Zooming Into a Zombie World
The conscious experience of thought experiments about conscious experience
Neuroscientists and philosophers largely agree that consciousness refers to “what it’s like” to be you or me. (Philosophers sometimes call this phenomenal consciousness to distinguish it from more functional aspects of consciousness such as attention.) That may not be a very technical definition, but it is evocative. I know there’s something it’s like to be me, and I’m pretty sure there’s something it’s like to be you. There’s probably something it’s like to be a bat1 and probably not anything it’s like to be a rock. There may be something it’s like to be some A.I. systems but not others, and we will need to wait for a well-established scientific theory of consciousness before we can say one way or the other. But popular thought experiments reveal the difficulty of agreeing on how far “what it’s like” can take us as a definition. We can easily get tripped up when trying to imagine what it’s like to be a character in a thought experiment.
In this post I will describe two such thought experiments that challenge my own imaginative limits: “Mary” and the zombie world. Both of them were created to show how consciousness cannot be explained in purely physical terms, but neither one works to that end as far as I can tell, although they do help to get us thinking about what “what it’s like” really means. I will try to describe my imagination of each thought experiment in enough detail for you to follow along and notice where your own imagination goes in a different direction.
Both of the thought experiments are classics in the field, which have had countless responses in the published literature. I have read some, though far from all, of those published responses, but what I describe here are my own responses developed independently. I will mention similarities to other responses where I am aware of them, and if they look familiar then I hope that I am at least adding some nuance to the discussion.
Mary’s Great Escape
Frank Jackson offered a classic thought experiment2 that every subsequent philosopher of consciousness has had to contend with. Jackson’s original formulation was light on the details, but to take it seriously we must imagine it more thoroughly. We are supposed to imagine a person named Mary who has been confined to a colorless room for her entire life. Mary has access to a computer with a black-and-white display. She is presumably served bland, gray food. Her clothing and furnishings are likewise drab. We mustn’t imagine Mary’s own biological functions, which would of necessity provide some color in her generally wretched life. And Jackson gives no clue as to the reason for her captivity: An abusive parent? A cult leader? A mad scientist?
In any case, Mary has, through her computer, become fascinated with the neurophysiology of color vision—something that she has never experienced for herself, and as far as she knows, she may never come to experience. So fascinated, in fact, that she studies all that there is to know about it, attaining a level of expertise rivaling all of the world’s experts combined. Mary studies every element of color vision, from wavelengths of light entering the eye, to rods and cones in the retina, the structure of the optic nerve, the functioning of the visual cortex, and the interplay of vision with memory, emotion, and motor coordination. There is nothing physical about color vision that Mary does not know in great detail.
The crux of the thought experiment comes when Mary is, at long last, released from captivity. Perhaps an elite police unit storms her captor’s compound, discovers her colorless cell, and breaks the door down. Mary is doubtless thrown into a state of shock: First startled by the noise, then terrified by the armed agents standing before her, then overcome with the joy of freedom mixed with the agony of her stolen youth—and somehow, in the midst of it all, a glimmer of color. That experience that she had studied so intently, yet that had been so elusive. Stumbling outside, she is met with the blue sky and the green grass and the flashing red lights of emergency vehicles surrounding the compound.
Jackson’s crucial point was that Mary thereby learns something that she did not know before her release. She already knew all about the physical functioning of the brain, but didn’t really know “what it’s like” to see in color until she did, in fact, see in color herself. The intuition that she has learned something new is supposed to elucidate the nature of consciousness. Whatever new facts that she has learned are, by definition, inaccessible to mere objective study, and may only be acquired through subjective experience.
Daniel Dennett and many others have critiqued the idea that Mary must be learning something new. For example, maybe we just aren’t imagining her studies in enough detail.3 If we are really following the rules of the thought experiment, Mary must have known every detail of how her own eyes, brain, and body would react to the specific colors outside her room, so she should not actually be surprised by any of it. We are also led astray by the emotional content of the scenario, which Jackson left out of his text but which anyone hearing the story will naturally fill in. Mary is not merely learning something new about color; she is escaping from captivity and facing the prospect of a long recovery period.
However, I don’t think we should escape from the argument that easily. I will leave it to the philosophers to argue over the definition of “knowledge,” but no matter how much she knows about color vision beforehand, Mary does experience something new when she actually sees color for herself. It is inescapable that the experience of studying color vision is distinct from the experience of seeing in color. “What it’s like” to be Mary studying vision is different from “what it’s like” to be Mary escaping from the room and seeing color for the first time. We use words like objective, extrinsic, and third-person to refer to the former situation, when one studies something without directly experiencing it. We use words like subjective, intrinsic, and first-person to refer to the latter situation, when one experiences something directly.
The distinction between first-person and third-person is very important to the understanding of consciousness, because consciousness is inherently first-person. What this argument doesn’t tell us is anything about the “physical” nature of consciousness, because every difference in Mary’s experience is accompanied by a physical difference as well. (This kind of dependence of one level of description on another, where every difference in one level involves a corresponding difference in another level, is known as supervenience.) In this case, when Mary is studying color vision from the third person, she lacks a particular kind of colored light landing on her retinas. When she sees color for the first time, her retinas are stimulated in a way they have never been before; signals travel down her optic nerve that have never traveled there before; patterns activate in her visual cortex and trigger responses in her amygdala that have never been triggered before. All of those are physical differences between the two experiences, so nothing in the thought experiment suggests that first-person experience is non-physical, despite being something more than merely studying an experience from the third person.
To drive this conclusion home, imagine a robot that has been fed the world’s entire knowledge about color imagery. It even has full knowledge of its own design and programming. But it does not have a camera connected. One day, an engineer plugs in a camera and the robot starts receiving video signals for the first time. In that moment, wires are energized that have never been energized before. Code paths are exercised that have never been exercised before. Logical conditions become true that have never been true before. The robot’s memory banks start filling up with visual data from its own camera’s viewpoint that has never been encountered by any being in the universe before that moment.
Something new has happened to the robot when its camera is attached, no matter how much knowledge it had beforehand. If the “Mary” thought experiment captures the essential meaning of consciousness in its distinction between first-person and third-person understanding, it would seem to open the door to A.I. consciousness as well. There is, of course, a huge open question as to the level of sophistication and richness in that experience. Humans have a depth of introspection and self-reflection that is unmatched by current A.I. technology, and the details of human phenomenology involve uniquely human emotions. (Phenomenology is a fancy word for all the details about the content of someone’s conscious experience.) But the essence of having “something new” happen when actually experiencing a phenomenon, as opposed to merely studying it, is unavoidable for any being with rudimentary knowledge-processing capabilities.
Mary in the Zombie World
Taking the “Mary” thought experiment seriously shows that some of our strongest intuitions about conscious experience are actually consistent with physicalism, but there are other challenges to consider. (Physicalism is the idea that the mind can be explained in physical terms.) One related challenge to physicalism mentioned by Jackson and developed much further by David Chalmers is that it seems like we can imagine a world that is physically identical to ours but in which there is no consciousness.4 This is known as the “zombie world” argument, because the people in the imagined world are philosophical zombies. They don’t act like typical zombies in movies; they act just like their counterparts in the real world. They’re just missing conscious experience. If we can imagine a zombie world, the argument goes, then philosophical zombies are possible, in a metaphysical sense even if not in our own universe, which means that consciousness is something more than physical.
However, try as I might, I cannot imagine a zombie world myself, at least not to my own standards. For me to be satisfied that I can imagine a zombie world, I have to be able to imagine some difference between a zombie world and its non-zombie equivalent. I haven’t figured out how to do that yet, and I would like to hear in the comments how you have done it yourself. Let’s give it a try here, so that you can see where I always get stuck.
First, to imagine the non-zombie world: I visualize the Earth from space, a blue-green marble flecked by white clouds. I can zoom in, seeing the meandering shoreline and a sprawling modern city. In the suburbs of the city, there is a large house with several police cars in front, their lights flashing blue and red. On the green grass of the backyard stands Mary, in stunned silence, having just emerged from the colorless room. At this point my imagination flips around from the third-person into Mary’s own first-person point of view, to imagine what it’s like for Mary in this moment. I imagine looking out through her eyes, taking in the sights and sounds of the outside world. I imagine feeling the excitement of escape, seeing the flashing lights and green grass and blue sky, hearing the shouting of the police, all from Mary’s perspective.
Next, to imagine the zombie world: I visualize the same Earth with the same continents and clouds. I zoom in to see the same shoreline and city. There is the same house with the same police cars, and Mary standing in the backyard looking just as stunned. Something new is happening to Mary in the zombie world just as in the non-zombie world, because colorful light is landing on her retinas and being processed through her optic nerve and visual cortex for the first time. But then I simply have to stop imagining. I cannot flip around into Mary’s own first-person point of view, because there are no first-person points of view in a zombie world. I have to tell myself no, she’s not conscious, so there’s nothing more to imagine.
This is the best that I can do, but in the end I have not actually imagined a difference between the two worlds. In the second case, I simply stopped myself from imagining the last part of the story. Everything is identical from the third person, but I haven’t imagined the first person one way or another. I am not imagining any difference in the world itself; I am choosing to lend my consciousness to Mary in the first case and choosing not to in the second case.
It is tempting to imagine Mary living in darkness and silence in the zombie world,5 but that’s not right. Darkness and silence are also conscious experiences, whereas she is supposed to have no experience at all. I have experienced temporary blindness due to a sudden drop in blood pressure, and it was a distinctive and terrifying experience. I was in a dance class in high school. My teacher noticed that I looked pale and suggested I go out for some fresh air. Making my way up the aisle of the theater, my vision suddenly went dark and all the sounds of the class became muffled. And yet I was still fully conscious. I tried to shout for help, but only let out a squeak that no one heard. Eventually I did make it to the door and into the fresh air, at which point my blood pressure returned as suddenly as it had dropped, and my vision and hearing came back to full intensity. (Don’t worry, I got the issue diagnosed and treated.)
To imagine a zombie world, we cannot imagine Mary being in the dark like I was. We must imagine no light, but also no darkness; no sound, but also no silence. No feeling, but also no numbness. No experience at all.
Imagining darkness, in contrast, would be like flipping my viewpoint around but not actually imagining Mary’s own experience. I would not be imagining her lack of consciousness, I would be imagining an unrelated experience, and merely pretending that I was seeing it from her point of view. I would be completely ignoring what’s new for Mary, which is happening physically even if I don’t imagine it from the first person. Imagining darkness would be imagining Mary being blind, which she is not—a physically different world. I might as well imagine spinning on a carnival ride, feeling nauseous from too much cotton candy. For the thought experiment to be meaningful, we must imagine the same physical world from a first-person point of view, not an experience of some arbitrary other world. From the first person, there is nothing but the first person to give us evidence of what world we’re imagining, so if we are not imagining the same physical world experienced in the same way by Mary then we are imagining some other world being experienced by some other person and merely declaring that it is Mary’s experience in the zombie world.
The closest I’ve found to my approach in the literature was a paper by Eric Marcus, which had some similar core points about imagining a difference between the worlds, bolstered by a more thorough treatment of the philosophical terms involved.6 Chalmers responded to Marcus by pointing out that imagining the absence of something does not require imagining the presence of something else that excludes the absent thing.7 However, I am not trying to imagine a zombie world in isolation and having trouble positively imagining the absence of consciousness. Rather, I am attempting to imagine any difference between a zombie world and a non-zombie world, and failing in that attempt. There is no difference in the worlds from the third person; and from the first person I am unable to even complete the task. If the presence or absence of consciousness makes no imaginable difference in the world, I don’t think we are taking consciousness seriously. (Chalmers further suggested that inverts—in which one person’s “red” is another person’s “green” and vice versa—support the same conclusions as zombies. Inverts raise other issues that won’t fit into this post, so I will constrain my imagination to zombies for now!)
Your Turn
Consciousness is real and deserves to be taken seriously. The “hard problem” really is hard, and consciousness is not an illusion.8 But try as I might, I cannot imagine a philosophical zombie world. That is, even though I take consciousness seriously—indeed, because I do—I cannot imagine a world that is physically identical to ours and yet is devoid of conscious experience. If I take consciousness seriously, then it has to mean something in relation to the physical world. And we do see in these thought experiments that whatever is new for Mary does correspond to physical changes. Those changes may just be physical correlates of consciousness—physical changes that correspond with conscious experience without being conscious experience—but the fact remains that the “Mary” and zombie world arguments do not end up describing anything beyond those correlates.
Rather than showing that consciousness is non-physical, both of these thought experiments show how consciousness is a special perspective that certain systems have on their world as they experience the world for themselves. That still leaves open the question of which systems have such a perspective and why they do. I have also avoided the question of what physical itself means. For the purposes of these thought experiments, a physical description of something is essentially synonymous with a third-person description.
If we consider the possibility of an immaterial substance that interacts with the material world to support consciousness, then worlds with and without that substance will be different from the third-person perspective as well as the first-person perspective due to the absence or presence of such a substance. Its interaction with the material world would cause changes to physical objects in the world, which Mary would therefore have to know about ahead of time and which would thereby contradict the zombie world argument. My conclusion is not that physicalism is necessarily true, but at least that first-person facts supervene on third-person facts.
I’m curious how others are able to imagine zombie worlds. Are you imagining darkness or some other variety of experience? If you go through the same imaginative sequence that I laid out, where does your imagination diverge from mine? Perhaps if we collect enough answers, we can produce a report on the phenomenology of such thought experiments themselves. We might make some interesting progress on resolving the related philosophical questions if we can clearly articulate how each of us is experiencing ourselves grappling with them.
something it’s like to be a bat: The “what it is like” phrasing was popularized by Thomas Nagel in a 1974 article titled “What is it like to be a bat?” in the journal Philosophical Review, which questioned whether humans can ever really know what it’s like to be a bat.
a classic thought experiment: “Mary” was introduced by Frank Jackson in a 1982 article titled “Epiphenomenal qualia” in Philosophical Quarterly. The general idea is known as “the knowledge argument” because it distinguishes between different ways of knowing.
we aren’t imagining her studies in enough detail: As Daniel Dennett put it in his 1991 book Consciousness Explained, if you imagine Mary learning something new, “you are simply not following directions!” (chapter 12, page 399).
in which there is no consciousness: This idea was mentioned by Frank Jackson in the same article as “Mary,” but is older. Later, it was central to David Chalmers’ 1996 book The Conscious Mind, with responses discussed in his 2010 book The Character of Consciousness.
darkness and silence in the zombie world: This imagery is often used by David Chalmers, such as in his 2010 book The Character of Consciousness where he wrote, “for Zombie Mary, all is dark inside” (chapter 10, page 318).
the philosophical terms involved: The relevant paper by Eric Marcus is from 2004, titled “Why zombies are inconceivable” in Australasian Journal of Philosophy.
that excludes the absent thing: David Chalmers’ response to Eric Marcus is in his 2010 book The Character of Consciousness (chapter 6, page 157).
really is hard … not an illusion: The “hard problem” phrase is due to David Chalmers, mentioned in the introduction of his 1996 book The Conscious Mind. The idea of consciousness being an “illusion” is typically attributed to Daniel Dennett, who entertained the thought in his 1991 book Consciousness Explained that zombies are “not just possible, they’re actual. We’re all zombies” (chapter 12, page 406), although to be fair his own use of the term “illusion” is specifically about the sense of having a continuous self or stream of consciousness formed from the more complex reality of processing in the brain, rather than about the existence of consciousness itself.
Great article, I enjoyed the slant on this one.
When you say “the fact remains that the “Mary” and zombie world arguments do not end up describing anything beyond those correlates.” the non-physicalist will reply – Yes, Exactly!
How is it you know about consciousness at all? Only because you have it. And no description of anything physical will take us beyond the correlates. That "is" the hard problem.
As to imagining zombies, I do it by thinking of a sophisticated robot, which you understand all the inner workings or mechanisms because you built it, but nowhere in that robot is an inner feel. So when it reports it has experience and it’s favourite food is pizza, it’s only the language centre stringing symbols together according to it's programming.
It’s easier to think this robot is a zombie than starting with Mary who is too similar to us. The exercise isn’t to imagine "what it is like" to be a zombie, because the answer is - nothing.
It’s only to show there is no contradiction in supposing we could build a robot identical to biological Mary which wouldn’t have inner experience. It would only go through the mechanistic motions.
And as you say - “That still leaves open the question of which systems have such a perspective and why they do.” That’s true, because the particular answers to those questions depends on your theory of consciousness.