Blog Coordinator

« Sinnott-Armstrong on Brain Blaming | Main | Causation via criterial resetting »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Thanks for another good post, Peter. I said before that I think the kind of non-randomness your view gives us is too weak. The kind of non-randomness we need for genuine libertarian freedom, it seems to me, has to involve a kind of agent-involvedness, or I-did-it-ness. In other words, it has to be ME who made the decision, or who controlled which option was chosen, or something like that. Your kind of non-randomness seems to me not to give us this result. It requires the chosen option to be of a certain KIND (e.g., an escape route) that’s dictated by my preferences, but it doesn’t require that it be ME who chooses the given option.

Second, I think that what you’re calling the Kane-Balaguer strategy CAN deliver the required sort of I-did-it-ness, and thus avoid the luck objection. Or at any rate, that strategy can deliver this result IF our torn decisions are undetermined in the right way. Suppose I’m trying to choose between two options, A and B, and suppose I’m utterly torn and then I choose A over B, while still feeling torn. Finally, suppose my choice was undetermined in the following way: the actual, objective moment-of-choice probabilities of A and B being chosen (given the complete state of the world and all the laws of nature) were 0.5 and 0.5--i.e., they matched how the probabilities FELT to me--and the choice occurred without any further causal input, i.e., without anything else being significantly causally relevant to which option I chose.

If my decision was undetermined in this way, then at the moment of choice, the ONLY event that was relevant to which option I chose was my conscious decision. In other words, the conscious choosing event WAS the event that settled which way I chose. So it was ME who did it. So the choice was non-random (and non-lucky) in the appropriate way.

In discussing this issue in your post, you seem to say that if my choice is settled by an undetermined quantum event, then it’s random and lucky and so on (and you’re certainly not alone in making this inference). But on the model I’ve got in mind, the only quantum events that are relevant to determining which way I choose are those that are PARTS of my decision. The decision IS a neural event, and the neural event IS a bunch of quantum events. So if the decision is undetermined, it is of course true that some of its quantum parts are undetermined. But so what? This doesn’t change the fact that the conscious decision itself is the event that settles which option I choose.

Here’s a different model: just BEFORE I choose, there’s an undetermined (and wholly non-mental, i.e., brutely physical) quantum event in my brain, and the outcome of this quantum event causally determines which way I choose (and, again, on this model, the relevant quantum event occurs milliseconds BEFORE my choice). I agree that on THIS model, my choice is random and lucky. But in my model (where the relevant undetermined quantum events are just PARTS of the undetermined conscious decision), the choice isn’t random or lucky, because the relevant undetermined event--i.e., the one that settles which way I choose--just IS the decision. I.e., it’s the conscious me-choosing event. So we get the result that it’s ME who controls which option is chosen, and so the choice is NON-random (and non-lucky) in the relevant sense. Or so it seems to me.


I like your model, especially the part where you state “that it’s ME who controls which option is chosen”.

I think you’ll likely agree that when an option is chosen by ME, physical events occur within my brain that are associated with which option I choose, and those events are partially controlled by ME. In order for physical changes to occur within a brain, it seems reasonable to believe that some kind of force is required. So here’s where I’m going with this… According to your model, it seems reasonable to believe that “ME” exerts some kind of force, and if free will in the ambitious sense exists, then those forces cannot result solely from a direct sum of preexisting forces. ME must exert new emergent forces which effectively control the option chosen.

Is that a reasonable extrapolation of your model?


I agree with you that criterial causation exists (i.e., that certain criteria/information/patterns are capable of causing events to occur). However, I’m thinking that causing an event is significantly different from controlling an event.

From a high-level modeling perspective, it’s okay to say that criteria controls an event (in the “weak” sense), but from a deeper analysis, the truth becomes evident: the control is exerted by the interpreting entity.

Here’s an example illustrating what I’m trying to say: A person and a bird interpret a STOP sign located at a street intersection differently. From a high-level modeling perspective, it’s okay to say that the STOP sign controlled the person. However, the bird interprets the STOP sign differently, thereby confirming that the sign is criteria/information which is causal in nature, but the control is determined by the interpreter.

So here’s the point I’m trying to make: Criteria is causal in nature, but criteria doesn’t exert control in the “strong” sense. Therefore I’m unsure how criterial causation can be the source of our free will (FW in the strong/ambitious sense).

Mark, Sorry for the slow response and thank you for your feedback and comments. I have been in grant-writing purgatory, but should emerge by Wed. But let me try to answer the comments you posted here and on the other thread here if I can.

“I want to argue that your way of understanding the first two requirements--(a) and (b)--is too weak”

I am sure some of the many compatibilists here would smile with a touch of Schadenfreude to learn that my views are now being labeled “weak” after I labeled the Humean definition of FW the same, and caught some flak from them for it.

Let me say, I am not committed to a hyper-strong notion of L-freedom. I am just trying to figure out how volition and decision-making in fact work in the brain. If whatever Nature evolved ends up meeting the conditions of L-freedom, great, if not, then that is also OK. As you said in your book, what Pluto is has nothing at all to do with our definitions of what a planet is. Same goes for what volition in the brain is and how we define FW.

A weak version of L-freedom would still leave us L-free, no? A maximally strong version of L-freedom can be ruled out as impossible logically, and not even available to God. Let us say that God wanted megalomanical levels of utter ultimate responsibility. He would then need to be able to fully decide his character at the time of any decision. But he would then need to be fully in charge of deciding the basis or grounds of that decision, ad infinitum. So even God, if he existed, could not be utterly ultimately responsible for his creations. I suppose, if total ultimate responsibility is His (or our) goal, then even God’s decisions would succumb to the luck argument, and we would have to conclude that God just lucked into creating this Creation versus some other. So I think we have to agree that the goal cannot be utter ultimate responsibility. But relinquishing that entails relinquishing utter self-creation and utter L-freedom. The best we can hope for is a constrained form of L-freedom. Now maybe the strong L-freedom you demand is not as strong as utter L-freedom. But even if it turns out that we have a weak form of L-freedom, by, for example being able to shape our character over time, that would be good enough for me.

It would not surprise me if in the end we are about as able to resculpt our brain circuitry as we are to resculpt our bodies. Both require practice and commitment. But no amount of going to the gym is going to turn me from a man into a woman or a giraffe. I can get stronger and healthier. I can fully realize my potential, but my potential is limited. What is given to me genetically is my potential. Potential is innate. How it is realized (or not) is in part up to us and the environment we find ourselves in (or better yet, put ourselves in: we are not plants, we are animals, and can usually leave to find better conditions that will help us better realize our potential). But there are limits. I can fully realize my potential, but I cannot realize anything that does not lie within my potential. I cannot will to grow wings and fly away, for example, or simply will myself to be as smart as Einstein or as good-looking as Tom Cruise. That said, I think there is nothing more common in this world than people who are not realizing their great, even if not infinite potential. I think that is in part because of a false belief that outcomes are determined by genetics, or worse yet, that outcomes are determined by factors outside one's control that go beyond genetics, determinism being a major one. But life outcomes are not determined, assuming indeterminism. Potential is determined by factors beyond my control, but there is still a lot of room for most of us to will ourselves into environments and new patterns of living that will foster the greater realization of our potential.

In my science over the past several years I have become interested in the brain mechanisms involved in the brain resculpting itself through training and learning. In one study we looked at Dartmouth students who decided to subject themselves to one of our most intense course sequences, Mandarin 1-2-3, over nine months. This involves four hours a day of classes, homework and drills. What we found after nine months was that the language learners' brains had white matter tracts connecting language and other frontal planning/decision-making areas that had become more organized. We think that this occurs in part because there are glial cells called 'oligodendrocytes' that myelinate neighboring axons more when more action potentials go by in them. This is a lovely feedback mechanism that allows the brain to commit resources to bettering circuitry that is in frequent use. Just as you only grow a callous where the rubbing is taking place, and just as you only grow stronger muscles that are used, your decisions (say to learn Chinese or to become more honest) can resculpt the basis of your future mental activity and decision-making, and you find yourself at the end of your self-imposed training able to speak Mandarin or possessed of the better character you had aimed toward when you made your New Year's resolution.

Mark, you seem to be the only person commenting who is open to libertarianism. The majority of comments anyway, have so far come in the name of defending compatibilism. So I especially appreciate that you are making comments, since my main point is that criterial causation in the brain as we find it affords a kind of l-freedom.

Let me now try to address your more specific comments, particularly:

“however we articulate requirement (b), it needs to involve some claim to the effect that the person in question has to be in control, or the author of the choice”

I agree. This is how I see criteria set by the agent. It has to be an escape route, not any utterly random thing or idea. So the agent harnesses randomness to his or her own intended ends.

“I am in broad agreement that L-freedom requires (a) indeterminacy, and (b) non-randomness. I also think there’s a third requirement: (c) the indeterminacy has to somehow generate (or enhance or some such) the non-randomness”

I am not sure I fully understand (c), but it seems to me that the core of what you are after is that the final decision comes down to ME making the decision. Now I am also not entirely clear what you mean by ME, given that there are only neurons up there, and no soul or homunculus. However, I will take you to mean by ME my consciousness; more precisely the plans, desires, and intentions of my consciousness.

This is why I think you wrote:

“there has to be multiple options that are all REPRESENTED IN MY HEAD…. Moreover, I have to have REASONS for at least two of the options; or at the very least, I have to consider them to be live options.”

And also why you wrote about my view that:
“The choice is (a) undetermined, and (b) non-random in the Humean sense that it flowed from my reasons. But it’s not L-free because the indeterminacy is totally irrelevant to the non-randomness.”

But I do think that voluntional processes in consciousness (what I take you to mean by ME) are the final arbiter of many of our decisions. Conscious agency plays a central role in specifying some criteria, but also plays a central role in the final stage of evaluating options that come up to consciousness, which might then get accepted, modified or vetoed.

Take the escape route example again. Options are generated, perhaps unconsciously, once the need to think of an escape route is set. These internally generated ideas, plans or representations are certainly constrained by our knowledge of possibilities afforded by the world, but they are not limited by them. For example, I can imagine things that are not physically possible but nonetheless conceivable, like escaping on the back of a dragon. That there is only one window to escape through, and the fact that I am ignorant of other actually existent escape routes, constrains the options I can generate as possible escape routes, but it does not only leave me one possible option in my mind. I can generate internal representations of diving through the window and landing in a crash on the street; I can generate the idea of slinking through the window onto the ledge of the building; I can generate the possibility of opening the window and crying for help; I can generate the possibility of grabbing my pursuer and throwing him through the window instead. My point is that facts about the world (e.g. that there are two objective escape routes, only one of which I know about) should not be confused with facts about representations in my head. The former can at best only constrain the latter.

Another example: I have criteria in mind for what counts as a good mate. Let’s say I find myself on a deserted island with only one other person there for the rest of my life,a woman. I will end up being her mate, if she will have me, because she is the only possibility in my world. But she was not the only possible woman who could have met my criteria had the world been different. And certainly I could have imagined many other ones who in fact were not available to me in fact. My internal representations are not constrained by what happens to be the case in the world. Imagination and fantasy are testimony to that.

Let me now address your claim that under my model conscious volitional, agentic processing does not play a role in making the final decision. I disagree. If by ME we mean such conscious decision-making, then consciousness plays a very central role. First it sets the needs to be met (escape route, integrity-enhancing New Year’s resolution). Then possibilities are generated, in part unconsciously. But then these are presented again to ME for conscious evaluation and deliberation. If an option is inadequate after being consciously evaluated (say the option was the plan to fly through the window, or to try to be more punctual) it can be rejected and the process of option generation can begin anew. But if an option comes along that is evaluated by ME consciously, it then does not get vetoed. It is ME who gives the decision the final green light. So I think the conditions you want are met, of ME being the final decider, by my model. There is no need to go to torn decisions, though I am fine with torn decisions also being a basis of L-freedom assuming (a) undeterminedness and (b) non-randomness. I am just sceptical that torn decisions are the sole basis of the possibility of grounding L-freedom in the brain.

I think decision-making is not something that happens at a moment. Philosophers sometimes write as if a choice happens at an instant T. But a decision is more typically a process that extends over a duration in the brain as we find it. There is feedback among multiple levels before a threshold is finally crossed, and the system commits to one option over others. Agency over an option emerges because there is conscious consideration of adequacy before there is any commitment to enacting an option. In the example I gave, there is first conscious deliberation concerning personal inadequacy and the need to resolve to better one’s character. Then unconscious processes follow their orders and generate possible resolutions. Think of these as the zombie slave processes in the basement who try to satisfy the parameters given them. Poof, the possible resolution ‘I will be more punctual come 2015’ pops into consciousness fully formed, ready for evaluation. Now conscious deliberation begins and assesses whether this option is adequate as an integrity-enhancing resolution. This can go on for a long time, playing scenarios out, imagining what kind of person one would become over the course of 2015, if one stuck to this resolution. If, at the end of this process, this option is rejected, it is an agentic rejection of that in part randomly generated option. The slave systems, if they have not continued generating options, can be given orders to try again, or to improve this option. When, after long agentic evaluation, deliberation and playing scenarios out in the mental workspace, the option ‘to be more honest’ passes the threshold of adequate integrity-enhancement, it only does so because conscious agency has not vetoed it, and has forced its modification (by sending earlier versions back to the slaves in the basement) until it met highest-level criteria for integrity-enhancement. What more role for ME in our choices could you demand? There is agency at the initial stage of setting these conditions, parameters or criteria to be met, rather than those, and there is agency at the final decision stage, when options perhaps generated outside consciousness are brought to consciousness for final approval or vetoing.

Let me just wrap up this overly long comment (sorry but it is almost 2AM and so no time for editing) by saying I think we should not talk as if there is a ME in there. That sounds like a soul or a homunculus. There are just neurons up there doing their thing, but part of that is consciousness, and conscious decision making makes the final call on what my New Year's resolution will be.

Is that still too weak for you Mark?

Let me repost something I posted here in response to Randolph Clark in May of this year:

I believe criterial causation offers an escape from the luck objection. Indeed, it is the escape that evolution stumbled upon in forming our types of nervous systems. Even if the universe is identical in all regards until the moment t of a decision, choice or outcome, decisions, choices and outcomes need not be utterly random at t. They are in part shaped by our prior acts of agency. How? Our nervous systems are capable of setting up the parameters of future responses to input before those inputs occur. As such, inputs will meet preset criteria which were imposed agentically. Since any decision will meet criteria imposed by the agent, the agent is in part responsible for any outcome that meets those criteria, even if which outcome is decided upon is also in part a matter of chance. For example, preset parameters might be identical until t, and in one possible world a man comes across a book on Japanese first at t, just by chance, and ends up learning Japanese, and in a nearby possible (imaginable) world, he happens to come across a book on Russian first at t, and ends up learning Russian. This is a significant outcome for him, chancy and the fact that he came across one book first by chance was certainly beyond his control. How is he then in any sense agentically responsible for learning Japanese versus Russian? He may have made a prior decision to learn the first language he comes across when he goes to the bookstore, which preset his responses to subsequent inputs. That it was the Japanese book that he came upon first was not something he could control or is responsible for. But he is in part responsible for having the kind of nervous system at t that would respond to the Japanese book in a certain way. In a sense we are all reprogramming ourselves continually, regarding how we will respond to future input. Agentic control and responsibility do not come from choosing one alternative over another by chance, but from having a given set of agentically chosen informational criteria in place before the decision is chancily made (because that decision happened to meet those criteria first and just by chance). Thus agency and responsibility do not arise from the particular decisions made, but from the criteria the agent sets up for making those decisions before they were ever made.

Let me add, this criterial way of making decisions based on future inputs gives rise to a two-way power in both senses. I might learn language A or B. And if an above-threshold book never comes along, I might never decide to learn a foreign language. So I might learn a language or I might not. And whether I learn language A or B, or don't learn any, is in part a matter of chance. Granted, some criteria we do not choose agentically. They come prespecified by genetic potential and how that potential is realized by multiple factors. For example, if I am a heterosexual man, I probably could not help but want to marry a woman. I had multiple criteria and some threshold in place that any potential mate would have had to supercede. But there are nonetheless, at least in principle, infinitely many ways to meet any finite set of criteria. That I married woman A versus B was in part a matter of good or bad luck. But it was not solely a matter of luck, because had a mouse or a kangaroo come along, I would not have married her. But since I did not agentically preset the criteria for marrying a woman (this was, say, preset in me from birth), I cannot be responsible for marrying a woman (e.g. rather than a tree or kangaroo). Responsibility only arises when the preset criteria were set in part agentically, which is to say, the chooser had some control over the criteria that were set.

One final question Mark: How do you counter Levy's claim that the Kane-Balaguer strategy of grounding L-freedom in torn decisions fails to overcome the luck argument?

Hi Peter,

As far as I can tell, you've misconstrue my argument. I didn't (and shouldn't) claim that torn decisions are wholy a matter of luck. Rather, I claimed that when 2 or more options are available to the agent, in a metaphysically demanding sense of 'available', selection between them is a matter of luck. So I can and do accept something like criterial constraints on choice, without my argument being weakened. Couple of minor points. There is of course some sense in which I think information plays a caual role. It is realised by perfectly ordinary physical processes, and those realisers latch together (due to our evolutionary history) so that their properties mirror semantic and synctactic properties. Further, I'm no incompatibilist: I don't think causal determinism rules out free will. My views with regard to determinism are pretty ordinary compatibilist views.

James: I'm assuming something like a token-token identity theory of mental events like decisions. So when I choose A over B, the ME-choosing event IS a neural event. I wouldn't say that it's emergent. It's just a neural event that happens to also be a mental event of a certain kind (and happens to feel a certain way). So there isn't a ME-event and also some brain events. The ME-event just IS a brain event, and it causes my muscles to move in a normal brain-to-muscle way.

Mark, I am relieved to hear you say that to James about ME-choosing events being realized in physical neural events. It was sounding a bit dualistic in your ME comment, but I realized that that was just a rough way of taking.

(I think that one of the Achilles heels of many libertarian accounts to date has been the notion that there has to be some almost supernatural step that breaks the chain of sufficient causes).

That said, that information is realized in particular physical events does not make it identical (token-token versions especially) to those physical events. Going way back to Leibniz's principle of the identity of indiscernibles, two things that are identical should have identical attributes. But information and its neural realization do not have identical attributes. For example, neurons have mass and momentum and information does not. Information can be created and destroyed, but the energy in which it is realized cannot be. Therefore, I think a straightforward identity theory does not work. Moreover, identical information can be realized in many different physical bases. If Information1 were identical to physical basis1, and also identical to a different physical realization, physical basis2, then by transitivity, physical basis1 would have to be identical to physical basis2, which it is not. The notion of supervenience gets around a lot of these issues, I think.

Mark, you say

"But in my model (where the relevant undetermined quantum events are just PARTS of the undetermined conscious decision), the choice isn’t random or lucky, because the relevant undetermined event--i.e., the one that settles which way I choose--just IS the decision."

I take it that randomness and luck are introduced by an event being undetermined, e.g., randomness is introduced by an undetermined quantum event. If such an event causes a torn decision to go as it does, then it's a matter of chance or luck which way I decided. So how can it not be the case that the relevant undetermined event at the agent level - the undetermined conscious decision - doesn't involve an element of chance or luck?

Your model proposes that "the indeterminacy somehow generated the appropriate non-randomness (or enhanced it, or some such thing)". I agree that this is what's necessary for libertarian agency to overcome the luck objection and be superior in some sense to deterministic agency by adding to one’s control, authorship and responsibility. But I'm not clear on how it's supposed to work, hence the above question. Can you sketch briefly how “the right kind of indeterminacy entails non-luck”?

You also say in a comment on a previous thread that "What’s needed, it seems, is that the person has reasons for the two options (or at least sees them both as live options), but the reasons DON’T settle the matter and don’t cause the person to choose one option over the other (because that would be an L-freedom-undermining sort of causal determination)."

If reasons don't determine the choice, then it doesn't seem the agent determines the decision for good reasons, which is a central desideratum in being effective agents, imo. On your model, we can't explain our choices as being rational in torn decisions, which you say might happen frequently. L-freedom seems to have the consequence that our acts become unintelligible to ourselves, even if we make them in some ultimate sense by generating non-randomness from indeterminism.

Tom and Mark, I agree with all of Tom's points. It seems to me, Mark, that your account leaves us deciding for this option versus the other in a torn decision simply as a matter of luck, without our agentic reasons for favoring these (sub)agentic reasons for option 1 over those (sub)agentic reasons for option 2. The thief in van Inwagen's example just finds himself stealing or not stealing for no reason specified by his (higher) reasons. So I have the same questions as Tom.

I do not think that my account has this problem because the criteria set are the agent's (they impicitly define our reasons for choosing that have to be adequately met by any possible satisfaction of those criteria). And once possibilities have been generated, in part randomly, the agent has veto power over whether they will be implemented, because they are evaluated consciously, using full powers of deliberation and playing things out in imagination (working memory) before a commitment is made to really go with this option. So whereas the Kane-Balaguer account seems to succumb to the luck objection, and lacks reasons for going with this set of reasons/motives versus the other set of reasons/motives (since it is just a matter of luck), the criterial decision-making account of grounding L-freedom affords reasons that are the agent's for going with this option versus the other, affords the agent indeterminacy-shaping power at both the criteria-setting stage and the final conscious veto-or-accept-or-modify stage, and lastly evades the luck argument.

That all said, if you can convince me and Tom that the Kane-Balaguer account does not succumb to the argument from luck, then great. Then there would be two ways of grounding L-freedom in the brain, yours and mine. We are in the end on the same side of this debate!


You said “when I choose A over B, the ME-choosing event IS a neural event”. I agree with you, that the ME-choosing event is a neural event, but I don’t think the ME-choosing event is controlled solely by the neuron level – it’s also partially controlled by something new which emerges at the “neural wave level” (i.e., our thoughts).

If all ME-choosing events are controlled solely by the “neuron level” of brain activity (vs. the “wave level”), and if all “neuron level” activity is controlled solely by the laws of physics (or else is random in nature), then how does your model explain that our ME-choosing events are anything other than predeterministic (or random) in nature?

I believe in ambitious/strong sense free will (so I’m in the same camp as you and Peter), but I don’t understand how FW can exist without the presence of new emergent forces.

Peter, I know you believe criterial causation is the answer, but as I argued in my last comment, I don’t believe criteria is capable of exerting control in the “strong sense”. If you have time to explain, I’d be interested in your argument(s) supporting the idea that criteria exerts control in the “strong sense”, not simply that criteria is causal in nature.

James, I think that your apparent way of thinking, where causation is tantamount to the application of some force on an object, is definitely the dominant way of thinking about causation. I also think it is an incomplete way to think about causation. And I think that attributing to information the capacity to exert forces on matter is a mistaken way of thinking about information's causal role. Don't get me wrong. I am all in favor of information being causal in the universe, just not as a kind of force. Information is not causal as a force, but as a set of constraints placed upon subsequent possible events. No force is applied once parameters for permissible future actions are in place. I think that you are running into trouble because you want information to exert a force on matter, which is leading you to mistakenly posit ghostly informational energy forms, such as your 'neural wave' theory. There is no evidence at all for anything like that, and it would anyway not solve the problem of FW, because it would just shift the problem from one level to the emergent level (that is, if we cannot ground FW and MR in physicalism, I do not see how we can ground it in the new kind of "informational energy" you seem to want to posit as emergent neural waves). To answer more fully, I will have to write a longer posting for everyone. I will do that tonight.

Peter, you say

"In contrast with Kane and Balaguer’s accounts of LFW, in the case of criterial decision-making, there is a higher, but non-determinative governing basis for making a choice."

Seems to me that this basis is determinative since the criteria select - determine - the option that's chosen. The option itself might be indeterministically generated, but the selection process is a deterministic function of applying the criteria, not a matter of chance, right?

"Under criterial causation the choice is not utterly random because it had to be an integrity-enhancing resolution. It is also not determined, breaking the chain of sufficient causes that underlies the ultimate responsibility-destroying regress, because a different integrity-enhancing resolution might have won out. The regress is broken by adding indeterminism. But the luck argument is broken by forcing any choice or action to meet criteria set by the agent."

I’m not sure how your proposal adds to the control and responsibility already present in deterministic accounts of agency. Given our initial, inherited, non-self-chosen characters and capacities, we *determine* (set) the criteria for self-forming resolutions and future action, which then *determine* the selection among options. As you put it, "we set these criteria versus others because of the kind of agent who we are". This agent determinism confers proximate, but not ultimate, responsibility for self-formation and choice.

I get that the options might be indeterministically generated, thus breaking the causal regress, but I don’t see how breaking the regress adds to control or responsibility. Control and responsibility require a law-like, reliable, non-chancy relation between character, capacities and action such that the action can be traced back to the agent. Of course, the agent too can be completely traced back to non-self factors involving chance and necessity.

I don’t see how inserting indeterminism at any point in the process of self-formation and choice adds to control and responsibility, making the agent ultimately responsible as defenders of LFW would have it.

Probably my comment ended up in the spam folder. I won't try to post it again, given k don't have a keyboard in front of me. The central point is that I never claimed that choice between metaphysically open options was wholy lucky; rather the claim was that the contrastive fact is lucky. You mention that, but then seem to forget it. Given that that's supposedly the locus of freedom, it's that fact that should be the focus. I can and do accept that there are constraints on choices which are not themselves lucky.

OK, there have been lots of responses to what I posted the other day. Here are some counterresponses (to Peter, Tom, and James) in more or less arbitrary order (and I apologise for the length of this, but I wanted to respond to all of the questions/comments/etc. that people have made to me):

1. Peter:

(i) I agree that there can be degrees of L-freedom; I’ve written about this in the past. I also think there can be degrees of agential authorship and control and so on. Your kind of L-freedom (in which the only required agential control lies in the fact that the agent sets the parameters of the TYPE of choice to be made) gives SOME control and L-freedom. But it strikes me as exceedingly weak, and it’s no more robust a kind of free will or control than we get in compatibilistic kinds of freedom. It’s also more or less identical to the very WEAKEST kind of control that we get from torn decisions. In a torn decision, MY reasons pick out the tied-for-best options. So even if the choice among those options was entirely random and selected by a mechanism external to my psychology, we would still get SOME authorship and control, because, again, my reasons picked out the tied-for-best options. So it seems to me that torn decisions already deliver about the same degree of freedom that your model delivers, even if the selection of a specific tied-for-best options is totally random. But I would like to get MORE L-freedom than that by getting a certain kind of authorship and control over the selection of the specific tied-for-best option. My claim is that if the selection is undetermined in the way that I described in my earlier post (and also in my book and other publications), then we get some authorship and control here and, hence, increase the degree of L-freedom that we have over these choices.

(ii) Peter, you wonder why I think we need clause (c)--i.e, why the indeterminacy needs to generate or enhance the non-randomness. Here’s why: Suppose that Jane is a fully deterministic creature who has Humean compatibilistic freedom. Her choices are clearly non-random--because they are causally determined by her reasons. Now suppose that event e is a quantum event that is part of some decision D of Jane’s. Since Jane is determined, and D is determined, it follows that e is determined. Now let Jane* be just like Jane except that the prior probability of e occurring is 0.999999999999. There was a tiny chance that e wasn’t going to occur, and if it hadn’t occurred, then D wouldn’t have occurred; but in fact, e did occur, and so D did too. Jane clearly doesn’t have L-freedom because she’s determined. Does Jane* have L-freedom. I think it’s clear that libertarians have to say that she DOESN’T--because the indeterminacy in her choice is totally irrelevant to the fact that it’s non-random and free. Libertarians think that we NEED indeterminacy for free will. So the indeterminacy has to MATTER--it has to be what DELIVERS non-randomness and L-freedom. That’s why we need clause (c).

(iii) Peter, I think that by “ME”, I mean more than my consciousness, because my subconscious beliefs and desires are part of me. Also, what’s actually present to consciousness at any moment is pretty limited. Lots of my personality traits aren’t things I’m consciously thinking of. So I think that *I* am broader than my consciousness. What is it exactly? I don’t know. I don’t have a view. I’d happy to let neuroscientists, cognitive scientists, and philosophers work this out. I will hasten to add, however, that the ME is NOT, on my view, supernatural in anyway. I am a full-blown materialist.

This relates to some comments you made about my response to James vis-à-vis the token-token identity theory. This is a gigantic topic; I could write 30 pages on it; I won’t do that, but let me say just a few words. When I said that I endorse a token-token identity theory of decisions, you should not have taken me to be endorsing a token-token identity theory of INFORMATION (or anything else). Decisions are EVENTS, so they are very different from information. It’s not entirely clear what a bit of information is. My own view is that talk of information is ambiguous between talk of PROPOSITIONS (which are abstract objects) and talk of things that express propostions (e.g., written sentences, sound waves, neural states or events, etc.). So if I wrote ‘snow is white’ on a piece of paper, there is the pile of ink, and there is the abstract object (the proposition that snow is white), and that’s it. Likewise, if I consciously think “Snow is white”, then there is the neural state/event, and there is the proposition, and that’s it. If this is what you mean by a neural state “realizing” information, then I’m fine with it. I would say that the neural state EXPRESSES the proposition, but whatever. What I think is an utter confusion is talking about a neural event realizing a pain (or a decision)--as if the pain (or decision) were a different thing than the neural event. There’s the CONCEPT of pain, which is an abstract object, but it would be weird to talk about the concept of pain being “realized” by a cut in my finger. If you want to talk that way, it’s fine with me, but it doesn’t strike me as useful. Anywow, as far as my SPECIFIC pain goes (when I cut my finger), ALL there is is the neural state. If you deny that, I think it’s a very quick slide into Cartesian dualism. My pain is in my head, and it is physical, and the only things up there are neural state and events. There is nothing else the pain could be except for a neural state or event.

(iv) Peter, regarding what you say about consciousness making the decision on your model. You talk about deliberation. That’s fine, but remember, Humean compatibilists can help themselves to deliberation. If I deliberate and decide that A is my best option and then choose that option, that’s perfectly compatible with determinism. Indeed, as far as I can see, determinism is compatible with all the freedom you could want in any decision where the agent thinks that there’s a unique best option. If my reasons favor A over B, then let my reasons settle it. This is precisely why I think that torn decisions are where the action is, vis-à-vis L-freedom. And as you say, we want agency at the FINAL stage of the decision. That’s why I think it’s important to save a way in which the agent is doing the choosing of the specific option in a torn decision.

(v) Peter, regarding your question about Levy’s claim that my view falls to the luck objection. Well, I’ve written an entire book chapter about how I think the luck objection is met on my view. And I summarized it in a previous post. In the briefest nutshell, I think that if our torn decisions are undetermined in the right way, then the ONLY event that has anything at all to do with determining which tied-for-best option is chosen is the conscious-choosing event, i.e., the neural event that IS my conscious decision. I think this makes the choice MINE, and under my control, in an obvious way. There is obviously lots more to say about this, but I can’t give my entire argument here.

2. Tom (and Peter, insofar as you said you agree with Tom’s points and echo his questions):

(i) You say: “If an [undetermined quantum] event causes a torn decision to go as it does, then it's a matter of chance or luck which way I decided.” I agree. But on my model, the torn decision isn’t caused by the quantum event (or by anything else). Or at any rate, it isn’t caused by anything IF THE CHOICE IS APPROPRIATELY UNDETERMINED. I said in my previous post what the relevant sort of indeterminacy is. Very roughly, this kind of indeterminacy requires that, at the moment of choice, which-tied-for-best-option-is-chosen is not causally influenced by anything. This is a bit vague. I would say that it’s not caused, period, but that’s not quite right, because it’s probabilistically caused by the agent’s reasons. But we can sort of say this: once the person is in a torn state, and is about to make a torn decision between A and B, which of these tied-for-best options is chosen is NOT CAUSED. So it’s not caused by the undetermined quantum event.

Now, there ARE undetermined quantum events that are PARTS of the decision, but that’s very different from being CAUSED by PRIOR quantum events. Think of it this way: when my car turns left, is it’s turning left caused by the fact that the particles that compose it are turning to the left? That would be a very weird way to talk. The car’s turning left just CONSISTS IN the particles turning left. Likewise for my decision. The fact that it’s undetermined just CONSISTS IN some of its quantum parts being undetermined.

(ii) You are probably scratching your head, thinking “But if NOTHING causes me to choose A over B, isn’t it true that the choice JUST HAPPENS? And isn’t that just luck?” Here’s my response: In a sense, that’s right. But notice that the event that “just happens” is a CONSCIOUS-CHOOSING EVENT. If I make a torn decision to choose A over B, and if you asked me, “Why did you do that?”, I might answer, “Well, I had reasons x-y-z for favoring A over B.” To which you might say: “Yes, but you had reasons p-q-r for favoring B over A. And you felt torn--i.e., you didn’t know whether you’re A-reasons outweighed your B-reasons. So why, given all that, did you choose A over B?” To this I might respond: “I don’t know; I just did. I had to choose right then, I couldn’t wait, and I was torn, so I just sort of randomly picked.”

I think we do this sort of thing all the time. And I don’t think there’s a problem with it, vis-à-vis the luck objection. There is a clear sense in which I chose randomly. But so what? The sort of non-randomness that’s required for L-freedom is compatible with this kind of randomness. The kind of non-randomness that’s required for L-freedom is an I-did-it-ness. And this is compatible with choosing arbitrarily, or “randomly”. What we have here is that *I* chose randomly. I did it. I might have chosen in an arbitrary way (or a “random” way), because I was torn, but it was ME who arbitrarily picked A over B. (Or, again, it was ME who did this, IF the decision is undetermined in the right way.)

(iii) Tom, you say: “If reasons don't determine the choice, then it doesn't seem the agent determines the decision for good reasons, which is a central desideratum in being effective agents, imo.”

If by this you mean that a SPECIFIC choice needs to be made for good (i.e., compelling) reasons in order to be L-free, then I disagree with this. What’s needed for a decision to be L-free is that the AGENT authors and controls the choice, not that the agent did this FOR COMPELLING REASONS. To me, requiring that all L-free choice be made for compelling reasons is (a) question begging--no libertarian should admit this, because as soon as we do, the game is over--and (b) out of touch with our concept of free will. Let me argue very quickly for claim (b).

Consider the following comment/argument: “Jane didn’t have good reasons for deciding to move to New York; it was really irrational; therefore, she didn’t choose of her own free will.”

This strikes me as an utterly bizarre thing to say. How could it follow from the fact that Jane chose irrationally, or not from good reasons, that she didn’t choose of her own free will? It doesn’t. In response to this, it would make perfect sense to say: “What do you mean? It was still HER who chose. She might have chosen irrationally, but she still chose of her own free will.”

You might think that a PERSON can’t HAVE FREE WILL (in some very general sense) unless he/she is capable of choosing for good reasons. But that is very different from saying that a specific decision can’t be free (or L-free) unless IT’S chosen for good reasons. The latter seems wrong to me, and that’s all I have to give up on my model. In other words, my model is perfectly consistent with the claim that a person can’t have free will unless he/she is capable of choosing for good reasons.

3. James: You ask how I can avoid determinism if I think that decisions are just neural events and that all brain causation is just neural causation without any new, emergent sort of causation. Answer: if indeterminism is true at the quantum level. That strikes me as a 50-50 bet right now. I think that our torn decisions are ultimately just bunches of quantum events. If all quantum events are determined, then all neural events and determined, and libertarianism is false. But if some quantum events are undetermined, then it opens up the POSSIBILITY that some neural events are undetermined, and if they are, then that opens up the further possibility that some torn decisions (which just ARE neural events) are undetermined in the RIGHT way. I argue in my book that, as of right now, there’s no good reason (empirical or non-empirical) for thinking that our torn decisions aren’t undetermined in this way. I think it’s an open question.

Mark, regarding the requirement that the specific choice in a torn decisions has to be made for compelling reasons: I think you're entitled to a third objection. Viz, it's ethically unrealistic to suppose that all torn decisions *have* compelling reasons that decisively favor one action over the other. This might be relevant when we think about moral responsibility in ethical dilemmas.

Paul, I agree. There are actually multiple things that one might mean by 'having reasons'. On one reading, it means having a psychologically real reason that's represented in my head and motivating for me. On another reading, it means something more objective--where I could have a reason to A even if I don't care about that reason at all. But the sense that I care more about is a peculiar phenomenological sense: I've got some reasons (that I care about) for A and for B, and I have no conscious belief as to which set of reasons is strongest. So I FEEL torn. That's what I mean when I say, in torn decisions, we have no compelling reason for choosing A over B.

Hi Mark: On your response to Tom and Peter, under 2ii, at the the end of the first paragraph, you make a crucial move: "To this I might respond: “I don’t know; I just did. I had to choose right then, I couldn’t wait, and I was torn, so I just sort of randomly picked.”" I think what you say here fits the phenomenology of certain torn decisions. And when I interpret the "I" in this quotation as referring to an agent-substance, and the picking as agent causation, I can make sense of what you say. But you're an event-causal libertarian. So how do we interpret this quotation, in particular the "I just sort of randomly picked" part, in event-causal terms?

Hi Derk. First of all, this is all assuming that the torn decision is undetermined in the right way, so that the objective, moment-of-choice probabilities of A and B being chosen are both 0.5, and the choice occurs without anything external to my conscious reasons and thought coming in at the moment of choice and causally influencing whether I choose A or B. Given this, I want to interpret the situation as follows. An event occurs in my head. Call it D. D is an undetermined neural event. But it's also a conscious-choosing event, or as I've been calling it, a Me-choosing event. Part of the phenomenology of D is that I've got reasons for both A and B and don't know which reasons are stronger--I feel torn--and I have the experience of just choosing. This is why I want to describe the situation by saying that *I* chose randomly. I get what you're saying--that if there's an agent-causal event, then we can say that *I* chose. But my claim is that there's another route to it being accurate to say that: if the undetermined neural event that settles which option is chosen IS a me-consciously-choosing event, then it is accurate to say that *I* chose. And, again, this is assuming the right kind of indeterminacy, so that nothing is MAKING me choose that way.

Mark and Derk--I don't man to interlope and please forgive me if I'm just being dumb, because that's entirely possible--but this I think is classically a crucial kind of case for marking out the efficacy of agency in these special circumstances. What tracks the phenomenology of feeling torn as epistemically .5 assessed as such even if one assumes an objective probability of .5? I guess I'm questioning "the choice occurs without anything external to my conscious reasons and thought coming in at the moment of choice and causally influencing whether I choose A or B", and whether "without anything external to my conscious reasons and thought" just stipulatively settles the question of the match between the epistemic and objective probabilities. Couldn't one be mistaken in feeling/believing the .5 probability though it is objectively otherwise? So assume the real objective probability--say .6--is why one chooses something but this is assessed epistemically as .5. Then one would be phenomenologically and epistemically mistaken as being *actually* torn. What guarantees the tracking between the subjective probabilities and what might actually be the case, even assuming the truth of metaphysical indeterminism? I'm assuming this must involve some sort of mind-body tracking in terms of reliability of epistemic claims about the metaphysical, but stipulating that problem away, particularly involving first-person subjective probabilities, seems problematic.


I’m a bit puzzled because on the one hand you say that the choice in a torn decision isn’t caused by anything and on the other that it’s probabilistically caused by the agent’s reasons. For a decision to be mine, I have to produce, make, or cause it in some sense, so the idea that it isn’t caused by *anything* doesn't seem tenable if you’re trying to secure a robust notion of libertarian agency. You say the appropriate non-randomness inheres in the I-did-it-ness of the choice, so *I* must have caused the choice in some sense. Ordinarily we cite an agent’s character and reasons, which she consists of as a particular, identifiable agent, as causes of choices.

You say that “There is a clear sense in which I chose randomly. But so what?” What I wonder about is how randomness in a choice adds to the control of and responsibility for choices made deterministically as a function of an agent’s character, reasons and capacities. On your view, the “I” that chooses randomly consists, at the moment of choice, of a decision process, happening in the context of conflicting neurally-instantiated reasons, that includes indeterministic quantum events which make the choice uncaused (“The lack of causation is a function of the fact the choice involves quantum parts that are undetermined.” And: “if the undetermined neural event that settles which option is chosen IS a me-consciously-choosing event, then it is accurate to say that *I* chose”). I don’t see how an agent that essentially consists in such a decision process is more in control and more responsible for a choice than in deterministic choosing.

Libertarians want a buck-stopping, robustly *I*-did-it agent that breaks the causal regress. But on your view it seems that buck-stopping agency shrinks, finally, to an undetermined neural event that just is the me-consciously-choosing event. But I wonder: is that event really *you* on any plausible construal of being an identifiable person who makes choices? And why is such a you more deserving of praise and blame than a non-libertarian agent?

Alan, on my view, we have no way of knowing from the armchair that the objective probabilities match the phenomenological probabilities. It's an open question for neuro-cognitive science (and for physics). For all we know, determinism might be true, in which case the objective probabilities are always 1 and 0. I have argued in my book that (a) what's needed for L-freedom is that the objective probabilities of our torn decisions match the phenomenological probabilities, and (b) it's an open empirical question whether this is true.

Tom, you raise a number of issues; let me respond to them (and thanks, by the way for your interest)…

1. Yeah, I was speaking a bit loosely when I said that it’s not caused by anything. Let me try to be more clear. On the libertarian view I’ve got in mind (and, again, I think it’s an open empirical question whether this kind of libertarianism is actually true), our reasons, thoughts, etc. cause us to be in a torn state in something like a Humean way. So if I’m trying to choose between A and B, and I decide that A is best and I pick it, then a Humean can say that this was caused by my reasons. Likewise, if I become utterly torn between A and B, my libertarian says that my reasons caused me to be in this state. My reasons can also cause me to make a torn decision at some time t (because, e.g., I have to decide NOW, and don’t have time to deliberate any more). And my reasons can cause it to be the case that I will choose either A or B (i.e., that I won’t choose some third option that I don’t like as much), and they can cause it to be the case that the (objective) probabilities of the two tied-for-best options being chosen are .5 and .5. All of this is perfectly Humean. Now, suppose that I choose A and that this was undetermined (with a .5 prior-to-choice probability). Then we can say that my choice was probabilistically caused by my reasons. But there is still an element of uncausedness here. Because we can say that when x probabilistically causes y, what this means is that it deterministically causes it to be the case that certain prior probabilities hold and that then the outcome is a result of chance. (E.g., “Scientist S running experiment E probabilistically caused electron e to collapse into a spin-up state in direction d” can be read as meaning that S’s running E deterministically caused it to be the case that there was some probability between 0 and 1 of e being spin-up, and beyond that it was a matter of pure chance that it collapsed into spin-up instead of spin-down.) So returning to my torn decision, the view is that my reasons deterministically caused me to make a torn decision between A and B with a .5 probability of each option being chosen, and then the fact that I actually chose A instead of B was uncaused.

2. You also wonder how this view delivers an I-did-it-ness. Well, the above is a large part of it. It’s MY reasons and thinking and so on that dictated that I should make a torn decision between A and B. The rest of it arises from the following: the torn decision was made BY ME. Why? Because it was a conscious choice. So in this case, since I’m torn, what I WANT is a random selection. I don’t want it to be dictated that A is torn (or that B is torn). I want it to be a random choice from A and B--because I’m torn and I have to choose now, i.e., I don’t have time to deliberate any more. Given this, the best-case scenario is that it be ME who does the just-picking. And my model gives us that--because the event that settles it just IS the conscious choice.

3. Finally, you wonder how this gives us more authorship and control than a deterministic model. Here’s how: Given that I’m torn--given that MY reasons dictate a 50-50 choice between A and B--if the choice is determined, then it is presumably determined by something external to my conscious reasons and thought (e.g., some wholly non-mental neural event in my brain). That external causation would be freedom-undermining. But on the model I’m describing, there is no external causation of this kind. The undetermined neural event that settles the matter IS my conscious choice. So this gives us more authorship and control (over our torn decisions) than we have on a deterministic model of torn-decision making. You wonder if this undetermined conscious me-choosing event is really ME. Well, let me just say this: it’s a lot more me than some wholly non-mental, non-conscious neural event is.

Hope this helps clear things up a bit.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan