My definition of “free will” is that I have the subjective experience of making choices. There’s no proving or disproving this, it’s reality. It doesn’t change the experience if the underlying process is fully deterministic or if there’s some level of randomness. I cannot conceive of how it could be anything else.
On the one hand that feels insufficient to me, but on the other hand if I try to argue against it I just get an “it’s turtles all the way down” situation. If I get up right now to go get some peanuts, that is because I decided to get some peanuts. I’m free to make that decision or not, but it will depend on the current state of my brain, and possibly some element of randomness. What caused that state of my brain? Well the previous state, and possibly some element of randomness. How else could it possibly be?
So, basically a finite state machine. Which is reasonable. But I assume that the state transitions are determined by the inputs in some way - which does not affect them being deterministic. How sensitive are the state transitions to small changes in the inputs?
It would also be interesting to see how the results vary if p =, say 99%.
I doubt that even the biggest opponent of a deterministic brain would say that p is as low as 50% - that would make addictive behaviors almost impossible to model.
Don’t get me wrong, interesting work. I’m just putting my paper review hat on here.
No it’s not about mesaurement accuracy or even our ability to calculate.
The initial value problem, where the intial state of a system is specified and you calculate the state of the system at some later time, is a very important class of problem in physics. For deterministic theories, the intial state always precisely determines the state at all other times.
General relativity is often thought of being determinsitic as it lacks any probabilistic elements. Time is treated quite differently in general relativity than older theories though, so even formulating an intial value problem may not be possible and you have to place what is actually a very restrictive set of condition to ensure an intial value formualtion works. Some important spacetimes like anti de Sitter spacetime are not amenable to an initial value formulation however.
Newtonain mechanics has long been thought of as being truly dterminsitic, however in 2003 John Norton describe a problem where a particle is balanced on the apex of a dome, where the curve of the dome obeyed a certain class of equation. It is consistent with the initial conditions that the particle can roll down the dome spontatenously starting at ANY TIME later. This means the later states don’t depend entirely on the initial state. Whilst the dome itself doesn’t seem unreasonable, it does rely on the particle being precisley balanced on the apex of the dome. You can add a condition tohow forces behave to make sure that the particle cannt sponatenously start to roll away, but that condition may not always be desirable.
Sorry I have not been replying much to the thread. I had a paper deadline on the 29th, another on the 31st, and two grant proposal deadlines yesterday. Needless to say, but I’ll say it anyway, I’m quite tired today. I have been reading the thread though, and the conversation has been interesting.
If the velocity of the particle could only vary discretely then clearly only the appplication of an instaneous force would move the particle from its position at the apex of the dome. However this limitation is not realistic and when looking at how a particle is affected by a force like gravity its velcoity needs to vary continuously. On Norton’s dome the non-deterministic nature of the solutions is directly related to the indeterminate nature of the 2nd deriviative of force wrt time at the apex.
Ah, interesting. Got invented long after I was in school. Too bad, I could have used that idea in some places. I worked on a lot of graphs. Need to read more.
But my question still stands about sensitivity to inputs.
I’m going to repeat what other posters have alluded to, but I think in plainer language:
The free will debate is silly, and the question of “Is the universe deterministic?” is a red herring in this context.
Forget our universe for a minute. Invent your own reality where agents have free will. How does that free will actually work? How are their decisions neither random nor based on prior data and brain states? (and not a combination of those things either)
Include souls if you like – how do souls make decisions? It doesn’t help.
We should appreciate that our decisions being a function of past data and neurology is the only kind of “choice” that even makes sense.
And, our aversion to our decisions being predictable by a third party, is irrational given that the only way I can predict your decision is by simulating your brain. Because your thoughts are an integral part of your decisions.
You still need a net force to accelerate it from rest. It won’t just spontaneously move in a perfectly Newtonian universe. If random forces randomly occur it’s sort of by definition a non-deterministic system. And a perfect dome with a perfect sphere perfectly balanced with zero external forces is sort of a non-realistic system to begin with.
A simulation can perhaps be elaborate enough to reproduce the rational component of reason and logic, but you end up leaving out biochemistry (the foundation of emotion, which is unmistakably part of the thinking process). That is quite difficult to factor in reliably.
The problem with this sort of argumentation is really that it just substitutes a mystery for an enigma. How does randomness actually work? How does determining future states based on past ones (i. e. causality) actually work? We don’t have any sort of account, and they actually invite many of the exact same problems as accounts of free will do. For one, if everything needs a cause, we’re immediately faced with an infinite regress; adding randomness doesn’t help—we’re still stuck with the question of what decided which possibilities were open, what set their likelihood, and so on. In fact, since indeterministic systems can be simulated by deterministic ones, it’s difficult to see how indeterminism could add any sort of explanatory power.
Ultimately, we’re more comfortable to accepting causality as a primitive, axiomatic notion (or formulated negatively, we’re less used to question causality), but this is an appeal to intuition in the end. We don’t know how free will can conceivably get off the ground; but we don’t know how causation, or randomness, does, either.
There’s a wide birth between things that are entirely random and entirely deterministic. Of course free will is a combination of those things. Everyone agrees that the past does influence our decisions. The question is whether that alone is enough.
I personally think that, even if our brains are deterministic, then what seems to be random must instead be chaotic–which means that small changes in the input cause huge changes in the output. And modeling a chaotic system would be difficult since any flaw in the model would make the results very wrong, as that would be a small change in input. Plus how likely are you to have perfect inputs?
What will most likely happen is just better models that give the likelihood of someone reacting–i.e. better versions of the same thing we already have. (See, for example, how they calculate likely votes based on demographic data.) And, in that context, it won’t really matter if the noise is due to randomness or due to chaos.
Ok, but I was not trying to suggest a full reductive description.
If you ask me how the characters in my video game make decisions I can tell you broadly how inputs get mapped to outputs, even if I understand little of how the underlying hardware works.
When it comes to free will, vague statements like “could have chosen differently” are part of the very definition, and a discussion of what that means in terms of decision-making is rarely addressed. That’s why I find it helpful to ask about how free will works in another universe that has free will by definition. Hopefully that helps people appreciate that the issue with free will is the concept itself, not anything about physical reality.
I don’t think this is a good argument. It’s akin to saying that evolution is a baseless theory because we don’t have good models for abiogenesis yet.
Whether we can answer the ultimate questions of causality has little bearing on our understanding of neuroscience.
I think I didn’t put my point clearly enough.
Right now, quantum phenomena do appear fundamentally random. Yes, there’s still the possibility of non-local hidden variables, and there are plenty of threads discussing that, but let’s say for argument right now that there’s randomness at that level and that does have effects on the brain.
Most of the “no free will” philosophers will handwave that because they’ll say that randomness is just something particles are doing and is hardly a deliberate choice of a conscious agent. I can cite some if you haven’t seen those sort of things, Cosmic skeptic, Matt Dillahunty come to mind.
They rarely however make the leap that this dismissal of quantum randomness illustrates that the argument over whether the universe is deterministic was a red herring: because even an indeterministic universe apparently also doesn’t have free will.
I think you misunderstood me. I’m not saying ee have no detailed, microphysical account of how causation works, I’m saying we have no account whatsoever of what causation even is—that is, when you say ‘B happened because A caused it’ I don’t really know anything more about how B made A happen than I do in the case where you say ‘B happened because I chose it’. Both include the same amount of mystery; we’re just too used to causation to notice it there.
To put it differently, what is it about an event A that makes it cause another event B? All we have, with Hume, is really just the observed regularity about B reliably following A. But that doesn’t mean that the next time A occurred, it could not be followed by C instead.
The intended target there was the regress argument against free will, which points out that to want something, we first have to determine what we want, and to do so, we first have to determine what we want to want, and so on, so free will doesn’t get off the ground. But the same argument can be leveled against causation—if A caused B, what caused A?
Me too. I only heard about them through a conference.
To answer your question, we are not very certain. We have 36 participants, and all of the scans are for resting state. Because of the data set size, we’re using leave-one-out (train with 35, test with 1, loop through each participant as “The One”). As a result, there’s quite a bit of variance in the resulting RGGs because the slice of brain activity (about 5 minutes) varies. The rules of the RGGs themselves are similarish. I have not formally compared them yet, but there seems to be about the same amount of stochasticity.
We need more brain scans, and we need them in a variety of states to really be able to test the validity of the algorithm. Sadly, that’s expensive, but we’ll see what happens with the grant we put in for last year to support this research.
While there is a variance with the data we have, the data we have is too limited.
Disagree about “same level of mystery”.
There’s a big difference between us having explanatory and predictive power for a phenomenon, yet still have the outstanding philosophical question of why, versus not having explanatory or predictive power.
If we can make good predictions and inferences about some phenomenon, that’s what we colloquially think of as understanding. To a scientist, the question of why an electron that absorbs energy then has a higher energy state might be considered unnecessary and ill-formed.
Thanks for clarifying but I think my objection still holds.
Our understanding of causality explains the general events of the universe and our world for the last 13.7 billion years and continues to be useful going forwards. Yes it becomes unstuck when we need to think of the very beginning of the chain.
But that’s a very different thing from a model that can’t explain any links in any chain, and has no predictive or explanatory power, which is where the idea that we have free will lies. What is the model for what this free will is?
It’s not a question of why, but simply of how—we just don’t know what it is that enables A to cause B. We have the observed regularity that B follows A, but that doesn’t mean we understand what it is that A does to produce B, and doesn’t even mean that we could claim that A is always, or necessarily, followed by B—it’s compatible with all that we know that it might suddenly be followed by C, instead.
That’s the same situation we’re in if we say instead of ‘A causes B’, ‘I choose B’. It’s an observed regularity of the universe that whenever I want something to happen, and have the powers (am unconstrained, etc.) to make it happen, then it will, in fact, happen.
But in the case of free will, that fact is taken to refute the notion; so, if we’re using the same standards, then either causation should just be as suspect, or the argument doesn’t work against free will, either.
We do have a very good model with predictive power: I want to have a sip of water, hence I predict that I will take a sip of water, and lo and behold, I just did so. We don’t have a model of how my wanting brought about that action (indeed, I don’t think it’s possible to have such a model), but that’s the same with causation.
