Given some recent chatter here, I offer my attempt to define some theses and the relationships between them. I ask some questions along the way, and I’m certainly not claiming they are as precise and complete as they need to be! Maybe they are just imprecise and incomplete enough to generate some more interesting chatter!
Determinism (D) = 1. A complete description of the system at one time, and of all the laws that govern that system, logically entails a complete description of that system at any future time. (This version is just past-to-future, though some define D to include present-to-past too. “The system” for all these theses is meant to pick out the scope over which the thesis applies, but notice that a system can only be deterministic as defined here if it is entirely “closed off” from any indeterministic system, which is why indeterminism in the micro-system surely falsifies D in the macro-system, unless the laws of the micro-system include some weird aggregation laws to ensure that the indeterminism never “percolates up.”)
Because the other theses below will be framed in terms of events and laws, here’s such a definition of D (I’d be curious to know if people think it is accurate and equivalent to the above):
2. For every event E in the system, there is a set of prior events which, as governed by the laws of that system, is logically sufficient for E.
(I wonder whether “logically sufficient” is the best modality to use here?)
(I wonder what the phrase about laws is doing here and how the thesis might work if that phrase was excised. I also wonder whether there is a good way to define “deterministic laws”; perhaps “D laws = all events governed by such laws are such that for every event E, there is a set of prior events which are logically sufficient for E”?)
(Brian Park’s definition of “determinism” suggests something like this, I think: “An event E in the system is ‘deterministic’ if there is a set of prior events which, as governed by the laws of that system, is logically sufficient for E.”)
Indeterminism (~D) = 1. A complete description of a system at one time, and of all the laws that govern that system, does not logically entail a complete description of that system at any future time.
OR: 2. For some events Ei in the system, there is no set of prior events which, governed by the laws of that system, is logically sufficient for Ei (even if for some or even most events in the system, there are sets of prior events logically sufficient for them).
Causalism (C) = For every event E in the system, there are prior events which, governed by the laws of that system, cause E.
C is clear consistent with D or ~D, though people, especially scientists discussing free will, sometimes use D when they really mean C (or perhaps L below).
Law-governed (L) = For every event E (or process Q) in the system, that event (or process) is governed by a law of that system.
L is pretty vague and clearly consistent with D or ~D, though D as written above entails L. On some combinations of views of laws and causation, (C & ~L) is possible and (~C & L) is possible (right?).
Predictability (P) = For every event E in the system, some other system could in principle predict E, given complete information about some relevant set of prior events and the laws that govern that system. (This is the Laplacean view of determinism. I think it helps to make the predicting system outside of the predicted system to avoid some possible paradoxes.)
I take it that P is not necessarily entailed by D, though many people make that move. For instance, if I understand the Bohmian interpretation of quantum physics, it says that the system is deterministic but in principle not (fully) predictable. And I think some people interpret non-linear dynamics to suggest it can make a D system ~P in principle.
D certainly does not entail P by a cognitive agent within the system, much less P in practice by any finite cognitive agent inside or outside the system.
I take it that ~D does entail ~P (at least putting God aside), even though indeterministic systems and laws might allow for levels of predictive accuracy in practice that are rarely achieved in practice in deterministic systems (see quantum theory).
I take it that P likely entails L but that L does not entail P. I don’t think P entails C and C does not entail P.
Mechanism (M) = 1. For every event E (or process Q) in the system which is at any “level” above the lowest level, there is a set of lower-level events (processes) which, governed by the laws of that system, are logically sufficient for E (or Q).
Or, more specifically: 2. For every mental event M (or psychological process Q) in the psychological system, there is a set of lower-level, physical (say, neurobiological) events (or processes) N which, governed by the laws of that system, are logically sufficient for E (or Q).
(M2 looks like a strong form of supervenience.)
We could also put M in terms of explanation and say: 3. For every mental event M (process Q), there is a set of lower-level events (processes) which, governed by the laws of that system, fully explain E (or Q).
I think it is very important to recognize that:
D does not entail M and M does not entail D.
For instance, M without D could clearly occur in a mechanistic system with indeterminism that seeps up from lower levels to higher levels, as seems to occur with quantum indeterminism. (I can’t easily imagine how a mechanistic system could be indeterministic at higher levels if deterministic at lower levels or vice versa, can you?)
And D without M could occur in a deterministic system that has higher-levels whose activity is not logically entailed by (or for M3 fully explained by) lower levels. For instance, dualistic parallelism with D relations among the mental events and among the physical events but no interaction between mental and physical. Or psychological processes that do not supervene on the neurobiological processes, but both levels are deterministic systems and they interact with each other deterministically (is overdetermination the only way to make sense of this? Markossian’s interesting account of compatibilist agent causation may be relevant here.)
Finally, M seems to entail C and L, though neither C nor L entails M.
M does not entail P nor does P entail M (though in practice M seems to be a good way to get increasingly close to P).
Since I’ve already gone on way too long (and probably said too many false things), I’ll stop here without sticking my neck out about what I think about the relationships between each of these theses and FW/MR. But I’d love to hear what y’all think!