What is free will, and how to make a system that has it?

Updated: Jun 5



Free will, which can be defined as the ability of an individual to choose or act independently, has always been a topic that is aloof from ordinary lives. Well, maybe we have free will; maybe we don’t. It just doesn’t matter.


With the current development of artificial intelligence, however, such philosophical inquiry starts to attract some attention from practical fields. For instance, should a self-driving vehicle take responsibility for its own car accident (that is, assuming that it can somehow)? If the vehicle said cannot make independent decisions or actions (i.e., does not have free will), then the answer to the question should be no. Nonetheless, if the vehicle does have such independency, then the answer may become yes.


Free will is closely related to self-aware as well, and some believe that a self-aware robot is a threat to mankind. Therefore, it will also be helpful to know that whether it’s possible to build a machine with free will or not. But before that can happen, we must have a deeper understanding of the phenomenon first, obviously.


In this article, we’ll begin by resolving a common misunderstanding about free will. Then, the elements that in our opinion plays an important role in free will are going to be discussed. Finally, a scale of free will is proposed, so that one can better assess how independent a system really is.



Free will does not originate from randomness


For many people, free will and determinism are two completely opposite ideas (such point of view is sometime known as “incompatibilism”). And from there, some people further links free will with the concept of randomness and quantum mechanics (in which everything is discussed through probability instead of certainty). The logic behind such train of thought is simple: The fact that a system’s actions can be predicted after knowing all the relevant variables in the initial state implies that the system’s fate has already been determined since the beginning; and in order to break free from the “determined destiny”, one must resort to some “randomness” to make the consequences less predictable.


Unfortunately, such logic is indeed unsound once being carefully examined, and here is the problem: Free will is about “control”. In other words, if you have free will, it means that you can “control” your own actions or fate. However, “randomness” is about “unpredictable”. And if part of a system cannot be predicted by anyone (including the system itself), the system is considered to be “out of control”, which is a state conflicting with “controlled” state associated with free will. In other words, making the fate of a system undeterminable does not give the “controllership” back to the system, and therefore randomness cannot be the origin of free will.


To better explain the contention above, let’s consider the following example. A man wearing a space suit is dropped from midair. And in this case, we presume that there is no air at all in the place where the man is dropped. Under such conditions, once we know the spot that the man starts his falling, we’ll have no problem to predict where he is going to land. Therefore, the man’s end result is determined, and he obviously has no control to his fate (i.e., no free will).


Now, let’s put the air back to the atmosphere and make the following two assumptions: (1) the weather is windy, and (2) the directions of the winds are unpredictable. In this scenario, the falling process of the man is going to be affected by the random wind, and therefore his landing venue will no longer be determinable. However, does it mean that the man now has control over his fate? The answer, of course, is a resounding “no”.



The key elements of free will


So, if randomness does not generate free will, what do then? Don’t worry. We’re going to answer this question in this section. But before we do, please check out the following diagram, which is a simplified scheme about a hypothesis of how our brain receives input and make corresponding output. We will base our discussion on this diagram:


Figure 1. Our brain will first turn external inputs and inner information into a “goal” through some cerebral mechanisms, and such goal will then guide us to form and execute output.

As you can see, the output of the brain is shaped by two kinds of causes: the first one is exogenous, while the second is endogenous. The exogenous part is marked “external input” in Figure 1, which includes command(s), instruction(s), suggestion(s), information, and so on coming from outside. The endogenous part, on the other hand, consists of two elements: the “inner information (i.e., our memories and thoughts)” and “inner mechanisms (i.e., the neural mechanisms capable of transforming all sorts of input and information into a goal and a final output)”.


It’s also worth mentioning that the brain does not directly transform external inputs and inner information into its output. Instead, it’ll first establish a “goal”, and then forming and carrying out its output accordingly. For instance, if you feel hungry (which can be considered as an endogenous signal), you can’t go eating immediately without first thinking about questions such as “what to eat” and “how to eat it”. In other words, before you actually do something, you must plan out your action(s) first, and such plan is exactly what we’re referring to with the term “goal”.


Besides the above elements, there is a special thing labeled the “sense of control” in Figure 1. This sense created by our cerebrum is a kind of “conscious experience”, aka “qualia”, which is similar to our sense of vision or hearing but more subtle (i.e., vision and audition are derived from real physical signals, while “sense of control” does not seem to have such physical correspondent). Since conscious experience is still a mystery for mankind, we actually have very little to say about it. Nonetheless, please note that it plays a very crucial role in free will because such sense is why we are able to “realize” that we have free will (as we know, by definition, we can only be aware of the things that we have conscious experience of), and such fact suggests that our sense of control may be highly correlated to (at least part of) the inner mechanisms said, which are responsible for constructing our goal and output, under normal circumstances (i.e., the sense of control inevitably appears while these neural mechanisms are operating).


With the discussion above in mind, we now have the following claim: The key elements pertain to our free will are the endogenous causes (including the inner information and mechanisms) and our inner goal shown in Figure 1. That is to say, the less exogenous factors are related to a system’s goal and output, the “freer” that output is, which we think is a very reasonable outlook regarding free will (since no exogenous factor means no external manipulation is affecting the system). Note that minimizing the exogenous factors may correspondingly increase one’s sense of control, in which an individual may feel more responsible for his or her own decision or action. However, this is not always the case considering that one’s sense of control can be tampered by certain measures (such as hypnosis), and therefore we do not consider it to be a key element of free will in our discussion.


There are several things to notice here. First, according to the above definition, a system with free will can be determinable. To put it otherwise, determinism and free will are not incompatible with each other. That is because the system must depend on some sort of mechanism to produce output (i.e., decisions, actions, etc.). And since a mechanism is involved, one should be able to predict the outcome theoretically once the mechanism is known (although such prediction can be very difficult, if not impossible, in reality since the mechanism said can be overwhelmingly complicated, like the three-body problem).


Second, free will is not an all-or-none property; instead, there should be a continuous scale for evaluating how “free” a system actually is, and that is exactly the topic for our next section.



The continuous level of free will


As mentioned before, in our opinion, “free will” should be a continuous phenomenon rather than a have-it-or-not property, and the level of free will should be estimated by the following two factors.


The first factor is the relatedness between the “external input” and one’s inner “goal” – the higher such relatedness is, the less “free will” the system is expressing. For instance, if an individual is told to open a specific door and think about doing exactly so, he or she is demonstrating a very low level of free will. On the other hand, if the individual decides to open a door different than the designated one upon receiving the same command, his or her level of free will should be considered higher than in the previous case.


Note that the “relatedness” aforementioned is actually affected by two other variables, which are also continuous. These variables are:


  • How detailed is the exogenous instruction?


The vaguer an external command is, the more details that the individual who receives the command must decide. For example, the command “open it” contains very little information. Open what? How to open it? When to open it? All the above details are unclear and must be decided by the command receiver, and therefore there is a large room for the endogenous causes to take part. Ergo the hazier the exogenous input is, the higher the level of free will of a system should be.


  • How unrelated is the involved inner information to the external world?


Note that some inner information coming from our memory is indeed not very “endogenous”. For instance, somebody tell you to open a specific window; you remember this command and execute it one week later when you suddenly recall the command. In such case, although the command comes into your mind from within (i.e., your memory), it is actually highly related to the external world, and therefore your level of free will should be considered rather low.


A way to break the link between the inner memory and external world is to process the memory and turn it into your own inner “thought”. For instance, by combining “ice cream” and “coffee”, one has invented a new product called “float coffee”. Note that since “float coffee” did not exist before the individual invented it, the idea was considered less exogenous at least for that very individual at that time. Thus, we should get the following conclusion: the more complex the internal processing we apply to our inner information, the more endogenous it will be.



The second factor is about how random the inner mechanisms are. As explained in the first section of the article, randomness is not a source of free will. In fact, it should be perceived as an opposite since it’s related to “out of control”. So, the more a system’s goal and output are based on randomness, the lower the system is on the scale of free will.



Footnote: Neurozo Innovation provides viewpoints, knowledge and strategies to help you succeed in your quest. If you have any question for us, please feel free to leave a comment below, or e-mail us at: neville@neurozo-innovation.com. For more articles like this, please join our free membership. Thank you very much for your time, and we wish you a wonderful day!

110 views