Compatibilism is the position that freewill and physical determinism are compatible.

This is in opposition to the idea that since everything has a cause, any act or decision a person takes has already been predetermined by an antecedent cause, and that causal relationship is dictated by the laws of physics. Hence nobody has freewill since causality and physics determine what our actions are, and we don't have the freedom to do otherwise.

This position is problematic, since it means that we don't have moral responsibility for our actions.

Compatibilists argue that this is a false dilemma. The problem is not whether we are free to act against the laws of causality and physics, but whether we are free to act according to our own motivations or not. As long as we are not being coerced into acting by outside factors and are doing so solely according to our own desires and wishes, then we have freewill.

Incompatibilists, conversely, argue that people have freewill only if they are free to act against the laws of causality and determinism. That given the exact same initial conditions, a person with freewill can choose to take different courses of action. This is called metaphysical freewill, as opposed to compatibilist freewill.

Now consider the 2 following scenarios:

  • A person is raised with a certain number of beliefs, and is sheltered from any other beliefs that contradict those they were raised with. That person's parents and community have indoctrinated them very well, and as an adult that person makes all of their decisions according to that belief system, even though they live in a open society that would have allowed them to do whatever they want. Per compatibilism, this person has freewill, since they are acting according to their own motivations regardless of how they got those motivations.
  • A intelligent robot (one that passes the Turing test) is programmed with a set of rules of behavior. It is then left to function autonomously in its environment, without any further input or manipulation from its programmers. Since it is well programmed and doesn't have any bugs, it always functions according to its original programming, even though it is autonomous and not coerced by any outside agents. By analogy with the above example, the robot has freewill, since it is acting according to its own motivations (i.e. its own internal rules).

My questions:

  1. From a compatibilist point of view, what is the difference between the person's motivations which they inherited from their parents' belief system and the robot's behavior rules inherited from its programmers?
  2. Would a compatibilist consider the robot to have free will - i.e. is my analogy correct ?
  3. Does the fact that humans can and do act against their original indoctrination, as opposed to robots who would always act according to their programs constitute an argument for metaphysical freewill?
  • This has been asked already philosophy.stackexchange.com/questions/3550/…, and the answer depends on which compatibilist you ask. – Conifold Jun 18 '15 at 3:37
  • what might such a robot answer when asked about free will? – nir Jun 18 '15 at 17:33
  • @Conifold This is almost a duplicate of the one you linked to, except for the 3rd question I posed. Does that warrant keeping this open? – Alexander S King Jun 18 '15 at 17:52
  • 1
    Depends on the answers, I guess. Since you are interested in mind-body problem and free will consider Davidson's anomalous monism. Mental events are physical events, but there are no laws correlating mental and physical, he makes it work en.wikipedia.org/wiki/Anomalous_monism See Davidson on free will here lightforcenetwork.com/sites/default/files/… – Conifold Jun 18 '15 at 20:11
  • Re "free to act against the laws of causality and determinism", that's a contradiction in terms. Thus one can dismiss the associated "incompatibilists" definition of "free will": essentially those people, if they exist, have chosen to define the term as something supernatural, something not meaningful. But that doesn't mean that there is only one alternative, and indeed as I understand the presented set of views, just "incompatibilists" and "compatibilists", that set is not exhaustive. The latter seem to define "free will" as something absolute. Lacking there is both relativism and degrees. – Cheers and hth. - Alf Jun 19 '15 at 22:53

An answer depends on how such a robot functions; one can think of various ways to construct a robot; for example it could be programmed to follow instructions such as if it rains outside take an umbrella; or it could be constructed to learn how to respond to its environment in an unsupervised manner, for example a neural network system which learns to play breakout https://www.youtube.com/watch?v=cjpEIotvwFY; in the former case it is clear that the system has no free will, and it may even be able to introspect and report that it has no free will on this matter; but in the later case the answer to the question is not so clear.

Since the former case is not interesting, lets consider the later, and suppose for the sake of argument that such a robot can pass the Turing test as a human; therefore the answers to your questions would be:

1) The robot may be lacking "behavior rules inherited from its programmers", and therefore the question may be irrelevant.

we can assume the robot is simulable by a Turing Machine, and that therefore it may be argued that its state evolves according to programmed rules, but not anywhere close to the sense of "If A happens then do B"; for example, in the breakout game video no one programmed the network to wait on the left side of the screen while the ball is bouncing behind the wall.

3) you write that humans "can and do act against their original indoctrination" and that robots "always act according to their programs", but if you imagine a robot consisting of 90B artificial neurons rather than a pre-programmed set of instructions, then the distinction between such a robot and a human is not so clear, and one can argue that the above descriptions are interchangeable; for example, that humans can be said to sort of follow a very complicated "program" and that robots seem to be able to act against their indoctrination, since it can be said their indoctrination is simply what they have learnt from their past interactions with their environment, and as a result there is no trivial function from situation to response.

2) indeed, it is not clear why one would ascribe free will to a system of 100 artificial neurons playing breakout; but you have postulated a system complicated enough to pass for a human in a Turing test; suppose such a system is made of 90B artificial neurons, which have learnt to respond in a way which seems appropriate for a human; assuming this system does not follow simple pre-programmed rules, then it is not obviously clear to me why a compatibilist should deny such a system has free will.

it might be fun to imagine a chat with such a machine on the subject of free will:

H: do you know what free will means?

R: yeah, it's that belief that one has a freedom to choose how to act in a given situation.

H: do you have free will?

R: well, knowing how I am constructed, I'd say that I can reasonably rule out libertarian free will; but I believe in compatibilist free will


A physical compatibility must use very precise definitions of "freewill" and "morality" in their philosophy. The position is unforgiving for sloppy definitions.

One challenge faced here is that "freewill" is often blindly assumed to mean "the ability of an agent to do whatever it wants." This has never been an effective model. For a test of why this fails so spectacularly, break into a bank after dark with no equipment or safe cracking training, and use your freewill to will yourself inside the safe. Clearly there are bounds to freewill. If nothing else, there are physical limitations.

An example of a definition which can hold up to such complexities is one where you isolate a system. A bounding volume can suffice. There are physical properties on the surface of the bounding volume. One could define the freewill of that system to be the ability to create behaviors along this boundary which are not predictable using only measurements outside of the volume (in human terms: if the movement of my arms is not 100% predictable using measurements outside my body, then I can claim to have freewill). Obviously there are limits to this. You know that the probability of my arms growing 10 feet long is very low, but you should get the gist of it.

As it turns out, with chaos theory, there are valid arguments that such a system can be unpredictable so long as the system is not driven into a stable state (a bullet to the brain pretty much ends freewill because we become predictable at that point, as we fall to the ground).

As for morality, this is where I find physical compatibility so alluring. If one claims that there is a valid model for an agent with freewill (such as the chaotic box from the example above), then the claim is that it is valid to model the behaviors of the universe either using traditional definitions of freewill or a physically deterministic process.

Thus concepts like morality are completely valid to a physical compatibilist because they start from the presumption that a physical construct can be created which meets our needs for defining freewill. Thus, any argument which hinges on freewill is valid to them, simply because they assert that a physical equivalent exists.

As for the computer example, if the physical compatibalist declares it to have freewill, by whatever definition they find valid, they find that it is valid to model that computer as though it had a mind with freewill, and is thus bound by morality exactly so far as we are bound by it.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.