I’m having trouble with the idea of free will and agent causation. If I have free will (in a true libertarian sense), does this mean that my actions are effectively random?

If they are, does this imply that basically everything must either be determined or random and that there is no third middle ground? I suspect that most people who believe in free will would not accept that their actions are effectively random. But then if they’re not random, they would presumably be determined.

From what I can understand, compatibilists believe that free will and determinism are compatible. But here, by “free”, correct me if I’m wrong, they only mean an agent’s ability to do what he wishes in the absence of impediments that would otherwise stand in his way. But this does not imply that those wishes themselves are free and those could still be determined, correct?

  • If they are, does this imply that basically everything must either be determined or random and that there is no third middle ground? - this is a common approach, but not one everyone agrees with. It makes sense to me, personally - I view it as like a determinism-to-randomness spectrum. Things are determined, random, or on the spectrum.
    – TKoL
    Commented Apr 7 at 8:24
  • As for your final paragraph on compatibilism, you're understanding it more or less correctly I believe - there's a famous compatibilist quote, from Schopenhauer. "A man can do what he wants, but not want what he wants"
    – TKoL
    Commented Apr 7 at 8:26

5 Answers 5


You might want to reflect upon the fact that humans do not usually behave randomly, so the answer to a literal reading of your question is straightforwardly no. However, let us suppose that free will involves choices between limited sets of alternatives; it is possible then that the choices and their timing are triggered by processes that contain steps that are chaotic to the point of being random. For example, if you are deciding whether to stop home and watch tv or go for a walk, there might be some kind of probabilistic process at work in your head that leads to the choice.


Short answer: Yes.

To an omniscient observer other than yourself, your actions are always either deterministic or random. To see what I mean, suppose you are offered a choice between A and B. There are two possibilities: Your choice is determined by your pre-existing desires and perceptions OR you generate the choice from nothing without basing it on pre-existing desires and perceptions. (The latter is libertarian free will, though according to metaphysical libertarianism, not every choice need play out that way. Libertarianism only asserts that some choices are not dependent on pre-existing conditions.) The omniscient observer knows your pre-existing desires and perceptions, so in the first case, he knows what you will choose before you choose it, and your choice is predetermined. In the second case, the omniscient observer does not know what you will choose, so to him it is indistinguishable from random. If you decided by coin toss (say a non-deterministic quantum coin toss), to the omniscient observer, there is no difference. In either case, he cannot know the choice in advance, nor can he tell the reason for the choice after it has been made.

The only difference between the coin-toss choice and a libertarian free choice is that in the latter case, you have the conscious experience of choosing A or B, while in the former, the coin (presumably) does not have that experience. Thus, the difference between random and free will is how it is experienced, not how it looks from the outside.

Your understanding of compatibilism is correct. According to compatibilism, the difference between a robot and a free agent is similar to the difference between a coin and a free agent - it is how the agent itself experiences the choice, and not how it appears to an omniscient observer. A robot does not experience making a choice of A or B, but the free agent does, even though (for the compatibilist) this choice is completely determined by what the free agent wants and how it perceives the situation (in very much the same way that a robot's choice is completely determined by its programming).

Addendum on Probability: Probability is simply a mathematical construct, which may be applied to real world scenarios that are completely deterministic (for example, a coin-toss or a die-roll), or to scenarios that are not (like quantum mechanics). Using probabilities or the mathematical concept of randomness to model something is not a statement on determinism. In this example of choosing A or B, the libertarian free will and the quantum coin toss have the exact same mathematical model, so they are empirically indistinguishable.


Free will is actually the very opposite of randomness. They are both excluded from determinism.

Philosophically random refers to everything that is not deliberately selected, decided or adjusted. Accidental and unintentional are synonyms.

Probabilistic events are partially random, as the cause does not determine the effect completely. We know the cause of radioactive decay, but the time of decay of an individual nucleus is unpredictable.

Freely willed actions are deliberately decided, they serve a purpose, they follow a plan.

True randomness is the inaccuracy between a cause and its effect.

Free will is the ability to self-cause voluntary actions in order to achieve a goal in the future. Agent causation thus inserts new causes to the causal flow of events.


It is semantics of course, but I would say arbitrary rather than random. A person in the presence of fire may do nothing, reach for an extinguisher, or reach for gasoline. Of the thousands of both rational and arbitrary actions, they will in some respect be related to the presence of fire. This is because consciousness is forever bound to the unconscious. But it is unlikely that their pseudo-unpredictable response would be truly random. I do submit that the strongest evidence of free will is the ability to engage in arbitrary thought or actions. Purple horse. Just thought I would throw that in... Arbitrarily.


No. Peter van Inwagen in his "Metaphysics" has convincingly shown that neither determinism, nor randomness are compatible with the free will.

He thus makes a flawed conclusion that free will is impossible. This conclusion is flawed because he does not consider the other options.

It is well possible that the most complete physical description of our universe is not stochastic (random or deterministic).

Thomas Breuer had shown that neither deterministic, nor stochastic physical theory can be universally valid, or in other words, cannot descride a system in which the observer is properly contained due to self-reference.

In simple words, this is a mathematical argument for non-self-predictability. No physical theory (even a future one) allows the observer to stochastically predict his own future or know his own past.

This basically means that from the physical point of view the most complete physical theory cannot be stochastic, instead it should deal with Knightian uncertainties, or imprecise probability.

This means that there are events whose probabilities are not determined by the physical states. In other words, events without physical cause.

This opens possibility that non-physical causes could include depending on interpretation, free will, input from the outside of the universe, God's will or fine-tuned initial conditions of the universe.

  • I’m not sure how this results in a possibility that isn’t stochastic or determined. There is no defined probability for a single case event anyways. This doesn’t mean that it is neither determined or random. Commented Apr 7 at 4:41
  • @Mikhail most of physical theories so far (including quantum mechanics) are stochastic: they can predict the state of a system in the future probabilistically, given the state at present. Breuer has shown that such theories are not universally valid, that is, cannot be applied to the system in which the observer it properly contained.
    – Anixx
    Commented Apr 7 at 4:44
  • If the universe is fundamentally unpredictable, then it is random. If it is not, it is determined. What is the third option? Commented Apr 7 at 4:45
  • @Mikhail this means, any universally-valid theory should be adble to do predictions of weaker strength than probabilistic. For instance, such theory should be possibilistic (en.wikipedia.org/wiki/Possibility_theory) or otherwise mathematically account for uncertain probability.
    – Anixx
    Commented Apr 7 at 4:47
  • @Mikhail if the universe is completely random, it is stochastically predictable: we can predict the probabilities. But the third option is that even the probabilities are not predictable. en.wikipedia.org/wiki/Knightian_uncertainty
    – Anixx
    Commented Apr 7 at 4:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .