Free Will and Emergence
Responsible agency and free will are not consistent either with determinism or indeterminism. This short statement is called the “Mind argument.†It has two parts. First, determinism is incompatible with free will : “If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born, and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us.†(Peter van Inwagen An Essay on Free Will (Oxford: Clarendon Press, 1983), p. 56.) (This is dubbed the “Consequence Argument)
The notion of something “being up to me†is that I exercise a certain type of control over my actions. I am responsible for these acts because they are my actions in the sense that I am responsible for causing them. For something to be my act, it has to belong to me the sense that the act arises from my own acts and not from something that just happens to me or happens by happenstance which is not in my control. The problem with determinism is that I don’t have control over the causes that lead to my acts.
However, the same problem with lack of control arises also with indeterminism. Say that our acts are the consequences of chemical states of our brains which are the result of neurons firing in certain patterns. However, lets introduce indeterminism into the picture here and say that whether neurons in our brains fire is the consequence of certain indeterministic quantum fluctuations that influence the chemical processes that determine whether our neurons fire. Could we be responsible if this kind of indeterminism is introduced into the picture? It is not up to us what the chemical states of our brains are and we certainly don’t and cannot control quantum fluctuations. Therefore, the consequences of these things (including our present acts) are not up to us. Of course if there is global indeterminism, matter are only worse. We could then have certain neurons firing and they may or may not result in specific acts. The result remains the same – our acts are not up to us because they are caused by events which are not up to us and that we don’t control.
However, the real problem is microphysical event causation, or the view that our acts are fully explained by events which are not within our control. So if our acts are the result of neural activity in our brains, and neural activity in our brains is the result of the chemical make-up of our neurons, then our acts are fully explained by events over which we have no control. For we do not control the neural activity in our brains; rather, that is controlled by the chemical make-up of our brains which is not up to us. The chemical make-up of our brains is also the causal result of laws that govern chemistry and physics. Further, it doesn’t matter whether these laws are deterministic or indeterministic, they are not up to us because they are not within our control.
The only viable way out of this problem of event causal lack of control, at least by lights, must recognize that it is a person or agent or organism as a whole that acts. Further, the acts of the agent cannot be fully explained by microphysical events over which an agent has no control. Rather, an agent must have a basic power of downward causation such that the agent causes events rather than the agent’s act merely being the result of such events. Agents must act and not merely be acted upon to use Lehi’s language in 2 Ne. 2.
However, such a view of an autonomous agent seems to me to require an emergent agent that is not merely the causal result of deterministic or indeterministic microphysical events. That is, the agent must be emergent in the sense that the complete description of the microphysical causal events (like neurons firing) doesn’t fully explain the agent’s act. Such a power of downward causation requires emergence of an agent that makes a causal break from the underlying causal base that gives rise to the agent’s powers. Otherwise, the “downward cause†is not really downward but just a causal pass-through of the lower level microphysical causes.
Why is this important? It seems to me that if we are going to assert that there are morally accountable free agents, then we must adopt a form of agent causal libertarianism. The agents that could have such powers must be able to be a cause without being fully caused to be what they are and do what they do. An eternal intelligence, it seems to me, just is such an agent. Nothing else fully causes its existence or to be what such an intelligence just is. In fact, I contend that only such an eternally existing entity could be such an agent cause. That means that if we truly believe that we are free and morally responsible, the Mormon world-view (along lines described by Kent) is the only game in town.
Blake, I agree with what you have said here with regard to free will and determinism. I do quibble with your definitions of “indeterminism” and “microphysical event causation”.
You imply that indeterminism entails the rejection of causality (the position that all events have a cause) where the standard sense of the term merely entails a rejection of causal determinism (the position that all events are completely determined by prior events and the laws of nature). Along the same lines and for the same reasons, microphysical event causation does not entail microphysical event determinism.
Comment by Mark D. — February 28, 2009 @ 6:08 pm
Mark: I agree with how you define determinism and indeterminism — but I didn’t define either in my post. I worked with very basic notions of cause and accountability. However, we can agree that indeterminism isn’t (necessarily) the absence of causes altogether. It is, however, the absence of sufficient causes prior to the agent’s act. Other causes may be necessary as a condition of the agent’s free act; but they are not sufficient without the agent’s additional input arising from exercise of the basic agent causal power.
On agent causation, there is not a sufficient cause for what is brought about by the agent until the agent adds his or her basic causal power of decision or choice to whatever other causes may be present. Based on our prior conversations, it seems to me that you are an event causal libertarian. I don’t believe that such a view is able to explain the kind of control necessary for an agent to be morally accountable. The “agent” lacks the kind of control that is necessary on such a view.
Comment by Blake — February 28, 2009 @ 6:54 pm
I’m in complete agreement with you on our need to adopt agent causal libertarianism to defend the idea of morally accountable free agents Blake. In fact I have argued in the past that Mormonism fails without these kinds of assumptions.
I appreciate how you tempered your arguments for emergent agents though. It seems to me that there is still room for the idea of beginningless irreducible agents. I personally like the latter in Mormonism because I think it fits better with the ideas JS taught in 1844. Of course I am willing to be talked out of that preference — I just haven’t been yet.
Comment by Geoff J — February 28, 2009 @ 8:01 pm
Since what I propose is precisely eternal, irreducible agents, and since I believe that is both what JS taught and also what the Book of Abraham teaches, I’m not likely to try to talk you out of it. Emergence is not inconsistent with that idea. The agent can be eternal. It is just that whatever conditions are essential to the emergence of such properties of mind (and certainly there are some conditions) must also be eternal.
Comment by Blake — February 28, 2009 @ 8:10 pm
I like the sound of that Blake. I suppose with this little data there is no reason for me to discount the idea of eternal, irreducible and emergent agent-minds…
Comment by Geoff J — February 28, 2009 @ 10:09 pm
Mark, Kane’s sort of event libertarianism is the view I’m most sympathetic. Although if one is looking at ontology in terms of causality then the point Blake makes is apt. i.e. if one is looking at libertarianism out of our intuitive sense of responsibility then Kane won’t work.
I’m pretty skeptical agent libertarianism is necessary for Mormonism. Primarily because I think our notion of responsibility and God’s need not be identical. (I know the counter-argument of the scriptures being written for us – but pushing that too far leads to a literalist hermeneutic which makes the scriptures become problematic IMO)
Comment by Clark — February 28, 2009 @ 10:35 pm
BTW kudos to Clark and his company for winning the major awards in the pan-European chocolate taste testing competition. His chocolate won the award for the best chocolate in the world! I am somewhat of a chocolate connoisseur (I buy Swiss chocolate as a rule from a very exclusive shop in Lugano Switzerland) and I have purchased a lot of Clark’s chocolate because it is indeed world class. However, I am quite sure that I am event caused to buy it because I find it irresistible and so it is doubtful that I am free when I buy it!
Comment by Blake — February 28, 2009 @ 11:18 pm
Congratulations, Clark.
Comment by Mark D. — February 28, 2009 @ 11:42 pm
Blake: I am having trouble understanding the leap from emergent libertarian free will to Mormonism being the only game in town.
Please let me know what disqualifies these other Games:
Hindusim: A self-existant eternal self or jiva is a principle idea hear.
Mainline Christianity: God could create all the components out of nothing from which the free self emerges.
Atheism: See the same answer to Mainline Christianity but replace God with Time and Chance.
Comment by Matt W. — March 1, 2009 @ 7:39 am
If it’s not clear, my last comment is a sincere question and not a debate.
Comment by Matt W. — March 1, 2009 @ 7:42 am
One more question:
Could it be said that the emergent agent we are now is inclusive of the capabilities we have gained from having a physical body and having the information we have in life. In other words, do I have more free agency with my physical body than I did without it?
Comment by Matt W. — March 1, 2009 @ 8:22 am
You seem to be assuming Blake is arguing for something in the post that he is not Matt.
His post is arguing that agent causal libertarianism is the only good option if we want to claim morally accountable free agents exist.
That specific argument does not assert Mormonism is the “only game in town”. Where in Blake’s post do you see him connecting the two?
Comment by Geoff J — March 1, 2009 @ 9:25 am
I should also note that the emergent thing is a not central to Blake’s post either — it is just a side note that he mentioned on his preferred way of explaining how agent causal libertarianism probably works.
Comment by Geoff J — March 1, 2009 @ 9:27 am
Matt: Look at ch. 11 of vol. #2 of Exploring Mormon Thought for my extended arguments that those who adopt creation ex nihilo are committed to a metaphysic that is inconsistent with free will (and that includes both compatibilist or incompatibilist views of free will). The first argument can be put in (very) shorthand as follows: If a person is created from nothing, then s/he is never the ultimate source or first cause of choices. However, to be free a person must be an ultimate cause that is not sufficiently caused by prior events of at least some part of of his or her decisions. Thus, persons created from nothing cannot be free.
A very truncated presentation of the second argument for the same conclusion is as follows:
(1) If a person is created from nothing, then it is not that person’s nature to exist (by def. from creatio ex nihilo).
(2) What does not exist by nature would cease to exist if God did not sustain that thing in existence (implication from 1)
(3) For each moment that a thing created ex nihilo exists, its essential and accidental properties are sustained (created) in existence (from 1 and 2)
(4) If persons are created ex nihilo, then each person’s essential and accidental properties are created ex nihilo in each moment that such person exists. (implication from from 2 and 3)
(5) Free acts are accidental properties of a person. (By definition)
(6) A person’s acts that are directly created by God cannot be free. (By definition)
(7) Therefore, no persons are free if they are created ex nihilo. (From 3, 5 and 6).
While I believe that agent causation can be compatible with a naturalistic (atheistic) world view, it is fairly clear that agent causation is not compatible with a view that reduces all explanations to prior physical events. I believe that at the very least a type of teleological cause implicit in the basic human power to creatively imagine various possible futures in a way that is not fully determined is necessary for free will.
It also seems that a person must be an ultimate source of acts rather the physical events that constitute a person’s body. My neurons aren’t morally responsible. I am, as person acting as an agent, morally accountable. That means that the prior events cannot be the full and sufficient explanation of human acts. However, I believe that the emergence from complex physical systems of creative agents having free will and moral responsibility is totally natural. The agent’s basic powers just aren’t reducible to complete explanation by reference microphysical events as physics attempts. Thus, I have a broader definition of nature than some philosophers who insist that naturalism is defined by the causal closure of what can be described by physics.
With respect to Hinduism, to the extent there is a view of an eternally existing agent that is isomorphic with an agent cause, then I see nothing that precludes such a view. BTW you might want to look at the arguments of Karl Potter and Roy Perrett who argue that only eternally existing agents can be free. See footnote 40, on p. 244 of my vol. #1.
Comment by Blake — March 1, 2009 @ 12:44 pm
Thanks Blake, that really helps.
Comment by Matt W. — March 1, 2009 @ 3:33 pm
Thanks guys. We were pretty excited to win the awards. We’ve been getting into a lot of the top restaurants in the country as well.
I do think agent-libertarianism is a pretty obvious reading of scripture, (intelligences as Cartesian minds ala B. H. Roberts), it becomes less obvious the closer one reads. (i.e. the lack of clarity in separating intelligence from spirit or using them a synonymous; the more idealist approach of Young and others; etc.)
My own view is much more one of being “agnostic” on the issue. As I said the view I’m most sympathetic to is actually Kane’s with perhaps a bit of Fischer’s semi-compatibilism thrown in to boot. I find some of Fischer’s analysis compelling even though I recognize many here do not. (i.e. is what is most important for responsibility is guidance control rather than alternatives) However I also see some of the obvious flaws in his approach.
To me the biggest problem I have with most agent-libertarian approaches is the same as Mele’s. I’ve become more and more convinced – especially the last year – that distinguishing agent-libertarianism from a certain class of event-libertarianism is a lost cause. That is there is no in-principle way of seeing a difference unless, as Blake suggests, the act of making a choice produces some trace telling us whether it was an agent or not. (A metaphysical position I have a hard time believing)
That said I think it important to work through the possibilities in our theology. That is to ask what we mean by responsibility if God has at least limited foreknowledge, thereby making agent-libertarianism impossible.
Comment by clark — March 1, 2009 @ 10:24 pm
I think it is rather more likely that “fore-knowledge” (in the strict sense of the term) is an oxymoron.
Comment by Mark D. — March 1, 2009 @ 11:30 pm
Clark: The problem I have with Mele’s “no-difference” argument is that it misconceives “luck” and conflates luck with lack of control where he hasn’t shown it. For those not initiated, Mele’s argument is essentially this: take the total state of the universe until a time tn when an agent acts. Let’s say that the agent buys one of Clark’s chocolate bars. Now if libertarianism is true, there is nothing in the prior conditions that sufficiently explains or causes the agent’s act until the agent acts. So he bids us to rerun the entire universe again up to tn when the agent acts. What is the likelihood that the agent will act in the same way? He believes that if the agent’s character and reasons play a role in the choice, then there ought to be some probability that the agent will act in the same way. There has to be some continuity of character and some force to the reasons that result in an action. However, the only difference in the explanation is the agent’s decision at tn. Thus, the agent seems just as likely to not buy the candy bar at tn as to buy it given agent causal descriptions. There doesn’t seem to be input by the agent other than a random act of choice disconnected from the agent’s history, character and reasons. Mele concludes that it is just a matter of luck that explains why the candy bar was purchased rather than passed over.
However, Mele also concludes that there is a lack of control necessary for responsibility on such a view. That assertion puzzles me. The agent clearly controlled whether the candy bar was bought or not because the agent’s choice or decision was a necessary cause that, together with whatever other causes existed, were sufficient for the agent to buy the candy bar. The agent is in complete control over the action. However, there does seem to be a disconnect between agent’s character and reasons and the act on such a description.
What is missing in this description seems to be obvious to me: it is simply far to idealized. It also seems to me that Mele makes the logical error of reifying character and reasons into things in space-time that act as efficient causes. They are not such spatial-temporal things. They are merely descriptions of the way persons in fact act over time or the way persons may rationalize actions.
On a process view of agency, it seems to me that the problem dissipates. The agent’s action consists in synthesizing the manifold of the data (stimulii) of the previous moments into a novel reality thru creativity that results in a choice. In other words, all of the data of the previous moments of the universe that act on the agent (with the agent as passive recipient like DNA information) and also the data that the agent forms into a conscious experience (like choosing to focus on the candy bars and their price) are then synthesized into a new moment of conscresence in the agent so that the agent literally embodies the prior data in a creative manner by adding a synthetic unity of experience to it that is not and cannot be found in the data as such without the agent’s creative organization of such data. These data are then synthesized by the agent in a new moment of creative act that results in a new conscrescence — a choice that embodies the agent’s creative organization of the data into a new whole. This choice is the creative input of the agent that results in a new reality that has never before existed and that cannot be explained unless one takes all of the data of the world to tn and adds the agent’s creative input to synthesize the prior data into a new consrescence of choice.
Thus, this new choice will reflect all of the prior data as a necessary cause into the agent’s decision with the agent’s decision as the last necessary cause that together are sufficient for the agent’s choice. The choice reflects the input of all of the agent’s past choices (what we call character by refying it) and also the agent’s activity of reasoning as a process (rather than mistakingly reifying it into a thing that acts as an efficient cause). The act of reasoning just is the agent’s creative organization of the data of the prior moments into a new whole having synthetic unity. This new synthetic unity is a new moment that is not completely explained unless we include the agent’s creativity to organize the data of the prior moments up to tn. The unity is that process of activity that is mistakenly reified into a thing we call “a choice.” In fact, it is a process and not a thing.
It follows that the agent is not fully determined, but acts as an agent whose causal input is the necessary condition to cause the choice. The agent has the necessary control because it is the agent’s active power to unify the data into an act of consciousness and thought that results in the agent’s choice. Nor is it merely a matter of luck because the act is in the agent’s control. Mele equates luck with any event that isn’t fully determined by the prior data. But the event, the agent’s act of choosing, is fully explained by the prior data together with the agent’s active power to organize data into a unified consciousness.
Further, it would be very difficult to argue that we don’t have the basic power to order the data of our experience into a unified whole of consciousness. Indeed, our consciousness of the world just seems to be such a unified experience that has an order and unity to it that cannot be found in the cacophony of the data that we take into our experience in making it into our momentary consciousness. All of this is very Kantian — but it is also vintage Hartshorne. It seems to me to solve the problem. Of course, for those who insist on thinking in terms of reified states as causes, this explanation will either not work or just seem strange. I claim that it is a logical error to reify character and reasons as Mele has done.
Comment by Blake — March 2, 2009 @ 7:50 am
I don’t think that is an accurate characterization of Mele’s position — at least not in his last book.
Rather he says given the state of the universe at Sn and the state of the universe after a choice is made Sn+1 that there is no way by looking at the universe to distinguish between that change being due to libertarian free will or randomness. There really is no need to generate probabilities for the argument to function.
From what I can see the argument doesn’t appeal to causality in the least. That is to me a topic you kept injecting into the discussion. For people like me who don’t feel that causality is fundamental this approach is quite nice.
If, as it appears you do, one views causality as fundamental then of course trying to translate Mele’s point into the language of causality will be confusing since it doesn’t depend upon causes at all.
Comment by Clark — March 2, 2009 @ 9:51 am
(To add, I probably should have cracked open Mele’s book before saying the above. But I honestly don’t recall causality playing any role in the argument. In any case even if Mele mentions causality then the argument without causality functions fine)
Comment by Clark — March 2, 2009 @ 9:53 am
Blake,
Of course, for those who insist on thinking in terms of reified states as causes, this explanation will either not work or just seem strange.
I suspect that if the answer truly rests with the explanation in #18 (I hope it does) that the reason it is hard to get across is that the fundamental differences in process ontology are not understood by the people you are trying to convince. Speaking for myself, as much as I am drawn toward process thought, I can’t say I really have my mind wrapped around it. I tried reading Process and Reality by Whitehead but couldn’t get through it (perhaps embarrassingly). The fact that he introduces a new vocabulary speaks to how thoroughly he rejects the ontology we are all used to, but when you start throwing around conscresences I would guess 1 of 1,000 here know what you are talking about and people are not likely to sign on to a view that still seems like it is talking mumbo jumbo.
Soooo, all this was leading to a suggestion. Maybe you could do a series of posts introducing us to the key concepts in process thought with the goal of making it consumable by a lay person like me. Alternatively, if you know of a good introduction aimed at an audience like me I would bite on that too.
Comment by Jacob J — March 2, 2009 @ 10:27 am
Clark: It is clear that if we rerun the state of the universe to a time tn+1, then there is a difference between the two: the agent’s causal input or act. so there is an easy way to distinguish the two universes. Further, as I describe the answer in process thought, the agent’s act is clearly not random because it arises out of the agent’s own prior states that literally embody the processes of reasoning and choosing in a synthesis of data of the agent prior to the act. That seems to me to more than satisfactorily answer Mele’s challenge.
So if Mele is saying that any act that reflects the past and reasons of the agent is “random,” then I suspect the notion of randomness at issue is vacuous. It amounts to just saying that the agent cause wasn’t fully explained by the prior states of the universe. But so what? That is just what the proponent of agent causation claims. How is that random if the act arises out of the agent’s basic powers in a way that reflects the agent’s reasons and past decisions (or character) as I explained above?
Comment by Blake — March 2, 2009 @ 1:22 pm
Jacob J. Jolly good suggestion! You can find an intro to process thought in the 2nd chapter of vol. 1 of Exploring Mormon Thought where I introduce Hartshorne’s thought. Does that do it (or am I presumptuous and egocentric to suppose you actually have it)?
Comment by Blake — March 2, 2009 @ 1:59 pm
Jacob J – “when you start throwing around conscresences I would guess 1 of 1,000 here know what you are talking about . . .”
I suspect that you are way too optimistic about that.
Comment by Blake — March 2, 2009 @ 2:16 pm
Blake, I definitely have your books and I will re-read chapter two this week to see if that gives me what I am looking for. When I first read vol. 1 it sucked me in and I read it in two days, so I’m sure I missed stuff.
Comment by Jacob J — March 2, 2009 @ 2:18 pm
Clark: “If, as it appears you do, one views causality as fundamental then of course trying to translate Mele’s point into the language of causality will be confusing since it doesn’t depend upon causes at all.”
How could one discuss agent causation without causation as a basic notion?
Comment by Blake — March 2, 2009 @ 2:52 pm
Well Blake that’s a rather telling question since it presupposes that the only way to talk about free will and responsibility is as agent causation.
But that presupposes what we are attempting to distinguish, which is a no-no. (As I pointed out in the thread we last hashed this out) All we can measure is the state at tn and the state at tn+1. The agent’s “causal input” is either already in tn or else it is something other than the state of the universe. If it is in the state of the universe then the “causal input” can’t be what the Libertarian proposes. If it isn’t then you concede the argument. (i.e. there’s no way to distinguish between the two)
Also, talking about rewinding the universe is a cop out since we can’t do that. Further even if we could it still wouldn’t tell us anything unless the randomness happened to be a nice gaussian distribution while libertarian choice isn’t. Yet it’s unclear what shape the curve would be for libertarian agent causation (or if it even makes sense).
Chance != Gaussian distribution.
Comment by clark — March 2, 2009 @ 8:21 pm
Except it presupposes a rather controversial ontology as the “solution” to the problem. Further it ends up being circular since it reduces to, “choice isn’t random because my metaphysics says it isn’t.” Worse yet it provides no way of discerning this.
It’s only vacuous because you have an ontology of what constitutes a reason and reasoning. However it’s precisely this that the question demonstrates as problematic. That is your ontology basically denies the reality of randomness of this sort. But it avoids the central question of why on earth we should believe Hartshorne’s metaphysics? (Certainly it is hardly convincing to most people)
I think the more interesting question would be whether the question could be responded to without merely gesturing to Hartshorne. That is can one answer the question on its own terms?
Comment by clark — March 2, 2009 @ 8:29 pm
Clark: Your response exposes a pretty important misunderstanding. If there is a difference at tn in the agent’s choice, on the process view that is because there is a difference in the way the basic data of the prior moments is synthesized by the agent in a process that starts in the past before the choice and results in a new moment of concrescence. It is a process and one cannot freeze the world at a moment to explain any act given process thought. So I agree that we cannot rewind the universe but for very different reasons. To rewind to tn cannot be done because we must rewind to earlier times when the free decision is in process. We cannot have the same history up to tn because the same result requires the same process and the possibility of a different result requires a different process of conscresence. So Mele’s challenge doesn’t apply well to process views of free action.
Further, as I said, all that is really being said by both you and Mele is that there is no sufficient explanation in the facts without the agent’s free choice. So what? Your dichotomy that an act is “either explained fully by the state of the universe in the prior moment” or “causal input cannot be distinguished” is just a mistake about what the issue is. There is a way to distinguish between the two possible worlds that are identical up to tn (in relevant time frames) — the agent’s exercise of a basic power. That is what is at issue, not whether there is a sufficient explanation for the way the agent exercises the power in the prior moments. (see http://www.umass.edu/philosophy/events/papers/Luck%20Happens%20Griffith.doc
see also here: http://www.indiana.edu/~scotus/files/MatterLuck.pdf
See also here: http://www.arts.cornell.edu/phil/homepages/pereboom/MelecommentPHILEXPLORATIONS.pdf
Look, if the agent’s decision isn’t fully explained except for “that is the way the agent synthesized the data of the prior moment at tn,” then there is still agential control and responsibility and free will that is precisely what we want. The agent’s act embodies the agent’s reasoning and entire history and acts to control whether the event occurs on my view. If that isn’t more than sufficient control for the kind of free will that we want, then it is difficult to see what could possibly constitute such control.
Clark: But that presupposes what we are attempting to distinguish, which is a no-no.
I believe this comment is mistaken. What we are distinguishing is the two different possible worlds, one where the agent choose to buy your chocolate freely and the other where the agent doesn’t. So what we are distinguishing is why there is a different possible world actualized by the choice, not why the agent makes the particular decision. As the libertarian holds, there is no complete explanation until the agent acts at tn and that decision is precisely the difference. It is that fact that provides the control and freedom necessary for a free act. That could mean that the past provides necessary conditions for several different possible worlds with the act of the agent determining which becomes actual.
I asked: How could one discuss agent causation without causation as a basic notion?
Clark responded: Well Blake that’s a rather telling question since it presupposes that the only way to talk about free will and responsibility is as agent causation.
You must not have read carefully — I asked how we could discuss agent causation without discussing notions of causation, not how we discuss free will in general.
Comment by Blake — March 2, 2009 @ 8:55 pm
Extremely difficult no doubt. But no one who thinks teleology is real looks at it that way. Instead, they look at the difference after some considerable time has elapsed.
Samples from an LAC process might well be Gaussian distributed without indicating much at all. A better indicator is non-trivial auto-correlation. One might well wonder how a micro LAC entity can generate a non-centered output for any significant period without violating energy conservation. A process that shifts outputs in relative time the way an electron “borrows” energy to tunnel through an energy barrier doesn’t have that problem.
Comment by Mark D. — March 2, 2009 @ 9:20 pm
Clark: “I think the more interesting question would be whether the question could be responded to without merely gesturing to Hartshorne. That is can one answer the question on its own terms?”
Well, I haven’t merely gestured to Hartshorne, now have I? I have given a Harthsornian explanation and description of a free act that is reasons responsive and controlled by the agent. That is quite enough to show that agent causation is quite possible.
I suggest that it both has been and can be further explained without assumes a process view. I believe that both Tim O’Connor’s and Randolph Clarke’s views of agent causation answer the question readily. Let me do it here without presupposing a process ontology (though just why I should have to adopt a metaphysic that you think is more to your liking is beyond me):
A person walks into a 7-11 and sees a a shelf full of Mars bars. Such a sensory experience causes (in a deterministic sense) the person to become hungry and desire to eat a Mars bar. She desires to eat a Mars bar. Seeing the status of the store, it “occurs to the person” (again in a deterministic sense) that she could either pay for the Mars bar or steal it. She then engages in a process of reasoning about whether to steal the Mars bar. That means that she uses a basic power to imagine possible futures such as stealing and getting caught, paying and not having much money left over. Such reasoning involves a basic power of imagination which is created by her creative engagement with the existing situation. She realizes what the consequences of her act could be. It is embodied in her memory and thought process. She then weighs her values in the continuing process of reasoning. She has a set of values but reevaluates them anew in this moment. She values her honesty and then realizes that given her values, paying for the Mars bar is in alignment with her moral values and stealing isn’t. This realization is embodied in her memory. She then creates the intention to procure the Mars bar by paying for it. So she proceeds to the check out and pays for it.
There are agent causes at several points. In the process of imagining possible futures she engages a basic creative agent power of imagination to assess the possibilities. That is not fully explained by the past data but is created in the moment in a creative interaction with the situation in which she finds herself. She also weighs her values. The weight she will give to her various options isn’t determined, but it is influenced by who she is and the past choices she has made. She reevaluates and this valuation is essentially an agent caused power of moral valuation. She also agent causes her intention to pay for the Mars bar so she acts in an ends directed manner to fulfill the reasons and values she has embodied in the process of deliberation. This ends-directed action is a free decision that results from the reasons she has considered and the means by which she will carry out choice.
If she had decided to steal the Mars bar, she would have had to have weighted the various possible options that she generated in her imagination differently. She would have created a different ends-directed intention to pick up the Mars bar and walk out of the store stealthily without paying for it. Why did she weight the respective values as she did? Because in that moment her process of reasoning and valuing led to that decision and her freely chosen intention resulted in the free action.
And there you have it. A reasons-and-value responsive decision that is free and within the control of the agent.
Comment by Blake — March 2, 2009 @ 9:22 pm
The way I look at it: The causal evolution of a non-LAC system can be modeled as a Markov process on a state vector X. In other words there is no memory in the system that is not encoded in the state vector (The positions and momenta of the particles or whatever is all there is).
Given any initial state Xo, we can model future states Xn as a probability distribution in phase space. After an adequate time period has elapsed, any viable LAC system must regularly and repeatedly end up in states for which the non-LAC process state probability is minimal to non-existent.
Establishing this occurs in actual life requires a plausible non-process related metric of apparent teleology. LAC is then established by the probabilistic, state averaged disjunction of that metric with the process state probability entailed by a neutral Markov process.
Comment by Mark D. — March 2, 2009 @ 10:02 pm
Mark: How does such a probability function give the agent control over her act? How could reasons figure into an act explained by such a probability function?
Comment by Blake — March 2, 2009 @ 10:04 pm
Blake: My point is that prototypical non-LAC systems (including the boundary case of a deterministic system) are governed by such probability functions. An LAC system (containing an LAC agent for example) is not.
There is no practical way of establishing a reliable probability distribution for what LAC agents do. The whole point is that such agents are not statistically neutral and prototypical non-LAC systems are.
Since we have no way of modeling what makes an LAC process special, we have to construct a plausible metric of what makes states (such as those containing the works of Shakespeare) special. We know such states exist and are overwhelmingly preponderant in real life.
It is not enough to say that such states are overwhelmingly small subset of state space. The object is to establish that a statistically neutral quasi deterministic process cannot make the transition from the overwhelming number of uninteresting, non-peculiar initial states to the kind of interesting states we now occupy with a non-infinitesmal probability. The corollary is something that is not statistically neutral (such as agent control) is required to make any such transition with greater than infinitesmal probability.
Comment by Mark D. — March 2, 2009 @ 11:14 pm
Mark: Doesn’t the LAC distribution relate only to populations of persons and not to acts of individuals? It seems to me that we cannot model what a free act is or whether a person is likely to do X or refrain from doing X at t in circumstances C based on a model that applies only to populations or numerous trials. Is that your point?
Clark: Since we couldn’t develop a model of libertarian acts, whether they would fall in a Gaussian distribution is something that no one could know — not even God. We could watch Martha in a 7-11 store and never be able to predict when she will buy a Mars bar rather than steal it. If every time we see her in a 7-11 she takes a Mars bar and pays for it, then what we know is merely that every time we see her she pays for it. It doesn’t follow that she cannot act out of character or that she just might surprise us some time. However, we wouldn’t be so surprised if Winona Ryder walked in and pilfered a Mars bar just because she has a known tendency. We still couldn’t predict what she will in fact do — and I know of no model that would allow us to predict any probability.
Further, isn’t a Gaussian distribution applicable only to populations or to numerous trials? That means that such statistical models commit the fallacy of distribution when applied to the probability of any given act occurring or not. The concept simply doesn’t apply. Was that your point?
Comment by Blake — March 3, 2009 @ 7:15 am
Blake, lots of comments and I don’t have time for a long engagement with them. A few brief responses.
1. While you were talking about agent causation the topic of the post was free will. My point is just that one perhaps shouldn’t only look at the question in terms of agent causation. (Which I think tends to bias answers towards thinking of self-contained agents) Sometimes the way we frame the question determines the answer.
2. I agree upon your point about distributions and agents. (The distribution of an agent choice) But that makes my point. It is in principle undetectable. There’s nothing in the produced phenomena to tell us whether it was produced by an agent. That’s because agent-choice as you characterize it is always a radically individual moment and not part of a pattern. It is by nature unknowable.
3. My point about Mele’s chance argument is whether one can answer it without just appealing to a metaphysics that answers it. That is can we talk about it in principle detectable? I don’t see O’Connor or Clarke really answering Mele’s charge. If you do, could you send me a reference to where they address it? I don’t have all of Clarke’s writings although I think I have most of O’Conner’s.
So I’m not looking for a metaphysics “on my terms.” Rather I’m just asking if Mele’s point is addressable without simply picking an overarching metaphysical position. If not, then the question then becomes why buy into one metaphysics above the other. And I can honestly say I have nothing remotely close to an overarching metaphysics of existence. In fact I’m pretty skeptical of such things such that I think appealing to them is inherently a negative.
4. I think most of your writings on this are dependent on Hartshorne. Certainly one can appeal to O’Connor, for instance. But often you invoke things I don’t think O’Connor would be comfortable with. While I think O’Connor wrong, I find his brand of emergence a bit easier to take.
5. How could one discuss agent causation without causation as basic. Causation could be an apparent feature. To make the physics appeal one could look at the evolution of a Hamiltonian equation where you’d have patterns that look like they are interacting with causes. However causes might not be a basic ontological feature but just a feature of how we describe correlations. Of course Hume was famous for pointing this out. We don’t have to be Humean about causation. But I’m skeptical that they are a fundamental feature of the universe. Certainly I don’t think we should conduct our analysis as dependent on their so being.
Anyway, my rejoinder about causation was just surprise that you can’t imagine talking about agent causation without causation as a basic notion. There’s lots of things we talk about without considering the subject to be basic. Causation in physics being the obvious example. That you don’t see how one can talk about agent causation without causation as basic to me says something about how you conceive of free will.
Comment by Clark — March 3, 2009 @ 11:19 am
Blake: There are numerous fundamental problems with deriving a probability distribution for the actions of LAC individuals. A big one is we have no
reason to claim that if the non-LAC factors are held constant and the individuality is changed that the distribution will be the same.
The probability distribution of the control case (no LAC) can be analyzed in strictly frequentist, non-Bayesian terms. Any distribution hypothesized for an LAC agent cannot be. If we constructed one, in principle it would make sense in subjectivist, Bayesian, “in desperation” terms.
My point is that given a sufficiently large time interval has elapsed (such that the probability distributions don’t look trivially similar to the non-LAC factors both models have in common), it is not important that there is no closed form or well defined distribution for the LAC case.
What is important, is that if LAC exists, net state transitions occur for which the non-LAC probability distribution is infinitesimally small. No differential analytic model of how or why systems containing LAC agents end up in such states is necessary, just a posteriori evidence that real world systems do.
I claim that systems exhibiting LAC do things that the equivalent LAC deprived system cannot do except by a series of sheer accidents. The only way to demonstrate that any actual net transition from a non-peculiar state to a peculiar state was almost certainly not the result of the control non-LAC process, is to show two things: that the non-LAC process shows a infinitesimal probability of making that transition successfully (due to series of coincidences required) and that the end state (works of Shakespeare for example) is in fact actually objectively peculiar. No LAC model probability distribution is required.
Comment by Mark D. — March 3, 2009 @ 9:26 pm
Blake et al: A more general argument goes like this. Suppose that there is an objective peculiarity metric alpha(X) over all possible state vectors X. The control case is characterized by transition probability density function tau(Xi,Xf,t) where Xi is the initial state, Xf is the final state and t is the elapsed time from Xi. tau(Xi,Xf,0) is the identity operator – nothing changes in zero time. tau(Xi,Xf,t) is the state operator (“Markov matrix”) characteristic of an elapsed time t. “tau” in general form encodes nothing other than the laws of physics that govern the control process strictly, and the LAC process approximately.
If you start out with an initial state probability density psi(Xi), (a delta function for a known initial state), psi(Xf) is nothing more than the integral of psi(Xi) with tau(Xi,Xf,t) over Xi. It is like multiplying a matrix times a vector. psi(Xf) for obvious reasons is generally “blurrier” than any well defined psi(Xi) if there is any non-deterministic character to the control process at all.
If the control process is deterministic (and energy conserving), psi(Xi) will eventually recur after the Poincare recurrence time of the system has elapsed for any arbitrary initial state Xi. If the control process has stochastic character, the psi(X,t) of the control process will blur out as the stochastics of the system gradually destroy our ability to differentially predict that any given state is more likely than any other.
If the system moves from a less peculiar to a more peculiar state, the pecularity metric alpha goes up. “tau” tells us where the control process will probably go. “alpha” tells us where it would be interesting if it did go.
We can calculate the expected increase in alpha from any initial state Xi and any elapsed time t as the integral of [alpha(X) – alpha(Xi)] * tau(Xi,X,t) over all possible final states X. The time derivative of that integral is the expected rate of increase in alpha as a function of Xi.
Our hypothesis is that the state average of the expected increase in alpha is infinitesmal, zero, or negative over all possible states of the control process. In other words, we expect the control process to flow to regions of lower alpha with overwhelming probability for all initial states.
If that is the case then the existence of any real world state with high alpha cannot be explained by the control process except with resort to past states of equal or greater alpha, which is the opposite of what we observe empirically.
I hesitate to mention that this alpha is *not* the inverse of the thermodynamic entropy. There are plenty of essentially zero entropy systems (pure crystals at zero degrees Kelvin) and very high entropy systems (heated gases) that do not exhibit any peculiarity at all, no more so than a typical brick or stone. A viable alpha must be high for an integrated circuit mask and low for anything resembling random noise, including the limit of empty space, regardless of temperature.
Comment by Mark D. — March 3, 2009 @ 10:34 pm
Mark: My shorthand for your argument (which appears to be sound) is that it is a category mistake to apply probability judgments to individual actions. That is sufficient to refute Clark’s reformulation of Mele’s argument based on Gaussian distribution = randomness. First,the distribution doesn’t apply to the kinds of judgments we are discussing. Second, even if it did, it is irrelevant because it doesn’t show that the agent lacks the relevant control to be both free and accountable.
Comment by Blake — March 4, 2009 @ 7:28 am
Clark: “I agree upon your point about distributions and agents. (The distribution of an agent choice) But that makes my point. It is in principle undetectable.”
This seems to me to be untrue. If one takes a power or transfer view of causation, then the action doesn’t occur without the exercise of the basic agent’s causal power. Now if you mean that science cannot verify it based on some concrete evidence — that depends on the relation between the exercise of the agent causal power and brain activity. I can think of several scenarios where such an exercise is detectable as a break with the underlying causal base and then reflected in brain activity. Indeed, that is what I take Walter Freeman and other neuropsychologists to be arguing about chaotic brain functions that are self organizing that result from top-down feed-back systems. So it isn’t undetectable in principle as you argue. It may be beyond our present science — but so what, most of human experience is beyond science.
What your argument from distributions shows is merely that you have applied a concept that is a category mistake when discussing the modality of individual agent actions.
Comment by Blake — March 4, 2009 @ 7:37 am
Could you clarify what you mean by “a power or transfer view of causation”? (It sounds like sneaking in a systematic metaphysics)
Note that I’m not talking about scientific verification but rather just in principle verifiable via inquiry.
When you say it could be “detectable as a break with the underlying causal base” that sounds like just a break with determinism, but not randomness.
I’m not making an argument from distribution, btw. To say that comparing agent action and random events is a category error misses the point which is how to distinguish categories. If they are in principle undetectable then that’s a big problem, whether according to one view it’s a category error or not.
Comment by Clark — March 4, 2009 @ 8:02 am
Put an other way, to argue for a category error when the categories are indistinguishable is itself an error.
Comment by Clark — March 4, 2009 @ 8:03 am
Blake: What you say in (#39) is indeed a position that I hold, but it is not the argument or a summary of the argument I am trying to make. The argument I am trying to make is that stochastically neutral processes do not generate the works of Shakespeare, nor explain the stepwise refinement of any subsystem (i.e. Shakespeare and his ancestors) that could, whether by natural selection or *any* other process considered in a stochastically neutral manner.
People have made this argument on a casual basis for centuries. The idea of stochastically neutral subsystem stepwise refinement is the contemporary answer given to such arguments. My response is that stepwise refinement is fine, but the idea that any such process is a stochastically neutral realization of the laws of physics is not.
The same idea applies to a snapshot of the state of Shakespeare just before he sits down to write. It is certainly possible that the non-LAC model can explain the the first few sentences that come to his mind. It cannot explain our obvious suspicion that the peculiarity of the end state (Shakespeare + play) is manifestly greater than the peculiarity of the initial state (Shakespeare + blank sheets of paper).
Determinism (under standard constraints) requires that any objective measure of pecularity is essentially a constant (due to Poincare recurrence). Thus the deterministic perspective, is, no, the pecularity of the end state (Shakespeare + play) is no greater than the peculiarity of the initial state, but rather the process has merely “distributed” (or made manifest) pre-existent peculiarity in Shakespeare’s brain onto the sheet of paper.
This is why determinism essentially entails Platonism. It implies that the universe has always been in a highly peculiar state, and all processes just unfold (reveal) pre-existent pecularity. The more subtle stochastic-deterministic view supposes that the introduction of blind, statistically neutral indeterminism will account for any empirically observed net increase in peculiarity (given an adequate time span).
My argument is an attempt to demonstrate that the introduction of statistically neutral randomness cannot improve the situation, but rather will make it worse.
Comment by Mark D. — March 4, 2009 @ 8:40 am
Just to clarify the point I’m making for everyone else.
It’s a category mistake to treat two independent categories as if they were comparable. Thus I can’t compare sound and color. To make the comparison or to treat say a color property as if it were a sound property is thus to make a category mistake. This is what Blake’s accusing me of doing by (to him) comparing free choice with random change.
My response is that the question isn’t about the categories but how to distinguish the categories. So to continue with my example I can’t compare sound and color but I can distinguish them. So if you were to ask me how I can tell sound from color I could produce a way of verifying them. (And this need not be a scientific verification – indeed we typically don’t use science to make such distinctions)
So what I’m asking Blake is how a divine being like God could distinguish choice from random change given that being’s knowledge of the constituents of the universe. Blake’s recent move is to say that this can’t be done logically because of category mistakes. But that doesn’t work. An earlier move he made is to say that God knows whether an agent is rational and by definition that entails knowing he’s made a choice. However my response was even if an agent is rational the changes they make in the universe aren’t always “free choices” but can be due to luck. So how do we distinguish the two?
Last year at my blog he answered this by saying that free choices always change the universe in a way that random change doesn’t. But this seems difficult to accept without simultaneously accepting a very controversial systematic metaphysics. Note I’m not myself adopting a metaphysical scheme – just being very skeptical that free choices must entail a change different from what random change can produce. Without that difference though then it seems to me that in principle there is no way to distinguish free will from chance.
Comment by Clark — March 4, 2009 @ 9:51 am
Clark: A power view of causation is elucidated here: http://reasoninglab.psych.ucla.edu/CHENG%20pdfs/Cheng%5B1%5D.PR.1997.pdf
A transference theory of causation is explained here: http://www.springerlink.com/content/p3760511v1h78x52/
Of course I prefer the theory of causation I elucidated in chapter 3 of my 1st volume is is a broadly Aristotelian agent causal theory.
However, since you are the one who is making the no-difference argument, you are actually the one who must elucidate a theory of causation on which agent causation isn’t a difference in the world. So you’re the one who needs some view or theory to make sense of your no-difference argument.
Here is how I see our conversation. You first complained that I erroneously dragged in probability as part of the statement of Mele’s argument. You then responded to my argument by using a probability argument that if we have a Gaussian distribution of agent-caused acts then it is random. So you then adopted the view that probability assessments are essential to the argument. You then acknowledged that making probability judgments about individual acts is a mistake — but you still insist it somehow shows that agent causation is nothing but random.
I respond that you equate randomness with any act that is not sufficiently caused or explained by prior causes. But that is just a bad argument because the agent causal view is precisely that an agent’s acts cannot be sufficiently explained or caused by prior events alone, one must add the agent cause. The basic agent power to cause events is the basis of the difference and the necessary additional cause to any sufficient explanation of a free act.
Nevertheless, you continue to insist that there must be a sufficient explanation in prior events or there is just no explanation unless it is just random. I respond that what is at issue is not such randomness (since randomness is a probability judgment that we both agree doesn’t apply), but whether the agent has adequate reasons-responsive control. I show how the agent has such control. You ignore it.
That’s how I see our conversation at present.
Comment by Blake — March 4, 2009 @ 9:52 am
Sorry – one additional point. There is an obvious assumption I’m making that not everyone will necessarily agree with. That is I’m assuming God’s knowledge is mediated the way ours is. So that there is no direct perception of universal qualities. If God knows whether a choice is libertarian free independent of knowledge of the stuff in the universe simply because he can know that sort of thing then of course my argument fails. Blake thus far hasn’t made that move and I honestly am not sure such a move makes much sense if one rejects creation ex nihilo.
Comment by Clark — March 4, 2009 @ 9:55 am
Clark: re: 44. So what I’m asking Blake is how a divine being like God could distinguish choice from random change given that being’s knowledge of the constituents of the universe.
God knows that the agent exercises the power in question. God knows it makes the difference because the prior events are not sufficient to explain what happens without adding the agent’s exercise of power. Therefore, God knows that the agent exercised the power and how it was exercised and how it made a difference — because things would have happened differently without it. That seems more than adequate to answer your query to me.
Comment by Blake — March 4, 2009 @ 9:58 am
Clark: I’m just puzzled. Why do you say that God doesn’t know that th agent exercises the power in question? It is an act of the agent that God knows in knowing all present truth-makers. It is a fact about the world because it is causal and makes a difference in which world is actual. By having present knowledge of all truth-makers, or of all propositions, or of all causes, or of all things that occur (all disjunctively), God quite clearly would know the agent’s exercise of the agent causal power.
Further, just how such exercise of power would follow from creation ex nihilo is beyond me since it clearly cannot follow as a matter of causal necessity. But if you assert that God can know what occurs in the world only if it is a necessary deterministic result of God’s initial first cause then I believe you are just mistaken. What would underwrite such a view?
Comment by Blake — March 4, 2009 @ 10:14 am
Clark: Let me add that God also knows that the agent exercises the power in question because he knows that there is a change in the accidental, intrinsic properties of the agent in the moment of decision. The agent changes from not having decided and intended to A to have determined and intended to do A. So God knows it directly. In layman’s terms, God also knows our thoughts and in knowing thoughts knows directly the exercise of the agent causal power. I remain baffled as to your assertion that God wouldn’t know it.
Here is what I think is happening. You’re still thinking of agent causation as if it were event-causal agency. There is no difference in events on such a view, so there is no difference. So you argument is fatal to the view you say that you find most persuasive, Kane’s view. But it just doesn’t apply to agent causal theory for the reasons I’ve elucidated.
Comment by Blake — March 4, 2009 @ 10:32 am
Arrgh. I had a comment all written and NCT went down and I lost it.
But that’s OK since it was mainly anticipating possible responses by Blake.
But this avoids the question. Let’s say there are three times: t1, t2, and t3. The “moment of decision” let us say occurs from t1 to t2. Thus according to you there is a change in the “stuff” of the universe between t1 and t2 which corresponds to “the accidental intrinsic properties of the agent.”
But why couldn’t the change from t1 to t2 be random? That’s what you’ve not established. Clearly you don’t think it is. But why? (And no, I’m not thinking in terms of event libertarianism)
What you are doing is saying we know the change from t2 to t3 isn’t random because of the change from t1 to t2. But surely you can see what that just begs the question…
Comment by Clark — March 4, 2009 @ 10:45 am
Let me add some of the points from my lost comment.
My assumption is that any change between two states can be accounted in terms of what is determined (i.e. due to the laws of the universe and constrained by them) and what is undetermined. LFW claims we can make a distinction in the indetermined area and say part is due to agent choice and part potentially to chance. (One could deny chance altogether but I don’t think anyone is seriously making that move) So the question then becomes, given the state of the universe between t1 and t2 how do we distinguish what is chance from what is non-determined non-chance.
Thus far you’ve not been able to provide any way of knowing this beyond assuring that God does know it. But the question remains how God could know it if he knows via mediation.
Comment by Clark — March 4, 2009 @ 10:49 am
BTW – what I outline isn’t fatal to Kane’s view. It is essential to Kane’s view. It’s just that people feel it doesn’t answer the question of responsibility because they feel chance is incompatible with responsibility whereas Kane doesn’t.
Comment by Clark — March 4, 2009 @ 10:50 am
Clark: But why couldn’t the change from t1 to t2 be random? That’s what you’ve not established. Clearly you don’t think it is. But why? (And no, I’m not thinking in terms of event libertarianism)
Clark, you’re the one asserting that it merely random. You have to show it is random; I don’t have to show it isn’t. But it isn’t random if it is reasons-responsive and within the agent’s control. I show how that is and you ignore it. It’s not random because it is the result of a reasons-responsive act by the agent. It isn’t random because it embodies the agent’s values and and acts of valuation. It isn’t random because it is the agent’s reasons, the agent’s values and the agent’s act. You just ignore that.
Comment by Blake — March 4, 2009 @ 11:02 am
If you accept Kane’s basic assertion that “chance” isn’t incompatible with responsibility, then what in the hades are we arguing about — since you don’t accept your own argument and I certain don’t accept it.
Comment by Blake — March 4, 2009 @ 11:03 am
Man. NCT keeps going down and losing my comments. (I’m now typing them in an editor first)
Blake, I think that for responsibility to be compatible one has to adopt a revisionist account. So I actually concede the linguistic and intuitive argument to agent libertarians.
My point is that what agent libertarians demand is inherently unknowable. Thus God can’t know it. So God doesn’t know if we are responsible in the way our language and culture demands. I think that has important implications for the free will debate in an LDS context.
Regarding your point in 53. I think you’re still missing what I’m saying. I’m not saying it’s necessarily random. (Although I suspect ala Kane that there isn’t ALFW – but I don’t think it’s knowable) I’m saying no one could tell whether it is random or not. One can assume a metaphysics that answers the questions but never know if the metaphysics is true.
Comment by Clark — March 4, 2009 @ 11:29 am
To add, if we can’t distinguish differences in indeterminism then my answer is “why can’t we all just get along.” That is overlook any difference that appeals to an inherently unknowable difference.
Comment by Clark — March 4, 2009 @ 11:37 am
Clark: To add, if we can’t distinguish differences in indeterminism then my answer is “why can’t we all just get along.†That is overlook any difference that appeals to an inherently unknowable difference.
Because your (Mele’s) argument is cogent against event-causal theories and not against agent-causal theories. I have given at least a half-dozen ways in which God could know whether a state of affairs or an event is the result of an exercise of agent-causal power (in #47, 48 and 49). You ignore these ways. We may not be in an epistemic position to verify the difference (at least currently), but that is hardly telling unless you’re looking for an empirical argument.
Comment by Blake — March 4, 2009 @ 1:26 pm
Clark (#56),
The proposition anything that is universally true is inherently unknowable is virtually a contradiction in terms. Non-tautological universal truths have real world consequences. Universal real world consequences are knowable in principle.
Your argument that ALFW (if it exists) does not have knowable real world consequences is essentially the same as saying that ALFW is a meaningless proposition. I can’t agree with that.
Comment by Mark D. — March 4, 2009 @ 11:04 pm
Mark, I’ll respond to Blake hopefully tomorrow. But what would you say is a real world consequence?
(I’d also say I’m not a positivist, so I don’t think saying something has no real world consequences is meaningless)
Comment by Clark — March 4, 2009 @ 11:34 pm
Clark: The proposition that something that has no real world consequences is essentially meaningless does not entail positivism (and the long list of baggage that school of thought encompasses). It is of course a proposition closely related to weak verificationism.
I disagree with the verificationist position on what legitimately constitutes knowledge. However, for our purposes, what constitutes knowledge is irrelevant. What is important is what constitutes truth.
Of course in principle there could be abstract entities whose existence has no real world consequences. The truth of the existence of such entities would not be tautological, but it would be trivial.
Your hypothesis is that ALFW has no real world consequences. If ALFW has no real world consequences, then the truth of ALFW is immaterial, not worth bothering about, no more than an entertaining diversion, a resource sink for frenzied minds. Not technically meaningless, but virtually meaningless – the substance behind the common caricature of philosophy and metaphysics as much ado about next to nothing.
Comment by Mark D. — March 5, 2009 @ 1:12 am
Let me put it this way. I think there are ways of verifying if it is false. (Say evidence for a block universe) I can’t think of any way of verifying it were true. Once again recalling that what it is being compared to is a more general indeterminism and not any particular frequentist model.
But your last part more or less gets it right. If it is unknowable then it’s at best problematic to focus in on. Therefore I think one should adopt a revisionist account of free will and responsibility that doesn’t depend upon an ontological rejection of “chance.”
Comment by Clark — March 6, 2009 @ 11:53 am
Sorry, to troll here. I’ll work on either fully participating or just plain lurking.
I was looking for Kent (MD)’s latest post and came upon this. 61 comments in (wow)…I skimmed at best. But I came across this and found it interesting and somewhat relevant to the line of thought being developed here:
Economist.com
Comment by David — March 6, 2009 @ 3:25 pm
Clark, My point is that everything with real world consequences is knowable in principle. The procedural details (confidence, probability, finite convergence) are immaterial. Those are all practical considerations.
I claim that the proposition that the status of X is unknowable in principle is equivalent to the proposition that X has no real world consequences. A theory with no real world consequences is comparable to the suggestion that somewhere out there the prime numbers 17 and 23 are engaged in a dispute about which is more important.
I further claim that the proposition that X has no real world consequences is practically indistinguishable from the proposition that X does not exist.
Finally I claim that the proposition that X has no real world consequences implies that all rational arguments concerning X are per se insupportable. If X has no possible relationship to anything we can have knowledge or experience of – not even in principle, all arguments concerning it are invalid.
Given the fact that the whole field of philosophy is based on the idea that the persuasiveness of philosophical argument extends somewhere beyond sophistry, I consider the claim that anything with real world consequences may be unknowable in principle to be rather dubious. Likewise the claim that anything unknowable in principle may have real world consequences.
Comment by Mark D. — March 7, 2009 @ 10:01 am
Mark, I guess I agree but that merely changes the question to what are the real world (measurable) consequences.
Comment by Clark — March 11, 2009 @ 7:56 pm
Hey guys – does Clark advertise his candy bars? – this could ruin everything!
Comment by chad — April 9, 2009 @ 11:23 pm
I don’t think Blake adequately answered Clark in 47, 48, and 49. I don’t think Blake adequately understood Clark’s point (presuming that I do understand Clark’s point) – if God knows by observing external phenomena (mediated knowledge), and events occur which do not follow deterministically from prior states, then how does one distinguish the outcomes due to chance from those due to intelligent action.
Intelligent action – the product of choice – is distinguishable from chance by virtue of specification.
Comment by Log — June 3, 2012 @ 12:19 am