To demonstrate the properties of sets of beliefs, Raymond Smullyan defines the following types of reasoners:

• Accurate reasoner:^[1][2][3][4] An accurate reasoner never believes any false proposition. (modal axiom T)

${\displaystyle \forall p:{\mathcal {B}}p\to p}$ 

• Inaccurate reasoner:^[1][2][3][4] An inaccurate reasoner believes at least one false proposition.

${\displaystyle \exists p:\neg p\wedge {\mathcal {B}}p}$ 

• Consistent reasoner:^[1][2][3][4] A consistent reasoner never simultaneously believes a proposition and its negation. (modal axiom D)

${\displaystyle \neg \exists p:{\mathcal {B}}p\wedge {\mathcal {B}}\neg p\quad {\text{or}}\quad \forall p:{\mathcal {B}}p\to \neg {\mathcal {B}}\neg p}$ 

• Normal reasoner:^[1][2][3][4] A normal reasoner is one who, while believing ${\displaystyle p,}$  also believes they believe p (modal axiom 4).

${\displaystyle \forall p:{\mathcal {B}}p\to {\mathcal {BB}}p}$ 

A variation on this would be someone who, while not believing ${\displaystyle p,}$  also believes they don't believe p (modal axiom 5).

${\displaystyle \forall p:\neg {\mathcal {B}}p\to {\mathcal {B}}(\neg {\mathcal {B}}p)}$ 

• Peculiar reasoner:^[1][4] A peculiar reasoner believes proposition p while also believing they do not believe ${\displaystyle p.}$  Although a peculiar reasoner may seem like a strange psychological phenomenon (see Moore's paradox), a peculiar reasoner is necessarily inaccurate but not necessarily inconsistent.

${\displaystyle \exists p:{\mathcal {B}}p\wedge {\mathcal {B\neg B}}p}$ 

• Regular reasoner:^[1][2][3][4] A regular reasoner is one who, while believing ${\displaystyle p\to q}$ , also believes ${\displaystyle {\mathcal {B}}p\to {\mathcal {B}}q}$ .

${\displaystyle \forall p\forall q:{\mathcal {B}}(p\to q)\to {\mathcal {B}}({\mathcal {B}}p\to {\mathcal {B}}q)}$ 

• Reflexive reasoner:^[1][4] A reflexive reasoner is one for whom every proposition ${\displaystyle p}$  has some proposition ${\displaystyle q}$  such that the reasoner believes ${\displaystyle q\equiv ({\mathcal {B}}q\to p)}$ .

${\displaystyle \forall p\exists q:{\mathcal {B}}(q\equiv ({\mathcal {B}}q\to p))}$ 

If a reflexive reasoner of type 4 [see below] believes ${\displaystyle {\mathcal {B}}p\to p}$ , they will believe p. This is a parallelism of Löb's theorem for reasoners.

• Conceited reasoner:^[1][4] A conceited reasoner believes their beliefs are never inaccurate.

${\displaystyle {\mathcal {B}}[\neg \exists p(\neg p\wedge {\mathcal {B}}p)]\quad {\text{or}}\quad {\mathcal {B}}[\forall p({\mathcal {B}}p\to p)]}$ 

Rewritten in de re form, this is logically equivalent to:

${\displaystyle \forall p[{\mathcal {B}}({\mathcal {B}}p\to p)]}$ 

This implies that:

${\displaystyle \forall p({\mathcal {B}}{\mathcal {B}}p\to {\mathcal {B}}p)}$ 

This shows that a conceited reasoner is always a stable reasoner (see below).

• Unstable reasoner:^[1][4] An unstable reasoner is one who believes that they believe some proposition, but in fact do not believe it. This is just as strange a psychological phenomenon as peculiarity; however, an unstable reasoner is not necessarily inconsistent.

${\displaystyle \exists p:{\mathcal {B}}{\mathcal {B}}p\wedge \neg {\mathcal {B}}p}$ 

• Stable reasoner:^[1][4] A stable reasoner is not unstable. That is, for every ${\displaystyle p,}$  if they believe ${\displaystyle {\mathcal {B}}p}$  then they believe ${\displaystyle p.}$  Note that stability is the converse of normality. We will say that a reasoner believes they are stable if for every proposition ${\displaystyle p,}$  they believe ${\displaystyle {\mathcal {B}}{\mathcal {B}}p\to {\mathcal {B}}p}$  (believing: "If I should ever believe that I believe ${\displaystyle p,}$  then I really will believe ${\displaystyle p}$ "). This corresponds to having a dense accessibility relation in Kripke semantics, and any accurate reasoner is always stable.

${\displaystyle \forall p:{\mathcal {BB}}p\to {\mathcal {B}}p}$ 

• Modest reasoner:^[1][4] A modest reasoner is one for whom for every believed proposition ${\displaystyle p}$ , ${\displaystyle {\mathcal {B}}p\to p}$  only if they believe ${\displaystyle p}$ . A modest reasoner never believes ${\displaystyle {\mathcal {B}}p\to p}$  unless they believe ${\displaystyle p}$ . Any reflexive reasoner of type 4 is modest. (Löb's Theorem)

${\displaystyle \forall p:{\mathcal {B}}({\mathcal {B}}p\to p)\to {\mathcal {B}}p}$ 

• Queer reasoner:^[4] A queer reasoner is of type G and believes they are inconsistent—but is wrong in this belief.
• Timid reasoner:^[4] A timid reasoner does not believe ${\displaystyle p}$  [is "afraid to" believe ${\displaystyle p}$ ] if they believe that belief in ${\displaystyle p}$  leads to a contradictory belief.

${\displaystyle \forall p:{\mathcal {B}}({\mathcal {B}}p\to {\mathcal {B}}\bot )\to \neg {\mathcal {B}}p}$ 

• Type 1 reasoner:^{[1][2][3][4][5]} A type 1 reasoner has a complete knowledge of propositional logic i.e., they sooner or later believe every tautology/theorem (any proposition provable by truth tables):

${\displaystyle \vdash _{PC}p\Rightarrow \ \vdash {\mathcal {B}}p}$ 

The symbol ${\displaystyle \vdash _{PC}p}$  means ${\displaystyle p}$  is a tautology/theorem provable in Propositional Calculus. Also, their set of beliefs (past, present and future) is logically closed under modus ponens. If they ever believe ${\displaystyle p}$  and ${\displaystyle p\to q}$  then they will (sooner or later) believe ${\displaystyle q}$ :

${\displaystyle \forall p\forall q:({\mathcal {B}}p\wedge {\mathcal {B}}(p\to q))\to {\mathcal {B}}q)}$ 

This rule can also be thought of as stating that belief distributes over implication, as it's logically equivalent to

${\displaystyle \forall p\forall q:{\mathcal {B}}(p\to q)\to ({\mathcal {B}}p\to {\mathcal {B}}q)}$ .

Note that, in reality, even the assumption of type 1 reasoner may be too strong for some cases (see Lottery paradox).

• Type 1* reasoner:^[1][2][3][4] A type 1* reasoner believes all tautologies; their set of beliefs (past, present and future) is logically closed under modus ponens, and for any propositions ${\displaystyle p}$  and ${\displaystyle q,}$  if they believe ${\displaystyle p\to q,}$  then they will believe that if they believe ${\displaystyle p}$  then they will believe ${\displaystyle q}$ . The type 1* reasoner has "a shade more" self awareness than a type 1 reasoner.

${\displaystyle \forall p\forall q:{\mathcal {B}}(p\to q)\to {\mathcal {B}}({\mathcal {B}}p\to {\mathcal {B}}q)}$ 

• Type 2 reasoner:^[1][2][3][4] A reasoner is of type 2 if they are of type 1, and if for every ${\displaystyle p}$  and ${\displaystyle q}$  they (correctly) believe: "If I should ever believe both ${\displaystyle p}$  and ${\displaystyle p\to q,}$ , then I will believe ${\displaystyle q}$ ." Being of type 1, they also believe the logically equivalent proposition: ${\displaystyle {\mathcal {B}}(p\to q)\to ({\mathcal {B}}p\to {\mathcal {B}}q).}$  A type 2 reasoner knows their beliefs are closed under modus ponens.

${\displaystyle \forall p\forall q:{\mathcal {B}}(({\mathcal {B}}p\wedge {\mathcal {B}}(p\to q))\to {\mathcal {B}}q)}$ 

• Type 3 reasoner:^[1][2][3][4] A reasoner is of type 3 if they are a normal reasoner of type 2.

${\displaystyle \forall p:{\mathcal {B}}p\to {\mathcal {B}}{\mathcal {B}}p}$ 

• Type 4 reasoner:^{[1][2][3][4][5]} A reasoner is of type 4 if they are of type 3 and also believe they are normal.

${\displaystyle {\mathcal {B}}[\forall p({\mathcal {B}}p\to {\mathcal {B}}{\mathcal {B}}p)]}$ 

• Type G reasoner:^[1][4] A reasoner of type 4 who believes they are modest.

${\displaystyle {\mathcal {B}}[\forall p({\mathcal {B}}({\mathcal {B}}p\to p)\to {\mathcal {B}}p)]}$ 

For systems, we define reflexivity to mean that for any ${\displaystyle p}$  (in the language of the system) there is some ${\displaystyle q}$  such that ${\displaystyle q\equiv {\mathcal {B}}q\to p}$  is provable in the system. Löb's theorem (in a general form) is that for any reflexive system of type 4, if ${\displaystyle {\mathcal {B}}p\to p}$  is provable in the system, so is ${\displaystyle p.}$ ^[1][4]

If a consistent reflexive reasoner of type 4 believes that they are stable, then they will become unstable. Stated otherwise, if a stable reflexive reasoner of type 4 believes that they are stable, then they will become inconsistent. Why is this? Suppose that a stable reflexive reasoner of type 4 believes that they are stable. We will show that they will (sooner or later) believe every proposition ${\displaystyle p}$  (and hence be inconsistent). Take any proposition ${\displaystyle p.}$  The reasoner believes ${\displaystyle {\mathcal {B}}{\mathcal {B}}p\to {\mathcal {B}}p,}$  hence by Löb's theorem they will believe ${\displaystyle {\mathcal {B}}p}$  (because they believe ${\displaystyle {\mathcal {B}}r\to r,}$  where ${\displaystyle r}$  is the proposition ${\displaystyle {\mathcal {B}}p,}$  and so they will believe ${\displaystyle r,}$  which is the proposition ${\displaystyle {\mathcal {B}}p}$ ). Being stable, they will then believe ${\displaystyle p.}$ ^[1][4]

• Epistemic modal logic
• Belief revision
• Common knowledge (logic)
• George Boolos
• Jaakko Hintikka
• Modal logic
• Raymond Smullyan

• Lindström, St.; Rabinowicz, Wl. (1999). "DDL Unlimited. Dynamic Doxastic Logic for Introspective Agents". Erkenntnis. 51 (2–3): 353–385. doi:10.1023/A:1005577906029. S2CID 116984078.
• Linski, L. (1968). "On Interpreting Doxastic Logic". Journal of Philosophy. 65 (17): 500–502. doi:10.2307/2024352. JSTOR 2024352.
• Segerberg, Kr. (1999). "Default Logic as Dynamic Doxastic Logic". Erkenntnis. 50 (2–3): 333–352. doi:10.1023/A:1005546526502. S2CID 118747031.
• Wansing, H. (2000). "A Reduction of Doxastic Logic to Action Logic". Erkenntnis. 53 (1–2): 267–283. doi:10.1023/A:1005666218871. S2CID 58939606.