ACADEMIA Letters

The Logic of What Might Have Actually Been: a

two-dimensional response to Salmon’s Woody argument
Benjamin Woolf, University of Bristol

In his paper ‘The Logic of What Might Have Been’, Nathan Salmon argues that the accessi-
bility relationship in modal logic is not transitive, and that S5, and a weaker system called S4,
therefore do not provide good models for metaphysical necessity.[1] To justify his position,
he asks us to imagine Woody, a table made from some material, m1, like wood. Given that
Woody is made of m1 in this world (world-1), there will be some other material (m2) which
it is possible to build Woody out of. Thus, in some possible world, world-2, accessible to
world-1 Woody is made of m2. Prima facie, there will be some other material, m3, which,
it is impossible to build Woody out of in world-1, but which it would be possible to build
Woody out of, had Woody been made out of m2. Thus, in some world (world-3) accessible
to world-2, Woody is made of m3, but in no possible world accessible to world-1 is Woody
made of m3.
In his own words: Given what Woody is made of, Woody “could have originated from a
matter slightly different from its original matter m* (while retaining its numerical identity, or
its haecceity) but not from entirely different matter. Wherever one may choose to draw the
line between what matter Woody might have originated from and what matter Woody could
not have originated from, it would seem that, by stretching things to the limit, we may select
some (presumably scattered) matter m such that, although Woody could not have originated
from m, m is close enough to being possible for Woody that if Woody had originated from
certain matter m’ that [it] is in fact possible for Woody [to be made of,] … then it would have
been possible for Woody to have originated from m, even though it is not actually possible.”
(Italics original)
The thought experiment appears to pose a problem for those wishing to defend S5. If we
let m1 be the material Woody is made of in the first world, world-1, and so on for m2 and m3,

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Benjamin Woolf, benjamin.woolf@bristol.ac.uk

Citation: Woolf, B. (2021). The Logic of What Might Have Actually Been: a two-dimensional response to
Salmon’s Woody argument. Academia Letters, Article 3395. https://doi.org/10.20935/AL3395.

1
then:

(1) ‘Woody is made of m1’ is true in world-1.

(2) ‘Woody is made of m2’ is true in world-2.
(3) ‘Woody is made of m3’ is true in world-3.

In addition, it is stipulated that Woody cannot actually be made out of m3. Thus:

(4) Necessarily(not(‘Woody is made of m3’)) is true in world-1.

By the definition of necessity and not, in all possible worlds accessible to world-1 it is
false that Woody is made of m3:

(5) ‘Woody is made of m3’ is false in all worlds accessible to world-1.

Under transitivity, if world-2 is accessible to world-1, and world-3 accessible to world-2,

then world-3 is accessible to world-1. Thus:

(6) ‘Woody is made of m3’ is false in world-3.

But, (6) contradicts (3).

Salmon’s solution to this dilemma is to drop the S4 accessibility relationship of transitivity.
Having done so, we lose the justification of (6), so the contradiction never occurs. However,
this comes at the cost of having to reject S5 as a good model of metaphysical modality. Given
that the latter thesis is defended independently,[2] many are unwilling to do so.
For the sake of argument, I will assume that we accept the intuition behind Salmons’ ar-
gument. However, I will argue that the thought experiment, when interpreted using a standard
formalisation of two-dimensional semantics, is, in fact, consistent with S5.
Two-dimensional modal logic is regular modal logic with the addition of an indexical.
This is achieved by slightly modifying the semantics. Instead of defining truth relative to one
world, truth is defined relative to two worlds: the world in which the sentence is said to be
true in (the evaluation world), and the world from which the sentence is being interpreted from
(the reference world). Formally: (Φ, v, w) = 1 means that Φ is true in world w, evaluated from
world v. Likewise, validity needs to be redefined as holding when either the formula is true
in every possible pair of reference and evaluation worlds, or only in those pairs in which the
reference and evaluation world are identical.
One motivation for the use of two-dimensional logic is that the addition of an indexical
increases the expressive capacity of our modal logics, and allows the introduction of new

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Benjamin Woolf, benjamin.woolf@bristol.ac.uk

Citation: Woolf, B. (2021). The Logic of What Might Have Actually Been: a two-dimensional response to
Salmon’s Woody argument. Academia Letters, Article 3395. https://doi.org/10.20935/AL3395.

2
operators like the actually operator. A proposition is ‘actually’ true if it is true in some world,
when evaluated from that world. Take the examples: (A) Necessarily, if the sky is yellow then
the sky is yellow. (B) Necessarily, if the sky is actually yellow, then the sky is yellow. (A)
is a trivially true tautology. (B) also appears to make a trivially true claim. However, under
classical modal logic we would have to take (B) as ‘necessarily (if P then Q)’, which is not
a logical truth. But, under its two-dimensional formalisation, ‘necessarily (if actually-Q then
Q)’, does make a true claim.
Let’s now re-formalise the thought experiment using an indexical. Firstly, we are asked to
imagine that, when evaluating from the actual world (world-1) Woody is made of m1. If we
use the same formalisation used above, we get:

(1’) ‘Woody is made of m1’ is true in world-1, when evaluated from world-1.

We are then told that, while still evaluating from the actual world, there is another world
(world-2) from which Woody could have been made out of m2. Thus:

(2’a) ‘Woody is made of m2’ is true in world-2, when evaluated from world-1.

Although not essential, I assume that we are also meant to believe:

(2’b) ‘Woody is made of m2’ is true in world-2, when evaluated from world-2.

Finally, we are asked to think that, from the perspective of world-2, there is another world,
world-3, where Woody is made from a third material, m3. Hence:

(3’) ‘Woody is made of m3’ is true in world-3, when evaluated from world-2.

We are told that from the perspective of world-1, it is impossible for Woody to be made
out of m3:

(4’) ‘necessarily(not(‘Woody is made of m3’)’ is true in world-1, when evaluated

from world-1.

For the sake of argument, we are assuming S5 and hence that world-3 is accessible to
world-1, thus:

(5’) ‘Woody is made of m3’ is false in world-3, when evaluated from world-1.

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Benjamin Woolf, benjamin.woolf@bristol.ac.uk

Citation: Woolf, B. (2021). The Logic of What Might Have Actually Been: a two-dimensional response to
Salmon’s Woody argument. Academia Letters, Article 3395. https://doi.org/10.20935/AL3395.

3
 (3’) and (5’), do not contradict each other. By introducing two-dimensional semantics, the
contradiction is therefore avoided, even though we are still using S5 to define the accessibility
relation.
One argument for the legitimacy of a two-dimensional interpretation is that Salmon em-
ploys the type of language which it is typically argued to be best formalised by two-dimensional
logic. This is seen in the quotation I included above, where he contrasts actual and non-actual
possibilities. Indeed, Salmon uses the word ‘actual’ 63 times in the essay. He also seems to
explicitly change the context of evaluation. For example, he writes: ‘if Woody had originated
from certain matter m’ that is in fact possible for Woody - [a] matter differing in as many
molecules from the actual original matter m* as possible, and sharing as many molecules
with m as possible, while remaining a possibility for Woody - then it would have been pos-
sible for Woody to have originated from m, even though it is not actually possible.’ We start
by evaluating Woody’s possible material in the actual world, then change to evaluating what
materials Woody could be made from in another possible world. By changing the world we
evaluate from we change what is possible. Hence, it seems at first impossible that Woody is
made of m*, but later possible that it was. Other sections seem to reflect similar changes of
reference. For example, he later writes about interpreting sentences ‘with respect to a world
w’.
Another advantage of this interpretation is that it allows for a formal proof of some of
Salmon’s other objections to S5. For example, Salmon objects that classical S5 does not allow
one to formalise relative possibilities. But relative possibility is a direct consequence of using
two-dimensional logic. Salmon additionally comments that ‘the ostrich approach [taken by
his opponents] misconstrues the simple modal term “necessarily” to mean the modally com-
plex concept of actual necessity … The simple modal concepts of necessity and possibility
simpliciter … are not the same as the concept of actual necessity and actual possibility, ne-
cessity and possibility according to the actual world.’ He clearly thinks that the traditional
necessity operator when using S5 is not sufficient to model our use of the word necessary,
but is instead some type of restricted actualised necessity. Instead, he thinks we should use a
more generalised type of possibility, which does not ignore what he designates as ‘impossible
worlds’. Again, the idea that there might be two types of possibility is a well-known feature
of two-dimensional logic. Davies and Humberstone [3], building on Evens [4], argues that
the traditional notion of necessity is superficial. They argue that there is a deeper form of ne-
cessity. Even if, from the perspective of some world, a proposition P is true in all accessible
worlds there may still be some other, accessible, world in which P is false from the perspective
of that other world (i.e. actually-P is false for this other world). It is thus argued that we need
to introduce a new type of necessity: deep necessity. A proposition is deeply necessary if it is

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Benjamin Woolf, benjamin.woolf@bristol.ac.uk

Citation: Woolf, B. (2021). The Logic of What Might Have Actually Been: a two-dimensional response to
Salmon’s Woody argument. Academia Letters, Article 3395. https://doi.org/10.20935/AL3395.

4
true in every world from the perspective of that world. To me, this is the essence of Salmon’s
critique.

Disclaimer
The author is dyslexic and apologises for any unnoticed and uncorrected spelling mistakes

References
1. Salmon (1989). The Logic of What Might have Been, The Philosophical Review 98, 3-34.

2. Williamson (2013). Modal logic as metaphysics. Oxford University Press.

3. Davies and Humberstone (1980). Two Notions of Necessity, Philosophical Studies 38,
1-30.

4. Evans (1979). Reference and Contingency, The Monist 62, 161-89.

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Benjamin Woolf, benjamin.woolf@bristol.ac.uk

Citation: Woolf, B. (2021). The Logic of What Might Have Actually Been: a two-dimensional response to
Salmon’s Woody argument. Academia Letters, Article 3395. https://doi.org/10.20935/AL3395.