The colossal problem of free will has captivated professional and amateur philosophers alike ever since the dawn of mankind, when we developed the capacity to ponder our identity and relation to the world. The problem is compelling because finding the balance between freedom and necessity is a question we all face, from planning our future to planning our evening. To be free is the founding principle of this nation, and its implication, that of moral responsibility, is the premise behind religious ideas that have inspired the greatest movements of history. To be informed of not having free will, to view ourselves as the characters of our favorite novel, enacting pages from the cosmic story of the universe, would not only be denouncing all the great ideas behind religion, the Constitution, it would be like discovering one is adopted, or bred synthetically out of a test tube, which is why the problem of free will is taken with extreme care and occupies a central role in ancient and modern philosophy. Unlike physics, the goal of philosophy, I believe, is not to fit theories and formulas to the universe, nor is it to classify all that is scientifically proven as gospel and all that is not as rubbish, but to view issues from a variety of perspectives, be it epistemic or metaphysical, and try to make sense of life and the destiny we all share. Thus, we study the central topic in philosophy of compatibility, whether we are responsible in the world that is deterministic. We do so by revisiting a classic stance compatibilism, in light of subsequent criticisms of it, and considering a more analytical take, to make concrete ideas that have merit.
A.J. Ayer, professor at Oxford University, advocated a compatibilist theory from which many of its ideas were used in subsequent modern discussions on free will. As this essay builds on Ayer’s idea of free will and constraint, I take his definitions of freedom (1) and constraint thereof in my proposal for a measure of free will. Ayer first discredits all contemporary attempts to attribute moral responsibility in an indeterministic world, classifying all actions as either random, in which we are surely not responsible, or causally determined, so that all fruitful attempts of deriving moral responsibility must either “show that men can be held responsible for actions which they do not do freely, or else find some way of reconciling determinism with the freedom of the will.” In taking the latter approach, he defines freedom as contrasted with constraint, and that the absence of the latter is equivalent to the presence of the former, and vice versa, so that “from the fact my action is causally determined does not necessarily follow that I am constrained to do it.” He juxtaposes situations whereby an agent acts out of the agent’s power (2), against those in which one acted while constrained, either because of some disability (3) or being “confronted with a still more disagreeable alternative” (4).
(1) Ayer considers many definitions on freedom, but emphasizes that freedom can’t be defined as self-contained, in a determined world or not, as some contemporary philosophers “have defined freedom as the consciousness of necessity.”
(2) A stronger belief of this is adopted by another group of philosophers, the libertarians, who believe the world is non-deterministic and derive moral responsibility from it. Agent-causal libertarians believe an “agent’s power” comes from the ability to be the first in the causal chain, to be the first of dominoes; event-causal libertarians adopt the former stance of Ayer’s ultimatum, that agents can be responsible for events they didn’t cause, and adopt responsibility despite not being the first of dominoes.
(3) In the case of a kleptomaniac, he is unable to impact his behavior with his thinking.
(4) Think of someone under gunpoint.
Ayer contrasts constraint from such “special causes” as distinct from behavior which is bound nonetheless by “the chains of causation.” The key but equivocal statement he makes is that, “in the given circumstances, the event which is said to be the effect would not have occurred if it had not been for the occurrence of the event which is said to be the cause, or vice versa, according as causes are interpreted as necessary, or sufficient conditions…” He finishes by the conclusion, “…first, that I should have acted otherwise if I had so chosen; secondly, that my action was voluntary in the sense in which the actions, say, of the kleptomaniac are not, and thirdly, that nobody compelled me to choose as I did… When they are fulfilled, I may be said to have acted freely.” Ayer makes compelling cases by giving examples of constrained individuals, and arguing the individual would have done otherwise had a constraint been removed.
The central problem with Ayer’s argument is the if-statement. As contemporary libertarians like Chrisholm, would point out, it may be the case that all three statements hold:
A) If A chose to do otherwise, he would have done otherwise.
B) A lacked the capacity to do otherwise.
C) A could not have done otherwise.
Indeed, validation of Ayer’s conditional analysis (which I now refer to as “could have, would have”) requires we travel to the nearest universe in which there is a relaxed version of the constraints that initially prevented an individual from doing otherwise to see if he did do otherwise, before attributing freedom and responsibility to him. Unless this can be done, I think it is futile to extend Ayer’s arguments beyond imaginative speculation, no more so than saying, “if it wasn’t against the rules to peer into the box, I would have seen the answer” after being fooled by a magic trick. Now, I concede imaginative speculation is at the core of philosophy, and that it is unfair to discount arguments like Ayer’s simply because they can’t be practically validated, for they brings otherwise unattainable insights into the nature of reality.
What follows is my attempt to extend Ayer’s “could have, would have” to a more concrete formulation by moving the time of speculation from hindsight to right before the scenario takes place — t minus epsilon, epsilon an infinitesimal, and t the moment the action was done. In extending Ayer’s idea, I first redefine the extent to which one “could have, would have” on a continuous rather than discrete (binary, to be exact) spectrum, and introduce various factors that can affect this function of freedom, in the form of a thought experiment. My hope is not to help validate Ayer’s idea, but to make it more tractable to our intuition, and perhaps convert more critics to reconsider Ayer’s conditional analysis rather than try to reject it altogether.
To decide how free Person A was at moment t, imagine a clone (5) A_c is created at moment t minus epsilon, and put to watch A’s actions through film. Have A_c rank possible actions A could take in terms of reasonableness, and associate confidence scores thereof. In cases like the kleptomaniac, A_c should, with near full confidence, assert A will take the purse. The key here is that A_c undergoes the same thought process as A. Standing as an outsider, A_c can replicate nearly all of A’s decision-making process, save for making the decision himself. The clone is also absent of any perturbation that may have spurred out of the situation at the instant of action.
(5) Such a clone would be subjected to all of A’s traits at birth and life development. Whole brain emulation, 3D printing, and synthetic biology, combined, might just bring this thought experiment to reality! For an expert’s take, read the chapter on paths towards strong AI in Nick Bostrom’s Superintelligence: Paths, Dangers, Strategies.
The ease with which the clone can guess should, in all reason, be an indicator to how constrained the original person A is. If the clone has great difficulty guessing what A was about to do (or better yet, if he is wrong in his guess), then A could be said to have very likely made what we agree to be a free action. In lieu of this, the metaphysical version of this is to have a cosmic spectator weigh multiple separate timelines, each representing a different action at time t, and having priors on the relative reasonableness of each timeline occurring. If the prior prediction that the timeline in which A does otherwise is high, that corresponds to a higher “could have, would have” score.
This sheds new light on the “could have, would have” statement, as it now allows us to both evaluate the level of constraint a person has. Similar to a chess game, a move made under the constraint of checkmate is one easily determinable by the spectator, whereas a move midway through the game, used to set up a strategy, with no endangered pieces on the board, is one that is much harder to predict.
One merit of this thought experiment is that it is no longer speculative, but reproducible. If we would replicate this on a trillion clones for actions that we conventionally view as free choices, like what flavor of ice cream to pick, several things can happen:
Clones, despite being replicas of one another, guess variedly in confidence, with many conflicting predictions.
This would be very telling for agent-causal libertarians and conclude the determinism thesis false, as that implies there is some transcendental trait of agents to be initiate a new causal chains in an action.
The clones more or less reach a consensus, but with significant perturbations in confidence (and so extends beyond what may have occurred in the epsilon time).
This may be telling for event-causal libertarians and also conclude determinism false, as that implies randomness, perhaps at the quantum level, are able to amplify itself to affect an agent’s decision process. Repeating this higher leverage actions may be more telling.
All trillion clones all reach the same prediction, with equal confidences.
For the context of this compatabilism argument, suppose this is the case, which isn’t very telling for the formulation we initially sought to obtain.
Thus, to generalize the thought experiment to satisfy the hard determinist, consider a rational being B as a judge, one who is well-studied on all matters regarding causation, who is of a level of intellect that he is able to predict reasonably well how agents act in decisions without bias, should he be given all the information of the person’s innate traits and life development. Using this as a baseline, we can then define the freedom of a person’s action on B’s difficulty in guessing what A would have done, relative to other actions in B’s training set, as a function of the distribution of outcomes B believe A will take, and the associated confidences thereof. The more varied this distribution (6), the less constrained B is, and so the more free he is.
(6) Unpredictability in action has been tied synonymously to freedom from control by subsequent philosophers like Daniel C. Dennett, in which he uses such a freedom to build his moral responsibility account (for Ayer, that’s more or less all we need). Examples he implied this to be the case include an RC helicopter that escapes the controllable range, or a pseudorandom generator.
To conclude, we extend off of Ayer’s characterization of free actions by formulating it in a more concrete, replicable setting, one that is more compatible with our intuitions, and one that may draw critics who discount Ayer’s approach altogether to reconsider the merits of it. In lieu of technological advancements mentioned in an earlier footnote, I think this may be the closest we can get to grasping the compatibilism argument built from “could have, would have.”
表裏一体 