If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Epistemology: Science, Can It Teach Us Everything?

Caspar asks: can science tell us everything there is to know about the world? He tells us about a famous argument that it can't, sometimes called 'the knowledge argument' or 'the Mary argument', due to philosopher Frank Jackson. If the argument is right, then there are certain aspects of the world that we can't learn about through science. In particular, we can't use science to learn what it is like to see red, or taste coffee, or have other experiences.

Speaker: Dr. Caspar Hare, Associate Professor, MIT.
Created by Gaurav Vazirani.

Want to join the conversation?

  • aqualine ultimate style avatar for user Flanny
    Isn't Mary's Room solely based on the premise that science needs language to communicate? The problem then does not lie with science, but with language. It can be imagined that science will invent a way to influence a person's eyes or tongue so that one can make a person know what red looks like or coffee tastes like.
    Following this argumentation, if science can transcend language, then may science be able to teach us everything?
    (44 votes)
    Default Khan Academy avatar avatar for user
    • marcimus pink style avatar for user Raymond Greenwood
      I was just about to make the same point. Humans cannot perceive really large numbers or really small objects or very far distances in space, because of our lack of the proper reference points in our personal experience. I believe that if science became expansive enough in its detail, then all knowledge could be explained by science if we were able to become precise enough in our description, even the experiential components. This is why I believe science is in its infancy. Just like when humans were working with sharp rocks and hand tools many years ago, they probably thought that computers or rocket ships were not possible or even could not fathom them.
      (14 votes)
  • male robot hal style avatar for user Giles
    This argument assumes that experience is not a part of science. However, the very nature of science is dependent upon experiments, in which the investigators experience the very thing they are investigating. The argument seems to be less that "science can't teach us everything" and more "there is only so much you can learn from books."
    (17 votes)
    Default Khan Academy avatar avatar for user
    • leaf orange style avatar for user mr.david36
      The argument presented here isn't about what one can find out, as one could do with experiments, but rather what can be taught. If an experiment did result in experience, that experience can't be described (at least to its fullest extent) by means of science, and therefore science can't teach o r describe everything in the world around us.
      (4 votes)
  • mr pants teal style avatar for user mblair73
    So if you were to learn EVERYTHING about red, as "Mary" has, would you recognize the color red when you first saw it, without being told it was red? Or if coffee had been described to you in every way, and you drank it without knowing it was coffee, could you recognize it when you taste it for the first time?
    (11 votes)
    Default Khan Academy avatar avatar for user
    • leaf red style avatar for user Noble Mushtak
      If Mary knows everything about an object type A and then is presented with an object of object type A without knowing it is of object type A, she must be able to examine it to the extent so she can determine whether or not it meets the requirement of being of object type A to determine whether or not it is of object type A.

      For example, let's define a blegh is something with a particular shape. Anything with that particular shape is a blegh and all bleghs have that particular shape.

      Now let's say that Mary knows everything about what a blegh is and is then presented with a blegh, but can only taste the blegh and can not otherwise examine it. In that case, Mary would not be able to determine that the object is indeed a blegh because she can not determine the presented object's shape. Even if all bleghs happen to have a distinct taste, there may be other objects that do not have the particular shape bleghs do, yet have the taste that bleghs do. In such a case, Mary might be sure that the object is a blegh from its taste, but can not show that it indeed is because she does not know the object's shape.

      On the other hand, if Mary is allowed to see the blegh, then Mary can determine the shape of the object, check that it is the same as required for being a blegh, and determine that the presented object is a blegh.

      I hope this helps you think!
      (9 votes)
  • blobby green style avatar for user Fabien Ninoles
    I'm not sure to understand the argument. Basically, it tell us implicitely that science is only about facts, not experience. Than, it says that science cannot teach you experience of a fact, since it's not a fact.

    I'm not sure that I can accept the premisse that natural science is only about knowledge. For example, should we consider ornithology as only composed of text and the equation of the wavelenght of the birds songs ? That would seem to me as only a partial knowledge of ornithology. But I would accept a teaching that only include "mocking" representations of the birds songs, like made by a human or a whistle. It would however be closed enough so that an ornithologist could recognize a bird by its song and not consider a new knowledge, only a new experience.

    Did I miss something here ?
    (5 votes)
    Default Khan Academy avatar avatar for user
    • leafers ultimate style avatar for user Justin
      Science attempts to use facts to explain a experience, not the other way around. Although, I think scientists can agree that because we all are the interpreter in our life experience its very important to use our observation in our own research, however its not valued so much in supporting the observation.
      For example, there would be no poetry in Mary's room even though it could be useful to a scientist studying color assuming they could not see the color. Because no matter how useful, if its not factual its not valid in a strictly scientific argument.
      (1 vote)
  • spunky sam blue style avatar for user Luke Thomas
    Is this not like the difference between knowledge .. and wisdom ?
    Knowledge = to know something in every way.
    Wisdom = to experience that knowledge.
    (4 votes)
    Default Khan Academy avatar avatar for user
  • leaf green style avatar for user Redjan Shabani
    I think it would be interesting also an introduction to Godel Incompleteness Theorem: "There exists statements not provable inside the set of chosen axioms". Once we have a new evidence, that evidence is added to set of axioms and than we can infer other "truths". Still there will be "truths" not provable inside the new system.
    Every exact science is simply a set of postulates and a set of propositions inferred/inferrable from those postulated by applying the rules of formal logic.
    The purpose of science is not to teach evidences, the purpose is to teach the possible consequences derivable from the evidences.
    (4 votes)
    Default Khan Academy avatar avatar for user
  • mr pink red style avatar for user JoeMcDo
    Lets say there are two different types of fruit, that are different in practically every respect, but taste exactly the same. If a person knew everything there was to know about both fruits, and had tasted one of them, could they understand the other as well?
    (2 votes)
    Default Khan Academy avatar avatar for user
  • male robot hal style avatar for user Chancellor Toledo
    Wouldn't it be implied in "premise 1" @ that being shown a red thing is required to possess a complete understanding of the color red, as science requires the use of all senses in order to be truly complete? Human knowledge is limited because we posses limited facilities of observation. of course we have created extraordinary means of observation (Xray, IR, Calculus, Telescopes) but all these means do is scale abstract concepts like bones inside living flesh and bodies of burning gases trillions of miles away to our level of understanding through our five senses. If you've ever played around with image processing, you've probably scaled an image up and down a couple times and every time you do it loses fragments of data as the program averages and then divides groups of pixels. The same could be said about science as we lose data using infinite values in finite computers. Science will never know everything but what it says with approaching infinite certainty is that ghosts, demons, and sky-daddies don't pull the strings, at least not the way you think of them. Argumentum ad consequentum doesn't fly.
    (2 votes)
    Default Khan Academy avatar avatar for user
  • purple pi teal style avatar for user Maria
    I'm kind of confused by this whole video. Experiences such as taste and color are unique to the individual. For instance, we can't know if the red I see is the red that you see, but we can both agree that tomatoes are red. Doesn't science depend on observations that are absolutely concrete, like 'this tomato is 3 inches across'?

    Also, I thought this video would be more about reconciling ethics with science. Science can tell us how to poison a person, but it can't tell us if poisoning a person is morally justifiable. Therefore, it can't teach us everything.
    (2 votes)
    Default Khan Academy avatar avatar for user
  • leafers seed style avatar for user Liz Spann
    I would argue that "what it's like to drink coffee" is not a fact at all. You cannot know what it is like to drink coffee--you can only know how you experience what it is like to drink coffee, at a certain moment, with a certain type of coffee, while you are in a certain physical state (eg, our tastes respond to stimulus differently under different circumstances--orange juice after toothpaste is much less pleasant, generally than orange juice prior to toothpaste; as a woman I experience spice differently at various times throughout the month).

    And what it's like to see the color red is likely also quite variable from person to person and under various circumstances. So again, it is not a *fact.*
    (2 votes)
    Default Khan Academy avatar avatar for user

Video transcript

(intro music) Hi, my name is Caspar Hare. I'm a professor at MIT, the Massachusetts Institute of Technology. Today we're going to talk about science. Can it teach us everything[br]there is to know about reality? Science is an impressive thing. Scientists have, for hundreds[br]of years, been able to predict the movements of comets,[br]the eclipses, and the like, decades in advance. In the 20th century, scientists were able[br]to transplant hearts, to send people to the moon, and do all kinds of extraordinary things, of which we can hardly even imagine what they'll be able to[br]achieve in the 21st century. There's an issue though. Just how impressive,[br]or how good, is science at describing the world that we live in? One strong claim that[br]we might make is that science can give a complete[br]description of reality. It can describe everything that there is to know or be known about the world. Now, at various points in history, people have disputed this. They've thought however[br]impressive science was, there would be no way that it could give a complete description of reality. There would be things like[br]angels, spirits, and the like that could not be described[br]in the language of science. Today, what we'll do is we'll[br]look at a modern argument to this effect. To the effect that are aspects of reality that cannot be described[br]in the language of science. Our first job is to explain what it is for a description of an aspect of reality to be complete. We'll do this by way of an example. Here's a description of what's going on[br]in a room (it happens to be this room): there are three people in the room. End of description. Was that a good description? Well, you might think that there are[br]things that the description left out. I said there were three[br]people in the room. I didn't say anything more about them. I didn't say whether they were[br]standing up, sitting down, whether they were riding on[br]little horses around the room. There were bits that my[br]description left out. I could have told you more. I could have said, "There[br]were three people in the room, sitting around a table." End of description. Still, my description[br]would have been incomplete. I didn't tell you their names. Maybe they were called Ben, Balthazar, Fitzgerald. Maybe they were called, as[br]they actually are called, Damien, Gaurav, and Caspar. Let me add that to the description. That their names are Damien, Gaurav and[br]Caspar. They're sitting around[br]a table, in this room. Still, there's lots that I've left out. I haven't told you what they're wearing, I haven't told you anything[br]about their heights, about their ages, about[br]all kinds of things concerning these people. These are all details[br]that I could fill in. Suppose I went on, and I added layer upon layer[br]of detail to the description, I said more and more about[br]the people in this room, what they're doing. And I said it in language[br]that could be understood by a natural scientist. I described their bodies[br]in the terms of biology. I describe the gravitational[br]effects on them in the terms of physics. I talk about them, I talk about their brains in[br]the language of neuroscience. I fill in all these details. Is it, in principle, possible for me to get to the point where there are no further questions of the form "You've left something out, tell me more" to be asked. Some people claim there is, it is possible to get to this point. Other people say it's[br]not, it's not possible to get to this point. However much I add, just detail,[br]detail, I add to the description, there are gonna be aspects of reality that are gonna remain undescribed. So, here we'll look at an argument to the effect that that's the case. There are aspects of reality[br]that cannot be described in the language of the natural sciences. Here is a thought experiment[br]originally made up by a philosopher by the[br]name of Frank Jackson. Imagine a woman, Mary, who lives in a black and white room. A windowless room whose walls are white, whose furniture is black. A room that contains no colored things. Imagine furthermore that Mary has grown up in this room all her life. She's never seen a colored thing. Her hands are painted gray. Her fingernails, black. The food she eats is[br]white food or black food. She's never seen anything that's blue. She's never seen anything that's red. She's only been confronted with black or white or gray things. Now, that's not to say that Mary isn't a bit of an expert when it comes to color. Scientists have taught her everything that they know about the physics, chemistry, and biology of colored things and color vision. So, for example, Mary knows,[br]with respect to redness, she knows the reflective properties of things that reflect red light. She knows the chemistry of surfaces that reflect red light. She knows the wavelengths of light such that we call light[br]of that wavelength red. She knows what happens when[br]light of that wavelength enters an eye, the way in which it[br]passes through the cornea and hits the retina. She knows what goes on beyond that. She knows everything[br]about the neurobiology, the neuroscience, of[br]the perception of red. We can imagine, indeed, we can expand this thought experiment so that she doesn't just know everything[br]that current scientists, the scientists around us nowadays, know about the vision of red. She knows everything[br]that science of any kind, any kind of language of natural science, could ever communicate with respect to the vision of red. She is a perfect expert when[br]it comes to the science of red. So, if you were to give her any kind of[br]exam, she would ace it, when it comes to the science of redness, what it is to see a red thing. Nonetheless, in spite[br]of all this knowledge, it might seem as if there's something that Mary doesn't know. To illustrate that, we can add to our thought experiment. We can suppose that, for the first time, Mary is about to be[br]confronted with a red thing, a ripe tomato. The scientists bring it in underneath a kind of veil, a gray veil. They put it on the table. The scientists are themselves[br]painted perfectly gray, wearing gray contacts. With no trace of flesh on them at all. They put the veiled object on the table. Mary stands before it. She's wondering, "What will it be like? What will it be like[br]to see the red tomato?" The veil is lifted. She gasps. There is the red tomato. Now, one thing it would be very natural for Mary to say in this context is "Wow, I've just learned something. I've learned what it's[br]like to see a red thing. That was something I didn't know before." Let's suppose that's right. If that is right, then it would seem as if there's stuff to know[br]that can't be described in the language of the natural sciences. That's what's nowadays know as The Knowledge Argument for thinking that there are aspects[br]of the world that can't be described in the language[br]of the natural sciences, that are nonphysical. That example was quite elaborate. We didn't need quite such an[br]elaborate thought experiment. Just as Mary doesn't know[br]what it's like to see red or what it's like to see blue before she actually sees a red thing or sees a blue thing, so one might think that somebody who doesn't know what it's[br]like to taste coffee doesn't know what it's like to taste[br]coffee prior to tasting coffee for the first time. No matter how much they learn about the science of taste. Let's try to put this[br]argument, what I just said, in a slightly more[br]rigorous and careful way. We'll put it in terms of an argument with premises and conclusions. Okay, premise one of the argument is this: before drinking the[br]coffee, you know all facts that are describable in the[br]language of natural sciences. Seems pretty good. Everything that a[br]scientist could tell you, you already knew. Premise two: after drinking the coffee[br]for the first time, you learned a new fact. That seems good as well.[br] you're like, "Whoa, that's what it's like to drink coffee. I always wondered.[br]That's something I now know: what it's like to drink coffee." Conclusion: there's some fact that's not describable in the language of natural sciences. And, it seems to follow that any description of the world that's purely done in the[br]language of natural science is going to be incomplete. Subtitles by the Amara.org community