HELP! MY SELF-DRIVING CAR IS A ZOMBIE!

This is the time of year to talk about monsters, and Sean Carroll does not miss the opportunity. A few days ago in Nautilus, he put out an article, “Zombies Must Be Dualists—What the existence of zombies would do to our philosophy of mind”, excerpted from his book, The Big Picture. http://nautil.us/issue/53/monsters/zombies-must-be-dualists-rp

Of course, we are not talking about horror movie zombies, undead beings created by the reanimation of a human corpse, as in Night of the Living Dead. Those zombies do not look or act like regular folks at all.

We’re talking about philosophical zombies, as popularized by Dave Chalmers. Those zombies look and act exactly like everyone else. They say the same things and do the same things. They are indistinguishable from everyone else. The difference is that they have no consciousness, no inner life. If a zombie stubs its toe, he cries out in pain. However, he feels no pain. He feels nothing, ever. There is no “what it’s like”—in Tom Nagel’s sense— to being a zombie.

The first question is, why aren’t we zombies? Every organism takes in physical stimuli from the environment, undergoes bio-chemical processes, and produces behavior, growth and reproduction, if all goes well. What does conscious experience have to do with it? All that stimulus and response is governed by the bio-chemistry. Consciousness seems like an add on, something over and above the physical system. Something outside the physical, and entirely dispensable.

In lacking consciousness, the zombie is just like any other ordinary physical object that is not an animal. Your laptop, for example, can do things which would require a lot of thought by any human. Your laptop can play chess at super-grandmaster level. For you or me to play any decent chess would take a lot of conscious effort. No matter how much we concentrated, your laptop would beat us easily with no consciousness at all, and without knowing or caring about its success. Your laptop is a zombie chess player.

A computer does not have a body, so would not be mistaken for a human being, and cannot reach full zombie status. However, under certain limited circumstances, it might be mistaken for a person. If it is playing chess on-line, the player on the other end might think he is playing a human. Until he gets blown off the board. For chess, and some other activities, your laptop could pass the Turing Test, by being indistinguishable from a human. Let’s call these kinds of computers “proto-zombies”. Another example would be a self-driving car, which can drive you to the airport (soon?) without consciousness. (“Driverless Cars Made Me Nervous. Then I Tried One.” https://www.nytimes.com/2017/10/22/opinion/driverless-cars-test-drive.html)

So proto-zombies are around already.

Could you prove you are not a zombie?  If zombies are specified as not being only behaviorally indistinguishable from the rest of us, but also physiologically indistinguishable, then no conceivable scientific test could distinguish a zombie from the rest of us. You cannot prove to the world at large that you are not a zombie. However, you know for yourself. Do you have conscious experiences and an inner life? Does it hurt when you stub your toe? Then you are no zombie. The real zombie will give the same answers you do, but he will be wrong. Carroll asks, “Why do zombies lie all the time?” They are not aware of lying, because they are not aware of anything at all. They can’t help saying what they say. Everything they say and do is produced by their bio-chemistry. (And we aren’t?) They have no conscious awareness of any difference between what they say and what they really think, so they are not really lying.

Carroll seems to question whether each of could tell whether we are zombies:

For that matter, are you sure you’re not a zombie? You think you’re not, because you have access to your own mental experiences. You can write about them in your journal or sing songs about them in a coffee shop. But a zombie version of you would do those things as well. Your zombie doppelgänger would swear in all sincerity that it had inner experiences, just as you would. You don’t think you’re a zombie, but that’s just what a zombie would say.

The zombie would write about his mental experiences, sing songs in the coffee shop, and swear with apparent sincerity, just as you would. That’s why you cannot prove to others you are not a zombie. But you can prove it to yourself: kick a stone. Did it hurt? You know you’re not a zombie. The zombie says he’s not a zombie, but he does not know that or anything else. He does not consciously know anything.

Zombies highlight our ideas about mental experience in fun ways. But there aren’t any, so what’s the point? Carroll puts it this way:

The big question about zombies is a simple one: Can they possibly exist? If they can, it’s a knockout argument against the idea that consciousness can be explained in completely physical terms. If you can have two identical collections of atoms, both of which take the form of a human being, but one has consciousness and the other does not, then consciousness cannot be purely physical. There must be something else going on, not necessarily a disembodied spirit, but at least a mental aspect in addition to the physical configuration.

It sure seems like zombies could possibly exist. That possibility is what we’ve been discussing: how you can prove you’re not one. So the answer to Carroll’s big question—can they possibly exist— is yes? And that makes it a “knockout argument against the idea that consciousness can be explained in completely physical terms.” There must be something it is like to be “the physical configuration” that is not fully captured by considering the physical configuration from the outside, as science does.

This does not mean that zombies are possible given the laws of physics in our world. If aliens probed your body and created an exact physical atom-for-atom duplicate of you on Planet X (see BQTA “Star Trek Transporter pt2, King of Planet X”), we assume that physical process would be sufficient to re-create all your mental states on Planet X. All your conscious feelings, emotions, memories. That body would not be a zombie. It would think he was you and wonder what the hell just happened to him. The physical carries with it the mental. The mental cannot exist without the physical—no disembodied spirits—but the physical can exist without the mental and usually does. Your self-driving car or your laptop, for example. The zombie concept seems perfectly coherent.

But Carroll argues that zombies are not as conceivable as I’m assuming. “According to poetic naturalism [what he calls his philosophy], philosophical zombies are simply inconceivable, because ‘consciousness’ is a particular way of talking about the behavior of certain physical systems.” This is hard-core behaviorism—there is nothing to ‘consciousness’ other than the right kinds of behavior—though Carroll does not use the term ‘behaviorism’. If behaviorism is true, then zombies are just as conscious– which is not at all– as everybody else, because they behave just like everybody else. End of zombies. End of problem.

However, behaviorism has fatal defects of its own as a philosophy of mind. The experience of pain is nothing other than pain behavior? But the feeling of pain and any sort of behavior can easily come apart. Pain without behavior: tough people who experience plenty of pain but never show it. Pain behavior without feeling pain: actors in a war movie. Therefore, pain is not the same as “the behavior of certain physical systems” and behaviorism is false. Zombies are back in action.

Carroll wants both to reduce consciousness to behavior, but also to insist that inner mental states are real. “The best way we have of talking about people and their behaviors makes important reference to their inner mental states; therefore, by the standards of poetic naturalism, those states are real, existing things.” He has an interesting view of language: if we can talk about it, it’s real. Well, we’ve been talking about zombies; does that make them “real, existing things”? Of course not. Being able to talk about something in a coherent way makes it conceivable, but not real and existing. Poetic naturalism may need new standards.

Does Carroll mean that consciousness is nothing more than “the behavior of certain physical systems?” Or that consciousness is identical to behavior? Or that talk about “consciousness” is really only talk about behavior? All of the above? Then subjective experiences don’t exist. There is only behavior. Carroll ends up insisting “There’s no reason for us to pretend subjective experiences don’t exist.”

Normally, we might think that consciousness and any associated behavior are two different things. You stub your toe and it hurts a little. You might react with some behavior—rub your toe or say “ouch”—or you might not. The sensation of pain is one thing, and the reaction to it is something else. Therefore, pain is not merely the pain behavior, and our talk about pain is not just “a particular way of talking about the behavior of certain physical systems.”

Carroll uses the phrase “way of talking” a lot whenever he is not discussing physics. For example, the poetic naturalist believes that mental states “are just ways of talking about physical states.”  So does that mean that there are not two things—mental and physical—but really only one thing, physical states (plus ways of talking)? Usually if we say “just a way of talking”, we mean the talk is not true, but it’s handy and not worth correcting. “The sun sets at 6 pm” is just a way of talking.  The reason zombies are supposed to be inconceivable is “because ‘consciousness’ [why the quotation marks here?] is a particular way of talking about the behavior of certain physical systems.” Does he mean that word ‘consciousness’ is just a way of talking, so like ‘sunset’ it’s not real? If Carroll is willing to claim that consciousness does not exist, that only physical systems exist, then that would indeed solve his zombie problem. But he isn’t. That leaves him in a muddle.

Carroll makes “one more good-faith effort to explain to an open‐minded property dualist how a poetic naturalist thinks about qualia”:    

What do we mean when we say “I am experiencing the redness of red?” We mean something like this: There is a part of the universe I choose to call “me,” a collection of atoms interacting and evolving in certain ways. I attribute to “myself” a number of properties, some straightforwardly physical, and others inward and mental.

This seems reasonable. You have physical properties and mental properties. This is “property dualism”: the belief in the separate reality of mental properties. But Carroll means to be debunking dualism in favor of poetic naturalism:

There are certain processes that can transpire within the neurons and synapses of my brain, such that when they occur I say, “I am experiencing redness.” This is a useful thing to say, since it correlates in predictable ways with other features of the universe. For example, a person who knows I am having that experience might reliably infer the existence of red‐wavelength photons entering my eyes, and perhaps some object emitting or reflecting them. They could also ask me further questions such as “What shade of red are you seeing?” and expect a certain spectrum of sensible answers.

There may also be correlations with other inner mental states, such as “seeing red always makes me feel melancholy.” Because of the coherence and reliability of these correlations, I judge the concept of “seeing red” to be one that plays a useful role in my way of talking about the universe as described on human scales. Therefore the “experience of redness” is a real thing. It’s a mouthful, and nobody would ever mistake it for a Shakespearean sonnet. But there’s a kind of poetry there, if you look closely enough.

Sure sounds like he thinks that everything mental is strictly a matter of behavioral dispositions and responses. A matter of having the right role in our social life. The phrase ‘seeing red’ may be a useful way of talking, but actually seeing red is an experience people really have. So the phrase ‘seeing red’ is not just a way of talking; it’s a potentially truthful way of referring to the real experience. Carroll talks as if  the phrase “I am experiencing redness” can never be literally true. But sometimes it can be.

By the way, the phrase “red-wavelength photons” is just a way of talking. Waves have various properties: amplitude, frequency, wavelength. But no color. (see BQTA “The Science of Color”.)

Though Carroll does not draw upon academic philosophy for support, he could have. The conceivability of zombies remains controversial. See https://plato.stanford.edu/entries/zombies/

But proto-zombies are all around us. Do animals have mental states? If we attribute feelings to a dog, is that just a way of talking or do dogs literally have feelings? This question seems to make perfect sense, even if we think the answer is obvious. In a sense, the question is whether dogs are zombies. We are inclined to grant consciousness to our fellow mammals, hence Tom Nagel can discuss “What Is It Like To Be a Bat” and not worry the answer is “nothing at all.” Mammals behave enough like people that if their behavior is not backed up by consciousness, they would be zombies, or at least proto-zombies.

As we go down to simpler animals, however, the question becomes tougher. What about social insects: ants, bees, termites, etc.? They do a lot of the same things we do. The work together to gather food. They communicate with each other. They cooperate to build homes for themselves. Do they have experiences and feelings? Is there something it is like to be an ant? Consciousness depends on neurons. How many are necessary? We have about 100 billion, so that’s enough. Ants have only about 250,000. That’s enough to create the neural activity necessary to power the behavior of ants, but is it enough to make them consciousness? If not, ants are zombies. Ants would have physical states, but no mental states. The question—are ants zombies?—makes perfect sense. Ant zombies are certainly conceivable.

Lastly, your self-driving car. For you to drive to the airport in a regular car takes a lot of consciousness. You perceive colors and sounds, you think about where you’re going. Maybe you get tired. You can’t drive while you’re asleep. You need consciousness. There is definitely something it is like to be the driver. Your self-driving car can do all the driving you can do, probably better. The self-driving car has all the driving behavior you do. Does it have any of the driving consciousness? In traffic, the other drivers may not be able to determine whether you are doing the driving or you have a self-driving car. The driving behavior is identical.

Is there something it is like to be a self-driving car? If not, the car is a zombie. As for animals, the zombie question makes good sense. Is there behavior + consciousness, or only behavior? Can your self-driving car feel pain if it crashes?

There is nothing in principle that prevents a computer from feeling pain” according to Sean Carroll. Maybe so. You are a physical system, your self-driving car is a physical system. You can experience driving, why not the car? So let’s suggest to the engineers at Google that they program some feelings into the car. It’s not enough just to have good driving performance. We want the car to feel everything. It’s not enough for the car to get you to the airport safely. We want it to feel glad about it. Programming the car to play a recorded “I’m glad we made it” is not enough either.

Take a basic function: stopping at a red light. You do this by having an experience of redness. Your self-driving car has no experience of redness or anything else. Instead, it accurately analyzes the wavelength of the light emitted from the traffic light and determines whether it falls within the appropriate range. If so, it gently stops. That performance is all the engineers can or should care about. They would have no idea how to give a self-driving car an experience of redness.

Your self-driving car is a zombie. 

Zombies are among us. There will be more and more as Artificial Intelligence advances. When will these computer zombies become conscious? When will they stop being zombies?

3 Comments

  1. A comment by Carter Gillies:

    I am not so sure Sean Carroll was committed to anything strictly in the behaviorist camp. One of the reasons I am not so sure is that I took him to continually suggest that these various things can be talked about based on certain manifest behavior rather than specifically stating that they ARE these behaviors.

    This is also why Wittgenstein is not a behaviorist. There is no discussion of what such things ‘really are’, and then reducing them to mere behavior. It is not a question of metaphysics or ontology. Which is also why I was suggesting that the question “What is consciousness?” is ultimately a bad question.

    For instance, what time is it right now where you are? What time will you be having dinner? What time do you usually go to sleep? What time of day is your favorite? Is there a season you enjoy the most? So Barry Stroud is in his nineties? If you can ask and answer any of these questions you obviously must know what ‘time’ itself is, right? So what is time?

    If you point to science for an answer it can’t be knowing the science that grounds our use of ‘time’ in our daily lives. There may very well be some scientific explanation that decides what it is and what it isn’t, but the normal human experiences are contingent and practical. When we talk about time, we are NOT doing metaphysics. We have these specific ways of talking about time, and they fit our lives in just these circumstances. Beware doing science with ordinary words!
    –Carter Gillies

    Like

  2. A reply by BQTA

    Carter Gillies has as deep an understanding of philosophy as you will find outside universities.

    His thoughts are in line with Barry Stroud’s, which is no bad thing. Stroud is anti metaphysics. He thinks reading behaviorism into Wittgenstein is way off base.

    Though Wittgenstein is not consistently anything, there is a strong strand of behaviorism in “Philosophical Investigations”. Nobody proudly announces that they are a behaviorist–though Dan Dennett comes close. Nobody proclaims the truth of dualism either. The arguments against such clear positions seem too fatal. Better to go to panpsychism, which nobody understands. Or to claim that any restatement of your position is a distortion. Better to remain elusive. See eg John Searle, “Why I am not a Property Dualist”

    What is consciousness? What is time? Tough questions. Meaningless questions? Stroud has fascinating arguments about the limits of metaphysics in his book “Engagement and Metaphysical Dissatisfaction”.

    Wittgenstein, Gilbert Ryle, Peter Strawson, JL Austin gave the mid-century push to the idea that the traditional problems of philosophy were transcended, by being shown to be mere confusions about language.

    My sense, however, is that metaphysics is making a comeback in academia. Tim Maudlin at NYU. Geoff Lee at Berkeley. Many others. They could give you an earful about time, which is too hard for me to think about.

    “Beware doing science with ordinary words!” Fabulous point.

    My response: science must come down to words and ordinary ones at that, if it is to claim to be the ultimate description of reality. Not just math, not just technical terminology, as useful as those can be in the short term.

    A main job for philosophy is to understand how science maps onto the human experience.

    Does a bat have an experience of redness? a dog? a self-driving car? Bats and dogs may not pick up the wavelength information in light, but the car certainly does. So dogs and bats have consciousness but not redness; and the car has ‘redness’ but not consciousness? These questions are meaningless? I don’t think Wittgenstein has shown that.

    Like

  3. Further comment by Carter Gillies

    Thanks for the compliment 😊 There is so much that I have no clue about (I think I amputated about 90% of what I had learned when I left grad school), but I am finding that the things Wittgenstein has taught me mostly stuck, and that they apply to many more situations than perhaps even he ever dreamed of. At least more than what he committed to writing…..

    And as with what we were saying about Wittgenstein dissolving certain questions not always being helpful, sometimes we really DO want to know where our questions lead us. Because it IS a question even if in a certain light asking may be misleading or a poor direction to follow. It would be less than human not to wonder “Where does this question fit?” And it is our privilege to ask what matters to us and expect answers of a certain sort regardless. It’s just not always the same as doing science.

    In other words, we are entitled to do philosophy! Because in the human sphere sometimes it turns out that certain answers matter even if the questions that led to them were off the mark. Like wandering into an unfamiliar part of your town and seeing surprising and undreamt of things. Sometimes the things are there to be found, even if we discover them by accident or mistake. And some questions invent possibilities that were not simply hidden from us. Asking some questions is not merely uncovering what is, but in fact creating what might be….. The difficulty, as always, is knowing when we are down rabbit holes and when we are actually doing ‘good’ work.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s