# Heterophenomenology

by Mentat
Tags: heterophenomenology
 P: 3,715 StatusX, Michael can only tell you that "red" is different from "green", in the first place, because he knows that there are two colors, and he knows that their names are "red" and "green". Unless he were capable of physically distinguishing between the two, he would have to disbelieve his mother about there being two colors. As it is, he can physically distinguish between the two colors (though not knowing which is which), and so he agrees that there are indeed "red" and "green" stripes. Now, when he gets out into the world, and see the colors for himself, he will still be able to distinguish between them. He won't know which is which (but then, he didn't know that to begin with...and neither did you, until someone told you), but the process of distinguishing and categorizing is merely a process of computation (no "extra ingredient" required).
HW Helper
P: 2,567
 Quote by Mentat Michael can only tell you that "red" is different from "green", in the first place, because he knows that there are two colors, and he knows that their names are "red" and "green". Unless he were capable of physically distinguishing between the two, he would have to disbelieve his mother about there being two colors. As it is, he can physically distinguish between the two colors (though not knowing which is which), and so he agrees that there are indeed "red" and "green" stripes. Now, when he gets out into the world, and see the colors for himself, he will still be able to distinguish between them. He won't know which is which (but then, he didn't know that to begin with...and neither did you, until someone told you), but the process of distinguishing and categorizing is merely a process of computation (no "extra ingredient" required).
That was my point. A computational model of his brain will only capture the fact that he distinguishes them. So according to this model, nothing would be different to Michael if when he walked out into the world, it was a normal one, or if it was one where stop signs are green and grass is red. The model would predict his inner subjective world would see the two identically. But clearly this isn't true, because his experiences of what he calls red and green would be switched.

If you want to preserve the completeness of heterophenomenology, you have to either claim there really is no difference to Michael between the two worlds, or offer a way heterophenomenology could account for a difference, by finding some asymmetry in the functional role of the colors.
Emeritus
PF Gold
P: 8,147
 Quote by StatusX That was my point. A computational model of his brain will only capture the fact that he distinguishes them. So according to this model, nothing would be different to Michael if when he walked out into the world, it was a normal one, or if it was one where stop signs are green and grass is red. The model would predict his inner subjective world would see the two identically. But clearly this isn't true, because his experiences of what he calls red and green would be switched.
I don't get this. Michael can ask around and find out which of the two shades he can distinguish is generally called red and which is called green. Asking around doesn't strain the computational-described brain at all, does it? This is how we learn colors ourselves; our brains provide us with distinguishable, recallable entities (congeries of neural activity) and we learn to give these names by social interaction.

Computer/software systems alreay can do what I have described here.
HW Helper
P: 2,567
 Quote by selfAdjoint I don't get this. Michael can ask around and find out which of the two shades he can distinguish is generally called red and which is called green. Asking around doesn't strain the computational-described brain at all, does it? This is how we learn colors ourselves; our brains provide us with distinguishable, recallable entities (congeries of neural activity) and we learn to give these names by social interaction. Computer/software systems alreay can do what I have described here.
I admit, it isn't worded very clearly. I meant if you had put him in an inverted red-green world instead, where everyone still calls grass green, etc, it would be no different to him than if you had put him in a regular world.

Maybe it's more clear if instead, you put him in a normal world for a day, erase his memory, and then put him in a red-green inverted world. Of course he won't notice a difference, but is it still a meanigful question to ask if there is one? (this subtlety is part of the reason I didn't go with this scenario first) And if so, is there a difference?
 Emeritus PF Gold P: 3,634 The ideal heterophenomenological neuroscientist (which can exist, since this is only a thought experiment) would look for the neural response that accompanies the human experience of "green" and "red" and be able to identify them in your hypothetical Michael. Even if he didn't know what they were, the ideal neuroscientist would. That is precisely why heterophenomenology does not restrict itself to using only verbal and behavioral reports as primary data. Let me reformulate your thought experiment. Imagine Michelle, Michael's sister. She has been raised completely without any knowledge of geometry. One day, after turning 10, Michelle is shown a sheet of paper with two figures on it, a circle and an ellipse. When the heterophenomenologist asks Michelle what she sees, she cannot answer. She doesn't know the words "circle," "ellipse," "flatter," "round," or any other word that refers to traits of conic sections, nor has she ever seen any of these things (just as Michael doesn't know "hue," "warmth," etc.). When formulated thus, do you still think this is a failing of heterophenomenology?
HW Helper
P: 2,567
 Quote by loseyourname The ideal heterophenomenological neuroscientist (which can exist, since this is only a thought experiment) would look for the neural response that accompanies the human experience of "green" and "red" and be able to identify them in your hypothetical Michael. Even if he didn't know what they were, the ideal neuroscientist would. That is precisely why heterophenomenology does not restrict itself to using only verbal and behavioral reports as primary data.
Yes, but does heterophenomenology include the specific neural responses as part of that set S of subjective data? Michael doesn't know what his neurons are doing, so how could that be considered part of his subjectivity?

 Let me reformulate your thought experiment. Imagine Michelle, Michael's sister. She has been raised completely without any knowledge of geometry. One day, after turning 10, Michelle is shown a sheet of paper with two figures on it, a circle and an ellipse. When the heterophenomenologist asks Michelle what she sees, she cannot answer. She doesn't know the words "circle," "ellipse," "flatter," "round," or any other word that refers to traits of conic sections, nor has she ever seen any of these things (just as Michael doesn't know "hue," "warmth," etc.). When formulated thus, do you still think this is a failing of heterophenomenology?
Just because she doesn't know those words doesn't mean she couldn't explain the difference. She could motion with her fingers, or draw them. But there is no way to explain colors besides identifying them with objects (red is the color of a stop sign) or by words we've associated with certain colors (red is the warmest color). That is precisely the trouble that expereinces present, because they can't be defined completely in terms of structure or function. While two shapes with the same structure necessarily are identical, two colors with the same structural and functional roles (red and green in this example) are not necessarily identical.
Emeritus
PF Gold
P: 3,634
 Quote by StatusX Yes, but does heterophenomenology include the specific neural responses as part of that set S of subjective data? Michael doesn't know what his neurons are doing, so how could that be considered part of his subjectivity?
Heterophenomenology considers all relevant data, including neural responses, behavioral responses, and verbal reports.

 Just because she doesn't know those words doesn't mean she couldn't explain the difference. She could motion with her fingers, or draw them. But there is no way to explain colors besides identifying them with objects (red is the color of a stop sign) or by words we've associated with certain colors (red is the warmest color). That is precisely the trouble that expereinces present, because they can't be defined completely in terms of structure or function. While two shapes with the same structure necessarily are identical, two colors with the same structural and functional roles (red and green in this example) are not necessarily identical.
If Michelle can draw another circle and another ellipse to explain the difference between a circle and an ellipse, then Michael can paint one sheet of paper green and the other red to explain the difference between green and red. Either way, they're just referring to what they've seen by relating it to something else they can see. This experiment can be twisted to include any manner of objects or qualities that a human can visually perceive.

By the way, what makes you think that the colors green and red have the same structural/functional roles and more than circles and ellipses do? Green and red result from different wavelengths of light and evoke different nervous responses. I've never understood why antiphysicalist arguments are so obsessed with color.
P: 732
 Quote by loseyourname If Michelle can draw another circle and another ellipse to explain the difference between a circle and an ellipse, then Michael can paint one sheet of paper green and the other red to explain the difference between green and red. Either way, they're just referring to what they've seen by relating it to something else they can see. This experiment can be twisted to include any manner of objects or qualities that a human can visually perceive. By the way, what makes you think that the colors green and red have the same structural/functional roles and more than circles and ellipses do? Green and red result from different wavelengths of light and evoke different nervous responses. I've never understood why antiphysicalist arguments are so obsessed with color.
I can write down the mathematical formulae for a cricle and an ellipse.
I know of no such formulae for the way-red-seems and the-way-green-seems. Do you ? (NB-- not talking wavelengths).
HW Helper
P: 2,567
 Quote by loseyourname Heterophenomenology considers all relevant data, including neural responses, behavioral responses, and verbal reports.
This experiment is ideal, and I'm saying that the functional roles of red and green are identical. That means that the neural circuits that cause red reactions and the ones that cause green reactions could be interchanged, and Michael would still behave the same. Whether his experiences would be the same is a further question, and to even ask it would be to assume the incompleteness of heterophenomenology.

 If Michelle can draw another circle and another ellipse to explain the difference between a circle and an ellipse, then Michael can paint one sheet of paper green and the other red to explain the difference between green and red. Either way, they're just referring to what they've seen by relating it to something else they can see. This experiment can be twisted to include any manner of objects or qualities that a human can visually perceive.
How could you meaningfully replace circles and ellipses for red and green in this experiment? That would mean you'd have to create a world where circles and ellipses are switched, but every structural and functional property is unchanged. That is logically impossible.

 By the way, what makes you think that the colors green and red have the same structural/functional roles and more than circles and ellipses do? Green and red result from different wavelengths of light and evoke different nervous responses. I've never understood why antiphysicalist arguments are so obsessed with color.
In general they don't. But the point of this experiment was to point out that heterophenomenology says that the only reason we see red as different than green is because they have different functional roles. If their functions were identical, they would be interchangeable. But we (or at least I) know from experience that isn't true, that red and green are different because they look different. In other words, heterophenomenolgy is built from bare differences, and experiences are not.
 Emeritus PF Gold P: 8,147 To Dennett's CADBLIND machine, which can input frequency information from a screen and store it internally as numbers, the red and green in the example would be two different numbers. It compares by subracting the numbers; if it gets a nonzero difference, they are different. The machine could answer the question that they are different, and the only thing it would need to answer more questions is a vocabulary. Function has nothing to do with it. Apples of identical shapes might be red or green.
Emeritus
PF Gold
P: 3,634
 Quote by StatusX This experiment is ideal, and I'm saying that the functional roles of red and green are identical. That means that the neural circuits that cause red reactions and the ones that cause green reactions could be interchanged, and Michael would still behave the same.
So you start off with an impossible situation. Okay, let's see where you're going.

 How could you meaningfully replace circles and ellipses for red and green in this experiment? That would mean you'd have to create a world where circles and ellipses are switched, but every structural and functional property is unchanged. That is logically impossible.
Why? Where is the contradiction in saying that Michelle's neural circuits can't be switched so that every time she is shown a circle, she sees an ellipse? She would still behave the same, and you'd still have the same problem.

 In general they don't. But the point of this experiment was to point out that heterophenomenology says that the only reason we see red as different than green is because they have different functional roles.
No it doesn't. Heterophenomenology says that we should remain neutral about whether our subjects beliefs about their experiences are correct until their claims can be verified in some way. It isn't a model of consciousness or cognition; it's just a methodology.

 If their functions were identical, they would be interchangeable. But we (or at least I) know from experience that isn't true, that red and green are different because they look different. In other words, heterophenomenolgy is built from bare differences, and experiences are not.
Please rephrase that. I have no clue whatsoever what you are trying to say here.
Emeritus
PF Gold
P: 3,634
 Quote by Tournesol I can write down the mathematical formulae for a cricle and an ellipse. I know of no such formulae for the way-red-seems and the-way-green-seems. Do you ? (NB-- not talking wavelengths).
Can you write down formulae for the way an ellipse seems or the way a circle seems? You completely missed my point anyway. It was only that if Michelle had no concept of these things (she doesn't know any geometry and can't write formulae), they would be foreign to her and she would have no way of describing the difference between a circle and an ellipse, any more than Michael could describe the difference between green and red.
HW Helper
P: 2,567
 Quote by loseyourname So you start off with an impossible situation. Okay, let's see where you're going.
Please explain which part of it is impossible. If you really think this is true, you shouldn't have tried to see where I was going, you should have stopped right here and explained why the premise is flawed.

 Why? Where is the contradiction in saying that Michelle's neural circuits can't be switched so that every time she is shown a circle, she sees an ellipse? She would still behave the same, and you'd still have the same problem.
How could she be made to respond to circles in the same way she responds to ellipses? They have different structures. A simple explanation of what an ellipse is could verify that her perception is wrong. With colors, there is no way to determine that what Michael is seeing is really red, or if he sees it as green but calls it red.

Of course, a heterophenomenologist would probably say there is no meaningful distinction here. But this experiment, just like any other argument to take the hard problem seriously, can only help us look to our own experiences for evidence. We know that red looks like something, and could logically imagine it looking like something else, or nothing at all, and still filling the same causal roles. All I can do is point out exactly what you're denying when you say heterophenomenology is complete. A perfectly consistent theory could probably be constructed that ignores experience and explains all behavior, but it would be wrong, or at least incomplete.

 No it doesn't. Heterophenomenology says that we should remain neutral about whether our subjects beliefs about their experiences are correct until their claims can be verified in some way. It isn't a model of consciousness or cognition; it's just a methodology.
Again, the argument is about whether heterophenomenology is capable of completely decribing the mind. You keep going back to the claim that heterophenomenology is neutral, but we aren't talking about that. I'm saying there are real phenomena it will never be able to account for, whether it makes a judgement about their existence or not.

 Please rephrase that. I have no clue whatsoever what you are trying to say here.
We don't just know that red is different than green, or that red is the color these things are, and green is the color those things are. We know what they look like. Rosenberg talked about this. Of course, you can deny this and still produce a perfectly consistent theory. If we ignored high speed objects, Newtonian physics would work fine. But ignoring data to preserve an ideology is not something I want to do.
HW Helper
P: 2,567
 Quote by selfAdjoint To Dennett's CADBLIND machine, which can input frequency information from a screen and store it internally as numbers, the red and green in the example would be two different numbers. It compares by subracting the numbers; if it gets a nonzero difference, they are different. The machine could answer the question that they are different, and the only thing it would need to answer more questions is a vocabulary. Function has nothing to do with it. Apples of identical shapes might be red or green.
Who's side are you arguing here? Michael is functionally identical to one of those machines, except there is no question he experiences. His experiences of red and green are not accounted by merely saying he can distinguish them. If he sees red and green anything like we do, there is clearly a lot being missed.
Emeritus
PF Gold
P: 3,634
 Quote by StatusX Please explain which part of it is impossible. If you really think this is true, you shouldn't have tried to see where I was going, you should have stopped right here and explained why the premise is flawed.
Inverted qualia without any associated neural difference is empirically impossible. Even Chalmers admits as much, which is why I continue to find it odd that he and others still use it as an arguing point. It is logically possible, sure, but it is also logically possible that gravity can be inverted or that any number of physical laws could be other than what they are. It's like the epiphenomenal ectoplasm argument. I can imagine a physically identical universe with epiphenomenal ectoplasm in it. Physicalism cannot account for this ectoplasm, therefore physicalism is false.

 How could she be made to respond to circles in the same way she responds to ellipses? They have different structures. A simple explanation of what an ellipse is could verify that her perception is wrong.
Why should inverted qualia be constrained to colors, though? For all we know, Michelle and Michael could have a cousin Michel that experiences the perception of a buffalo every time he sees a horse. He still calls it a "horse" and uses the same descriptive terms we use. His neural responses are even exactly the same as ours. Logically possible, right? Just as much as inverted colors. But it aint gonna happen and its in fact rather irrelevant.

 With colors, there is no way to determine that what Michael is seeing is really red, or if he sees it as green but calls it red.
Same with anything. When aunt Marsha looks at the moon, she might really be seeing Mars. What she calls craters are actually the canals. What she calls the man in the moon is actually Mons Olympus and that George Washington face. How can we determine what she is really seeing if she says the same things we do when describing the moon and her neural activity is exactly the same?

 We know that red looks like something, and could logically imagine it looking like something else, or nothing at all, and still filling the same causal roles.
There are better arguments supporting the hard problem than this. Your argument is analogous to saying that we can logically imagine animism being true without any change in natural patterns and therefore any theory of natural causation is imcomplete if it cannot account for animal and vegetable spirits. If you're going to make an argument from direct experience, stick with direct experience. You've never directly experienced inverted qualia nor is there any reason whatsoever to believe that such a phenomenon could actually occur without any ability to detect it.

 All I can do is point out exactly what you're denying when you say heterophenomenology is complete.
Can you please find an example of me making such a claim? I'm pretty sure that all I've claimed is that heterophenomenology is the best method we have and that it can account for any phenomenon that any other known method can.

 A perfectly consistent theory could probably be constructed that ignores experience and explains all behavior, but it would be wrong, or at least incomplete.
Kind of like a theory of inverted qualia? Internally consistent and so logically possible, but can never actually be the case?

 Again, the argument is about whether heterophenomenology is capable of completely decribing the mind.
Since when? The only argument I've made is that if heterophenomenology can't do it, there is no other method I've ever heard of that can.

 You keep going back to the claim that heterophenomenology is neutral, but we aren't talking about that. I'm saying there are real phenomena it will never be able to account for, whether it makes a judgement about their existence or not.
I know, and you are saying this based on arguments that fail. Address the failings of these arguments or come up with better ones. The more important thing for you to do is to show me a method that can account for what you think heterophenomenology cannot. If you can't do that, you may as well complain about the imcompleteness of science because it cannot answer ethical questions. At least people that make those complaints try to think of methods that can answer ethical questions.

 We don't just know that red is different than green, or that red is the color these things are, and green is the color those things are. We know what they look like.
I think you're going to have to develop this concept of what it means to "know" a fact such as this. I'm not convinced yet that such knowledge is even factual knowledge. You certainly aren't using the word "know" in a way that it is typically used. It seems more accurate to say that we are "acquainted with" the quality of color perception, or something to that effect. Why don't we forget this inverted qualia stuff and move down this road? This one might actually get us somewhere.

 Rosenberg talked about this. Of course, you can deny this and still produce a perfectly consistent theory. If we ignored high speed objects, Newtonian physics would work fine. But ignoring data to preserve an ideology is not something I want to do.
Well, I'm glad to hear that you value your integrity so much. So do I.
HW Helper
P: 2,567
I have to admit I'm a little confused as to your where you stand, loseyourname. You don't really seem to be claiming qualia aren't real, or that they're within the scope of heterophenomenology. You're just saying that heterophenomenology will do as well as any other method in explaining them. If that's the case, I don't see why you have a problem with this thought experiment. I'm just trying to point out what it is that the method is missing. I'm not suggesting another way to answer these questions. I don't know if one exists that will answer them, but that doesn't mean we can't ask them.

There are still a few points I'd like to clear up though:

 Quote by loseyourname Your argument is analogous to saying that we can logically imagine animism being true without any change in natural patterns and therefore any theory of natural causation is imcomplete if it cannot account for animal and vegetable spirits. If you're going to make an argument from direct experience, stick with direct experience. You've never directly experienced inverted qualia nor is there any reason whatsoever to believe that such a phenomenon could actually occur without any ability to detect it.
There is no primary data for animism, but there is for qualia. And I don't quite understand how anything can be proven about qualia, such as inverted qualia being impossible. But even if it somehow could be shown empirically, it is still a priori possible that red and green be switched while preserving functional roles. Don't you see the importance of this? You can't switch the moon for mars, or circles for ellipses and preserve every last functional and structural role. It isn't even logically possible. You can have very confused people, but there is always a way to clear up the confusion by spelling out the terms precisely. If Mary points to a circle and says that one side is longer than the other, she is wrong and can be corrected. If Marsha points to the moon and describes the face she sees in enough detail (maybe excrutiating detail, down to the grains of sand, depending on the extent of her delusion), we'll find out she's talking about something else.

The reason is that we can verbalize structural and functional differences. But how could you ever verify that you see red the same way I do? There is no way, which is why the logical possibility that we could see it differently arises in the first place. The logical possibility is all that's important here. I'm not saying they could actually be different in this world, I'm saying that nothing in the definition would make this logically impossible.

Now, getting back to Michael, I have asserted he responds to green and red in exactly the same way. The neurons attatched directly to his eye obviously react differently, but the way they are connected to the rest of his brain is completely symmetrical, so that they could be interchanged without an effect on his behavior. Don't think this is possible for a human? Fine, use a machine like the one selfadjoint mentioned, as long as you're willing to admit that machine can experience. (unless you think an inherent assymetry to color processing is a necessary condition to experience) Now, I think that Michael (or the machine) has specific, distinct experiences of red and green. If so, then switching every instance of red and green in the world will not affect his behavior, but will affect his experience.

Let me go over this once more, questioning any assumptions I've made. We have a boy (or machine) that responds to two colors, but does so symmetrically. That is, if you switched every instance of the two colors, he (it) would behave identically. If this is agreed to be possible (for a very simplistic human, or if not, a machine), then we need to ask does he have an experience associated with the two colors? If so, switching the colors changes the experience. So the only question left is, has heterophenomenology allowed for this change? I don't think any reasonable interpretation of the method could. If every single aspect of his behavior is identical, everything the theory could possibly talk about is identical.

The only possible loophole is in that nerve right behind his eye. But the functions of the nerves are identical, so a switch between worlds only changes which of them is being activated in a given situation. For simplicity, we might assume the two circuits are mirror images of eachother, flipped about Michael's center. Then the question is, could reflecting Michael about his center (I'm starting to feel bad for this kid) have a meaningful impact on his subjective world, as heterophenomenology describes it?

Any reasonable physicist would have to say there could be no change. And yet, if Michael does have distinct experiences accompanying these colors, there must be one. And if Michael can tell them apart, he must have distinct experiences. This leads into the second issue:

 I think you're going to have to develop this concept of what it means to "know" a fact such as this. I'm not convinced yet that such knowledge is even factual knowledge. You certainly aren't using the word "know" in a way that it is typically used. It seems more accurate to say that we are "acquainted with" the quality of color perception, or something to that effect. Why don't we forget this inverted qualia stuff and move down this road? This one might actually get us somewhere.
This is a pretty deep issue. It seems to me that what we know is exactly what we experience. We know facts because we experience the thoughts about those facts. (note that I'm talking about what the experiencer knows at a given instant of time) Since what we experience can be correlated to what our neurons are doing, you might think we are restricted to the knowledge "in" our neurons, ie, that which could be extracted in a detailed scan of our brain. But we also "know" what the experiences are like. We know what it feels like to know a fact or see a color.

This may not be the traditional defintion of knowledge, but I think on a little reflection you'll see it's accurate. You can't justifiably claim that what we know is what is in our neurons. How do we have access to those neurons? This might seem like a ridiculous question to an eliminativist, who would simply say we are those neurons. But there really is a seperation. If we were just our neurons, why don't we know what all our neurons are doing, or for that matter, what our stomach cells are doing? How do we even know for certain that we are made of neurons and not chinese people? This might be getting pretty far from the original topic, but these are all important issues to the overarching mind problem.
P: 732
 Quote by LYN Can you write down formulae for the way an ellipse seems or the way a circle seems?
Yes -- in that it is the same thing. (Well, a circle looked at from
an angle appears elliptical,but that can all be dealy with mathematically too).

 You completely missed my point anyway. It was only that if Michelle had no concept of these things (she doesn't know any geometry and can't write formulae), they would be foreign to her and she would have no way of describing the difference between a circle and an ellipse, any more than Michael could describe the difference between green and red.
And *my* point (as per the original Mary gedanken) is that a complete
knowledge of maths, science etc leaves you with specific gaps in understanding experience. You would be able to predict what a dodecahedron looks like, without having seen one , but not what colours look like
or tastes taste like.

 Inverted qualia without any associated neural difference is empirically impossible.
I agree. It's a useless way of arguing for quaia anyway, since it implies they are epiphenomenal.

 I think you're going to have to develop this concept of what it means to "know" a fact such as this. I'm not convinced yet that such knowledge is even factual knowledge. You certainly aren't using the word "know" in a way that it is typically used.
Au contraire, people are always saying "you don't know what it is like"
(about childbirth, for instance) -- meaning that you have no first-hand experience.
Emeritus
PF Gold
P: 3,634
First off, I was trying to say that you can't write an equation to describe the experience of seeing an ellipse or a circle. That's the argument with colors, right? Because certainly you can write an equation that describes the physical underpinning that causes that experience. I fail to see a difference between the two categories of experience.

 Quote by Tournesol Au contraire, people are always saying "you don't know what it is like" (about childbirth, for instance) -- meaning that you have no first-hand experience.
Yeah, I know. We were just talking about this in a class I'm taking on why humans have such a propensity for war. What I was saying is that I don't think what you are talking about is factual knowledge. It is more of a personal acquaintance. I'm not going to go so far as to say it's a misusage of the verb "to know," simply because it is so commonly used, but clearly there is a distinction.

I'm reminded of an argument against the omniscient of God. It is said that God does not know what it is like to ride a bike. He has no legs or body, and even if you are Christian and believe that God took human form 200 years ago, there were no bikes. Since God doesn't know what it's like to ride a bike, God doesn't know everything and so cannot be omniscient. The theist responds by saying that this is nonsense; having the experience of riding a bike is not in the category of factual knowledge.

Anyway, let's get back to what we were talking about in my class earlier today. We read Susan Sontag's Regarding the Pain of Others over the weekend, a short book about photographic depictions of wartime atrocity and how these effect the viewer. Certain people in the class were arguing, drawing from Sontag, that unless you've been directly exposed to war, you'll never know war, you'll never understand war. I got to thinking, however, about all of the research I've done in the past about the US Civil War. I know the name of every commanding officer and the death tolls from every major battle. I've read the letters written by Abe Lincoln, describing his decision-making process. In fact, forget about me and consider men who are actually scholars of the Civil War, who have spent their entire lives looking at every single motivating factor that went into the creation of this conflict. These men know the Civil War far better than the rural farmer in western Georgia that had her farm blazed over during Sherman's March and saw her husband, both sons, and a brother die. They understand the war better, even though they never experienced it.

It seems compelling to conclude, in light of these and other cases, that there are really two different things meant by the verb "to know."