Despite approaching the problem from several different directions, he found it so intractable, so resistant to illuminating solution, that he was forced to conclude "No one of the approaches turns out to be fully satisfactory, nor indeed do all together."

Nozick admits that "Over the years I have spent more time thinking about the problem of free will — it felt like banging my head against it — than about any other philosophical topic except perhaps the foundations of ethics."

van Inwagen's Consequence Argument

Philosophers often treat the topic of free will as a problem about punishment and responsibility: how can we punish someone or hold him responsible for an action if his doing it was causally determined, eventually by factors originating before his birth, and hence outside his control? However, my interest in the question of free will does not stem from wanting to be able legitimately to punish others, to hold them responsible, or even to be held responsible myself.

here is the standard determinist objection to free will

Without free will, we seem diminished, merely the playthings of external forces. How, then, can we maintain an exalted view of ourselves? Determinism seems to undercut human dignity, it seems to undermine our value.

Our concern is to formulate a view of how we (sometimes) act so that if we act that way our value is not threatened, our stature is not diminished. The philosophical discussion focusing upon issues of punishment and responsibility, therefore, strikes one as askew, as concerned with a side issue, although admittedly an important one.

The task is to formulate a conception of human action that leaves agents valuable; but what is the problem? First, that determinism seems incompatible with such a conception; if our actions stem from causes before our birth, then we are not the originators of our acts and so are less valuable. (We shall look later at what assumptions about value underlie this reasoning.) There is an incompatibility or at least a tension between free will and determinism, raising the question: given that our actions are causally determined, how is free will possible?

and here is the standard randomness objection

Some would deny what this question accepts as given, and save free will by denying determinism of (some) actions. Yet if an uncaused action is a random happening, then this no more comports with human value than does determinism. Random acts and caused acts alike seem to leave us not as the valuable originators of action but as an arena, a place where things happen, whether through earlier causes or spontaneously.

Clearly, if our actions were random, like the time of radioactive decay of uranium 238 emitting an alpha particle, their being thus undetermined would be insufficient to ground human value or provide a basis for responsibility and punishment. Even the denier of determinism therefore needs to produce a positive account of free action. On his view, a free action is an undetermined one with something more. The problem is to produce a coherent account of that something more. Once that account is formulated, we might find it does all the work, and that it is compatible with determinism and sufficient for our value purposes; in that case, the something more would become the whole of the account of free will.

How is free will possible? Given the tension between causal determination and randomness on the one hand, and valuable agent-hood on the other, how is valuable agenthood possible? The problem is so intractable, so resistant to illuminating solution, that we shall have to approach it from several different directions. No one of the approaches turns out to be fully satisfactory, nor indeed do all together.

One line of approach motivated the previous chapter's investigation of knowledge. Couldn't action be connected to something exactly as belief is to the truth when it constitutes knowledge? We want our beliefs to track facts, and the desirability of this is not undercut (but indeed may be aided) by our belief's being caused in a certain way. Wouldn't it be similarly desirable if our actions also were connected to something, tracking it? Might not the causation of action too aid rather than undercut this tracking and so contribute to the desirable mode of action or at least be compatible with it? If this is to be plausible, what the action tracks will have to be some evaluative fact or characteristic. We shall pursue this line of thought to (see to what extent we can) defang determinism without denying it, in the second part of this chapter. In the third and last part, we shall investigate issues of punishment (despite their not being central to the free will problem); in particular, we will formulate a rationale underlying punishment in retribution for a wrong.

First, however, we shall try to delineate an indeterminist view of free will. If some such view could be made to work, we would welcome it most. Seeing its difficulties might prepare us to view what the second part lays out, although causally determined and hence "second best", still as desirable and a form of free will nonetheless, the best instantiated realization of it. On the other hand, if the parts of this chapter came in a different order so that we first saw the inadequacies of the determinist picture, this might induce more tolerance of the indeterminist account, difficulties and all. Parts I and II of this chapter, it is only fair to warn the reader, contain more thrashing about than any other chapter of this book. Over the years I have spent more time thinking about the problem of free will—it felt like banging my head against it—than about any other philosophical topic except perhaps the foundations of ethics. Fresh ideas would come frequently, soon afterwards to curdle. (This is especially evident in Part II.) The presentation of many of these ideas and approaches may spur the reader to a success that has eluded me, or at least lift her spirits temporarily on this most frustrating and unyielding of problems. We approach the issue of free will from many directions. If we cannot solve the problem, at least we can surround it.

Making some choices feels like this. There are various reasons for and against doing each of the alternative actions or courses of action one is considering, and it seems and feels as if one could do any one of them. In considering the reasons, mulling them over, one arrives at a view of which reasons are more important, which ones have more weight. One decides which reasons to act on; or one may decide to act on none of them but to seek instead a new alternative since none previously considered was satisfactory.

After the choice, however, others will say we were caused to act by the considerations which were (or turned out to be) more weighty. And it is not just others. We too, in looking back at our past actions, will see which reasons swayed us and will view (accepting) those considerations as having caused us to act as we did. Had we done the other act, though, acting on the opposing considerations, we (along with the others) would have described those considerations as causing us to do that other act. Whichever act we do, the (different) background considerations exist which can be raised to causal status. Which considerations will be so raised depends upon which act we do. Does the act merely show which of the considerations was the weightier cause, or does the decision make one of them weightier?

Nozick describes an adequately determined will

The reasons do not come with previously given precisely specified weights; the decision process is not one of discovering such precise weights but of assigning them. The process not only weighs reasons, if (also) weights them. At least, so it sometimes feels. This process of weighting may focus narrowly, or involve considering or deciding what sort of person one wishes to be, what sort of life one wishes to lead.

What picture of choice emerges if we take seriously the feeling dial the (precise) weights to be assigned to reasons is "up to us"? It is causally undetermined (by prior factors) which of the acts we will decide to do. It may be causally determined that certain reasons are reasons (in the one direction or the other), but there is no prior causal determination of the precise weight each reason will have in competition with others. Thus, we need not hold that every possible reason is available to every person at every time or historical period. Historians and anthropologists delineate how certain ideas and considerations can be outside the purview of some societies, some of whose reasons would not count as reasons for us. (Yet, there does remain the question of whether an innovator couldn't have recognized as a reason something outside the purview of others in his society.) Psychology, sociobiology, and the various social sciences, on this view, will offer casual explanations of why something is or is not a reason for a person (in a situation). They will not always be able to explain why the reasons get the precise weights they do. Compare the way art historians treat style; not every style is equally available to every artist in every period, yet within a style creative choices are made, and some artistic revolutions introduce new stylistic possibilities.

It is neither necessary nor appropriate, on this view, to say the person's action is uncaused. As the person is deciding, mulling over reasons R_A which are reasons for doing act A and over R_B which are reasons for doing act B, it is undetermined which act he will do. In that very situation, he could do A and he could do B. He decides, let us suppose, to do act A. It then will be true that he was caused to do act A by (accepting) R_A. However, had he decided to do act B, it then would have been R_B that caused him to do B. Whichever he decides upon, A or B, there will be a cause of his doing it, namely R_A or R_B . His action is not (causally) determined, for in that very situation he could have decided differently; if the history of the world had been replayed up until that point, it could have continued with a different action. With regard to his action the person has what has been termed contra-causal freedom—we might better term it contra-deterministic.*

* The notion of contra-causal human freedom (though not the term) originated with Philo. In his view, God, in creating the world, reserved for himself the power to upset laws by working miracles, and gave to man a portion of that same power—although man's 'miracles' are not worked with respect to laws that he himself created. (See Harry A. Wolfson, Philo, Harvard University Press, Cambridge, 1947, Vol. I, pp. 431, 436.) The Epicureans denied causality altogether, and Chrysippus held that causality, by its nature, stopped at the will of man.The Philonic view is the first to place absolute free will within a world of some causality which otherwise would apply and which is suspended. (See H. A. Wolfson, Religious Philosophy, Harvard University Press, Cambridge, 1961, p. 196; The Philosophy of the Kalam, Harvard University Press, Cambridge, 1976, p. 733.) It became a matter of controversy within Christian theology whether humans naturally retained this gift of free will from God, or whether, after Adam's fall, God withdrew it as a matter of course and bestowed it only as a matter of divine grace. (See H. A. Wolfson, "St. Augustine and the Pelagian Controversy" in his Religious Philosophy, pp. 158-176.)

Nozick describes soft causality

Thus, we draw a distinction between an action's being caused, and its being causally determined. Some philosophers would deny this distinction, maintaining that whenever one event causes another, there holds a general law in accordance with which it does so: some specification of the first event (along with other conditions which hold) always is and would be followed by an event of the same type as the second. It is a metaphysical thesis that the root notion of causality, producing or making something happen, can operate only through such lawlike universality. If this were correct, and if a law could not hold only at that (moment of) time, then causality necessarily would involve causal determination: under exactly the same conditions repeated, exactly the same thing would have (again) to happen. According to the view that distinguishes causality from causal determination, an act can be done because of something and have a cause even though in exactly the same conditions another act could have been done. It is common, in retrospect, to see what caused us to act as we did. Although we can retrospectively identify a cause, this does not mean our action was causally determined; had we acted differently in that situation (as we could have) we retrospectively would have identified a different cause — R_B instead of R_A.

Is this conception of decision as bestowing weights coherent? It may help to compare it to the currently orthodox interpretation of quantum mechanics. The purpose of this comparison is not to derive free will from quantum mechanics or to use physical theory to prove free will exists, or even to say that nondeterminism at the quantum level leaves room for free will. Rather, we wish to see whether quantum theory provides an analogue, whether it presents structural possibilities which if instanced at the macro-level of action — this is not implied by micro-quantum theory — would fit the situation we have described. According to the currently orthodox quantum mechanical theory of measurement, as specified by John von Neumann, a quantum mechanical system is in a superposition of states, a probability mixture of states, which changes continuously in accordance with the quantum mechanical equations of motion, and which changes discontinuously via a measurement or observation. Such a measurement "collapses the wave packet", reducing the superposition to a particular state; which state the superposition will reduce to is not predictable." Analogously, a person before decision has reasons without fixed weights; he is in a superposition of (precise) weights, perhaps within certain limits, or a mixed state (which need not be a superposition with fixed probabilities). The process of decision reduces the superposition to one state (or to a set of states corresponding to a comparative ranking of reasons), but it is not predictable or determined to which state of the weights the decision (analogous to a measurement) will reduce the superposition. (Let us leave aside von Neumann's subtle analysis, in Chapter 6, of how any placing of the "cut" between observer and observed is consistent with his account.)

Nozick appears to associate the collapse of the wave function with the decision. This is too extreme. Quantum uncertainty need only generate alternative possibilities for action.

Our point is not to endorse the orthodox account as a correct account of quantum mechanics, only to draw upon its theoretical structure to show our conception of decision is a coherent one. Decision fixes the weights of reasons; it reduces the previously obtaining mixed state or superposition. However, it does not do so at random.

Nozick put forward an analytic theory of epistemology that did not depend on foundational truths that could "justify" true beliefs. Instead, he thought that true beliefs must "track" external "cases" in the world. His theory is thus "externalist" and a form of reliabilism.

Nozick's Four Conditions for S's knowing that P were:

1. P is true.
2. S believes that P.
3. If it were the case that (not-P), S would not believe that P.
4. If it were the case that P, S would believe that P.

1. If P weren’t the case and S were to use M to arrive at a belief whether or not P, then S wouldn’t believe, via M, that P.
2. If P were the case and S were to use M to arrive at a belief whether or not P, then S would believe, via M, that P.

Other epistemologists who are reliabilists include: Frank Ramsey (1927), David M. Armstrong (1973), and Fred Dretske (1989).