FreeWill - Principle of Alternative Possibilities

COULD HAVE DONE OTHERWISE?
One of the many claims made by incompatibilists, either it be libertarians or hard determinists, is that if determinism is true, then we could not have done otherwise, and if we could not have done otherwise then we can't be considered free agents, let alone morally responsible for whatever action we do. They argue that given a certain situation, we must have the ability to do something other than what we did. This principle has been titled the "Principle of Alternative Possibilities" (PAP) by the philosopher Harry Frankfurt. [1]

PAP: A person is morally responsible for what he has done only if he could have done otherwise

This principle at first appears very intuitive, so intuitive that some philosophers have been said to consider it an a prior truth. How can someone be blamed for doing something, if they could not have done otherwise? One thing to point out is, what people who argue this 'could have done otherwise' principle mean is, that in the exact same circumstances one could have chosen differently. By exact same circumstances, we mean same the state of affairs – every event in the cosmos. The same stars burning at the same temperature, the same emotional state of the agent, every position and velocity of every object, even every atom, in the cosmos.
However if determinism is true, then we could not have done otherwise in the sense given above. If the universe we live in is a mechanistic machine governed by uniform laws, as the determinist believes, then given the same circumstances, the same outcome will occur. If God cranked the universe's wheel of time back to 3pm the same events that took place at 4pm will occur. So under the assumption of determinism, we could not have done otherwise. This means that if determinism is true, and the PAP principle is true, then the compatibilist conception fails, since the compatibilist conception of freedom is that we are free and morally responsible in a real sense, even under the assumption that determinism is true.
However the Principle of Alternative Possibilities is false. You can be considered free and morally responsible even if you could not have done otherwise. I will also argue that the concept of an agent doing otherwise or choosing differently in the same circumstances is impossible even under the assumption that determinism is false. Not only this, but I will show that even if we assume an agent can be said to 'do' otherwise, this would only render the agent unfree and by extension not morally responsible for the act.

Is PAP true?

So then, can a person be considered free even if the person could not have done otherwise? Harry Frankfurt has argued a person can, and does so with a thought experiment that takes many forms. His exact argument can be found in his original essay, but the variations will do. The thought experiment goes as follows:

"Bob passionately wants to kill the president, and plans to do so when he comes and speaks in his hometown on Monday. A fringe political group discovers this, and has an invested interest in the president being dead. The night before the president's speech, the group kidnaps Bob and plants a neural-disruption chip in his brain. If Bob has second thoughts, and decides against killing the president, the chip will activate and in turn cause him to kill the president. Bob wakes up, and kills the president. The chip never activated."

The question is - did Bob kill the president voluntarily, of his freewill? It seems obvious he did. Would he be considered morally responsible for what he has done? Again, yes. Could he have done otherwise? No. The moment that he would have had second thoughts which would have altered his course of action, the chip would have activated rendering him unable to not kill the president. The mere presence of the chip prevented the possibility of doing otherwise, yet he's considered morally responsible for what he did. Now, if he had second thoughts, and the chip was activated then of course he could not be considered free in his assassination. Now some may object to this thought experiment on the grounds that a chip like that may never exist making it cute at best, but nothing of a real challenge to the truth of PAP. There are more down to earth versions. Consider the following:

"Imagine that while Bob is driving, he sees Paul starting to cross the road ahead of him. Bob decides that he does not want to slow down for Paul, and continues on at the same speed. Paul, who is unable to get out of the way of Bob's car, is run over. As a result, Paul is killed instantly.

Assuming that Bob is a rational and sane human being, it seems clear-cut that he is morally responsible for what he did. Now imagine that, unbeknownst to Bob, the road that he was traveling on was covered in ice. Due to the ice, even if Bob had tried to stop the car from hitting Paul, he would not have been able to and Paul would have been killed anyway. Let us further imagine that Bob does not try to stop the car. It seems that we would still want to hold Bob responsible for what he did even though he would have run Paul over regardless of whether or not he hit his brakes. If Bob had hit his brakes, then one might argue that he would not be morally responsible for what he did. However, the fact of the matter is that Bob did not hit his brakes. It seems, then, that Bob is morally responsible for what he did even though he could not have done otherwise."

It seems that the ability to do otherwise is somewhat irrelevant. It seems that what matters is that the person at hand was aware of what's going on, and acted on his desires. In both cases, these men would be considered morally responsible because they were quite aware of what they were doing, and of course they wanted to do what they did.

The Principle of Alternative Possibilities states that a person is morally responsible for what he has done only if he could have done otherwise. However, this is not necessarily the case as shown by Frankfurt. It seems there may be other conditions that must be met for a person to possess freewill.

Doing otherwise?

As explained above, some argue that 'do otherwise' means 'could have chosen different under the same circumstances.' Not just similar circumstances, but the exact same circumstances, and not similar times but at the same time. This means that when the golfer misses the put and says "Ah, I could have made that" and picks the ball up, placing it in the same position, in order to simulate the circumstances, he's not actually 'doing otherwise' in this sense. All factors must be the same, the same stars burning at the same temperature, the same emotional state of the agent, every position and velocity of every object, even every atom, in the cosmos. This is one of the conditions that must be met for the CD (could done otherwise) principle. Another condition is that "I could" or "you could" have done otherwise. In other words, doing otherwise means choosing otherwise, there must be a connection between the act and the agent.

Let's first sketch what 'doing' looks like (choice making).

Richard is driving to his friend David's house which is out of town. As he drives down the road, he encounters a fork in the road where he has two options. The path on the left (A) is a little faster but has an 'under construction' sign, whereas the path on the right (B) takes longer but is construction free. As Richard drives along he thinks back on his last trip to David's where he decided to take the faster path, but as a consequence ran over a large screw which resulted in a flat tire. Richard then decides to take the longer, but safer, path.

This is simple, but what actually happened? What events took place? How can this event be broken down? First we have Richard the agent. An agent is a being with intentions and goals. Not only is he an agent, but he is a person. Richard is the totality of his character, beliefs and desires.
Second Richard's driving is for the sake of some goal, namely to arrive at his friends. This means all his actions are corresponding to a desired end. Richard encounters the fork in the road. Data from the world travels to Richard's eyes and eventually to his mind. After this happens an extremely fast and complicated set of processes begin. After the light travels to his Occipital Lobe (the part of the mind that deals with vision), each object is recognized (road, sign etc) and associated with certain meanings. With the available data he then reasons on what action corresponds best to the goal. Since there are two options that best correspond to the goal, he then produces a virtual future for each course of action where he reasons what each action will entail. He taps into his memory of the last visit and remembers what occurred when path A was taken. Path (A) entails he will reach his desired goal faster, but with the risk of getting a flat. Path (B) entails him reaching his desired goal slower, but with the benefit of not getting a flat. With the two virtual futures in his mind, he then selects which of those virtual futures he wants to make an actual future. After all of this is reasoned out, he makes the choice of taking path B.
A voluntary action (doing) can be therefore reduced to: An agent receives input from the world, examines acquired data, reasons through what possible actions entail (ie: producing virtual futures), selects which virtual future best corresponds to the goal, and eventually acts rendering the virtual future an actual future.
The fundamental aspect of this process that makes the act Richard's 'doing' and not the act of someone else, or something else is that there is a real connection between which future is actualized and Richard himself. Which future is actualized is determined by Richard's character along with his goals.
Now lets say 10 minutes after Richard chose path B, we were to 'rewind' time so to speak to the same time in the same circumstances. If determinism is true, then the same outcome of course will occur. All variables will be the same, Richard's emotional state, what's in his field of vision, the position of his head etc. Richard with the same goals, same state of mind, will of course choose path B. What conditions are required for him to do otherwise in the same circumstances? First let's consider what conditions are required for not him doing otherwise, but just the otherwise part – the part where something other or something different can occur at time t under the same state of affairs.
If the cosmos is uniform in its behaviors governed by laws that in some sense are eternal and unchanging, then determinism is a logical consequence. The only condition that must be the case for determinism to be false is to invoke random events. Random being either uncaused or in some sense 'in principle unpredictable' events. This appears to be the only option, since if the cosmos is uniform in it's behaviors, the only way a different outcome is possible is if you change one of the variables, however once some variable is changed it's no longer the exact same circumstances, but a variation of those circumstances. Imagine we rewind time to time t (time t being a specific time), and let everything play out. If you are to allow for some kind of random or uncaused event, imagine say 10 atoms freak out and change their course, then a different outcome is possible.
Let us now apply this to the aspect of doing otherwise. Again PAP states that a person can only be considered free or morally responsible if he could have done otherwise in the same circumstances. The important part is not whether a different outcome is possible at time t but that an outcome is different by virtue of the agent doing. In our example with Richard picking a path, we have many steps of the way. We have Richard receiving input from the world, examining acquired data, reasoning through what possible actions entail (ie: producing virtual futures), selecting which virtual future best corresponds to the goal, and eventually acts rendering the virtual future an actual future. The part that can be said to be 'the act of doing' is when Richard selects the virtual, or imagined, futures causing his future actions (moving his arms, etc.) to be in correspondence to the goal. This of course can be further reduced to events going on inside the brain. Richard's character, his beliefs, desires, memories, etc. all are stored in his brain, and play a role in which path he chooses. All of this exists by virtue of a huge nexus of neurons communicating with each other via electric impulse.

If we are to imagine what is required to change which course of action Richard took at time t, via the use of a random/uncaused event, then we'll no doubt discover it to be some event in the brain. One of the events in the long chain of events must be 'randomized' for him to do otherwise.

The main events are: (1) an agent receives input from the world, (2) examines acquired data, (3) reasons through what possible actions entail (ie: producing virtual futures), (4) selects which virtual future best corresponds to the goal, and eventually acts rendering the virtual future an actual future. The red represents mentally produced events, or mental events.

Richard is at the fork in the road, and all the events listed above are taking place. Now imagine a few of the signals traveling in Richard's brain, say from the part of his mind that contains memories to another part, are disrupted due to a random event causing them to travel to another part of his mind, or to stop traveling altogether. (Keep in mind this is exactly what is required for PAP to work) Richard then ends up still picking path B but turning his steering wheel one inch more to the right than before at time t. Although in a technical sense he did do otherwise since Richard did something different in the same circumstances, but what most supporters of PAP really mean is that a person can make an entirely different choice.

What would it take for Richard to choose path A, or another path? In our thought experiment above we imagined that a few signals were disrupted, but it's been estimated that the human brain contains between eighty five - one hundred billion neurons (85,000,000,000 – 100,000,000,000) so we'll certainly need more than a few connections to be disrupted for an entirely different 'choice' to come out. Let's assume that Richard made the choice on his own, without the intervention of indeterminism, puts 12 billion neurons to use. It seems that to alter the path Richard chooses, we'll have to disrupt at least half of those connections.

The problem here is, random or uncaused events can not be attributed to Richard. If during Richard's reasoning process, a random indeterministic event occurs, resulting in a different outcome from time t, then it no longer is true the choice is the result of Richard. You could not attribute the doing to Richard since the doing is the result of random events. This problem that arises when you attempt to invoke indeterminism to save freewill has been pointed out by the philosopher Daniel Dennett responding to the philosopher Robert Kane:

"The problem is compounded for Kane, since he has to figure out some way to get the undetermined quantum event to be not just in you but yours. He wants above all for the decision to be "up to you," but if the decision is undetermined - the defining requirement of libertarianism - it isn't determined by you, whatever you are, because it isn't determined by anything. Whatever you are, you can't influence the undetermined event - the whole point of quantum indeterminacy is that such quantum events are not influenced by anything. [...] But in order to do this, there has to be more to you than just some mathematical point, you have to be someone; you have to have thoughts - memories, plans, beliefs, and desires - that you've acquired along the way..."

Daniel Dennett - Freedom Evolves: pg – 123

The Principle of could done otherwise states that we can only be free if we could have done otherwise, however for there to even be a 'do otherwise' there must be a do-er responsible for the otherwise – as we have seen the only possible condition for 'otherwise' is to invoke random or uncaused events. When random or uncaused events are introduced then the link between the do-er and the otherwise is lost. Statements such as "Richard choose path B" presuppose that there's a connection between the agent and the choice, there must be a causal connection, a real connection, between the agent and the choice otherwise the agent is not responsible for the doing. So on closer examination the statement "You're only free if you could have done otherwise" in the senses listed at the top of the chapter entails a contradiction, since one can only be considered a the 'do-er' if what was 'done' was directly linked to the 'do-er' and not the result of random events.

Thus, PAP entails the impossible. It's impossible in short because for an agent to 'do' is only possible under determinism, and 'otherwise' (in the senses above) is only possible under indeterminism.

Don't blame me, I did otherwise!
So far I have argued that not only is PAP false, but entails somewhat of a contradiction. Here I will argue that even if we grant 'doing otherwise' is possible, it would only render the agent not responsible for the action. As said above, if the action is the result of random events this severs the connection between agent and choice. A consequence of this is that the agent can't possibly be held responsible for what occurred. An interesting thought experiment has been published by the historian Richard Carrier. Although the quote is pertaining to libertarian freewill, it touches on the PAP principle in the sense of doing otherwise in the exact same circumstances.

"Image two parallel universes, identical in every detail, and imagine a man in each universe, identical in their character, knowledge, desires and everything else, standing in totally identical circumstances. Now imagine that one of these men chooses to kill his wife, but the other man chooses not to. What could possibly explain this? Since the two situations and the two men are identical in every respect, there can be no cause whatsoever for either man's choice. This is what Moreland's theory entails.

But this has an unacceptable consequence. For it means that neither these men's desires, nor their knowledge, nor their moral character - nothing at all - can be blamed for having caused their choice. Moreland even agrees: "no description of our desires, beliefs, character, or other aspects of our makeup and no description of the universe prior to and at the moment of our choice...is sufficient to entail that we did it" (138-9). But this means that we could not even say that the first man was evil and the second good, since that would assume the first man's badness caused him to kill, while the other man's goodness caused him to refrain. But these men are identical, so one cannot be evil and the other good. Moreland might say he is evil or good after the deed, but that means we could not say he did what he did because he was a good or bad man. In fact, we could not say at all why he acted. What quality in either man that is uniquely a part of "him" can be blamed for causing his particular choice? There is none.

Now imagine that this man is you, and in one universe you kill your wife, in the other you do not. What would you think of yourself then? You would know that nothing causes your actions - not your character, nor your environment, nor the surrounding circumstances, nor your knowledge, not even your love of your wife. Nothing. Your choice to kill or refrain is purely a result of happenstance: whichever the universe you are in is a mere luck of the draw. Imagine how you would feel, having learned that it is nothing but the result of unpredictable randomness whether you kill your wife or not at this very moment. Shocking, yes? Imagine that you refrain from killing, but could step into a time machine, run the universe back a million times, and watch yourself again each time, and then saw that sometimes you killed and sometimes you didn't, even though each time all the circumstances including your thoughts, desires, character, everything were the same. There would be no rhyme or reason to why you did this one or the other. It would be a mere shake of the dice. This is the nightmare of a world that Moreland's theory describes." - Richard Carrier - A Defense of Metaphysical Naturalism

If we could do otherwise, it would most probably be like having Tourettes Syndrome. Having no control over our actions, being the victims of the random, and being reduced to uncontrollable random machines. Thomas Pink has mentioned this issue with conceptions of freewill that rely on indeterminism

"There is yet another problem that faces the libertarian. Is libertarian freedom something we could ever exercise, in the only way that freedom ever can be exercised, though what we deliberately do, in genuine and intelligible action? Libertarian freedom seems to reduce what we do to no more than blind, undirected motion – to the equivalent to jerks and spasms." Thomas Pink - Free Will: A Very Short Introduction – pg 83

What do we mean?
Some of those reading this may be convinced in what has just been said, and some of you may already be familiar with the topic and had been convinced long ago – yet some of us will continue to say 'sorry, I could have done it but…' or 'I could have done otherwise' casually in our daily lives. It rolls of the tongue, and we mean something when we say it. What do we actually mean? Sometimes language can be a difficult thing to get around in discussions of philosophy, and getting a clear understanding on what words mean, and what we mean should be an important goal when we consider certain ideas. The principles that have been talked about above have dealt with doing something different in the same circumstances at the same time, yet on closer examination we discover that in practice we run into trouble, yet we will continue use that kind of language.
Is it possible we mean something incoherent? An interesting example of this might be the idea of changing the future. You'll hear politicians say it all the time – "Let's change the future!" – change the future? From what to what? The future is what will happen, what are you going to change? That may not be a good example, since one just a moment of reflection we discover that what people are really getting at is the idea of changing the anticipated future. If things are to continue as they are now, then we ought to expect X, thus changing how things are now will alter what we are to expect. Another possible example is the idea of having a mind - or more specifically the statement "I have a brain." Our scientific understanding of the mind is that it is the manifestation of a kind of brain. The self in a way resides in the mind, which is the product of the brain. So to say "I have a brain" is somewhat like saying "my brain has a brain." Some have even argued that we have a cognitive dispassion to divide the world into two kinds of things – minds and objects. So notions of a disembodied self come so natural that we find our self speaking in a way that presupposes the truth of dualism. Another possibility, one that I have speculated on, is that since the great majority of our ancestors held certain religious or superstitious views, our language and way of speaking naturally reflects that.
What I think we may mean when we say such things is not that we could have done differently given the exact circumstances, but rather, we could have done otherwise if we had known better. We could have done otherwise if we had chosen differently. That's utterly uncontroversial, since it's quite obvious that different outcomes are entail by different circumstances. So could it be that all we're really saying is "If I knew what I know now, I would have done differently" or "If I had chosen differently, I would have done differently" – It's hard to say.
1. Harry Frankfurt, "Alternative Possibilities and Moral Responsibility," in Free Will, ed.
Gary Watson ( Oxford Univerity Press, 2003), 167.
2. Shariff, G.A. 1953. Cell counts in the primate cerebral cortex. J. Comp. Neurol. 98:381–400

By: David Campbell
Originally written: Sunday, June 10, 2007