Expansion postulates in Belief revision logic

65 Views Asked by At

I've just started looking into epistemic logic and belief revision approaches, and I'm struggling already with proving some basic properties.

I am given the following definition:

Let $\mathcal{L}_0$ be the set of propositional formulas, generated by some set of atoms $P$, and the classical connectives. Let $Cn(.)$ be classical consequence operator, i.e. $Cn(\Sigma) = \{ \sigma \in \mathcal{L}_0 | \Sigma \vdash \sigma \}$. A belief set $\mathcal{K}$ is a set of propositional formulas closed under $Cn(.)$, i.e. $Cn(\mathcal{K}) = \mathcal{K}$.

I'm looking at the expansion postulates:

Type: $(\mathcal{K} \oplus 1): \mathcal{K} \oplus \varphi$ is a belief set

Success: $(\mathcal{K} \oplus 2): \varphi \in \mathcal{K} \oplus \varphi$.

Expansion: $(\mathcal{K} \oplus 3): \mathcal{K} \subseteq \mathcal{K} \oplus \varphi$

Minimal action: $(\mathcal{K} \oplus 4):$ if $\varphi \in \mathcal{K}$ then $\mathcal{K} = \mathcal{K} \oplus \varphi$

Monotony: $(\mathcal{K} \oplus 5):$ For all $\mathcal{H}$, if $\mathcal{K} \subseteq \mathcal{K}$ then $\mathcal{K} \oplus \varphi \subseteq \mathcal{H} \oplus \varphi$.

Minimal change: $(\mathcal{K} \oplus 6): \mathcal{K} \oplus \varphi$ is the minimal set satisfying $(\mathcal{K} \oplus 1 )- (\mathcal{K} \oplus 5)$

These postulates I assume are meant to be intuitive representations of adding new beliefs to our sets.

I am now trying to prove the following:

A function $\oplus$ satisfies $\mathcal{K} \oplus 1- \mathcal{K} \oplus 6$ iff $\mathcal{K} \oplus \varphi = Cn(\mathcal{K} \cup \{\varphi\})$.

One difficulty I'm having is about what I should assume about $\vdash$. I'm a bit confused in terms of whether I am meant to assume the ordinary natural deduction properties, or a specific system. Would really appreciate someone taking a look at my reasoning and helping with the bits I can't solve.

My attempt so far:

$\Rightarrow$. Assume postulates 1-6. I first want to show that $Cn(K \cup \varphi) \subseteq K \oplus \varphi$. By postulate 2 and 3 I have that $\varphi \in \mathcal{K} \oplus \varphi$ and $\mathcal{K} \subseteq \mathcal{K} \oplus \varphi$ which means that $\mathcal{K} \cup \{ \varphi\} \subseteq \mathcal{K} \oplus \varphi$. Now I'm not sure if I'm allowed to do this, but I take the consequence set of each to get:

$Cn(K \cup \{\varphi\} \subseteq Cn(K \oplus \varphi) = K \oplus \varphi$ by postulate 1.

That's all I've got to so far, I couldn't manage to prove the other subset inclusion.

$\Leftarrow$

Postulate 1 is trivial by the equality. Postulate 2 is straightforward since $\varphi \vdash \varphi$ and so $\mathcal{K} \cup \varphi \vdash \varphi$. Postulate 3 holds by similar reasoning: $\mathcal{K} \subseteq \mathcal{K} \cup \varphi \subseteq Cn(\mathcal{K} \cup \varphi))$. Postulate 4 because if $\varphi \in \mathcal{K}$ then we get $\mathcal{K} \oplus \varphi = Cn(\mathcal{K} \cup \varphi = \mathcal{K})$ which is true by definition of belief set (and 1). Postulate 5 I think must be because of some sort of transitive property of deductive systems but couldn't flesh out the proof exactly. Property 6 I don't really know.

So just to recap: I need help in proving the set inclusion: $K \oplus \varphi \subseteq Cn(K \cup \varphi)$ as well as proving postulate 5 and 6.

Thank you

1

There are 1 best solutions below

1
On

$Cn$ is the classical consequence operator, so $\vdash$ is the classical consequence relation.

As for the other inclusion, you should try to prove that $Cn(K\cup\{\varphi\})$ satisfies axiomes 1 through 5 and appeal to the minimality of $\oplus$.

To prove property 6, use what you've done in the previous part to show minimality.

Property 5 for $Cn(K\cup\{\varphi\})$ follows from the fact that $Cn$ is increasing, because $\vdash$ is: you can add hypotheses, it cannot (in classical logic) reduce the set of theorems you can prove. Essentially this comes from "$\Sigma\vdash \sigma_0 \iff $ some finite subset $\Sigma_0\subset \Sigma$ satisfies $\Sigma_0\vdash\sigma_0$"