What is determinism?
(Following chapter II of John Earmans A Primer on
Determinism)
A world is a four-dimensional spacetime.
It encompasses the whole history of the world
throughout all of space.
The actual world is the collection of all events that
have ever happened, are happening now, or will ever
happen.
A possible world is a way that the actual world
could have been, a possible alternative history of the
entire world. (Of course, parts could be the same as the
history of the actual world.)
A physically or nomically possible world is a
possible world that conforms to the basic physical laws of
our universe. For the time being, we will pretend that
these basic laws are the laws of Newtonian mechanics.
We can then assign times to events absolutely
that is, in a way that is independent of the state of
motion of the clocks used to assign the times.
So we have a meaningful notion of the state of
the world at time t. It is the entire set of events
that happen at the given time.
So let W stand for the collection of all physically
possible worlds.
Defining Determinism - Earman
Then the world w W is Laplacian deterministic iff
for any other world w' W, if w and w' agree at any time,
then they agree at all times.
One can refine this definition into two
subcomponents:
The world w W is future (or past) Laplacian
deterministic iff for any other world w' W, if w and w'
agree at any time, then they agree at all future (or past)
times.
Perhaps a world can also be partially deterministic
that is, deterministic with respect to some magnitudes
but not others.
But this definition may be problematic. Consider, for
example, a hypothetical particle called the freeon, which
is supposed to be indeterministic in a world in which the
magnitudes that characterize the rest of the ordinary
matter are deterministic.
Does the Freon interact with ordinary matter
or not?
Some further applications of the definition of
Laplacean determinism:
First, the acts of the self in the theory of agency (or,
what Hobart calls libertarian indeterminism) are
indeterministic. As Taylor notes (51): No antecedent
2
Defining Determinism - Earman
conditions were sufficient for his performing just that
action.
Taylor seems to think that the theory of agency is
deterministic, but that may be because of the confusing
claim that acts of the self are self-caused. From this it
might seem to follow that they are caused, and so
determined, as Taylor uses the term. They are clearly
Laplacean indeterministic, and so indeterministic in the
primary ontological picture of determinism that we derive
from William James.
Second, the space invaders example shows that
Newtonian mechanics is indeterministic, if the
mathematics works as claimed.
7. Fear and loathing
There are various ways in which Earmans heuristic
definition might be trivialized. It is instructive to see how
they might go.
First it is worth noting that (x)Fx is logically
equivalent to ~(Ex)~Fx. So if (x)Fx to be true, then
~(Ex)~Fx is also true, and so (Ex)~Fx must be false.
Consider a universe which is empty, which contains
no objects at all. In such a world (Ex)~Fx is false, no
matter what property F we choose to consider. Of course
(Ex)Fx is also false.) So in this world (x)Fx is true (as is
(x)~Fx). Both are said to be vacuously true.
3
Defining Determinism - Earman
It is not easy to define what a property is. Suppose,
for instance, that there can be properties of the form __
is at place p at time t in world wi. Nothing in any other
possible world (that is, no world other than wi) will have
this property. So no other possible world in the set W
can agree with world wi at time t. But since there is no
other world that agrees with w at t, it is vacuously true
that for any other world w' W, if wi and w' agree at any
time, then they agree at all future (or past) times.
That is, if there were properties of the odd form
described above (Earman calls them indexical), every
nomically possible world would automatically, vacuously,
be Laplacean deterministic. Thats too quick a way to
settle the question of determinism. The right thing to do
is to disallow indexical properties like the one used in this
argument.
The definition is also trivialized if one allows past-
involving and future-involving properties. Does my car
now have the properties of being made in 2003 or being
in my garage 5 hours from now? If there are properties
like this, then no distinct nomically possible worlds will
agree at any time, and once again our world satisfies
vacuously the definition of Laplacean determinism. Such
properties, then, must not be permitted.
