Vladimir Vasyukov

Institute of Philosophy RAS

High School of Economics
Moscow
vasyukov4@gmail.com
 J. Hintikka notes that “what is needed for
the logical necessity of a sentence p in a
world w0 is more than its truth in each one of
the arbitrary selected set of alternatives to w0.
What is needed is its truth in each logically
possible world”.
Yet nothing prompts us whether all such
worlds belong to alternatives of given world.
Maybe the way out would be replacing logical
(metaphysical according to Kripke) modalities
with some suitable variant of transcendental
modalities.

This proposal was realized by appealing to

Kant’s conception of transcendentality (“we can
cognize of things a priori only what we ourselves
have put into them”) introducing a family of binary
(accessibility) relations on a set of possible worlds
and combining respective modal logic with L.
Humberstone’s inaccessibility logic. It seems that
this approach will be pertinent for intuitionistic logic
too especially in the framework of Kripke semantics
generalized by van Dalen.
 Relativization
of Modal
Transcendental Logic
Transcendental Modalities
Agency

Transcendentality
in Intuitionistic
Logic

Is Alethic
Modal
Logic
Possible?

OUTLINE
 The secret thoroughly kept from the
public eye by logicians studying semantics of modal logics

… what is needed is a
truth of a sentence in
each logically
possible world yet
(alas!)

nothing prompts us whether

all such worlds in Kripke
semantics belong to
alternatives of given world
 What means “logicality” of the world?
G. Sher:
every operation is logical
if it is invariant with respect
to every bijection between
domains

Z.N. Mikeladze:
this concerns also
relations between
individuals
A. Tarski:
logical notions are only those
which determine objects invariant
with respect to any bijective (one-
one) mapping of the “world”
(either “universe of discource” or
‘”class of all individuals”) onto
itself
 But will we always be
able to define these
properties?

Jakko Hintikka: I.Kant

what is needed here is

perhaps some notion of
transcendental
possibility that limits the
range of worlds in our
frames to those which
conform to the general
presuppositions of our we can cognize of things
conceptual system a priori only what we
ourselves have put into them
 Splitting of the modal operators

ΩA means «А is logically necessary»

□A means «А is, in a sense, logically necessary»

Semantics of thes╞ ΩA modal

relative iI tW
logic with t  t╞ A)
(sRisuch
operators shoulds╞ □A 
exploit iI
not onetW
but a(sR it  t╞ A)
family of
binary (accessibility) relations on a set of
possible worlds.

We can introduce the transitive closure of all

relations
R = (R1R2…Rт) (полагая I = {1,2,…,n})
and rewrite the truth-condition for A as
s ⊧ Ω A  tW (sRt  t ⊧ A)
L. Humberstone’s
inaccessibility logic:
is obtained by
introducing into the
language new operators
▲ and ∎ (▲A means “A
is logically inaccessible”,
Formally: ∎ A means “A is
s ╞ ▲A  iI tW (not sRit logically inaccessible
 t ╞ A) in a certain sense”)
s ╞ ∎ A  iI tW (not sRit
 t ╞ A)
or s ⊧ ▲A  tW (not sRt  t ⊧ A)
 Transcendental inaccessibility

If following Hintikka one consider an operator of

logical necessityas an operator of transcendental
necessity then the operator ▲ should be called
an operator of ‘”transcendent” logical
inaccessibility or simply logical transcendency I. Kant
But as soon as we posit since it presupposes
the unconditioned that is
(which under no
what is really at issue)circumstances,
in that which liesinoutside
no sense we will be able to
the sensible
world, and hence in thataccess
whichaispossible world
outside all in which
possible the respective
experience,
then the ideas come to be transcendent;
formula will be true. ∎ do
they not serve should be
respectively
merely to complete thecalled an operator
empirical of “transcendent”
use of reason (which always logical
remains an idea, never to be completely
inaccessibility in acarried
certainout, but or logical
sense,
nevertheless to be followed), rather they separate themselves
transcendency in a certain sense.
entirely from it and make themselves into objects whose matter is
not drawn from experience, and whose objective reality rests not
J. Hintikka
on the completion of the empirical series but on pure concepts a
priori.
 The duals ▼ and ♦ of operators ▲ and ■
 s╞▼A  iI tW (not sRit & t╞ A)
 or s ⊧▼A  tW (not sRt & t ⊧ A)

 s╞ ♦A  iI tW (not sRit & t╞ A)

▼A means “A is serially logically

transcendent” and ♦A – “A is serially
logically transcendent in a certain
sense” (by analogy with the seriality
property st[sRt])
 An operator of logical transcendentality

⇓A means “A is logically transcendental”

⇓A = def ■A  A
Transcendentality of А means that А is transcendent in a certain sense
and logically possible in some sense

A = def ▼A  ΩA
Serial transcendentality of А means that А is serially transcendent
or logically necessary
 Transcescendental Agency

agent i knows, that А

In case of epistemic modalities instead of
speaking ”in a certain sense” we should
speak of agents and sRit substantiates the
semantic interpretation of an expression
”agent i knows that A”, formally KiA.

KiA

sRit
 Knowledge: individual and common

KA means «Everybody knows, that А»

(common knowledge)
KiA means «An agent i knows, that А»
(individual knowledge)
 Belief: individual and common

BA means «Everyone believes, that А»

(common belief)
BiA means «An agent i believes, that А»
(individual belief)

s ⊧ BA  tW (sRt & t ⊧ A)

s ⊧ BiA  tW (sRit & t ⊧ A)
 Ignorance: common and individual

s ╞ ▲A  iI tW (not sRit  t ╞ A)

or s ⊧ ▲A  tW (not sRt  t ⊧ A)
s ╞ ■A  iI tW (not sRit  t ╞ A)

s ⊧ ₭A  tW (not sRt  t ⊧ A)

s ⊧ ₭iA  tW (not sRit  t ⊧ A)
₭A means «Nobody knows that А»
(common ignorance)
₭iA means как «An agent i do not know that А»
(individual ignorance)
 Disbelief: common and individual

s ⊧ SA  tW (not sRt & t ⊧ A)

s ⊧ SiA  tW (not sRit & t ⊧ A)
s╞▼A  iI tW (not sRit & t╞ A)
or s ⊧▼A  tW (not sRt & t ⊧ A)
s╞ ♦A  iI tW (not sRit & t╞ A)
 Conceivable reliability

A = def ₭iA  BiA

A means «Conceivably
reliable, that А» (an agent i
do not know, that A, and
⇓A =
yet he believes, that ■A  A
defA)
 Entire reliability

A = def SA  KA

A means «Entirely reliable,

that A» (either no one
convinced, that A, or
everybody knows, that A)
 Kripke semantics
generalized by van Dalen

Such semantics is based on the conception of mathematics

(and hence logic) as a mental activity of a mathematician (or
logician respectively) S. His mental activity is structured in
linear time (runs through 0,2,3, …) where at each time S has
acquired a certain knowledge which increases monotone in
time. This process should be presented as treelike picture of S’s
possible histories. Each node of the tree represents a stage of
knowledge of S and we have assigned a set of sentences Si
subject to the condition that Si increase
 We can introduce the transitive closure

More of all relations o = (1  2  . . .  n)

(supposing I ={1, 2, . . . , n}) and rewrite the

formally truth-condition for ⊩   and  as

a⊩   if bo a (b⊩  b⊩)

a⊩ if bo a (b⊮ )

Here  stands for “logical” implication, 

for implication for agent i,  for “logical”
negation and i for negation for agent i. It is
obvious that we will have (  )  ( i )
and   i .

We replace  with the family {i}iI and define

the following conditions:
a⊩   if iI bi a (b⊩  b⊩)
a⊩ i  if iI bi a (b⊩  b⊩)
a⊩ if iI bi a (b⊮ )
a⊩i if iI bi a (b⊮ )
Now we enrich our language with the operator K and Ki
where K means “Everyone knows, that ” (common knowledge), and
Ki means “An agent i knows, that ” (individual knowledge).

Then we add two conditions:

we would a⊩
beKable tooaread
iff b (b⊩ )them as
follows: a⊩ Ki iff iI bi a (b⊩ )
B means “Everyone believes that ”
(common belief),
Bi means “An agent i believes that ”
(individual belief).
Among axiom schemes for such
operators will be K  B, Ki  Bi.
Defining dual operators for K and Ki with the
help of definitions
a⊩ B iff boa (b⊩ )
a⊩ Bi iff iI bi a (b⊩ )
 Pursuing an analogy with
transcendental modalities (the
possible worlds conform to the
mind: in knowing, it is not the mind
that conforms to possible worlds
but instead possible worlds that
conform to the mind) we also need
to exploit the axioms of “Kantian
transcendental apperception”
The next step is to define some new
(where the self and the world come
conditions based on
together) or “Humean” one (all
the “inaccessibility” relation:
a⊩ ⋑ if b≱o a (b⊩  b⊩)
experience is the succession of a
a⊩ ⋑i if iI b≱i a (b⊩  b⊩variety
) of contents):
a⊩ ~ if b≱o a (b⊮ ) ( ⋑)  (  )
a⊩~i if iI b≱i a (b⊮ ) ~  
 The same way as before we enrich our
language with the following “inaccessibility”
operators and conditions:
a⊩  iff b≱oa (b⊩)
( means “Nobody knows that ” –
common ignorance)
a⊩ i iff iI b≱i a (b⊩)
(i means “An agent i do not know that
” – individual ignorance)

Dual operators would be defined with

the help of definitions
a⊩ S iff b≱oa (b⊩)
(S means “Nobody believes that ” –
common disbelief)
a⊩ Si iff iI b≱i a (b⊩)
(Si means “An agent i do not believe
that ” – individual disbelief)
 Finally, we can obtain explication of
reliability by way of the definitions:
⊞ =def i  Bi
(⊞  means “Conceivably reliable
that " – an agent i do not know
that , and yet he believes that )
⊛ =def S  K
(⊛ means “Entirely reliable,
that ”- either no one convinced
that , or everybody knows that )

A co-reflection   K, typical for

intuitionistic epistemic logic, in our Kantian
version could have a form   ⊛ being
“transcendental” consequence of axiom like
⋑⊛. Hence, in terms of the Brouwer-
Heyting-Kolmogorov (BHK) semantics it
means that if  is constructively true, i.e. has
a specific proof, then  will be entirely reliable
THE
END
THANK
YOU
FOR YOUR
ATTENTION