Free Will in Psychology Today

Yet another piece on free will by a psychologist can be found here.  This time, the author is Joachim Krueger--a social psychologist from Brown University.  In light of the gathering number of popular science pieces on free will, I thought this might be a good opportunity to get everyone to list the ones they know of in this thread.  I actually had someone contact me the other day when I posted the the link to the "free choice" study asking me if there was a rough and ready list for getting one's feet wet with respect to what psychologists are saying about free will (or lack thereof).  Since this is something that comes up here at the Garden, I figured this would be a perfect place to get the ball rolling.  Once the comment thread is flowering, I will compile it into a bibliography and post it here for others to use.  It might aslo be helpful if you state in your comments whether the psychologists take a pro, anti, or neutral stance on the existence of free will--which would make it easier for me to clasify the bibliography.  Since I know at least two Gardeners are already working on a paper about the contemporary interest amongst psychologists in free will, I assume they will take the lead here! :)

UPDATE:

The piece by Krueger is a response to an earlier post by Baumeister which can be found here.  Another response to Baumeister can be found here. 

Comments

[tnadelhoffer]

Al and Chris,

Thanks for posting this. It's precisely what I was after. I will actually turn into a free standing post so that in the future people can ad to it without having to wade through this already lengthy comment thread! I hope all is well around my old stomping grounds in Tallahassee!

[Paul Torek]

Tom,

If I understand your argument correctly, it goes something like this:

(1) Assuming fatalism* is false, various deterministic relations hold between events in agents and the world, such as reasons responsiveness, guidance control, etc.
(2) Assuming fatalism* is true, agents STILL stand in the same deterministic relations, have the same properties and powers, etc.
(3) The deterministic relations mentioned in (1) underwrite compatibilist free will in a non-fatalistic* scenario.
(4) Therefore, the compatiblist should likewise attribute free will to the fatalistic* scenario.

I reject (2). Properties and relations don't always survive in possible worlds where we change the physical necessities (i.e., the laws of nature). For example: if you posit radically different physics of liquid-solid-gas interfaces in world W, you can render incoherent the attribution of "wetness" to anything in W. Changing the attribution of logical necessities is, on its face, an even greater change than changing attributions of physical necessities, so we should be on the lookout for properties and relations to be affected by this change.

In particular, fatalism* threatens to make all deterministic relations, qua deterministic relations, go away. It deprives us of one way to make sense of the claim that these relations are deterministic. Suppose all possible worlds in which our laws of nature are true, are worlds where cause C is followed by effect E. That seems like a powerful and important point. But if fatalism* is true, it's a trivial point. It holds for every pair of events C and E, provided that they are actual events.

[Tom Clark]

Paul, thanks. Not sure I get this, but it seems what you’re saying is that global necessity of the sort in which everything has to happen as it did (in a fatalistic* world) subverts the relative, deterministic necessity of the sort which holds between events, the antecedent of which didn’t have to happen in a merely deterministic world. That is, the reliable cause and effect relationships embodied in laws of nature which we call deterministic cease to exist under fatalism*. This is interesting since the two worlds we are considering are identical in their events, it’s just that in one the past and laws had to be as they were, in the other not. If the events are the same, then it seems we should be able to extract the same deterministic laws of nature in a fatalistic* world, since those laws are simply a matter of the observed conjunctions between events. If so, and compatibilist agent powers (ultimately) depend on such laws, it seems that agents in both worlds would have the same powers. But I’m likely missing something here.

What’s also interesting is that fatalism* isn’t what Kip calls True Fatalism, the idea that “State 5 happens *regardless* of state 4. State 4 could be anything, state 5 doesn't care. Whether I throw the brick or don't throw the brick, the window still breaks.” Or, as some people reportedly believe, since my death is fated to be what it is no matter what I do, whether I die in a car crash is independent of how I drive. That we have to put an asterisk after fatalism in this discussion suggests to me that Eddy’s definition of it in footnote 8 isn’t capturing what we (the folk) normally mean by it, namely the idea expressed in these examples. It’s this more folkish and I think standard understanding of fatalism that seems to me to abrogate deterministic necessity, by making consequents bizarrely independent of antecedents.

[Paul Torek]

Tom,

Events, along with properties and relations, are among the things I doubt can be the same, in a fatalistic* world. Laws of nature also seem to lose something, or even everything, in a fatalistic* world.

The thermostat is set to 72. I have a dime in my pocket. The temperature of the room is 72 Fahrenheit. The thermostat setting is causally relevant and the dime is not. If there were no dime in my pocket, the temperature would still be 72. If the thermostat had been set to 65, the temperature would not be 72.

Fatalism* nullifies the last two statements. Arguably, it nullifies the last three.

I agree, though, that True Fatalism (also) abrogates deterministic necessity. I also agree that True Fatalism is, to the folk, probably a clearer threat to freedom than fatalism*, and also more prominent on their radar screen.

[Kip]

Paul:

My claim is just that agents in fatalistic* worlds have just as much power as agents in merely deterministic (deterministic A) worlds. This is because the agents in both worlds have their powers in virtue of traditional compatibilist capacities such as reasons-responsiveness, and not in virtue of other branching paths (or laws of nature) being actually contingent. In other words, there is *no* sense in which, by plucking an agent from a merely deterministic world, and inserting him into a fatalistic* world, we've impinged upon that agent's powers *at all*.

You note that "Fatalism* nullifies the last two statements." That may be true. But an agent with the same molecule-by-molecule composition, who wants to make, and does make, the exact same sequence of decisions and actions, and whose decisions and actions have the exact same effect on the world he lives in---indeed, where the entire history of the universe looks exactly the same---has the same powers as the same agent in a merely deterministic world.

It amazes me (perhaps through no fault but my own) that some compatibilists disagree with the above. To assert otherwise is to hinge free will on actual contingencies, as opposed to conditional/conceptual contingencies. In other words, to assert otherwise is to hinge free will on exactly the sort of categorical contingencies and "forking paths" that compatibilists have criticized libertarians for prizing.

[Paul Torek]

Kip,

You're helping yourself to a claim of molecule-for-molecule compositional sameness. But that property of molecular composition is one of the properties (and relations and events) whose cross-world identity I question. I gave arguments, including the example about wetness and the point about causality and counterfactuals, to underwrite my suspicions. Maybe my arguments are no good, but when you help yourself to the kind of identity claim that the arguments target, I feel that you're trying to start the race at the finish line.

To my mind, properties like "being an oxygen atom", laws like "tends to take on a -2 valence when oxidizing metals", and counterfactuals like "if exposed to magnesium at high temperature, will react", are three sides of the same coin. It's a strange coin, I admit, but I hope it's not a strange claim.

If you smash that third side, you smash the whole coin.

[Mark Smeltzer]

Paul's interpretation of fatalism* seems correct. Fatalism* paints a picture that is the same as Zero's paradoxes about motion: a world in which all of our perception is illusion. In such a world, none of the objects we perceive have any sort of individual existence -- there are no parts in such a world; just one big, indivisible whole.

Determinism may, in general, allow us to think of the world as one big whole, but it is a whole that (hopefully) allows us to make real distinctions regarding its parts in order to assert the existence of individual objects within the world.
Without that ability in a given world, we can't even talk intelligible about the agents or their actions within that world -- all of it would be equally illusory.

[Tom Clark]

Paul, Mark,

According to Eddy's definition of fatalism, "the actual state of affairs (or the actual past or laws) are necessary (could not be otherwise)." So it isn't as if discrete events and objects cease to exist in such a fatalistic* world, nor do laws cease to hold, it's just that they couldn't have been otherwise. In such a world it's still the case that room temperatures are causally related to (constantly conjoined with) setting thermostats but not to putting dimes in pockets.

Eddy writes: "Most philosophers agree that even in a deterministic universe, natural events are still contingent; because past events (and laws of nature) are not necessary, neither are the events they deterministically cause." But I'm wondering: do we know for sure that our world isn't fatalistic*? If it is, then compatibilist freedoms and powers as you think of them don't exist.

[Kip]

Paul,

You wrote:

"But that property of molecular composition is one of the properties (and relations and events) whose cross-world identity I question."

I don't see any argument above for why we can't conceive take molecules from one world and put them in a fatalistic* one. You allude to arguments above, but these arguments only mention "events," "properties," "relations," and "laws of nature." No molecules (or even things).

So I don't understand how you "question" the "cross-world" identity of molecules in my thought experiment. Or why. I'm not even sure what you mean by that.

And I don't understand how "laws of nature" lose something (or "everything") in a fatalistic* world.

I think the following example might help better understand each other:

Imagine fatalistic world* A* and merely deterministic world A. Suppose that the only difference between A and A* is that A* is fatalistic*: the laws of nature "had to happen" and the first state of the universe "had to happen" (so every later state "had to happen" too).

A and A* have the same laws of nature. A and A* have the same first state of the universe. Indeed, in a physical sense, A and A* look and unfold exactly the same in every way.

So my question is:

1. Why (how) do you question the cross-world identity of the molecules of a person in A and A*? What do you mean by that? It seems to me that, to object to a hypothetical thought experiment like mine, you would want to show that it's logically impossible or incoherent somehow. So how? It seems perfectly consistent and logically possible to me.

2. In your February 22, 2009 at 07:37 AM post, you wrote:

"So, as I see it, your [February 19, 2009 at 09:36 PM] thought experiment is caught in a contradiction."

What contradiction is that?

3. In the same post, you wrote that you believe fatalism* is false. Why do you believe that? How can you be sure?

As far as I can tell, merely deterministic worlds, and their fatalistic* counter-parts, looks physically identical. So I'm not sure what evidence you could have that we live in one versus another. Maybe I am using "fatalistic*" in a different sense than you?

[Mark Smeltzer]

Tom,

I overlooked the part in paratheses in Eddy's definition, in italics here: "the actual state of affairs (or the actual past or laws) are necessary (could not be otherwise)". My comments assumed something more bland like this: "the actual state of affairs are necessary (could not be otherwise)".

This seems to represent something very different from: "the initial state of affairs and the laws are necessary (could not be otherwise)". So, I don't see how they can be packaged together in a meaningful label... the prior represents the kind of world that Zeno was worried about, and the latter does not.

[Paul Torek]

Kip,

I will try to explain my perspective at greater length. I am afraid I am not the best person to attempt this. Thinking about fatalism* makes my brain hurt even more than, for an interesting comparison, working with three-valued logic. It's hard to keep from relapsing back into familiar modes of thought.

Let's start by noting just how strong fatalism* is. Consider a possible universe whose only mass/energy consists of two electrons repelling each other. Now consider another world where three blue cats exist instead of the two electrons. Now imagine a universe which contains exactly one tape-measure.

No problem, right? Wrong! - according to fatalism*. Each of those universes is impossible, it claims. There ARE no other possible worlds; this is the only one. Here then is why I think fatalism* false: it rules out possibilities that seem perfectly coherent.

Here, it matters a lot what kinds of "possibility" you think there are. Nomological possibility is not the kind we're talking about here. (It would be redundant to point out that worlds where our laws of nature don't hold are not nomologically possible.) Logical possibility is another kind. And, as Cihan mentioned, some philosophers think that "metaphysical possibility" is a middle ground third kind. But I don't claim to understand any middle-ground kind, so I take fatalism* to be talking about logical possibility. And denying logical possibility to any scenario other than actuality seems absurd.

Now, as to why your thought experiment was caught in a contradiction: you propose inserting an agent's reasons-responsive mechanism from a regular world, A, into a fatalistic* world, A*. But according to fatalism*, world A doesn't, can't, exist. You can't get there from here. You are relapsing into familiar modes of thought. You are forgetting just how totalitarian, so to speak, fatalism* is. You can't transport anything from A to A*; at most one of them is accessible.

Lastly, I can try to answer your first question, why molecules in a fatalistic* world would have to be different from molecules as we know them. That goes back to the way properties, including the property of being oxygen, are enmeshed with natural laws and counterfactuals. Call the fatalistic* world's nearest thing to oxygen, Substance O*. Something's being O* supports no counterfactuals. It doesn't, for example, tend to react(*) with magnesium(*). It just happens to do so, in the one and only possible world. Substance O* is a facade. It is to real oxygen, what someone who goes around grimacing and yelling "ouch!" is to someone who is actually in pain.

If we were interested in superficial behavior, there would be no problem. But compatibilists have explicitly talked about dispositions, powers, etc.; and we have used counterfactuals quite freely. I don't see why we should quietly tolerate a view, fatalism*, which puts all such talk in question.

Now, there is an easy way out of my argument here. Tom Clark points it out.

Tom,

You're right of course that constant conjunction still applies in a fatalistic* universe. And if that's all there is to our laws of nature, then those laws could still obtain in a fatalistic* universe. Humeans about causality should not accept my arguments. I am not a Humean.

[Kip]

Paul,

Thanks for your response. I understand if my arguments are growing tiresome (people say I argue too much). I do hope, however, that you find this issue about fatalism*/determinism half as interesting as I do.

There's a whole lot to respond to.

First, you wrote:

"Each of those universes is impossible, it claims. There ARE no other possible worlds; this is the only one. Here then is why I think fatalism* false: it rules out possibilities that seem perfectly coherent."

But coherence doesn't imply possibility. You're relying on that premise; I see no reason whatsoever to grant it. I can imagine how the Holocaust wouldn't have happened. That doesn't mean that its not-happening was possible.

"And denying logical possibility to any scenario other than actuality seems absurd."

Here, again, we misunderstand each other. I don't mean logical impossibility. How could I, when I agree that non-actual worlds are logically consistent. When I imagine that the Holocaust didn't happen, I don't break any laws of logic.

What I am talking about is more like metaphysical possibility. You say that you dismiss this concept outright. That would explain why we're not understanding each other.

"But according to fatalism*, world A doesn't, can't, exist. You can't get there from here. You are relapsing into familiar modes of thought. You are forgetting just how totalitarian, so to speak, fatalism* is. You can't transport anything from A to A*; at most one of them is accessible."

It's a thought experiment. I'm not bound by the traditional impracticalities of the world. I, thought experiment creator, stand above both of the fatalistic* and merely deterministic worlds that I've imagined. Thus, I am not bound by either their fatalistic* or merely deterministic rules. I can pick and reinsert things from either, into the other, like a god. Impractical? Yes. Logically impossible, no. And that, I think, is all I need for the thought experiment to *make sense*.

Compare: Imagine standing over our world (as we know it) as well as Captured-World (which is different). In Captured-World, you reach down like a god, grab Thomas Jefferson just before he sits down to write the Declaration of Independence, and insert him into my apartment in our world. Then Thomas Jefferson, suffering from huge amounts of future shock, joins me for a walk around Washington DC and talks about the excessive expansion of the federal government.

Impractical? Yes. Logically impossible, no. And the same applies for fatalistic* and merely deterministic worlds. You say that "You can't get there from here." But, with my amazing thought experiment powers, it seems that I can get there from here. That is, I still don't see the logical contradiction that would make the thought experiment non-sensical.

"Something's being O* supports no counterfactuals. It doesn't, for example, tend to react(*) with magnesium(*). It just happens to do so, in the one and only possible world. Substance O* is a facade. It is to real oxygen, what someone who goes around grimacing and yelling "ouch!" is to someone who is actually in pain."

"Tending to react with magnesium" is not an essential attribute of oxygen. Oxygen happens to do that, in our world, with our laws. But if you took it and put it in another world, with different laws, it might lose that property. It would lose that property, but not its identity. (Compare: Kip breathes air; but if you put him in the bottom of the ocean, he no longer breathes air; that doesn't mean he's not Kip any more). The fact that we both, by stipulation, are considering taking the exact same physical molecules of oxygen, and inserting them in world A* from world A, shows that the molecules are the same.

"Now, there is an easy way out of my argument here. Tom Clark points it out."

I'm not looking for an easy way out. And I don't think my position hinges on Humeanism.

[Paul Torek]

Kip,

I prefer not to say that I dismiss metaphysical possibility outright. Rather, I think it collapses into logical or nomological possibility.

Even thought experiments have their limits. Kip the Capturing God needs to decide how many worlds he rules over. If it's many, none of them can be fatalistic*. After all, the definitional rule of fatalism* is This Is The Only Possible World. Not even a Capturing God can break that rule if He wants access to A*, the one and only fatalistic* world. I don't think this depends on my skepticism about Third Way metaphysical possibility, but let's see.

Let's suppose that metaphysical possibility is narrower than logical possibility and broader than nomological. Let's suppose that an ordinary merely-deterministic world with RR agents is logically possible but not metaphysically possible. The Capturing God takes one of these agents and inserts it into the one and only metaphysically possible world A*. But this changes A* into A**, which is a logically possible but metaphysically impossible world.

Where does that get us? It doesn't demonstrate that it's metaphysically possible for a free agent to exist in a fatalistic* world. And isn't that the point?

While we are pondering multiverses of possible worlds, let's also do justice to the fact that reasons-responsiveness is not a single-world feature. Rather, it describes a whole neighborhood of (nomologically) possible worlds. So it seems you should not just transfer a single agent from one world A, but a whole panoply of agents from worlds A, A', A'', etc. into the neighborhood of A*. But A* hasn't got a neighborhood. That's the problem.

The molecular-identity problem is a variant of this same problem. At least it is, on a non-Humean view of causality, combined with the view that properties are intimately tied to causal regularities.

My explanation was too short. I don't want to suggest that every single tendency of oxygen is essential, nor the tendency of reacting with magnesium in particular. Rather, it is all of oxygen's tendencies taken together, with some weighting of relative importance, that matters. I hope you wouldn't insist that in a world where no molecules react with any other matter, nor take gaseous form at room temperature, nor play a role in breathing, nor can be broken into atoms with atomic number 8 - etc., etc. - nevertheless could have oxygen? There are limits even to what a God of Counterfactual Worlds can stipulate.

But what if O* molecules do do all these things in A*, yet cannot be said to tend to do any of these things? Well, I think that is just too thin a basis for calling O* oxygen. (Put another way: if fatalism* is true, common sense is radically wrong about basically everything, including the nature of oxygen.) Your mileage may vary.

Suppose your mileage does vary, and you feel that the term "oxygen" can be applicable even in the absence of full-blooded counterfactuals. Still, I think you should look more kindly on my claim that "reasons responsiveness" is intimately tied to counterfactuals.

[Mathew Iredale]

How about the March/April 2007 issue of Behavioral Sciences & the Law? It was a special issue on free will.

http://www3.interscience.wiley.com/journal/114203221/issue

[Kip]

Paul:

Some responses. I'm not sure if we're making progress.

I am of two minds about your responses. First, I would like to just keep arguing. In that case, I could try with variations on the thought experiments I've used, as discussed below.

Second, I could abandon the attempt to explain how the thought experiment could still work, and simply try to convince you that the difference you mention is irrelevant---as I am 100% sure it is. That is, even if the particular thought experiment doesn't work for you, you should still agree that agents in fatalistic* worlds have as much power as those in merely deterministic worlds, on principle alone. The arguments for this point are many, the most important being that agents in both worlds do the exact same thing, perceive the world the exact same way, and live lives that appear 100% indistinguishable (even if you want to call oxygen "oxygen" in one world and "oxygen*" in another).

I have to say that I find your responses telling. It is a fascinating question to me, whether the free will debate doesn't get resolved because of some fundamental mistake by one party, or because of mere semantics and word games. Responses like yours suggest that it's not merely semantics: it seems to me that one of us has to be 100% wrong.

I'll proceed with prong 1. But my hopes of convincing you are growing dimmer and dimmer. I 100% believe that the powers compatibilist agents have are entirely irrelevant to categorical contingencies at the Big Bang or any other time. Whether the state of the universe could have been different at T1, whether the law of gravity could have been different, is absolutely irrelevant to a compatibilist agents' freedom.

Proceeding with prong 1, you wrote:

"Kip the Capturing God needs to decide how many worlds he rules over. If it's many, none of them can be fatalistic*. After all, the definitional rule of fatalism* is This Is The Only Possible World. Not even a Capturing God can break that rule if He wants access to A*, the one and only fatalistic* world."

Consider a meta-world. A meta-world is a collection of worlds. I know you will immediately object: "Well, if it's a meta-world, then the sub-worlds aren't really worlds"---put this objection to the side for a moment and see if the thought experiment can still work for you.

The meta-world is divided into two major sub-worlds, A and A*. A* is fatalistic* with respect to sub-worlds. That is A* has one, and only one world, in it, and the first state of the universe, and the laws of nature, in that single world "had to happen."

A is merely deterministic but not fatalistic*. That is, there is a world in A, but its first state and laws of nature didn't "have to happen." They could have been different.

[What does this mean? It means whatever Eddy and others meant when they said that compatibilists deny that the past and laws of nature had to happen. One way to think of this is: suppose that a God stands over the meta-world and can hit the "restart" button over and over. As he hits restart over and over, world A refreshes with randomly different first states and laws of nature, but world A* is always the same.]

Now suppose there is a God-type being ("Socrates") that stands over the meta-world, including A and A*. Suppose he selects compatibilist agent Joe in world A, then clicks "Edit" and "Cut." Then he selects world A*, and clicks "Edit" and "paste."

I submit to you the following:

1. There is nothing incoherent about the above thought experiment. It is amazingly impractical and fantastic, but violates no law of logic.
2. The thought experiment shows that Joe has the same power in fatalistic* worlds as he has in merely deterministic worlds.

[If you deny 2, because you do not believe that A* above is truly fatalistic*, I submit that, even if it is not truly fatalistic*, it is *not relevantly different* than a truly fatalistic* world: there is no principled difference between A* and the "genuinely" fatalistic* world of your choice that would advance your point (that Joe has more power in merely deterministic worlds).]

The above is just one way of conveying the same point. Another is to consider not a meta-world, but rather a single world, which is restarted as a fatalistic* world.

Suppose that there is a single, merely deterministic world A. It has already had its first state, and its laws of nature are set, but these didn't "have to happen." Now Socrates hits "Pause", selects Joe, clicks "Edit" and "Cut."

Now, to prove Kip's point about fatalism* and determinism, Socrates then clicks "Lock First State of Universe and Laws of Nature." Now, whenever Socrates hits "restart" the first state of the universe and the laws of nature will always be the same. They "have to happen."

[If you say, Socrates is God-like, he can do anything, so he could later unlock these things, so the world is not truly fatalistic*, imagine that he is less powerful than an omnipotent God. Suppose he has the power to lock the laws of nature and "throw away the key," to permanently bind himself and everyone else to the first state and the laws of nature.]

Now that Socrates has locked these things, imagine that he restarts the universe. The first state is precisely the first state he picked out, which "had to happen" because it is locked in. The same is true for the laws of nature. Now suppose that Socrates clicks "Edit" and "paste" to paste Joe into this world.

I submit that, as in the example above:

1. Although amazingly fantastic, the above thought experiment is not incoherent.
2. Joe has just as many powers in this fatalistic* world as he had in the merely deterministic world.

If you disagree that the above world is fatalistic*, because Socrates restarts the universe after "locking" the laws of nature and first state, I submit that it is not relevantly different than a "genuinely" fatalistic* world.

What do I mean when I say it is not relevantly different? I mean this: even if there is a difference between the world A* above and a "genuinely" fatalistic world, they are not differences that provide Joe with any more power or freedom in merely deterministic worlds (in comparison to fatalistic* worlds). From Joe's perspective, it does not matter whether the world was first deterministic and later had its first state and laws "locked", or whether Socrates never existed and the first state and laws were always locked from the start. Joe experiences the world, and has the exact same powers, in either case. In other words, it is not *in virtue of* such differences between "genuinely" fatalistic* worlds and worlds like A* that compatibilist agents have their powers and freedom.

There are more thought experiments where these come from, and maybe I am not designing them in the best way to make my point. But the point remains, and I think you can reach it without any thought experiments, just on principle alone (as I tried explain in the beginning). If the above doesn't persuade you, it might be that nothing will.

[Kip]

Quick revision:

Above I wrote:

"It means whatever Eddy and others meant when they said that compatibilists deny that the past and laws of nature had to happen."

This isn't quite right. Eddy and others didn't deny that these things had to happen. They simply asserted that "determinism" doesn't imply that they had to happen. And I agree with them about this.

[Mark Smeltzer]

Kip,

Having thought about this a bit more, I think I was wrong in my assessment of fatalism* from the quote that Tom provider. After Tom's post, I thought I had made an error of recollection regarding the definition of fatalism, but now I think I was spot on.

I think what I am going to say here aligns with Paul's thoughts, and might be a nice way of summarizing things and making it a bit more accessible.

Let's start by reviewing the meaning of fatalism*, for any world W:

1. Let W designate a world in which the laws of nature are fixed at all times T in W.
2. Furthermore, let W only designate worlds in which the initial state of affairs at time T0 could not have been otherwise.

Any instance that W refers to is thus a fatalistic* world. Agreed?

Okay, now here's the tricky part: there are multiple readings of "could not have been otherwise" because we can distinguish between logical and metaphysical necessity, and both senses might apply to a W world.

For instance, any world in which brane theory is true would be W world: metaphysical branes determine both the initial state of affairs and laws of physics in each universe created by the branes' collisions. However, it is plausible to think that the initial state of affairs and/or the laws of physics in those worlds, while metaphysically necessary, might be logically contingent.

Moreover, if the type of necessity in a W world is of the logical type, then Paul's worries (and mine) come to bear: every event is thus a logical extension of some set of first principles -- there is no cause and effect, no molecules, no physics, no objects, no properties, and no relations; just the first principles and their logical entailments.

Everything we want to "point out" in a W world derived from logical necessity is just fantastic illusion, and is an utter waste of metal effort. The only thing we could possibly talk about with any degree of sensibility in such a world would be the first principles themselves (and even that is questionable if they are reasonbly complex). Thus, any attempt at meaningful comparison between features of a W world derived from logical necessity to features in a world not derived from logical necessity is doomed to be a wasted effort.

Fatalism* has to be split up by at least this amount. And, luckily, that amount ought to be just the amount needed to change your (Kip's) mind.

[Paul Torek]

Kip,

I've been thinking that maybe I should change my view about responsibility in a fatalistic* world. But not as you suggest. Rather, I should probably endorse a radical version of Wolf's Asymmetry Thesis about agents in a fatalistic* world. That is, when they do respond to reasons, they are perfectly free, and when they fail to do so, they are perfectly unfree.

You can probably see why, but I'll explain anyway. Like Fischer, I see people in the actual world as moderately reasons-responsive. We're almost never guaranteed to do what we have most reason to do. Even when our agency functions well, there are usually some possible worlds in the neighborhood, in which we fail to do so. But, the positive flip side of this is that when we fail, there are often many nearby possible worlds in which we succeed. We tend to be reasons-responsive. And in virtue of this we rightly say "I could have done better," when we fail.

That's how it is (arguably) in the actual world and its neighborhood. But that's how it most certainly isn't in the fatalistic* world. When people succeed, that had to happen. When they fail, that had to happen. Agents in A* are never moderately reasons-responsive. They are perfectly rational at some moments, and perfectly irrational at others.

So I imagine you'll still be unhappy with my view.

I really like Mark's remarks about a W world derived from logical necessity. Another way to put it is that all the "facts" in W, for example that "Joe exists", become theorems. How similar is any Joe you know, to a variable or term in a theorem? Isn't that world a little too logico-mathematical to cozy up to? Can you put yourself into the shoes of a mathematical construct?

On prong 1, your constructive argument by thought-experiment, I really do think you're violating laws of logic. What you stipulate in one place you take back (contradict) in another. For example, powers. If Joe has certain powers in A, this has implications for nearby nomologically Possible Worlds (PWs). But in A*, there are no other nomologically PWs, because there are by definition no other metaphysically PWs - nearby or otherwise - and every nomologically PW has to be a metaphysically PW. So necessarily, if we place Joe in A*, we strip away his powers.

Maybe not all his powers. Some powers might be underwritten by Joe's successful performance in a single PW plus the assertion that his success had to happen. But gone are any powers relating to anything that didn't actually happen.

I think your intuitions are being skewed by the phrase "possible worlds". It's easy to imagine transfer between (literal) worlds, but we should instead think of possible scenarios; ways that all facts might have been. Even Everett's Many Worlds Interpretation of QM counts as a single "possible world". Now that's bad enough, but when we start describing one possible world in ways that also define other possible worlds or the lack of them - as with fatalism* - things get really ugly. It's really easy to make some innocent looking remark that actually contains a contradiction.

[Kip]

Paul,

You wrote:

"We tend to be reasons-responsive. And in virtue of this we rightly say "I could have done better," when we fail.

That's how it is (arguably) in the actual world and its neighborhood. But that's how it most certainly isn't in the fatalistic* world. When people succeed, that had to happen. When they fail, that had to happen. Agents in A* are never moderately reasons-responsive. They are perfectly rational at some moments, and perfectly irrational at others."

We simply have a fundamental disagreement about what is important, and should be important, to the semicompatibilist about reasons-responsive mechanisms.

There are two possible things that are important:

A. The fact that, in a metaphysical sense, the past (e.g. at the Big Bang) "could have gone" differently, and the laws of nature could have been different, and therefore, in a real sense, if an agent had *really* been presented with different reasons to act, the agent would *really* have acted differently.

In other words, it is important, for compatibilists, to have genuine access to genuinely (categorically) branching paths, similar to those libertarians prize. Even if the branching took place *way* before the agent was ever born.

B. The fact that, in a merely conceivable sense, we understand that, we can imagine that, if an agent had been presented with a different reason to act, the agent would have acted differently. No genuine access to categorical branching paths is necessary, not now (as in libertarianism), and not at the Big Bang (as in your rejection of fatalism*).

Quite simply: I feel quite confident that what is important about semicompatibilism is B, and not A. You think it is A, and not B.

We should ask John Fischer. I bet John would say that B, and not A, is important to semicompatibilism (at least he should). If only because, the whole point of semicompatibilism is: we can still be morally responsible even if we *couldn't* have done otherwise. Fischer doesn't believe that the (categorical) ability to do otherwise is important to moral responsibility. That's part of the reason why he constructed his semicompatibilist system.

Not only do I completely lack your intuitions about this, but I find your responses fascinating. Do many other compatibilists think like you? Is that important (essential) for them to adopt compatibilism? (If free will is prima facie difficult to reconcile with determinism, it's even more prima facie difficult to reconcile with fatalism, of any kind).

Now, I've tried hard to explain why B, and not A, is important for semicompatibilism. But I get diminishing returns for my efforts. So I'm not going to persist and address all of your points in detail. We'll have to agree to disagree.

But I'll add one last thought experiment/analogy for you to chew on. It's an old one, from Locke (and Frankfurt):

A key part of your disagreement with me is that you believe it is logically inconsistent for an agent to have the exact same RR mechanism in a fatalistic* world as in a merely deterministic (MD) world. We couldn't even call oxygen "oxygen" any more, on your view.

Consider a mechanism M that, if (conceivably) given an input X, gives response Y. I want to distinguish between two ways in which the mechanism can fail to give response Y:

1. M is defective: even when it is (conceivably) given X, it doesn't give Y. It malfunctions.
2. M is perfectly healthy, but is simply never given *access* to X. Therefore, M never produces Y.

It seems to me that, in a fatalistic* world, all that you can say (by analogy) is 2: the agent's mechanism never gives Y because it never has *access* to X.

But it seems to me that you need 1 in order to nullify the moral responsibility that semicompatibilism considers. Yet fatalistic* worlds do not rule out 1. In a fatalistic* world, we can still imagine that an agent would have done something differently, if the past had been different, even if we know that the past could not have been different.

For example, we can imagine that, if Joe was presented with kids trapped in a burning building, he would rescue you them, because that's just the kind of guy Joe is (courageous, heroic), even if we know that Joe will never, and can never, be presented with a burning building. The fact that Joe is never presented with a burning building doesn't mean Joe isn't courageous and heroic, he's just as courageous and heroic as he always was.

You have emphasized over and over that the mechanism can't be the same. Oxygen is not "oxygen" in fatalistic* worlds. What I am saying is that: change of access doesn't change the mechanism. If mechanism M gives responses X, Y and Z for inputs A, B and C, respectively, but M only ever sees A, and therefore only ever gives X---that does NOT mean the mechanism has changed. M is exactly the same. Only its environment has changed. So you can (arguably) still call oxygen "oxygen" in fatalistic* world (which look exactly like our world). But even if you insist that oxygen is different in one world than the other, I maintain that it is not *relevantly* different (relevant to our discussion of moral responsibility).

Locke had a thought experiment about this (which I'll probably butcher, trying to recall from memory): if a man tries to open a door, but doesn't know it's locked, is he free to open the door? What if he decides to not open it, and never tries to: can he be morally responsible for not deciding?

The agent in a fatalistic* world is exactly like the agent in Locke's example. The agent can have the same RR mechanism (or a not relevantly different mechanism) in both the MD world and the fatalistic* world, just as the agent in Locke's example has the same brain whether the door is locked or not. The agent in the fatalistic* world doesn't have "genuine" access to other possible worlds. But that doesn't matter, because he has access to all of the possible worlds he actually chooses, and his mechanism is otherwise the same (or not relevantly different). The lack of access doesn't undercut his responsibility.

In fact, the whole soul of compatibilism--which your responses here seem to violate (or so it seems to me)--is that genuine access to different worlds is irrelevant to free will.

[Mark Smeltzer]

Kip,

The only logically possible worlds that satisfy W (see my previous post) are worlds in which the initial state of affairs, and all thereafter, are logically necessary. In such a world, time and identity aren't even plausible features.

The reason this is the case is that a W world cannot derive from (mere) metaphysical necessity. It is logically incoherent because that would require that the metaphysical properties that necessitate the initial state of affairs and the laws of physics are also necessary - since we have exhausted metaphysical necessity at this point, those metaphysical properties must be logically necessary for the story to work. If the metaphysical properties of a W world are logically necessary, that logical necessity trumps any further discussion of necessity in W.

A world derived from logical necessity is the kind of world that Zeno was worried about: there is no change, no movement, no objects.

Thus, your thought experiments are hopelessly flawed (and I think Paul would agree, and has stated as much) insofar as you are attempting to compare features inside W worlds to features inside non-W worlds. In the strongest sense, there is no such thing as "oxygen" in a W world, let alone things like people or choices. I don't know how to make this anymore clear.

I think this "debate" has gone on for so long because Paul made the mistake of indulging those flawed thought experiments by attempting to show the internal inconsistencies. However, there is this glaring external inconsistency that trumps that whole discussion.

[Kip]

Mark,

While I understand Paul's constructive criticism, even if I don't agree with it, I don't understand yours. It includes many inferences that I don't follow, such as:

"every event is thus a logical extension of some set of first principles -- there is no cause and effect, no molecules, no physics, no objects, no properties, and no relations; just the first principles and their logical entailments."

"Everything we want to "point out" in a W world derived from logical necessity is just fantastic illusion, and is an utter waste of metal effort. The only thing we could possibly talk about with any degree of sensibility in such a world would be the first principles themselves (and even that is questionable if they are reasonbly complex)."

"The only logically possible worlds that satisfy W (see my previous post) are worlds in which the initial state of affairs, and all thereafter, are logically necessary. In such a world, time and identity aren't even plausible features."

Etc. Each of these is a premise followed by a conclusion, and I don't see how each conclusion follows from its premise. My understanding is that:

1. A fatalistic* world would look exactly the same as our own world.
2. Therefore, we cannot tell, by science or investigation, if we live in a fatalistic* world or not.
3. Therefore, we could be living in a fatalistic* world, for all we know.
4. Yet it makes perfect sense for us to say that oxygen exists, and tends to react with certain other chemicals, etc.

Based on my understanding 1-4 above, I don't follow any of the inferences in your post. Perhaps, however, my inability to follow your inferences is entirely my own fault.

[Mark Smeltzer]

Kip,

(4) only works if we reject Zeno's conclusion and presume that change does occur in our world and that we are not living in a fatalistic* world.

In a world where Zeno's paradox is true, then all perception (whether about choices or molecules) in that world are illusions. I'm not sure what is confusing about this...

It sounds like the underlying problem here is a gross conflation between mere determinism and fatalism*. They are far from the same thing!

[Kip]

Mark,

Do you agree with 1-3?

If so, are you asserting that, for all we know then "all perceptions ... in [our] world are illusions"?

[Paul Torek]

Kip,

I certainly don't want to speak for all compatibilists. I'm sure some would agree that the "soul of compatibilism" is that "genuine access to different worlds is irrelevant to free will." But I think at least some compatibilists, Kadri Vihvelin for example, would agree with me that counterfactuals are important.

Neither your (A) nor your (B) appeals to me (not that I've concluded (A) is false, but its emphasis seems misplaced). Instead I propose

(C) Our reasons-responsiveness involves robust powers and dispositions, the possession of which implies a set of behaviors in nomologically possible scenarios.

Now since another term for "possible scenario" is "possible world", this does imply that we need "access to possible worlds." But only because that is concomitant to possessing robust powers and dispositions in this world. (By "robust" I roughly mean, something that doesn't just state the fact of constant conjunction, but explains it by positing a deep causal structure.) And if we take a Stalnaker-ish rather than Lewis-ish view of possible worlds, this doesn't seem too heavy a burden to bear.

Note that on my view, we're not hanging our hat on the fact that the Big Bang could have been different. That it could have been different might be a consequence of our claim that the world could have been different at the time the agent acted and the agent would then have done what there was reason to do - but not a terribly interesting consequence.

[Mark Smeltzer]

Kip,

Sure, I don't have a knockdown argument against (1) - (3), but it goes further than that:

If P is logically necessary, then P is must be true in all possible worlds. And if the entire course of history H is logically necessary, then all possible worlds have the exact same history H (which seems to indicate that there would be only one possible world).

Does that sound very plausible to you?

There are some arguments I find compelling to think that change exists, but at the end of the day, I haven't yet found an argument that is not logically possible that change does not exist.

That bottom line is that we all presume change in the actual world, and there's no effective way to escape that presumption. If there is change in one possible world, it would seem that fatalism* is false, and therefore not a concern.

[Tom Clark]

Mark,

As to the existence of change, I'm pasting below the abstract of a paper (don't have the paper itself) by Vesselyn Petkov at Concordia University, "The Nature of Spacetime and Free Will." Not definitive of course, but it seems like it bears on this discussion. He's got another paper "Time and the Reality of Worldtubes" available here in which he discusses Zeno's paradox. In it he also somewhat amusingly defends physics against philosophy, to wit: "I also wanted to emphasize again the crucial role of conceptual analyses for (i) any advancement in fundamental physics and (ii) deep understanding of physical theories, because it appears that such analyses are regarded by some physicists as old-fashioned and even belonging to philosophy." :-)

Here's the abstract for the spacetime and free will paper:

The analysis of the consequences of the theory of relativity clearly shows that reality is a four-dimensional world represented by Minkowski spacetime (or other relativistic spacetimes when gravity is taken into account). There could be hardly anything more counter-intuitive than the four-dimensional Minkowski world since it is a 'frozen' world in which the whole history of every physical body is entirely given as the body's worldtube. The view of such a free-will-free world, imposed by science, in which nothing happens constitutes, I think, the greatest intellectual challenge that humankind has ever faced. The reason is that the four-dimensional representation of the theory of relativity is not just a possible interpretation as most physicists appear to assume; it is the only one that does not contradict the experimental evidence. I will specifically show that the kinematical relativistic effects and more importantly the experiments which confirm them are manifestations of the four-dimensionality of the world, as Minkowski advocated, and for this reason they would be impossible if reality were a three-dimensional world (where free will can exist). Hermann Weyl's conjecture, that it is our consciousness that creates the illusion of the passage of time and of free will, and its implications will be discussed in depth.

[Mark Smeltzer]

Tom,

Since your post lacks commentary, I can only guess as your motive (that aside, I thank you for posting the reference to Petkov's paper)... and I have to wonder if it has something to do with the same problem that Kip is currently dealing with: distinguishing between between determinism and fatalism*.

Determinism may, for all I know, entail that all space-time facts in a world exist contemporaneously, as would be the case in a block-universe, but this is not the same thing as fatalism* because there are many possible block worlds. However, there can be one and only one fatalistic* world (and this world would have to be part of it, if it exists).

Are we agreed here?

Kip,

I have a short proof that "works for me" that you might find interesting:

1. If there exists a world in which fatalism* is true, then that world is the only logically possible world (P -> Q).
2. There are at least two logically possible worlds (~Q).
3. Therefore, no world exists in which fatalism* is true (~P).

QED. Right?

(An argument for (1) can be extracted from some of my previous posts.)

[Cihan]

"If there exists a world in which fatalism* is true, then that world is the only logically possible world (P -> Q)."

I'm relieved that somebody else thinks like me. (I mean isn't this sort of obvious?) I was sort of beginning to doubt that I was missing something.

[Cihan]

I SWEAR I am missing something.

"Define fatalism as the conjunction of the following: A. determinism, B. the laws of nature had to happen, and C. the past had to happen. Let's call this fatalism*."

How is this different from ordinary, Oedipus-kind fatalism again? I don't get the distinction between fatalism* and fatalism.

(Perhaps fatalism* is the complex conjugate of the other?)

[Tom Clark]

Mark,

You wrote in response to Kip that "4 only works if we reject Zeno's conclusion and presume that change does occur in our world and that we are not living in a fatalistic* world." On the block universe view, change is simply a matter of how human consciousness carves up reality into successive instants, it isn't an objective feature. The block universe idea is sort of a synchronic (a-chronic?) take on determinism in which we see that even in non-fatalistic* worlds objective change is impossible. All events just *are*. I'm wondering if this poses any problems for compatibilism (it obviously does for libertarianism, seems to me).

You also wrote "That bottom line is that we all presume change in the actual world, and there's no effective way to escape that presumption. If there is change in one possible world, it would seem that fatalism* is false, and therefore not a concern."

It seems as if the block universe view might defeat or escape the presumption that change occurs in the actual world. Since (on this view) change doesn't exist, it can't falsify fatalism*, per your statement above. But let's stipulate, as you do, that there are other possible block universes. One question is whether these possible universes enable change, in some real sense, in the actual block universe that wouldn't be possible without them (that is, if fatalism* were the case). I hope this question, if indeed it's well-formed, makes sense.

[Mark Smeltzer]

Tom,

A possible world containing one or more block universes is compatible with the type of change that I am worried about: namely, realism about facts and realism about causal relations between those facts (e.g. if P2 where true instead of fact P1, then Q2 would be true instead of fact Q1). In other words, I am concerned with preserving the logic of counterfactuals.

This sense of change, more or less equated with the logical recipe for counterfactuals, doesn't rely upon folk intuitions about what counts as "change" and the "passage of time" (which certainly run contrary to the time-semantics of a block universe).

Regarding my usage of "other possible block worlds", I want to be clear that I meant "other possible worlds which contain block universes". That is significant because, in possible world semantics, there's no reason why a possible world cannot contain many universes (as would be the case in any possible world in which brane theory is true). Taking that into account, I'm not sure what to make out of your question... Does the short argument I posed to Kip help clarify my stance at all?

I apologize for the looseness of my language. In this medium, I find myself constantly pit against the inclination to rush my responses even though I tend to end up writing 10x what I could have written if I had been clear from the get ;)

[Paul Torek]

I call a verbal flag on the plays which move from Einsteinian physics to the denial of change, "all space-time facts in a world exist contemporaneously", etc. Instead, the proper response is to admit that time is merely one thoroughly equal dimension among space-time dimensions, and that precisely which direction is temporal depends on one's inertial reference frame. Change remains real, and the increase of entropy is an objective fact. Stephen Hawking explained in his Brief History of Time why the subjective arrow of time necessarily coincides with the entropic arrow of time, so time as we don't quite know it (but can still recognize it) remains safe.

[Kip]

Some responses:

Paul, you wrote:

"I'm sure some would agree that the "soul of compatibilism" is that "genuine access to different worlds is irrelevant to free will." But I think at least some compatibilists, Kadri Vihvelin for example, would agree with me that counterfactuals are important."

This paragraph jumps between two ideas: from "genuine access to different worlds" to "counterfactuals." They are relevantly different: counterfactuals are not (and were not) necessarily possible.

It is a counterfactual that I ate a hamburger today. That does not mean that I had, or ever had, genuine access to a world in which I ate a hamburger today.

This is important because you can still think of counterfactuals in fatalistic* worlds. Return to Joe, our heroic person. Joe lives in a fatalistic* world in which he never encounters a burning building. But we know that, *if* Joe had encountered a burning building, he would have rescued kids. That's just the kind of guy Joe is.

We don't need genuine access to different worlds to make sense of the above. But counterfactuals like the above are sufficient to ground semicompatibilist moral responsibility.

What part of the Joe example don't you agree with?

"Note that on my view, we're not hanging our hat on the fact that the Big Bang could have been different. That it could have been different might be a consequence of our claim that the world could have been different at the time the agent acted and the agent would then have done what there was reason to do - but not a terribly interesting consequence."

If it is a consequence of your view, then your hat is hung. And I, for one, consider the consequence very interesting. Because, as I've explained many times, whether the Big Bang could have gone 50 different ways, or just one way, is completely irrelevant to compatibilist free will.

[Kip]

Mark,

You wrote:

"If P is logically necessary, then P is must be true in all possible worlds. And if the entire course of history H is logically necessary, then all possible worlds have the exact same history H (which seems to indicate that there would be only one possible world).

Does that sound very plausible to you?"

First of all, when I say that the past "had to happen" I am not talking about logical possibility. I am talking about actual possibility.

The scope of logical possibility is larger than the scope of actual possibility.

Logical possibility only forbids things like "round circles" and "married bachelors." Actual possibility might forbid much more.

For example, the following world is logically possible:

The earth revolves around two stars;
Rain is black like oil instead of clear;
People compulsively join hands together and sing Kumbaya every day at 12:34 pm EST;
The stars are colored like Skittles;
People live to the age of 500;
The gravitational constant is 6.78 instead of 6.67 m^3 / kg*s^2.

Etc. That's just off the top of my head.

All of that is logically possible---it violates no law of logic. That doesn't mean, however, that it is, or ever was, actually possible. It doesn't mean that *it ever could have been the case* that, for example, the stars are colored like Skittles, or that the gravitational constant could have been different. For all we know, the gravitational constant we have is the only one we ever could have had.

[One way to think of this: God rewinds the universe and restarts it over and over again, and every time, the gravitational constant shows up the same.]

So, to answer your question, no, it doesn't sound plausible to me that only one history of the world is logically necessary. It doesn't sound plausible to me because the laws of logic, as far as I can tell, cannot dictate what the first state of the world, or the laws of nature, can be. There's nothing about modus ponens or tollens that says "the first state of the universe has to look like X" or "the gravitational constant has to be Y."

The laws of logic are conditional in nature. If X, then not ~X. If the gravitational constant is g, then (in accordance with other factors) planets will move along trajectory Z. But the laws of logic don't say what the first state and laws of nature are in the first place.

But, as I explained above, fatalism* is not defined in terms of logical necessity. Nor would it make much sense for it to be so defined. Fatalism* is defined in terms of actual necessity, which might have a much, much narrower scope.

And yes, I do believe it is quite plausible that only one world is actually possible. (That is, I have no reason to believe that isn't the case.) It's logically possible that the gravitational constant could have been 6.78. But, for all I know, it only could have ever been 6.67.

[Kip]

Cihan,

You ask what the difference between fatalism* and fatalism is.

First, fatalism*. Consider the sequences of time slices of this universe's history. T0, T1, T2... etc.

Each time slice presents a certain state of the world. For example, when I throw the rock through the window, the window breaks.

Fatalism* is merely the claim that the sequence of time slices we have is the only one we ever could have had. The first state had to happen, the laws of nature had to happen, and therefore the later states had to happen: the only sequence did.

But note (and this is crucial): the states in this sequence can still be arranged in accordance with regular laws that give rise to cause and effect. For example, in the sequence of this universe, whenever I throw the rock through the window, the window breaks. There is no part of the sequence where I throw the rock hard enough at the window, and it (magically) doesn't break.

(Of course, a fatalistic* world doesn't have to be organized to give rise to cause and effect, but it can be.)

In contrast to mere fatalism*, consider True Fatalism ("fatalism**"). Fatalism** says less about whether the entire sequence of events had to happen, and more about whether certain events had to happen *regardless of previous states*.

Return to the window. Suppose that when I throw the rock at the window, the window doesn't break. Even if I throw it really hard. Even if nothing otherwise prevents the window from breaking. It just. Doesn't. Break. Magically. Because it was fated, since the beginning of time, to always not break at time T, and it doesn't care what happens at T-1. A hurricane could fall over the window. NASA scientists could detonate a thermonuclear bomb on the window. It doesn't matter. The window doesn't break.

That's true fatalism: no matter what I do, event X always happens. It doesn't care what I do, it laughs in my face and mocks me. Cause and effect break down. Our actions are not efficacious.

I hope that helps explain the difference between fatalism* and fatalism**. Note that fatalistic* worlds can also be fatalistic**, but need not be. And note that, for all we know, our world is fatalistic*: even though it is arranged so that cause and effect arise, it might be that this sequence of events is the only one that ever could have happened.

[Tom Clark]

Mark,

Thanks for the explanation/clarification re change and counterfactuals, and no apologies necessary, especially given my own ambiguities. I guess I was trying to see if there's anything empirical that bears on whether we are in a fatalistic* world or not. Near as I can tell from this thread (not too near!), there isn't. Still, I have the intuition that this world, for all we know, might be ultimately necessitated (couldn't have been different in its past and laws) and that in such a world we could still make sense of counterfactuals as logical possibilities, as we are doing right now. Compatibilist reasons-responsiveness would still exist as well, since agents would still be considering various possibilities in advance of choices. Change, at least as Paul describes it following Hawking, would also still exist (but apparently not the sort of change you say is necessary to preserve the logic of counterfactuals).

Your short argument to Kip seems strictly conceptual, and I'm suspicious of strictly conceptual, armchair arguments that aim to establish metaphysical conclusions. A prejudice in favor of empiricism on my part no doubt.

[Cihan]

Kip,

As I suspected several posts ago, I think you have a third notion of possibility - actual, logical and some third one.

Mark and I (and I think Paul, too but I'm not so sure of that) interpret "had to happen" as (being intended to mean) across all logically possible worlds - that is why I agree with his argument. Here is where I think the disagreement comes up from.

First, let's contrast actual possibility and logical possibility for an event - i.e. your breaking the window - as I interpret these terms and then for natural laws.

Actual possibility: You throw a small pebble at the window and break it. In a *deterministic* world, it is not *actually possible* that the window doesn't break. (I'm using the present tense but these claims should be tenseless really.) In some *indeterministic* world, it is *actually possible* that the window doesn't break. (It's a small pebble after all.)

I'm doing this in the spirit of thinking that most philosophical disputes could really be disputes in language.

Logical possibility: Whatever the laws of nature are, it's always *logically possible* that the window doesn't break.

That is to say, *actual possibility* is indexical on what kind of world you are on - the scope of possibility is constrained by the past, the state of the universe at some time and the laws on that world. What's *actually possible* in some world may not be *actually possible* in some other world.

Logical possibility is NOT indexical across which world you are on/talk about. It's an across-the-worlds measure. Something is either logically possible or not - and the past or laws of nature can have no effect on it.

After all this introduction, now I can put my fingers on where the disagreement lies.

You talk about the *actual possibility* (or *actual necessity*) of the laws of nature and the first event in fatalistic* worlds. (Of course, I naturally didn't understand this and thought you were talking about logical possibility - as did Mark, or so it seems.)

But given that *actual possibility* is defined in terms of laws of nature and the first event, what could the *actual possibility* of the laws of nature/the first event mean?

To me, this could mean three things: Either there are natural meta-laws that make the (first-order) laws actually necessary, or in the terminology of Robert Nozick, laws of nature are self-subsuming, or perhaps the laws of nature are the result of math/logic.

The third option is just logical necessity.

The first option seems to lead to a regress. We could ask about the *actual possibility* of meta-laws and so on.

The second option doesn't seem too attractive either. Suppose there were self-subsuming natural laws - borrowing Nozick's example, laws that would reflect their own truth such as some principle P:"any sentence less than 10 words is true." If P were true, it would be an indication of its own truth, since P itself is less than 10 words. Apparently, this sort of quality is what we are looking in *actually necessary* (i.e. had to happen) laws.

However, if there were such *actually necessary* laws, it seems to be they would hold across all possible worlds and thus we get plain old *logical necessity*.

Thus, I don't see whether it's a meaningful question to ask about the *actual possibility* of the laws of nature/first event. (*Actually possible* with respect to which natural laws?)

[Brian Parks]

Cihan,

Suppose that "Cihan did X." Now, let's ask the question, is it *logically* possible that "not Cihan did X"?

Considered in isolation, there is no contradiction or internal inconsistency in the statement "not Cihan did X." So the answer would seem to be yes, it is logically possible that "not Cihan did X."

The question I am struggling with is, SO WHAT? OK, we grant that "not Cihan did X" does not contradict itself. How is that fact relevant to the free will debate?

Let me put the point differently. Cihan can't make a square circle. He also can't create 10E10000000J of energy from absolutely nothing, or make gravity a repulsive force, or choose X when his brain is in physical state Y, or whatever. Why is the first kind of impossibility--the logical kind--prohibitive of his free will, but not the second kind--the physical kind?

For our purposes, is Cihan any *more* able to create 10E10000000J of energy from absolutely nothing than he is to make a square circle?

[Cihan]

Brian,

I don't know - you'd have to read the compatibilists to see what kind of difference it makes.

One thing I do know is that compatibilists sometimes depend on counterfactual tests such as "If the agent had wanted/decided otherwise..." (or perhaps different ones as in the case of John Fischer' semicompatibilism) to determine whether an agent is responsible and logical possibility is a pre-condition for such counterfactual tests.

You write that:

"Let me put the point differently. Cihan can't make a square circle. He also can't create 10E10000000J of energy from absolutely nothing, or make gravity a repulsive force, or choose X when his brain is in physical state Y, or whatever. Why is the first kind of impossibility--the logical kind--prohibitive of his free will, but not the second kind--the physical kind?"

I think you are somewhat highlighting the intuition behind the consequence argument in disguise. ("We can't change the past or the laws of nature and everything we do is a consequence of the past and natural laws. So we can't be morally responsible.")

All I can say is that you might want to read the compatibilists (and perhaps responses to the consequence argument) and see why they think logical impossibility would be a threat to moral responsibility but not physical impossibility - and I have already given a partial answer, i.e. the counterfactual tests don't work when you have a fatalistic world (or a logically necessary events).

[Mark Smeltzer]

Tom,

You said, "Your short argument to Kip seems strictly conceptual, and I'm suspicious of strictly conceptual, armchair arguments that aim to establish metaphysical conclusions. A prejudice in favor of empiricism on my part no doubt."

Rationality seems to necessitate counterfactual logic. Moreover, empiricism itself must be grounded non-empirically, and in fact presumes real counterfactual possibilities in order to weigh evidence for and against logically possible scenarios.

In fact, if counterfactual statements are necessarily false, everything that can be known can only be known a priori, if at all -- since the first principles are all that exist in such a world and they (obviously) cannot be known empirically.

It would seem that anyone interested in empirical knowledge would want to endorse the premises and conclusion of the little argument I presented.

Kip,

You said, "And yes, I do believe it is quite plausible that only one world is actually possible. (That is, I have no reason to believe that isn't the case.) It's logically possible that the gravitational constant could have been 6.78. But, for all I know, it only could have ever been 6.67."

Cute, but impossible. If this world is logically necessary, then it is not logically possible for the gravitational constant to have been different.

That is the point of the argument I presented: either we reject the language of possibility (or at least choose illusionism regarding possibility), or we reject the possibility of a logically necessitated world. We can't have both.

If we reject the idea of a logically necessitated world, then we are rejecting the possibility of (and therefore the need to discuss) fatalism*.

[Kip]

Cihan,

You raise some fascinating points, and I think you get to the heart of the matter.

Let me recap (for understanding) and then respond:

"Kip, regarding fatalism*, you say the 'past had to happen.' What do you mean by that? There are several things you can mean. But none of them support your view about fatalism*.

First, you could mean 'had to happen' because the meta-laws so say. But this leads to a regress.

Second, you could mean 'had to happen' because the laws and first state are 'self-subsuming'.

Third, you could mean logical necessity."

I'll address these points in turn.

1. Is a regress fatal to that view? Why can't it be meta-laws all the way down? An infinite tower of meta-laws saying "it had to happen that way."

Why are people so opposed to the idea that the universe had to happen the way it happened?

Note that logic doesn't imply that the universe had to happen. So there goes logical necessity.

That would leave actual (metaphysical) necessity. But now you are saying that we can't have actual/metaphysical necessity either. In other words, you are saying that there is *no* sense of necessity in which the universe had to happen. And all of this from the comfort of your armchair!

2. I'm not sure what this means. So I can't comment on it.

3. Logic doesn't imply that the first state or laws of nature had to happen, as I've explained at length.

I'll end with the following remark (a little burden shifting):

I did not raise the idea of the laws of nature and first state having to happen. That was raised in the critiques, by compatibilists, of N+K, who said:

"The data is flawed, because the folk could have interpreted the questions to mean that the past and laws had to happen. But that's fatalism. So they were testing fatalism-compatibilism, and not determinism-compatibilism."

I've tried to say: "Wait a second. Even if the first state (and therefore the past) and the laws had to happen---in the sense of your choice---that wouldn't affect compatibilist powers. Therefore it shouldn't disturb the incompatibilist-friendly results."

And now, many responses later, you are saying: "'Had to happen'? In what sense do you mean 'had to happen'? That sense doesn't make any sense (ha)!"

To which I respond: if you don't like the sense of 'had to happen' I've used, just substitute the sense of 'had to happen' of your choice. Use, for example, the sense that Eddy used when he mentioned the first state and laws having to happen in his article. *That's* the sense I'm using.

If you say that Eddy's sense doesn't make any sense, if you say that there is no sense (can be no sense) in which the first state and laws had to happen, then it seems that Eddy's critique of N+K didn't make much sense either (because he referred to the first state and laws having to happen just as much as I did).

[Cihan]

Kip,

I'll try to be brief.

"I've tried to say: "Wait a second. Even if the first state (and therefore the past) and the laws had to happen---in the sense of your choice---that wouldn't affect compatibilist powers. Therefore it shouldn't disturb the incompatibilist-friendly results."

No - if had to happen is some sort of *logical necessity*, I think it does affect compatibilist powers, as compatibilists define them. That's why Eddy criticized the use of 'had to happen' in the first place. And that's why many so compatibilists spend time distinguishing between fatalism and determinism.

"And now, many responses later, you are saying: "'Had to happen'? In what sense do you mean 'had to happen'? That sense doesn't make any sense (ha)!"

Apologies on my part but I noticed that you and I were using 'had to happen' in a different sense until very late. I'm a little slow and you'll have to bear with me. For me, 'had to happen' always implied 'across all possible worlds' or 'logical necessity'.

"To which I respond: if you don't like the sense of 'had to happen' I've used, just substitute the sense of 'had to happen' of your choice. Use, for example, the sense that Eddy used when he mentioned the first state and laws having to happen in his article. *That's* the sense I'm using."

Gladly. My interpretation of 'had to happen' is 'necessity' as it's usually used in philosophy - i.e. being true across all possible worlds. But then, as I tried to explain to Brian, compatibilists save face. (If everything happens happens *logically necessarily*, then imagining "if you wanted otherwise" is like imagining "if the square was round".)

And I can't find another meaningful notion of 'had to happen' for laws of nature. Even if you talk about meta-laws (forget about the regress right now), then the compatibilist will just say 'if the current state of the universe, the past, the laws AND the meta-laws were slightly different' in defining their conditionals - and compatibilism saves face again. So, even if there was some third way of meaningfully talking about the contingency/necessity of natural laws, I don't think this would effect the compatibilists' analyses.

[Paul Torek]

Kip,

Joe is a heroic person. In situations allowing heroism, Joe is constantly conjoined with heroic action. Nevertheless we cannot say that, *if* Joe had encountered a burning building, he would have rescued kids. Joe lives in a fatalistic* world in which he never encounters a burning building. Therefore there is no counterfactual situation in which Joe exists and encounters a burning building. This follows from the fact that there is no counterfactual situation.

"If Joe had ... he would have ..." commits to a counterfactual situation, a.k.a. a possible world. That's ALL that possible worlds are - consistent stipulations containing a counterfactual. To entertain the possibility is to commit to the possible world.

Note that this point still applies even if there is a "third kind" of possibility and you couch your point in terms of that kind. All that matters is that your kind of possibility is at least as wide as nomological possibility. If fatalism* is true in Joe's world, there is no nomological possibility of Joe's rescuing children from a burning building - which makes nonsense out of any statement beginning "IF Joe were to encounter a burning building, ...".

[Arjan Haring]

Great discussion, but did someone already post a overview of the resources on the psychology of free will?

[Mark Smeltzer]

Kip et al,

Fatalism is necessity without causality.

Suppose we have a set of world slices W0, W1, W2, W3, etc. to Wn.

There exist two possible necessitarian worlds (worlds in which all world states are logically necessary): one is fatalistic and one is not.

The fatalistic necessitarian world is constructed as follows:

1. Necessarily, W0.
2. Necessarily, W1.
3. Necessarily, W2.
4. Necessarily, W3.
5. ...
6. Necessarily, Wn.

The non-fatalistic necessitarian world is constructed as follows:

1. Necessarily, if W0, then W1.
2. Necessarily, if W1, then W2.
3. Necessarily, if W2, then W3.
4. ...
5. Necessarily, if W(n - 1), then W(n).
6. Necessarily, W0.

The difference between these two worlds is explained by how the world states are derived. In the fatalistic world, each world state is "fated" because it has no other explanation: the past states do not explain later states (and thus it is logically incoherent to attempt to back trace events in such a world).

In the non-fatalistic world, each world state has a causal explanation: facts from the prior world states come to bear on the subsequent world states (perhaps due to laws of physics or what not), all the way back to the first world state W0.

Consequently, fatalism is incompatible with any theory of identity over time, any theory of physics, etc. So it shouldn't be surprising that free will and moral responsibility are also logically incompatible with fatalism.

[Kip]

Mark,

You say that: "Consequently, fatalism is incompatible with any theory of identity over time, any theory of physics, etc. So it shouldn't be surprising that free will and moral responsibility are also logically incompatible with fatalism."

But, as far as I can tell, for all we know, we live in a fatalistic world, as you define it. You define the fatalistic world as follows:

1. Necessarily, W0.
2. Necessarily, W1.
3. Necessarily, W2.
4. Necessarily, W3.
5. ...
6. Necessarily, Wn.

For all we know, that describes our world perfectly. In other words, there's no way to know whether every state of our world is necessary in the way you describe.

So it seems to me that compatibilist free will is generally compatible with fatalism, at least the kind that you describe.

[Mark Smeltzer]

Kip,

Only something like Hume's "compatiblism" is compatible with proper fatalism, because it finds its roots in sentimentalism and does not require causality. Kant calls Hume's compatibilism a wretched subterfuge for this very reason: Kant sees that genuine freedom requires causality in order to escape the void for mere sentiment.

For instance, Fischer's account strongly requires causality because Fischer says that the action must stem from the agent's own MRR mechanism. If fatalism is true, then no actions stem from anything because in a fatalistic world, all events are brute facts.

Regardless of whether we "might" live in a fatalistic world, we have no good reasons to think that we do, and since it is a fundamentally untestable question, we will never make any substantive progress on that front. However, we can make a very strong case for causality -- that is a big part of what Kant was trying to do, and all of modern science depends on Kant being right. If Kant was wrong about causality, then science is grasping at straws.

[Kip]

Mark,

First, let's distinguish between No-Causation-Necessitarianism (NCN) and Causation-Necessitarianism (CN), as you define them above.

Your recent comments are very interesting to me for a variety of reasons.

1. You already seem to grant that CN is compatible with compatibilist free will (CFW). This is already interesting, because CN is stronger than determinism - it rules out metaphysical possibilities that are still open in merely deterministic worlds. So, even if NCN is not compatible with CFW, the fact that CN is, is interesting.

2. You define fatalism as NCN. However, I doubt that is the only accepted definition of fatalism. For example, Eddy defines fatalism as follows:

Determinism entails that BOX[(Po & L) -> P]—i.e., necessarily, given the actual past state of affairs (Po) and the actual laws of nature (L), there is only one possible present state of affairs (P). But determinism does not entail (fatalism) that BOXP (or that BOXPo or BOXL)—i.e., that the actual state of affairs (or the actual past or laws) are [is] necessary (could not be otherwise).

And it seems that the above definition is consistent with CN as well as NCN.

3. You seem to say that CFW hinges upon causation, where causation is something stronger than perfect correlation. This is interesting too, because it means that we might not have CFW. One of the advantages of CFW has been, traditionally, that we know we have it. For example, Fischer wants to make moral responsibility safe from physics. If you're right that CFW hinges upon causation-more-than-perfect-correlation, then we can no longer be so sure.

4. I tend to think of NCN and CN as much more similar than you do. In other words, CN seems just as fatalistic to me as CN does. It's not clear to me what CN adds to NCN that makes it so much safer for free will. You define CN in terms of laws, but these laws can be extrapolated from NCN worlds that look identical to the CN worlds. I think it's strange to think that free will hinges on this sort of obscure distinction.

[Kip]

Mark,

BTW I have a 10 page paper on this idea (free will and fatalism) that you may be interested in reading. It's based on conversations with people at the Tallahassee conference. I ran the argument by almost everyone that I met there.

[Mark Smeltzer]

Kip,

All theories of free will (except Hume's) turn on theories of causation: if causation does not exist (the other side of the causation coin is identity-over-time), free will does not exist.

(1) CN is definitely stronger than determinism, but it is still just plain old determinism: rules plus an initial state. The difference is that in a CN world, both the rules and the initial state are necessary.

(2) CN may be stronger than determinism, it may be the case that Eddy call would call CN fatalism, and it may be the case that CN rules out *some* types of possible worlds semantics, but CN is still compatible with other types of possible world semantics: for instance, agents within a CN world would still be able to conceptualize of other possible worlds.

(3) Fischer wants us to be as sure of moral responsibility as we are of science. In a NCN world, science and all forms of identity-over-time are ruled out, so we wouldn't be too worried about losing free will. Another way to think about the problem of NCN is that since it rules out multi-world-slice-objects, it certainly rules out persons, and if there are no persons, there surely cannot be any free will.

(4) CN allows for the existence of identity over time and for the existence of persons. So, CN is certainly quite a bit different from NCN. So the reason for the distinction is certainly right out in the open: for free will, we first need persons. Right? Moreover, in an NCN world, it is logically impossible to derive any normative laws since each world slice exists apart from the rest (that is true by stipulation -- if there are any normative laws, then it cannot be an NCN world).

I would be interested in reading the paper when I have time. Is it online in PDF anywhere?