Frank Jackson's Knowledge Argument

| | Comments (6)

This is the 51st post in my Theories of Knowledge and Reality series. The previous post discussed a different kind of dualism from Descartes' interactionist substance dualism. To avoid objections against Descartes' view, some philosophers propose epiphenomenalist property dualism. They argue for this view based on facts about the first-person perspective that can't be reduced to third-person facts accessible to science. This post looks at Frank Jackson's formal argument for that thesis. 

Consider someone who can discern shades of red better than we can. Show him two tomatoes that look the same color to the rest of us, and he'll be able to see them as two different shades. He will consistently separate the same tomatoes as being red-1 and the others as being red-2, no matter how many times and how well you mix them up again. There's something about his perception of colors that we can never know, even if we know everything about his brain and how it works. There's a fact that remains - what his experience is like for him.

Consider a color scientist named Mary who has never seen red. She lived in a black and white environment with special contact lenses all her life, so she'd never seen most colors.  Then she went on to learn the neuroscience of color perception. She now knows everything there is to know from science about color perception. She knows what color words apply to which wavelengths of light. She knows what goes on in the brain when people see various colors. But she's never seen red. Then she takes off the contact lenses, and someone gives her a tomato. She now sees red for the first time. Does she learn something? Jackson says she does - what it's like to perceive the color red.

  1. Mary knows every physical fact about color perception.
  2. There's a fact about color perception that Mary learns when she sees red - namely, what it is like to experience seeing that color.
  3. Therefore, there are more than just physical facts (so materialism is false).

There have been a few traditional ways of resisting this conclusion.

  1. If materialism is true, maybe we shouldn't expect Mary to learn anything new.  If this is right, we should expect her to see red for the first time and say "Ah! That's exactly what I expected it to look like."  That seems highly implausible.
  2. David Lewis suggests that Mary doesn't learn a new fact but just gains a new ability - how to recognize red from within. She could identify red before in different ways, and she's gained a different way to identify it. It's like learning a new language, only more complicated. You can say the same facts in a different language once you've learned it, but hearing something in German that you already knew in English doesn't mean you've learned a new fact. Some philosophers call this implausible also, since language learning is just translating things we knew into different representations, but this is a totally new experience. There's got to be something more to seeing red than just having a certain ability.
  3. Some have suggested that Mary gains a new concept but doesn't learn anything new. She has a new way to express what she already knew - in terms of color experiences now, whereas before she just had the concepts involved with wavelengths, brain waves, neurons, and human behavior. But is this going to be successful? Mary seems to gain some new knowledge about color perception. Gaining just a concept doesn't seem enough. Something about the new experience seems to suggest more than just gaining a new way to think about something she already knew.

In the end if Jackson is right, you get dualism. You might think it's the best of both worlds. It avoids the simplicity arguments against dualism, since it doesn't require actual things in the world that are non-physical. It just requires some feature of me, a physical being, to be a non-physical property. So the view is called property dualism. The standard dualist view, substance dualism, holds that there is a real thing that's part of me - an immaterial soul or mind. Also, this view avoids conservation law problems. According to our best science, matter and energy can't be created or destroyed. If something comes in from outside the physical  order and interferes, this law would seem to be broken. But property dualism just says there are features of physical things that it wouldn't be right to call physical. The natural order of things continues on as normal. Nothing outside the natural order needs to come in and affect the physical world. So someone can honor dualist intuitions and have a view that's not materialist but seems to avoid the dualist's problems. Some people think they're trying to have their cake and eat it too, but Jackson and Nagel see this as the best of both worlds.

The other way around the Knowledge Argument is to deny the first premise. Mary doesn't really know all the physical facts about color perception. She does know all the impersonal facts, facts you can know independently of experiencing the color through perception. But maybe these experiential facts are still physical facts, just not impersonal ones. This does get out of the argument, but for some reason many materialists don't take this way out. It might be because they see people who take this line as abandoning one of the motivations for being a materialist in the first place. The whole idea was to get a theory according to which you can understand all of reality in scientific terms. That's why we want to avoid dualism, since that goes beyond science. This approach abandons that idea. Science can't capture all the truths, even all the physical truths. The other ways of avoiding Jackson's argument try to hold on to that notion. This one abandons it. It could be right, but as a materialist view it seems less in line with materialism as a whole, since it loses one key reason for being a materialist.

One response to this argument might be that it's not in principle impossible to get all the facts, even first-person ones. We lack the technology, but it seems possible with virtual reality. We could give someone the same brain state as someone else. This might take a lot of work, and it might be difficult to get the person to remember it when  you restore them to their previous brain state, but it seems in principle possible to give one person the same inner feeling another person has, provided we figure out how to manipulate neurons, transform brain matter to match how another person's brain is physically arranged, and so on. It probably wouldn't take changing the whole brain, just the parts necessary for conscious experience. This does rescue at least some of the idea that science can in principle capture all facts about the universe, and any investigator could eventually in principle do what's necessary to know any fact. It would take something far more radical than just what I described above, though. After all, we would have to be able to experience for ourselves what it's like to be a bat, a bee, or any other organism that has conscious experience, even the ones with minimal experience. To get ourselves so that we could do that, we might have to modify our brains so radically that we're not really us anymore, depending on your view of personal identity. So this response has something to say, but it's not clear that it goes all the way.

Another hesitation a materialist might have at this response is that this isn't what people meant by science capturing all the facts about the world. The original idea was to list all the facts resulting from external, third-person investigation, measurable entities you can quantify. If you can't simply list off all the facts, even if you have the potential to have all the possible first-person experiences anything could have, then you can't even in principle give a scientific account of the world in third-person terms. It's that kind of description of the universe that many materialists want science to come up with, and if Jackson is right that these first-person facts are additional facts, that ends up being impossible.

In the next post, I'll look at one further mind-related issue before turning to personal identity: artificial intelligence.


Hello Jeremy. I was glad you devoted space to the response which says that the third person facts don't exhaust the physical facts. Given that so-called impersonal (or objective or third-person) facts are derived from intersubjective agreement about first-person experiences, why would one think the physical world is somehow constituted exclusively by impersonal facts? The stance, sometimes called Russellian monism, dissolves the conceivability argument as well, without committing one to an ontologicial dualism.
Best regards,
- Steve Esser

I don't think I would be so quick to dismiss response #1 as implausible. It's really no more implausible than the thought experiment itself. I have to side with Daniel Dennett on this one; we're dealing with a Grade A "intuition pump" here. We can't put so much trust in our knee-jerk assumptions about what someone with superhuman knowledge can and can't do.

Science is a set of descriptions, it is not the things it describes so maintaining that "science is wrong" because a theory of the motion of a ball is not the same as a ball itself is just a crazy non-sequitur. If conscious experience is a real thing, like a ball is a real thing, then all the descriptions in the world, scientific or not, will not be that conscious experience. (See Superscientist Mary and types of physicalism ).

Your point about using science to manipulate the brain so that it has particular experiences seems valid to me. However, what we really need is a theory of how the brain avoids the regress that is implicit in materialist theory (See Materialists should read this first). If we can discover this theory we will make a quantum leap forward.

Not sure what you mean. I don't see anything here confusing theories with the things they're about. The argument takes there to be facts about what it's like to experience something, and the question is whether those facts are physical. If you can know all the physical facts and then learn these, then they aren't physical facts. This isn't a confusion of the experience with facts about the experience. It's simply a question about whether those facts (about the experience) are physical facts.

There's also no assertion of science being wrong here, just that it would be incomplete if there are facts about first-person experiences that it can't include because it's only done from the third-person perspective.

I missed this reply (3 years ago). I did largely agree with your piece. However, you seem to have misunderstood my first point, which was additional and complementary to your article. In my first point, which was that science is just a set of descriptions, not events in themselves I was drawing attention to the practical nature of physical knowledge.

We obtain physical knowledge by performing experiments. Scientists have direct contact with events through these experiments. The philosopher who thinks he has physical knowledge by reading and believing the descriptions given by scientists does not truly have "physical knowledge", he has knowledge of the relations between reported physical phenomena. He might swell his head by calling his book-learning "facts" but he knows very little of events.

Physical knowledge is not about "facts" it is about events. Philosophers have "facts", the world has events. As you point out in the article, Mary has her own event, her red. A description in a book is not this red.

A description is not "red" at all.

Now revisit your comment: "The argument takes there to be facts about what it's like to experience something, and the question is whether those facts are physical."

Facts ABOUT = symbols describing
Experience something = an event in experience

Therefore you seem to have argued that: there are symbols describing an event in experience and the question is whether these symbols are physical. This seems to be a trivial truth unrelated to materialism etc.

Lets try substituting events for "facts", the argument then becomes: there are events in experience, and the question is whether those events are physical. This makes more sense.

Lets carry on with "events" for facts, you say:

"If you can know all the physical facts and then learn these, then they aren't physical facts."

Substituting events:

If you can know all the events and then learn new events then they aren't physical events. Well, no, you could not have known ALL the events in the first place. Which is the substance of your article - I agree.

You say:

"This isn't a confusion of the experience with facts about the experience. It's simply a question about whether those facts (about the experience) are physical facts."


"This isn't a confusion of the events in experience with symbols about these events. Its simply a question about whether these symbols/events are physical symbols/events."

But events are events - whats all this about "physical" events?

Using vague words like "facts", "physical" and "knowledge" just confuses everything. There are events and these can be related to one another by symbols. We call the catalogue of events and relationships "scientific knowledge", the catalogue is obtained using the "scientific method".

Wait, are you saying that we could never make a distinction between physical events and mental events if dualism turns out to be true? So dualism is false, and we can know that a priori? That's a pretty significant claim you're making. I don't think most materialists would want to say such a thing. Dualism at least seems possible, even if it turns out to be false.

Leave a comment


    The Parablemen are: , , and .



Books I'm Reading

Fiction I've Finished Recently

Non-Fiction I've Finished Recently

Books I've Been Referring To