This blog is the first in a series that will do a deep dive into what is generally meant by the term ‘autonomy’ as it is applied both to humans and to robots. I will give an overview of the debate around free will and determinism and offer some thoughts about how an immaterial spirit or mind could interact with a physical brain and body to facilitate freedom of choice in a deterministic context. These and other related topics are discussed in more detail my recently published book ‘Rise of the Moral Machine: Exploring Virtue Through a Robot’s Eyes‘
Autonomous Robots
Believe it or not, autonomous robots have been with us for over seventy years. Seventy years! The first robot that was recognised as being autonomous was developed by William Grey Walter in 1948. Walter built two tortoise-like robots which he called Elmer and Elsie. These robots used photocell and contact switch sensors to enable them to navigate their way around obstacles and follow a light. The light source provided a goal for the robot to act on (i.e. move towards the light), and the contact switch gave the robot a basic touch sensor that signalled whenever the robot bumped into an obstacle. These robots are described as autonomous because they performed these acts without direct human intervention. In other words, the robots could sense their environment and make decisions on which actions to take in response to what they were sensing.
There is a delightful implicit contradiction in the term ‘autonomous robot’. The term ‘robot’ originates from the English translation of Karel Capek’s 1920 play “Rossum’s Universal Robots”. Capek used the Czech word “robotnik” to describe the fictitious humanoid machines that featured in his play. Robotnik means “slave” and comes from the Old Church Slavonic word for servitude. So according to this the phrase ‘autonomous robot’ can be expanded to mean something like “the freely acting, self governing slave”. The implicit contradiction in this phrase nicely characterises the much debated territory that we are just about to enter, namely the concept of free will itself.
There are those who would argue that free will is an illusion and that the phrase ‘autonomous human’ carries just as much of an implicit contradiction as does ‘autonomous robot’. According to this view, whilst we appear to be self-governing and capable of acting freely, we are actually slaves to causal determinism in much the same way as robots are. This debate hits at the very heart of what we have come to understand about what it means to be human and what it means to be moral beings.
Available on Amazon
(Click here)
“If you wish to get real insight into the fascinating issues surrounding robots and morality, then this is the book to read. “
John C. Lennox MA MMath MA(Bioethics) DPhil PhD DSc
Emeritus Professor of Mathematics, University of Oxford
Emeritus Fellow in Mathematics and Philosophy of Science, Green Templeton College
More information here.
Free Will
In order to proceed with this discussion, we need to delve deeper into what we mean by acting freely. The source of freedom and the capacity for original action and creativity in the human self is commonly associated with the will.
I have found Prof Dallas Willard’s description of the will particularly helpful. He places it in the context of five dimensions of the self, organised into concentric circles, with the outer circle representing the soul, which in some sense contains the other dimensions through its integrative function. Moving inwards, the next circle represents the social context from which flows information and experience, that passes through the body into the mind and is presented to the will through thought and feeling.
Willard places the will at the centre not only because its function depends on this inward flow of information and experience, and the outward flow of action, but also because it steers the content of what is coming in and what is going out. The will is ultimately responsible for the condition of all of the dimensions of the self through which experience and action flow.
This steering operates at different times scales and in different ways, and is mediated by three primary functions of the will: (a) to give rise to original thought or new ideas, (b) to choose what the mind focuses on, and (c) to initiate the execution of what currently holds the attention of the mind. It is important, at this point, to be clear about what we mean by original thought or new ideas. These terms are not describing thoughts that nobody else, including ourselves, have ever thought before – although these are not ruled out. No, in this context we are using the terms to describe thoughts that did not arise from the automatic thought habits that we have developed. They are thoughts that have not automatically emerged from what is currently the focus of attention in the mind.
Autonomous Robot =
“the freely acting, self governing slave”
Let’s illustrate this with an example of shopping habits. Imagine there’s a person, let’s call him David, who normally goes to the nearby corner shop to get his groceries, and that this is the thought that habitually comes to mind whenever he thinks about going shopping. On one particular day, however, he has an inclination to go to the superstore instead and see what other brands and offers they have. David’s will, in this case, would bring this new thought to the attention of the mind. So, the mind now has a choice placed before it: Go to the corner shop as usual, a choice which would automatically come to mind for him, or explore what the superstore has to offer. What David actually does will depend on a number of factors that could come into play. But if his will is strongly set in favour of going to the superstore, and it is not overruled by the power of the habitual thought, then David will set out to go to the superstore.
It is important to note that whether David went to the corner shop or the superstore, in other words, whether he followed what his habitual thinking led him to do or whether what his will led him to do, both are ultimately outcomes of the operation of his free will. To understand this we must return to how the inner dimensions of the human self develop over time.
David’s habit of going to the corner shop for his groceries had a beginning and that beginning was through the operation of his will. Suppose we turn back the clock to when David moved into the house he is now living in. Having just moved in, David needed to buy groceries. He noticed that there was a corner shop nearby and he decided (act of the will) to go and investigate. He likes the shop and the shopkeeper. After a few more visits David decides that he should shop there regularly. And so whenever he needs more groceries, he habitually thinks of going to the corner shop. The key point is that this habit was formed through the engagement of David’s will at various points in his past.
In this way, habitual thoughts, as well as ‘new’ thoughts, are primarily a product of human free will, that operates in the context of other influencing and constraining factors. This leads us on to a key point about free will, which is that the human will is never totally free. It is always constrained by (a) the way information flows to it from its social context, and (b) the capacity of the mind and the body.
What is presented to the mind, and hence to the will, about the individual’s body and environment is always filtered. It is first of all filtered by the capacity of the senses to detect what is going on in the environment. Our senses and nervous system are tuned to detect certain signals from our environment (sights, smells, touch, temperature, taste etc.). But these are limited by what they can represent. So, for example, our eyesight is limited to a sub-range of the available radiation signals that flood our environment. Similarly our hearing is limited to a sub-range of sound pitches.
Furthermore, it is well known that our senses can be fooled. There is a classic study in psychology that illustrates this called ‘the rubber hand experiment’ in which a human participant is sat at a table with one of their hands underneath the table and hidden from view. On the table in front of them is a life-sized rubber hand. The experimenter uses feathers to simultaneously stroke both the rubber hand and the participant’s hand that is under the table. The participant can see the rubber hand being stroked, and can feel their own hand being stroked at the same time. After being exposed to this for a period of time, the experimenter will stop stroking the hand that is under the table, but will continue to stroke the rubber hand. The participant will report that they can feel their hand is still being stroked, even though it is only the rubber hand that is being stroked. Their mind has been fooled into detecting this stroking sensation because that sensation has been associated with the visual cue of the rubber hand being stroked.
Surprisingly, though, our sensory perceptions are rarely fooled in this way. We have remarkably stable and accurate information presented to our minds from our senses. But that is not the end of the story, because what we sense, and how we interpret it depends very much on how those sensory signals are filtered and processed together.
Some people, for example, are attuned to detecting certain things that others would simply not observe, even though they are witnessing the same event, and receiving the same light, sound and touch signals. The neural circuits in our brain adapt over time to filter out the key bits of information that is most interesting and relevant to us as individuals.
A keen bird watcher, for example, will instantly recognise the musical phrases of a particular kind of songbird, which to many others would fall into the category of background noise. An expert wine taster will pick up senses from their taste buds and recognise a good quality wine, which many of us would not be able to distinguish from low cost products.
So the brain of every adult has become attuned to highlight some sensory signals and not others. This has an impact on what we perceive and as a consequence how we react to what we perceive, which in turn sets the extent of the freedom within which we can act. We cannot choose to act in response to some- thing if we are unable to perceive it.
But our will is not just limited by what we can perceive, it is also limited by what is physically possible in the moment the will wishes to act. My youngest son, Joel, has always had big ambitions, even from an early age. When he was about 5 years old he came to me and asked “Dad, can we build a rocket out of cardboard in the garden? Then I could get into it and you could set fire to the bottom of it so that I could fly to the moon!” How do you begin to explain to a 5 year old that, not only will this cause him significant harm, but that it also wouldn’t work. That really is rocket science! What he wanted to do was far beyond what he and his dad could do with a cardboard box and some matches in the garden.
Joel continued to conceive of many other ambitious and impossible schemes, and he was so utterly convinced that they could be done that we had a hard time explaining to him the limitations of what could be achieved in reality. He is older now, and, sadly, has grown out of that phase. I do miss it. But he now has a better understanding of the limitations of what can be realistically achieved. Through experience, his will has adapted and refocused his ambitions in other directions.
In the next blog, we will begin to take a look at free will in the context of a world governed by causal determinism.