_1 \lor \neg \varphi_2 \Leftrightarrow \neg (\varphi_1 \land \varphi_2) = \neg \varphi $

Suppose $\varphi = \varphi_1 \lor \varphi_2$. By inductive hypothesis we can assume $\varphi_1^* \Leftrightarrow \neg \varphi_1$ and $\varphi_2^* \Leftrightarrow \neg \varphi_2$. Hence: $\varphi^* = (\varphi_1 \lor \varphi_2)^* = \varphi_1^* \land \varphi_2^* \Leftrightarrow \neg \varphi_1 \land \neg \varphi_2 \Leftrightarrow \neg (\varphi_1 \lor \varphi_2) = \neg \varphi $

  • $\begingroup$ This what i meant for the problem that i asked @Taroccoesbrocco in the comment $\endgroup$ May 10, 2018 at 12:32

The induction step corresponds to the cases $\varphi = \lnot \varphi_1$, $\varphi = \varphi_1 \lor \varphi_2$ and $\varphi = \varphi_1 \land \varphi_2$, where $\varphi_1$ and $\varphi_2$ are well-formed formulas that only have the logic symbols $\lnot$, $\lor$ and $\land$. By induction hypothesis, $\varphi_1^*$ and $\varphi_2^*$ is already defined (this is the "magic" aspect of definitions by induction), so you just have to set:

  1. $\varphi^* := \lnot (\varphi_1^*)$ if $\varphi = \lnot \varphi_1$,
  2. $\varphi^* := \varphi_1^* \land \varphi_2^*$ if $\varphi = \varphi_1 \lor \varphi_2$,
  3. $\varphi^* := \varphi_1^* \lor \varphi_2^*$ if $\varphi = \varphi_1 \land \varphi_2$.
  • $\begingroup$ Perfect just one more thing,can it be proven that for every well formed formula $\varphi^*=\neg{\varphi}$ using structural induction? $\endgroup$ May 10, 2018 at 11:57
  • $\begingroup$ @user3133165 - You can't prove that $\varphi^* = \lnot \varphi$ if "$=$" is interpreted as syntactic equality, since for instance, given an atomic formula $X$, you have $(X \land X)^* = \lnot X \lor \lnot X \neq \lnot(X \land X)$. But you can prove by straightforward structural induction that $\varphi^* \equiv \lnot \varphi$ where "$\equiv$" stands for logical equivalence (i.e. same truth table). $\endgroup$ May 10, 2018 at 12:17
  • $\begingroup$ I think I've explained incorrectly trying to simplify it,I think it will be better if i open another question for this $\endgroup$ May 10, 2018 at 12:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .