Could we ever build a computer that is conscious?
Abstract
Given the similarities of a computer in terms of its structure and operation to that of a human brain, I argue that it is plausible for us to conceive that given enough advances in technology that we could one day build a computer that is conscious. My main argument of why it is conceivable for us to assume this, stems from a physicalist approach, based on the belief that consciousness is formed as a result of bio-physical processes. In this essay I look at the ways in which we define the term “conscious”, continually pruning its definition to the form that something can only be considered conscious if it sentient, has a high-level overview of its own cognitive states and most importantly its sensory input is accompanied by qualia. In conclusion I identify that in order for computers to achieve the final necessary condition of qualia; would require the development of the other aspects of intelligence, such as self-awareness and emotions.
The purpose of this essay is not to prove whether or not computers should be considered as conscious artefacts, but to convince you that it is conceivable for such a conscious artefact to one day exist. I am going to attempt to answer this question by looking at the features which scientists believe give us as humans this very special and unique experience which we refer to as consciousness. In doing so, what I hope to extrapolate are the principles that make entities like humans conscious, and to be able to effectively apply these principles to computers.
In order to be able to answer this question, we must first define what we mean by the term “conscious”. In order for something to be considered conscious at a basic level it must firstly, be sentient; “capable of sensing and responding to its world”.
Julian Jaynes suggests in his book Origins of Consciousness in the Breakdown of the Bicameral Mind that there could have existed some millennia ago a race which met these conditions (of sensing and responding to the world around them) yet whom were not conscious. Jack Copeland forms an interesting analogy to help understand Jaynes’ theory;
“Envisage yourself driving along a familiar stretch of road, while completely engrossed in thought or conversation with a passenger. Because you are absorbed in your conversation or reverie, little or no information about the road filters into your consciousness. Yet the data still flows from your eyes and ears to your brain, and even though you are not consciously aware of doing so you continually plan manoeuvres, judge distances, read road signs, select gears and so on.”
I myself have been subject to such an experience on a number of occasions, and so would agree that sentience alone is not sufficient to qualify something as being conscious. If it were then arguing that we could build computers that are some day conscious becomes a much simpler task, as machines such as smoke alarms who by purpose are sentient could be considered as being conscious under this pretense. However I would argue while sentience is not sufficient on its own, it still remains a necessary condition. Without sentience, an entity cannot plan, reason, deliberate etc… regardless of whether such processes occur consciously or not.
In which case an entity cannot be considered conscious by simply processing environmental stimuli, input and output. It must also be able to internalise these inputs. In other words, in humans our perceptions are our inputs, and perceptions occur at an unconscious level. It is only awareness of these perceptions which give them a conscious context. Consciousness could therefore be seen as a sort of internal monitoring system for the brain, and it is this internal monitoring system which allows for us to give consideration to our sensory stimuli. In gaining access to these internal monitors, an entity is able to give consideration to them in order to formulate a belief. For instance when formulating a plan to do something, we internalise the variables which are input, such as a proposed date, and we may access other variables stored internally, such as other dates in order to give consideration to the proposed output. The result is an output which we believe to work, ie: a plan.
So in pruning my original definition of consciousness, an entity can only be considered conscious if it is sentient and has a high-level overview of its own cognitive states and processes. These characteristics are what philosopher David Chalmers has dubbed the “easy problems” of consciousness. Chalmers defines the easy problems as those which are “directly susceptible to the standard methods of cognitive science, whereby a phenomenon is explained in terms of computational or neural mechanisms.” he goes onto say that “although we do not yet have anything close to a complete explanation of these phenomena, we have a clear idea of how we might go about explaining them.” In other words, we have a reasonable scientific idea about how to approach solving these problems.
Some philosophers such as Daniel Dennett and Douglas Hofstadter argue that if something is capable of passing the Turing Test, then it is necessarily conscious. The Turing Test, which gets its name from Alan Turing who first proposed it, is a test to determine whether something is operationally intelligent. The basic setup of the test involves three participants; a human volunteer who converses via keyboard input with another human (who is assumed to be conscious) and a computer. The volunteer has no other information (such as voice or appearance of the participants) other than the text responses which appear on screen to determine whom they are conversing with. The goal of the Turing Test is for the computer participant to convince the volunteer that they are human. If the volunteer fails to identify which participant is human and which is a computer, then the computer is said to have passed the Turing Test.
Under our current criteria, computers today have demonstrated on some level the characteristics as defined in our definition for consciousness. For instance a computer’s operating system is able to access and monitor its own internal states and processes and respond accordingly – for example, allocating more memory to a process when it requires it, responding to external inputs and forming outputs. Does this mean that the computer is consciously aware of these states, and that it has somehow formed a belief as a result? No. The reason why we still do not consider these processes as an indication for consciousness is because of a missing ingredient: experience – dubbed by Chalmers as “the hard problem.”
I disagree strongly with the view that the Turing Test is sufficient enough to say that something is conscious. Firstly such a claim is assuming that the term “intelligent” and “conscious” are synonymous, which in my opinion is a false assumption to make. Secondly the Turing Test does not account for the missing ingredient of experience, which I believe to be fundamental to the definition of consciousness. For instance, when we are consciously thinking about something and when we consciously sense something there is a subjective aspect to this process. This aspect is experience. It is this subjective consideration which enables us to internalise the sensory input we receive from our environment at a conscious level. This is where the true essence of consciousness lies, or as Copeland puts it, in the “ineffable feel of it all”. What Copeland is referring to is the feeling we get when our senses respond to something. For instance when we see the colour red, we feel warmed by it. Not because the object which we are observing is emitting more heat than another object, or not because somehow the colour red itself intrinsically emits more heat than another colour, it simply has a warm feeling to it. These properties which our senses pick up on are referred to as qualia.
So in pruning our definition for consciousness once more, an entity can only be considered as being conscious if it is sentient, has a high-level overview of its own cognitive states and processes, and who’s sensory interactions with its environment are usually accompanied by qualia of some description.
The disagreements in philosophy surrounding qualia usually pertain to which mental states contain them, and whether such subjective experiences occur physically inside us or occur externally to us. This forms one of the central arguments to the Mind-Body problem. “The mind-body problem concerns the explanation of the relationship that exists between minds, or mental processes, and bodily states or processes.”
A dualist approach to the mind-body problem proposes that such phenomena occur in some respects non-physically. René Descartes famously proposed that the mind is a non-extended, non-physical substance to which he attributes the characteristics of consciousness and self-awareness, separate from the brain itself. Thomas Nagel proposes a similar view that qualia or subjective experiences which he claims transcend bio-physical processes by nature.
In his paper What is it Like to Be a Bat? Nagel proposes that an entity is conscious if there is something like to be that entity, and a mental state is conscious if there is something it is like to be in that state.
Nagel states that “we know that most bats (the microchiroptera, to be precise) perceive the external world primarily by sonar, or echolocation, detecting the reflections, from objects within range, of their own rapid, subtly modulated, high-frequency shrieks. Their brains are designed to correlate the outgoing impulses with the subsequent echoes, and the information thus acquired enables bats to make precise discriminations of distance, size, shape, motion, and texture comparable to those we make by vision. But bat sonar, though clearly a form of perception, is not similar in its operation to any sense that we possess, and there is no reason to suppose that it is subjectively like anything we can experience or imagine. This appears to create difficulties for the notion of what it is like to be a bat”.
In other words, what Nagel is suggesting is that a bat is the only thing which can experience what it is like to be a bat. We as humans are limited by our own subjective experience to form any real conception of what a bat’s subjective experience is like. Nagel indicates that all we can conceive is “what it is like for [us] to behave as bats behave” not “what is it like for a bat to be a bat.”
But importantly I would argue qualia are caused in us, not in the environment. I am not however suggesting that they are processes which occur on their own, more likely that they are the bi-product of other physical processes occurring in the brain. In either case, not being able to subjectively experience another entity’s qualia is not sufficient proof that it does not exist bio-physically. For instance if we consider pain, the distinctive properties of pain are caused within us, not externally to us. In this case I would take a physicalist approach to consciousness, and disagree with the view that consciousness is caused by anything else other than a series of bio-physical processes. As expressed by Michael Gazzaniga; “consciousness is an emergent property and not a process in and of itself.” Therefore I conclude that consciousness is an emergent property of a bio-physical process.
I come to this conclusion because all of the aspects by which we have defined consciousness can be attributed to physical processes occurring in the brain. Using techniques like “functional MRI, scientists can almost read people’s thoughts just from the blood flow in their brains. They can tell, for instance, whether a person is thinking about a face or a place or whether a picture a person is looking at is of a bottle or a shoe.” A person’s consciousness can be altered and skewed by physical manipulations. For instance, if you analyse the effects that chemicals such as alcohol or LSD have on the brain, these chemicals can profoundly alter the way in which people think, feel and reason.
This is what Francis Crick dubbed as The Astonishing Hypothesis in which he supports my view that our “thoughts, sensations, joys and aches consist entirely of physiological activity in the tissues of the brain.”
The basic make up of the human brain and computers are in principle predominately similar, although the implementation of the brain is incredibly more complex. But each have a multitude of connections, which send signals to its different parts. Both take input from their external environment through their sensory equipment. At present computers are restricted majorly by their processing power, and by comparison to the human brain do not even come close to the amount of instructions they send and receive. The brains ability to also truly parallel process and run concurrent states give it an edge.
So if the brain through its processing is somehow able to spawn consciousness, is it not plausible that with the same amount of processing power and with a similar construction and operation that one day a computer could also do the same? I certainly think that it is a real possibility.
Whether or not computer technology advances enough to reach a stage where it is on par with the human brain is another question, one of which cannot be completely answered. But in principle if computers were too advance to similar levels of complexity, and have a similar level of processing power to that of the human brain, then we may find that consciousness is created simply as a bi-product of the processes which occur. That in itself poses an interesting philosophical question; is the very fabric of our consciousness, simply a side effect of the processes occurring inside our brains.
To conclude, consciousness can be thought of as being either a process or processes in itself or the bi-product of such processes occurring in the brain. Technologically speaking it would be plausible to think that an artificial structure which performs in an identical way to the brain could form a consciousness. A computer though would only ever be considered as being conscious if it is sentient, has a high-level overview of its own cognitive states and processes, and who’s sensory interactions with its environment is accompanied by qualia. In order for computers to achieve the final necessary condition of consciousness; qualia, would require the development of other aspects of intelligence, such self-awareness and emotions – as I believe these aspects are particularly intrinsically linked to the conditions of consciousness. In any case I think it would be difficult if not impossible for us as humans to identify whether a computer is capable forming qualia, due to it’s subjective nature as Nagel has indicated. No human can ever know what it is like to be a computer. In which case, under these conditions we may never truly know whether a computer is in fact conscious, and without a subjective experience to draw on we would find it difficult to assume so either.
Bibliography
- Armstrong, D. 1981. “What is consciousness?” In The Nature of Mind. Ithaca, NY: Cornell University Press.
- Copeland, J. 1993. “Conciousness” In Artificial Intelligence: A Philosophical Introduction. (164). UK: Blackwell Publishers.
- Chalmers, D J. “Facing Up to the Problem of Consciousness” Australian National University Press.
- Turing, Alan (October 1950), “Computing Machinery and Intelligence”, Mind LIX (236): 433–460, doi:10.1093/mind/LIX.236.433, ISSN 0026-4423, http://loebner.net/Prizef/TuringArticle.html, Accessed: 09/04/2010
- Kim, J. (1995). Honderich, Ted. ed. Problems in the Philosophy of Mind. Oxford Companion to Philosophy. Oxford: Oxford University Press.
- Nagel, T. 1974. “What is it Like to Be a Bat?” In Philosophical Review. 4:435-450.
- Gazzaniga, M. 2009. “Your Brain; Consciousness” In Time Magazine – Your Brain: A Users Guide. 17.
- Pinker, S. 2009. “Your Brain; Consciousness” In Time Magazine – Your Brain: A Users Guide. 15.
Leave a Reply
