> Indeterminism at the level of psychology is required for free will and alternative possibilities. That is entirely compatible with determinism at the fundamental physical level.
Not sure I understand/agree with that. Can you build non-determinism out of deterministic pieces?
> Can you build non-determinism out of deterministic pieces?
Yes. Pseudo-random generators are deterministic on lower level, yet they effectively non-deterministic in the sense that you cannot easily predict next value.
Or you can just take a system with large number of interacting particles - even if behavior of each of them is determined, the system as a whole is hard to predict.
Converse is also possible, you can build effectively deterministic system on top of a non-deterministic one. We do that with computers, we have digital computers (deterministic) on top of solid-state physics, which is non-deterministic due to interactions of electrons and the material.
This is not really a paradox, since at the higher level you only consider macrostates, so the fact that the microstates, which you consider at the lower level, are (non-)deterministic doesn't really matter.
> in the sense that you cannot easily predict next value
Easily or _in principle_?
Do not conflate our physical constraints with the actual mechanism behind the scene. Free will is not about our own ignorance or inability, it is the principal thing.
These are constraints from computational complexity.
My point is the question "is it deterministic" is wrong without considering the level of detail you're looking at. Even in deterministic universe, a roll of dice is considered to be undeterministic, and this is good enough for casinos.
If you're asking "in principle", then it might as well be unknowable because, as I pointed out, how we might observe the system to be (our model of it is deterministic or not) doesn't have to correspond at all whether it is actually deterministic or not at the lower level. So there always can be another level which can behave differently.
Well put. Compatibilist arguments typically hinge on some bait and switch and this is a popular one. The author does this exact thing in his weather example:
> At the level of individual air molecules, there is no such thing as weather. Perhaps the system at that very fine-grained level of description would indeed behave deterministically according to classical physical laws, but as you move to a more macroscopic description, you abstract away from this microphysical detail. That is not driven by ignorance on our part, but by the explanatory need to focus on the most salient regularities.
> When you consider the macroscopic weather states, the system is not deterministic, but stochastic, or random.
The high level weather pattern is just like the output of the pseudorandom number generator. It appears stochastic to someone who is only given the high level description, but it is still deterministic in the actual world. Given the Laplace's demon description of the system, there is no room for alternative possibilities.
I think the author is only added the part I have bolded to try and deflect this. He is saying "we don't want the Laplace's demon description of the system, we just want the high level precis." Weirder still, he seems to suggest that getting the Laplace's demon is a realistic possibility, which obviously it's not.
But the fact is it doesn't matter whether it's possible for a human to get the Laplace's demon description, or whether you actively pursue it or whatever - if your accept that it's there, as List seems to, then it fully determines the higher level phenomena and this idea about higher level indeterminacy is moot.
I agree that they are different because determinism is a (mathematical) property of the model, while the predictability (with respect to class of models) is observed.
I guess it depends on how you define randomness. Chaos Theory, for example, starts with deterministic parts, but the outcome can be unexpected; this is largely a result of scale and feedback loops. Take the weather, for example: if you could track literally everything, then arguably you could predict the weather with 100% accuracy.
Yes, free will and determinism are compatible. It's known as Compatibilism, and most philosophers are actually Compatibilists. I often debate free will on reddit [1], so I'll reproduce the relevant argument here:
Firstly, to set the stage, understand that the "free will" debated in philosophy around questions of moral responsibility is not the same "free will" as typically used in science, eg. one such definition is where experimenters are free to set up their measurements independently of the system they're measuring. This is a common confusion.
What matters for the philosophical free will debate is whether there is a coherent definition of free will/choice which can make sense of our language of volition and moral reasoning, and which can serve to justify moral responsibility. Note also that "moral responsibility" does not necessarily entail "punishment" (which is a question of justice). This is another common confusion.
Finally, note that you make choices according to your nature, but to also have a choice in your nature would be logically circular.
So is there at least one definition of free will that can satisfy all of these criteria? Compatibilism is the most widely accepted approach among philosophers, and seems to match most people's intuitive moral reasoning [2]. For a rough example of what this might look like, consider a definition like "the moral responsibility of an agent capable of general learning is proportional to the amount of information it has learned, and a freely willed choice is one made based on internal reasons and not forced by other agents".
Note how this is perfectly compatible with determinism, how it makes sense of why we don't hold babies responsible but we hold adults responsible, it makes sense of how environmental [and] social factors that can impede learning moral lessons can diminish moral responsibility, it strictly defines what "free" means in the context of choice, and "moral responsibility" reduces to "moral feedback", ie. instructing what was done wrong. It's not perfect, but it should suffice as an introduction to Compatibilist-style reasoning.
This does not cover justice, which is what we must do in response to moral culpability, and which bring in further assumptions.
I have the same confusion about his statement as you do. His weather analogy is also strange because the weather only takes one path, presumably determined by physics and initial conditions. I doubt many people would argue there's agency directing local weather (although some might).
That is not non-determinism; it's just difficult to predict determinism.
I postulate that we cannot build non-determinism out of deterministic pieces - we can just get it complex enough that we can't accurately pre-determine what will happen. With enough data, and a system of sufficient predictive complexity, we can still predict everything, I'd wager.
It is not possible universally. I like the proof of this, it is simple and resembles to many similar no-go theorems (like the halting problem).
1. Imagine a finite formal system which universally can identify if an input is regular (compressible) or random (not compressible) / So we are using the Kolmogorov-complexity definition of randomness here...
2. Assume the above formal system can be described using L bits (e.g. it is an L bit long program)
3. Lets take the first (e.g. smallest in numerical ordering) 2×L (or 10×L, 100×L, ... take it as large you want) long bit string that the program classifies as random bit string
4. "3." is a definition of a 2×L long string in L bits (the formal system itself) + constant
5. The 2×L long string is therefore compressibe so it is not random - but our original algorithm is misclassify it as random (it is in its definition) - so the algorithm is not universal...
Not sure I understand/agree with that. Can you build non-determinism out of deterministic pieces?