Tuesday, May 20, 2008

Should a Russellian be a Panpsychist?

Emmett Holman has an interesting article in the latest Journal of Consciousness Studies that’s also very timely in light of the discussion in the prior post. Surveying several approaches to a Russellian theory of mind (RTM, to use his abbreviation), Holman observes that most have been panpsychist, but at least one is presented as a version of physicalism, and some are neutral monist interpretations. Holman takes a look at the landscape to see what the advantages of competing approaches might be. He doesn’t come to a strong conclusion, but thinks that it’s not clear that a panpsychist approach should be preferred.

Setting the stage, Holman outlines the basis for interest in RTM: “The advertising for RTM is that it constitutes just the insight needed to break (what many see as) the current impasse on the mind-body problem.” He explains in particular that RTM is a response to dissatisfaction with “mainstream” physicalism, defined as the stance that “the mental supervenes on the physical as the physical is characterized by physical theory” (emphasis original).

According to RTM, the problem is that physical theory characterizes the entities in its domain strictly extrinsically, in terms of causal, functional and other relations in which they stand to each other and to our experience. Nowhere does physical theory inform us as to the intrinsic nature of the entities; and it can only be in virtue of this intrinsic nature that the entities have the causal powers or dispositional properties they exhibit. The question arises: can we know what this intrinsic nature is?

Well, the Russellian notes that conscious experience acquaints us with intrinsic phenomenal properties. In taking this point further, there are several possible views to take of the intrinsic nature of basic entities.

1. First, a monistic panpsychism posits that this intrinsic nature is essentially mental or phenomenal, and thus the mental dimension of human experience derives from the inherent mental quality of all nature.

2. A Russellian version of physicalism might say that the mental emerges or otherwise is a consequence of a configuration of physical (defined here minimally as non-mental) intrinsic properties.

3. Finally, a neutral monist would say the mental derives from basic intrinsic properties that themselves are neither mental nor physical.

Looking critically initially at the second (physicalist) option, Holman notes the objection that it might be seen to offer an epistemic gap just as large as conventional mental/physical gap. Well, perhaps not, since a combination of non-mental intrinsic properties giving rise to mental intrinsic properties may be seen as superior to trying to get mental intrinsic properties from non-mental extrinsic ones.

Still, isn’t the panpsychist option superior? Here, however, Holman looks critically upon another “gap” faced by this option. Since the idea of an elementary particle having mental states like we have seems ridiculous, advocates of panpsychist theories typically speak of the basic intrinsic properties as “proto-phenomenal” or “proto-experiential” etc. Elementary entities are sometimes said to have a “low-level” version of experience or mentality, or have properties that are in some way just analogous to human mentality. As long as we lack an account of how the simpler properties give rise to the more robust ones, a troubling gap persists here.

How about a neutral monism? Holman sketches his own version of how a neutral monist approach to the nature of these intrinsic properties might go. “The relevant fundamental intrinsic property [would be] a determinable property which is itself neither experiential nor non-experiential”(emphasis original). Its determinates can manifest degrees of experientiality along a zero to one scale.

Before pursuing this line of thought further, Holman circles back to see if any panpsychist accounts introduced concepts more amenable to coming in degrees than simply the concepts of the phenomenal, the experiential, or the like. One concept he sees which may have promise is the notion of “subjective unity”, which he finds, among other places, in Gregg Rosenberg’s work. Conscious states differ from the rest of nature in that they have (quoting Rosenberg) a “complex but not composite” character. Conscious states can be very rich and diverse (including various sensory modalities) yet they have a special kind of unity which binds them together. They can’t be simply decomposed into parts the way physical systems can be. The components can only exist as part of the unified state.

If this idea has merit, then maybe subjective unity can be “scaled down” gradually or in degrees in way that appears more intelligible than was the case for the notions of mentality or experience as such. In fact this could perhaps support a neutral monism in the following sense: the most basic entity considered alone has no experiential value. But once multiple entities are considered, they can be participants in a subjective unity with an attendant degree of experientiality.

Holman doesn’t take this past a sketch, and he also considers some other variations very briefly. He concludes that more work needs to be done on Russellian theories and it is not clear to him which avenue is superior. But he maintains that it is not clear to him from his survey that panpsychist versions are inherently to be preferred. At the same time, he doesn’t’ think much of the “physicalist” option either. Holman’s sympathies appear to lie with neutral monism, but he awaits a more full-fledged theory to consider here.

Postscript 1: Like many others who have written about this subject, Holman’s main references to Russell himself are to 1927’s Analysis of Matter. I note that Carey Carlson’s book relies much more on 1948’s Human Knowledge, which I have not read. I wonder if Carlson’s much greater emphasis on the importance of viewing RTM though the lens of an event ontology vs. an object ontology can be linked at least in part to this fact. I’ll have to get the book.
{UPDATE 14 June 2008: Having read both books -- the event ontology is firmly in place in both. The basic framework did not change from 1927 to 1948. To pick up the neutral monism from Russell without picking up the event ontology on which it rests is simply to miss too much of the point.]

Postscript 2: Note that Holman’s lone example of the “physicalist” version of RTM is from an older paper from Daniel Stoljar. Stoljar’s position in his more recent book (Ignorance and Imagination) has evolved and he doesn’t specifically champion this version of RTM. His endorsement of physicalism in the book relies on the epistemic argument that we just don’t know enough to say that the non-mental couldn’t be the supervenience base for the mental. The unknown “experience-relevant non-experiential facts” could be intrinsic, extrinsic or both. I have a series of posts on this book under the label Stoljar.


Mike Wiest said...

Hi Steve,

Sounds like a very interesting article. Thanks for your nice summary, seeing as I can't get the article for free.

I think your summary is also a very nice, concise, non-partisan summary of the space of candidate solutions to the mind-body problem.

You said: "Well, the Russellian notes that conscious experience acquaints us with intrinsic phenomenal properties." I don't know where Russell got this idea (I'm curious to find out) but it sounds like Schopenhauer's doctrine that we know the intrinsic character of the physical world through our own experience--aka Kant's thing-in-itself. By identifying that nature as "will" instead of "proto-consciousness," he (1) directly tied it to causality as in the RTM, and also (2) gives us a way to imagine how things (brains, electrons) can be more or less conscious while still being essentially mental.

I love the "subjective unity" approach to generating consciousness at different scales (electrons, brains). That's where my money is. Now we just need the concrete specific theory to fill out Rosenberg's skeleton. And a way to test it experimentally. I hope I live long enough to see it...

Steve said...

Hi Mike: I thought you would like the subjective unity idea. Tying binding to experience and having them scale up together seems like a good idea.

Anonymous said...

Hi Steve,

I very much enjoyed your two last posts. I've already told you about my position - that the views based on the dispositional/categorical distinction seems implausible, but let me try to put my thoughts in more clear way.

IF the dispositions of the particles of matter can be described in mathematic way, what we get is a logical possibility that anything that is happening in the world can be simulated by making a model based merely on those equations.

But if this is possible, then in such model, which would model the development of the life on Earth, to the present day, you would get philosophers discussing the consciousness the way we do.

So, I would see how on the intristic/dispositional properties account on might argue against possibility of normal philosophical zombies. For example one can say that there can't be dispositional properties without the categorical(intristic) ones, and further say that maybe it is metaphysically necessary that the intristic properties are exactly those and not some others.

But, the account like this, won't remove possibility of simulation-zombies. And, as I have pointed before, it will be very weird that again in this simulated case, we have the same 'discussing the consciousness' phenomena, which we need to be accept are fully explainable in the terms of the form of the equations.

This might not be epiphenomenalism, but it surely it is very close to it.

Steve said...

Hi Tanasije:

Here's how I would respond, using the stronger causal version of RTM where there is no causation without the intrinsic aspect.

To address your point we would need to take a closer look at your simulation:

If you succeeded in building a full and complete model which could independently interact with its environment and be indistinguishable from a human, then it would not be a zombie! It would have first person experience! (If you can build a complete physical simulation of the world, you are a deity!)

If you had in mind some partial simulation, there is no guarantee of similarly robust experience, but that doesn't seem like a problem for the theory to me.

Anonymous said...

Sorry, it seems I was not clear about the kind of simulation I had in mind (or maybe I didn't understood your response)...

I was thinking of a case of simulation of e.g. history of our planet, through a computer in which we would only put the data about the dispositional properties of the particles, and physical laws which under which properties develop.

This would be simulation of dispositions, so it wouldn't be simulating the categorical properties. But yet still, in this simulation we would get simulated-philosophers talking about the problems of consciousness, arguing that besides dispositions we need categorical properties and so on...

So, it is there, that I think categorical phenomenalism gets into similar problems as epiphenomenalism.

Steve said...

Hi. I'll need to respond more completely when I have time, or I fear you'll miss my point.

But briefly: First, I think you underestimate what would be involved in such a simulation. I'm not sure the amount of data one needs to put in is less than the same order of magnitude as the amount of data contained in what one wants to simulate.

Anyway, let us assume it can be done. The simulation must be causally implemented in some way. If it is done in a (huge) digital computer, then the physical system of the computer would be a real subject of experience. If you replicate the causal structure, you replicate the experience.

This seems absurd, but I think that is only because we underestimate the challenge of the simulation -- this will especially be true if our grounding in quantum mechanics needs to be considered to fully capture life and consciousness (as I personally think is likely).

Anonymous said...

I merely talk about the simulation as a thought experiment, so talking about theoretical possibility. (I can think of few theoretical problems of simulating closed physical system, especially with digital computers, but let's talk about this, only if you think those issues are important in relation to categorical phenomenalism.)

It is causally implemented in some way, but still it is simulation. So, for the categorical phenomenalism to claim appearence of consciousness in this case, you need to claim, not just that:

1)two systems with same dispositional properties will necessarily have the same categorical properties

(which I think is OK), but also...

2)the simulation of what will happen to the dispositional properties, in a purely external way (be it through calculation in computer, or series of hand-written notes, etc...) will necessarily have same categorical properties with the thing simulated.

The causal structure of the thing, and the simulation is different one. While in the case of the thing itself, we could say that the reason for the dispositions, and changes of the properties of the things (position, speed, etc...) comes from the categorical properties; in the case of the simulations, the causal change comes from different place - the engineered system which is created in order to calculate the predicates of the given system. So, to say, given the dispositions are only simulated, we can't use the principle (1).

Though, of course, if we return to the possibility of simulation, categorical epiphenomenalist can I think do several things: point to QM phenomena (be it entanglement, indeterminacy, etc...), and say that those things can't be simulated, and that those are important dispositional properties which if left out in the simulation will not given the prediction that we need.

Mike Wiest said...

I'm not familiar with this "dispositional/categorical" terminology, but if I read those as something like "extrinsic/intrinsic" respectively, then here's my take on Tanasije's question about simulation.

It seems to me the crux of the matter is where Tanasije says "the causal structure of the thing and the simulation are different." If that's the case, then those differences could explain why one system is conscious and the other is merely an unconscious simulation of people arguing about consciousness. In a panpsychist framework the simulation would have a certain phenomenology, but if its causal structure was different from the simulated system, then there is no reason to expect the phenomenology to be the same.

To evaluate your proposition #2, we have to explicit about what "simulation" means. If it is just a representation of all the interactions and their timing and so on, there is no reason to expect the simulation to have the same intrinsic properties as the simulated system. Actually, even the dispositional properties of the simulation could be different, if they just represent (in the mind of some interpreter) the simulated dispositional properties. So for example my hand-written notes to estimate when a bomb will explode and how much energy will be released, don't themselves explode.

Perhaps we could salvage propositions 1 and/or 2 by changing the wording to "two systems with ALL the same dispositional properties will necessarily have the same intrinsic properties." If this claim is true, then it would explain why computer simulations and notebooks aren't conscious even if they represent certain aspects of brain or psychological dynamics.

Another way to approach this would be to say, a simulation is only a simulation because it is interpreted as such by some observer. Therefore such an observer-depended status could not amount to consciousness, which I take to be an intrinsic property. The only exception I can see would be where the "simulation" was actually a "reproduction" such that "the causal structure of the thing and the simulation" would be the same.

A good example of this would be if consciousness does depend on quantum phenomena. A classical computer can simulate quantum phenomena, but can't for example reproduce the nonlocal correlations that occur in the quantum dynamics. It can simulate those nonlocal dynamics, but everything will take longer in the simulation than in the real system--which just one indication that the causal structure of the simulation is different from the real system. This could be the very causal structure that is responsible for the "subjective unity" of consciousness, so if that is missing from your simulation, then so is the consciousness!

Mike Wiest said...

So I guess I'm saying that how you implement your simulation IS relevant to this question about whether a simulation is conscious or is a zombie. If it is a PERFECT simulation, then it is a reproduction, and it will also be conscious. If it is only a representation of certain selected aspects of the dynamics, then it may or may not have the causal structure that makes it conscious.


Anonymous said...

Hi Mike,

I fully agree with what you said, and to me it seems that categorical phenomenalism should in fact buy the view that non-locality and non-determinism play roles in consciousness, and that it can't be simulated.

The problem with simulation is that, yes you can easily say... what you get in the simulation is simulation of people who act as if they are conscious, but they are not. That is they discuss consciousness, write books about consciousness, and so on...

But what is not clear, given that we removed the explanation that they act that way because they are conscious, if we will be able to give a plausible story of why such things appear.

I have wrote something here on this issue, trying to show that we will get into problems when we actually try to give an explanation for such things happening.

So, to repeat it does seem to me that categorical phenomenalism (or the Russelian way of solving the problem) needs to buy into the thesis that non-locality and indeterminacy are phenomena which have something to do with consciousness.

Steve said...

I was traveling the last couple of days and am catching up now. I appreciate the continued discussion which has clarified some things for me.

I agree with Mike's point that if a simulation falls short of replication (of causal structure), then its not clear the thought experiment has any value.

In terms of replication, I would not say it's impossible even if causality is rooted at the quantum level, as I believe is true. Some future technology utilizing quantum computing could work; or even using alternative biological parts - themselves utilizing non-trivial quantum effects - could maybe work.

We're somewhat in sci-fi realm here, but I think its important to note that the view here is different from a traditional dualist who would say replicating consciousness is impossible because you need an immaterial soul. Its also different from a modern property dualist who has no basis for asserting whether or not a replication would succeed: just because the epiphenomenal mental properties contingently attach in our case gives no guide to whether they would follow in a replication.

Mike Wiest said...

Hi Guys,

I have only glanced at Tanasije's posted argument, and I certainly have not absorbed it yet (I have trouble with zombies in general), so probably I should keep my mouth shut.

But there was a point I tried to make in my previous comment that I'm not sure if I got across. It's just this: something is only a simulation because some observer interprets it that way.

I don't see that there is really an issue or problem about explaining why the unconscious "people" in a simulation are talking about consciousness. There are no people, and they aren't talking about anything. There are just a bunch of molecules that make up a computer or a book or a robot, and some human looks at that and thinks "that picture on the computer reminds me of how a real person acts" or something like that.

We can straightforwardly explain all that "simulated behavior" in terms of the behavior of the physical parts of the simulator. I understand that this is where Tanasije is worried, because in the panpsychist RTM picture all behavior is supposed to be caused by "consciousness" (or whatever we call it). But it still is. The (proto-)consciousness of the electrons and protons is postulated to underlie all their physical interactions. In this picture we don't have to choose whether causality is implemented by the mental or the physical, because the mental is postulated to be the basis of the physical. So we get to keep the physical interactions that explain...ALL behavior including our own conscious behavior, and we get to say that our consciousness is causally efficacious because it is what implements or actualizes the events that others observe as physical behavior.

I can sympathize with the feeling that this solution is too easy, which seemed to bother BDK. It almost seems like a cheat. Like, if it's that simple, why were the most brilliant minds completely stumped for thousands of years? But I think it might just be right...

Anonymous said...

Hi Steve and Mike,

First, let me say that your view does seem to make more sense than physicalism or either substance or property dualism. So, I have sympathies for it.

I think I didn't do very good explanation of the simulation-argument, so I will try to put it in more clear way.

Say that we can distinguish the categorical properties (be them pro-phenomenal or whatever) and the dispositional properties of the matter.
Also say that we do have all the external properties of the matter in certain part of reality at certain time, which are important for the physical behavior of this matter. And say that now we build a simulation, which carries out calculations (according to the physical laws) of what will happen with within this part of reality.

What we are simulating and predicting is what will happen with the "external" properties of the matter. No phenomenal or proto-phenomenal properties are included in this.

So, what happens is that, given that we allow possibility of such simulation, the simulator will calculate (or predict) what will happen in that part of reality. And it will predict that there will be such phenomena (purely in the dispositional realm), where people (e.g. us) will discuss whatever we are discussing now.

As Mike says, there won't be real people, nor there will be real talking going on. But notice that whatever happens to the dispositional properties will be fully explainable by the rules which we put in the simulation and the starting state. Let's mark this explanation ETD (explanation through dispositions)

So, what I'm worried is this... Given that we do such simulation about what happened on Earth in past e.g. few millions years, I don't see a plausible EDT which would explain the behavior such as we do in fact show. That is, discussing consciousness, writing books about consciousness, and so on...

In the common-sense we have a ready explanation, and that is - we do those things because we are conscious. But, note that in the simulation, that explanation is out of question. EDT will be merely in terms of the starting state + physical laws.

So... if one (same as me) finds the possibility of EDT implausible , as long any view (be it property dualism, or categorical phenomenalism, or consciousness as being view) is alright with possibility of such simulation, it makes for me that view implausible too.

The good thing, is that categorical phenomenalism can go around this problem and deny the possibility of simulation.

Hope I succeeded in clearing up my thoughts.

Steve said...

Hi Tanasije:

That is helpful, thank you. I see now why you commented that it is important to the Russellian view that QM is involved in the process.

Your idea is that initial conditions + deterministic laws are enough to simulate physics. That is the vision of classical physics. I'm confident that this set up cannot simulate the world. I believe QM is needed to describe our world for reasons independent of the mind/body problem.

Discussion of gets us perhaps off-topic, but I'll summarize a couple of points.

1. QM is our most fundamental law and there is no reason to believe it ceases to describe nature once we reach the macroscopic scale. Just because we don't observe phenomena like superposition effects doesn't mean QM is still not the relevant physics. (I also think QM will survive while GR will be seen as approximately true in the formulation of quantum gravity -- I obviously have no expertise here, but I would bet money on it).

2. Classical physics doesn't accommodate real causality. Those deterministic laws famously work equally well when the time dimension is reversed. QM can accommodate asymmetric causality and the quantum measurement event is the best candidate for the basis causal process.

Mike Wiest said...

OK, that simulation scenario seems like a fairly strong way of stating the epiphenomenalism problem. We (think we) can imagine zombies or simulations that embody all the physics without the phenomenology.

I don't think I was really denying the possibility simulation, so I'm not sure we're safe with the RTM+. I have to admit I don't see that quantum mechanics automatically gets us off the hook. It seems we can imagine the laws of quantum mechanics being implemented without phenomenology, even probabilistic wave-function collapses (that's how most people think of them!). So even if we can't have a deterministic simulation it's hard to even understand "talking about consciousness" as one of the possibilities. Similarly, we can imagine nonclassical holism in superconductors and such without any mental aspect, so what does the consciousness actually DO?

I guess the answer on the table is "everything." But that answer seems to stop the discussion and any hope of empirical testing...but maybe that's the answer.

Maybe the alternative is that at some point we're going to have to refine the assumptions that Tanasije stated (that I usually accept too). I.e. maybe you can't distinguish the dispositional from the categorical "all the way down." Though I don't see a way to do this that doesn't just boil down to adding more dispositional "hidden variables," the causal openness of quantum mechanics seems like the only hope...

By the way, what is "categorical phenomenalism"? Also, why isn't the view we're talking about (panpsychism) a dual aspect or dual property view (especially with the assumption about distinguishing categorical from dispositional)?

Mike Wiest said...

Woops I see Tanasije was questioning the categorical/dispositional distinction from the start. But I remain unclear about how to abandon it...

Anonymous said...

Hi Mike,

Categorical phenomenalism is a name for the idea that what we figure out through physics is the dispositional properties of the objects, but that while doing that we don't know the categorical properties - the ground/base/reason of those dispositions.

As for why I said that QM might give a possibility to circumvent the simulation issue for categorical phenomenalism- I think it would go something like this...

1.We deny indeterminacy of the collapses, and say that there is a reason why the collapses happen in this and not some other way (so we buy principle of sufficient reason)

2.We deny that the reason for collapses happening in this and not other way is a physical reason, i.e. we deny that what way the collapse would happen can be figured out just on base of the dispositional properties (which is compatible with some interpretations of QM)

3.We say that the reason for the collapses happening this and not the other way is related to the categorical properties. (the nature of which may remain problematic - are they proto-phenomenal, are they holistic in nature, etc..)

Of course, the problem remains of how to test this, however it does seem at least theoretically that a test is possible. First, if the functioning of the brain can be simulated by classical physics, then we don't have reason to believe that QM has anything to do with it, and it seems to me, categorical phenomenalism becomes implausible view. If we are lucky, and if we figure out that QM events has something to do with right functioning of the brain, we try to simulate the indeterminacy of the collapses with a pseudorandom generator? If we do this, and still the brain doesn't function as it should, than we can conclude that collapses happening this and not that way, has a role in the functioning of the brain. (Of course, as a separate physical question we have assumption (2))

Though,of course we are long way from any practical possibility for these kind of tests.

As for the distinguishing categorical phenomenalism from the dual-aspect views, or from fully fledged dualisms... I'm not sure I'm knowledgeable enough to answer that, but I will put here some thoughts, and Steve can correct me.

Seems that categorical phenomenalism isn't "normal" dualism because the dispositional properties are not separate from the categorical properties... Given that one knows categorical properties, one will be able to deduce the dispositions (so, it is kind of other way from physicalism).

About dual-aspect views, I guess we can distinguish them from categorical phenomenalism this way... in the dual aspect view the consciousness and the physical are two aspects of one and the same thing. So, the first is not based in the second, nor the second in the first, but they have same "ground" -they are two aspects of the same thing, in the way color and size are aspects of the same apple. I in fact, buy something like this view. (I think dual-aspect views, because of the simulation-issue also like categorical phenomenalism need to buy into prediction that QM has something to do with the functioning of the brain, and that how collapses happen, has also to do with the functioning of the brain)

Steve said...

I like your analysis there Tanasije. Regarding the dual aspects: what we do assume in the causal-RTM approach is a pluralism of events. The categorical part is the inside of the event, the dispositions arise from how the event relates to others (I think intrinsic/extrinsic works fine as terminology also).

So the duality arises naturally from the fact that there are many events, as contrasted with a dual aspect view where a single fundamental entity has two different properties underpinning the physical and mental.

Mike Wiest said...

Thanks, guys, this is helpful. Who says it's impossible to make progress in a discussion about consciousness?

I like the quantum example as a toy model of "categorical phenomenalism" (CP). But it seems like it could be distinct from the causal-RTM model as just described by Steve.

It seems to me that causal-RTM IS a dual aspect view (according to the def you both just agreed on), if we just regard the "causal nexus" as the single fundamental entity, and the multiple events as emerging from the nexus (or "matrix").

I have been tending to think in "parallelistic" terms as we've discussed causal-RTM, i.e. every physical degree of freedom has a corresponding mental degree of freedom. I think I got this from Spinoza. If I'm getting the terms right, it seems like Spinoza would be a dual aspect panpsychist.

Causal-RTM could be parallelistic like this, if the intrinsic is completely fixed by the extrinsic. (Is that what "supervenience" means in mainstream neurophilosophy?) Anyway if we assume that for causal-RTM, it won't fit the profile for categorical phenomenalism.

That is, doesn't Tanasije's way to avoid the simulation/epiphenomenalism problem
REQUIRE that the phenomenal space is bigger than the physical space? By that I mean the phenomenal state determines the physical, but NOT vice versa (as would be the case in a psychophysical parallelist scheme). The quantum collapse toy model gives us a way to imagine the extra "phenomenal" degrees of freedom that would determine the physical outcomes.

However, I'm worried about whether the theory is conceptually coherent as soon as we relax the parallelistic/supervenience assumption. My worry is this, if we are able to enumerate the phenomenal states and their effects on the physical dynamics, and represent them with communicable symbols...then why aren't they physical degrees of freedom?

If the answer to that last question is about how the phenomenal are different from the physical (vs. they are two sides of one event as in causal-RTM), then don't we have a true ("substance") dualism? Not that there's anything wrong with that...!

I'm not sure if Steve was envisioning the supervenient causal-RTM (sc-RTM) or the causal-RTM where there is more inside the event-monad than meets the extrinsic eye(s) (CPc-RTM). If we pick sc-RTM we won't satisfy Tanasije's simulation criterion, so we'll have to settle for the faith-based solution to the epiphenomenalism problem. But if we pick CPc-RTM we have to come up with a good ontological distinction between the physical degrees of freedom and the INDEPENDENT spiritual degrees of freedom...can we do that?

Mike Wiest said...

Sorry to add more words again, but on reviewing Tanasije's definition of CP, I see that it also could have "supervenient" versions and versions with extra phenomenal degrees of freedom. Am I right to think that this is what determines whether the theory (CP or causal RTM) passes Tanasije's simulation test?

Steve said...

Hi Mike: Wow, those are challenging questions.

I’ve been thinking that the intrinsic event determines the extrinsic relations, but not vice versa. So I must think there is more going on within the event itself than the residue it leaves in its relations with other events.

As you mention, if we see the event as an actualization of possibilities (=quantum measurement or wave function collapse), it involves a process of selection from a set of choices. This is the part which is hidden from the extrinsic analysis. You ask whether what’s going on here can be represented with abstract symbols/mathematics. The possibilities can – we’re supposing that those are the wave functions. But the collapse itself I don’t think so – it’s a free choice. (If this isn’t deemed “physical”, so be it. It’s certainly natural).

What do you think?

Mike Wiest said...

Hi Steve,

You suggest that the wave function delineates the possibilities, but if physics only covers the extrinsic variables then the wave function is incomplete because it doesn't represent the phenomenal possibilities.

I think your "free choice" would avoid this particular problem because it doesn't appeal to new extra variables to explain the choice. On the other hand I literally don't know what it means. We don't want to call it random (or is that ok?) but we also can't explain the choice in terms of desires and values and so on unless (a) those are indexed by physical brain states or (b) we postulate that they are the extra non-physical variables that determine the outcome. If (a) then we're back to supervenience and physicalism and "extrinsic epiphenomenalism"; if (b) then we may loose the genuine "freedom" you wanted, and we're back to the problem of coming up with an ontological basis for distinguishing the independent phenomenal variables from the physical variables.

I'm glad Tanasije mentioned the principle of sufficient reason. I find it a very interesting and attractive principle, but I was under the impression that modern philosophers just stopped talking about it at some point. Did they have a discussion and drop it because of quantum mechanics? I would like to see that discussion! Anyway, if we believe that principle then I don't think we can accept Steve's free choice. (On the other hand Schopenhauer considered the principle of sufficient reason to be fundamental and did have a scheme that allowed for some putative real freedom...)

As you know I've entertained a collapse scheme where nonlocal entanglements determine each quantum choice, so that the choices are random from any local point of view but determined from God's point of view outside spacetime. I think a minimalist version of this would respect supervenience, so I don't think it would pass T's simulation test.

There's something else I'd like to mention regarding a way to distinguish extrinsic and intrinsic degrees of freedom. There are a lot of places in quantum theory where the terminology is very suggestive of bringing mind into the picture (to the annoyance of many of the positivistic persuasion). One of the less familiar places where this happens is I think relevant to our discussion, because it gives a concrete way to think about internal and external degrees of freedom in the spacetime events of causal-RTM.

In gauge field theories, the type of quantum field theories that make up our Standard Model of elementary particle interactions, one talks about "internal symmetries" and "internal degrees of freedom." These internal degrees of freedom define what kind of interactions the different kinds of particles participate.

At risk of taxing your patience I'll try to say how the "gauge principle" could be a concrete realization of the causal-RTM picture in which internal variables determine interactions with outside events. The "gauge principle" is kind of a generalization of relativity that says physics shouldn't depend on your point of view or reference frame. In other words, your physics should be "invariant" under changes of origin and measuring system. What happens in our system is the same for someone using a cm-ruler as for someone else using an inches-ruler. "Gauging" a theory means that you enforce a much stronger LOCAL invariance: you allow for different "rulers" at every spacetime point.

The beautiful amazing thing about these gauge field theories is that
by enforcing this weird requirement that your metric can stretch and twist at every spacetime point but the physics has to be invariant under those transformations, the forces/interactions just pop out! Einstein gauged the special theory of relativity and gravity popped out. The other forces pop out when you gauge the invariances that pertain to the "metrics" on the "internal variables" like spin and charge.

So the physical interactions like electromagnetic forces arise as a way of enforcing consistency among the infinite local "viewpoints," when those viewpoints or reference frames are "free" to float with respect to each other.

So, finally, I'm suggesting that maybe we have already discovered the internal, phenomenal degrees of freedom, from the outside. They are fundamentally distinguished from the extrinsic variables simply by being the non-spatial variables. One problem is that space-time is taken for given in this picture...another is that I think this is still a supervenience-respecting picture so I don't think it will pass T's simulation test.

Mike Wiest said...

ps. What really defines something as "dispositional"? I felt I had to drop that terminology in my last post because I can't tell if the "internal " variables (phenomenal or otherwise) count as "dispositional." If a variable can be indexed by numbers, does that make it dispositional?

Steve said...

1. With regard to your first question, I’m not sure, but I’m guessing that the phenomenal properties only need to come in during the process of choice.

2. I’m not sure I share your concerns in your second paragraph. “Free is free”, meaning the outcome cannot be derived from other known variables. Free vs. random: I think that what is free from a first person perspective might be viewed as random from the perspective of a third-person observer. While hidden-variable theories and determinism have not been completely (logically) ruled out, it is my impression that the so called “no-go” theorems in QM strongly indicate that measurement events are free (or random) in this sense.

3. The classical version of PSR fails in QM. But if you supplement your “reasons” by counting the free choice as one, then you can have a revised PSR I think. I haven’t read anything in recent memory about PSR in a quantum world.

4. I don’t like the idea of non-local entanglements *determining* the local choice. I would rather say that entangled systems are larger groupings which together represent a correlated set of possibilities to be chosen together. Or from the perspective of the entangled system, it implements a coordinated simultaneous free choice when interacting with another system (or something like that).

5. Your thought regarding gauges is interesting and I’ll have to ponder that.

PS: I’m not that sure about the terminology of dispositions either. I think they are supposed to be properties which give an object the power to affect something else or otherwise give rise to a manifestation. Given the Russellian analysis of physics – that we only know physical entities through their interactions with other things, extrinsic might be better than dispositional (it matches better with third-person vs. first-person also). Intrinsic properties might be seen as dispositional in some sense, since they could be seen to ultimately confer the power to affect other events.

Mike Wiest said...

Hi Steve,

I know you don't like determinism but there's determinism and there's determinism! I used the word "determine" in my non-local collapse proposal, but remember this would not mean the future is determined by the present or even by all of history. So for anyone in time, freedom would be absolutely real, and/or quantum indeterminacy would still be irreducible even in principle.

I don't know if this scheme is consistent in detail, but if it is I think is preserves both the PSR and irreducible uncertainty or real freedom.

I like the "coordinated simultaneous free choice" formulation too though. I just get stuck about how to get past the randomness puzzle: if "free" means undetermined (vs. Spinoza's def: self-determined), but randomness is not freedom, how do we capture the motivation, the optimality, the adaptivity, the sufficient reasons for free actions?

Maybe we could index different levels of quantum freedom in terms of the spatial scale over which a system could optimize, harmonize and coordinate its global configuration, FREE from local constraints....

Penrose had a nice example of a liquid crystal behaving in this way, forming patterns that could not be constructed locally piece by piece, but had to arise from a larger scale quantum state.

Maybe that's a solution for keeping indeterminacy/freedom AND keeping reasons/motivation for our free choices: the stable "determining" factor is some goal we have, but a quantum dynamic in the brain frees the neurons from local mechanical constraints in trying to find an optimal route to the goal...

Would that be freedom in your view? Or do I have to specify how free you were when you formed your goal??


Steve said...

Hi Mike. Those are good thoughts on the question of freedom – it’s a challenging subject for me.

I think having reasons/motivation/goals is consistent with freedom. These are shaped and constrained by past events and the adjacent environment, but there are always multiple possibilities (including whether to participate as part of a larger system), and the final choice is not determined (thus is free).

On randomness: i'm taking the view that there is no such thing in the fundamental ontology. That is just how things may appear to an observer.

Mike Wiest said...

Yeah, I also tend to think of randomness as reflecting ignorance rather than "things happening for no reason." Which I think means we believe the PSR. Maybe Einstein's "no dice" doctrine means he did too. But it's non-trivial to defend that view against the putative irreducible randomness of quantum dynamics.

But maybe your solution could be to identify any residual quantum randomness as free acts, which look random but DO have a reason...?