Responding To The External World Problem
Michael Vossen

The canonical argument against knowledge, first presented by Descartes in 1641, is still very much alive in contemporary epistemology. This is not to say that work in epistemology is stagnant, on the contrary, there has recently been much progress on the problem, and many responses have been put forward. In this paper, I will look at the response from content externalism, otherwise known as semantic externalism, and argue that it is currently the best response to the external world problem. Here, I take the best response to the external world problem to be the response that concedes as little as possible of our conception of ourselves as knowers and our beliefs about logic. This response also needs to be justified, and true or likely to be true. I will then analyze the problems with content externalism, suggest some possible revisions and responses to these problems, and mention how it compares with some of its competitors. First, though, I will clearly present the canonical argument as it appears today, along with the motivations for its premises.

The canonical argument claims that I don’t know that any beliefs based on my sense perceptions are true. Further, this undermines all knowledge I base on sense perceptions. This means that I don’t know that there is a glass of water in front of me, even though I am looking at one now. The reason I don’t know that their is a glass of water in front of me, the argument claims, is because I don’t know that it is not some sensory illusion, such as a hologram, a dream, or some image or sensation placed in my mind by a deceiver or scientist. Further, all of my perceptual experiences could be deceptive, not just my perception of a glass of water. And if I don’t know that my perceptions are not sensory illusions, then I don’t really know that my beliefs based on them are true. Thus, I don’t know that any of my sensory induced beliefs are true. This is the intuitive basis for the canonical argument. Contemporary epistemologists will put this argument into logical form as to elucidate its assumptions better. In logical form, the canonical argument looks like this:

C1. For a subject S to know some proposition P, S must know that they are not in
a skeptical scenario.
C2. S cannot know that they are not in a skeptical scenario.
C3. S cannot know that P2

Here, we can analyze the premises. The reasons for holding the first premise night not immediately apparent; why should anyone believe premise C1? Well, premise C1 is based on the principle of epistemic closure. It seems true that if a person knows that p and knows that p entails q, then that person knows q. For example, if a person knows what a bachelor is, and knows that being a bachelor entails being an unmarried man, then the person would know what an unmarried man is. With regards to my belief that there is a glass of water in front of me, if I know that p (there is a glass of water in front of me) and I know that p entails q (the glass of water is not a hologram), then I should know q (the glass of water is not a hologram.) If I don’t know that the glass of water is not a hologram, then I clearly don’t know that it is a glass of water. Thus, the epistemic closure principle, and the belief that something being B entails it not being a sensory illusion of B are the motivations for accepting premise C1. Finally, it should be noted that if the closure principle fails, then the current model of deductive thought is probably undermined as well. Problematic consequences with truth conditions seem to occur when this premise is rejected.

The details of premise C2 will vary comes from scenario to scenario to make the situation look more plausible, but the common theme behind this premise is to show that it is possible for all of a person’s sensory experiences to be illusions. In other words, her sensory experiences in a skeptical scenario would be identical and indistinguishable from her experiences in the real world. This premise is generally motivated by reference to real life counterparts, such as very explicit dreams, holograms, and scientist’s ability to reproduce sensory stimuli in the brain through chemicals. The skeptic will then use these to create a skeptical scenario that is or at least seems physically possible.

Finally, step C3 is the logical consequence of one and two, and if logic holds, then premise three is guaranteed to be true so long as premise C1 and two are true. Of course, if premise three is true, then we don’t know that any propositions based on our senses are true. Since our knowledge of the external world is based on these sensory perceptions, then if step C3 is correct, we do not have knowledge of the external world. As an aside, I think it’s interesting that the consequence of the canonical argument seems to deny its second premise. Take P to be the second step of the canonical argument. If the conclusion holds, then S cannot know that S cannot know that they are not in a skeptical scenario. This is only the case if premise C2 is not a-priori knowable, since the conclusion only challenges our knowledge based on sense experiences. However, we seem to use our knowledge about what is physically possible, which we obtain through sense experience, in affirming premise C2. If this consideration holds, then it seem like the canonical argument is self-refuting. 3

In addition to these considerations, it will also be helpful to keep Austin’s epistemic circularity considerations in mind for evaluating responses to the argument. Austin notes that you can’t use sense experiences or propositions based on sense experiences in responding to the skeptic, or you will beg the question against them. For example, a person could argue that we should accept our pre-skeptical model of perceptual beliefs because they somehow provide a better explanation for our beliefs than a skeptical scenario. Arguments of this form could take many shapes; they could argue that skeptical scenarios are in some way parasitic on our ordinary beliefs, that the skeptical scenarios assume more entities than our ordinary beliefs, that our ordinary beliefs are more likely to be physically possible, etc.

The problem with this type of response is that in arguing that one explanation is better than another, some explanations are considered more likely to be true. This is to say that theories that are simpler, have a lower number of entities, are physical possibility, etc are likely to correspond with our beliefs about the world. This seems to beg the question against the skeptic, for in favoring one theory over another it assumes that the world corresponds with our beliefs founded from sense perceptions. To put it another way, we think that some considerations are more likely to be true because we can deduce from our sensory founded beliefs that they are more likely. If we were in a skeptical scenario though, our sensory founded beliefs could lead us to deduce that different considerations are more likely to be true. Thus, we implicitly rely on our sensory beliefs to justify our response, and beg the question in responding to the skeptic.

Given that the canonical argument is a deductive argument to which the conclusion’s truth follows from the truth of its premises, and that an acceptance of the argument’s conclusion is incompatible with our intuitive self-conception and everyday experiences, we have three options in responding to it.4 We could either accept the conclusion, or deny one or two of the premises. If we deny premise C1, we loose the principal of epistemic closure, and thus must re-think our beliefs about logic. If we accept the conclusion, then our self-conception as knowing the external world is lost, as well as our beliefs about it. And the second premise seems obviously true to many people after reflection, such that there is doubt as to how we could reasonably reject it. We are thus faced with a dilemma as to how to minimize the loss of our intuitively held beliefs in responding to the skeptic. As mentioned earlier, I take the best response to external world skepticism to be the response in which the loss is minimized, the response is justified and true or likely to be true. Any response that can successfully challenge premise C2 would be the ideal response, as we would lose nothing in denying it. After that, responses which concede as little as possible are preferred, though they would still seem like a loss to the skeptic.

Content externalism is a response to skepticism that challenges premise C2. First developed by Hillary Putnam in his paper “Brains in Vats”, this response argues that by reflecting on the “...preconditions for thinking about, representing, referring to, etc” we can determine that skeptical scenarios, or at least the scenario presented by Putnam, are conceptually impossible.5In particular, Putnam thinks that the contents of a subject’s thoughts and utterances are determined, at least in part, by the relations they bear to circumstances in their environment.6 Putnam argues that beings who haven’t experienced trees before don’t really have the concept of trees; to obtain the concept Q, a subject must have in some way casually interacted with Q. Putnam leaves open how this is done, which leaves open the possibility for causal chains such as one subject transmitting a concept to another among other things. Putnam uses two thought experiments to motivate his claim.

In the first, Putnam asks us to imagine a planet with beings much like us, the difference being that the inhabitants of the planet have never experienced trees before. Putnam then has us imagine that a spaceship passing over the planet drops a piece of paper, on which paints have spilled in such a way as to form a perfect representation of a tree. Imagining that the planet’s inhabitants find the paper, Putnam asks whether the inhabitant’s mental image of the object represented by the picture represents a tree to him. Putnam answers that no, the mental image does not represent a tree to him, though it would to us. Similarly, Putnam asks us to imagine that a person is hypnotized such that they could speak fluent Japanese when they spoke, such that the person would correctly converse in Japanese when spoken to. Putnam thinks that this person, although he can speak correctly, does not know the meaning of his words. By combining the conclusions of these thought experiments, Putnam concludes that someone can think words which are in fact a description of X in some language and simultaneously have the appropriate mental image, but neither understand the words nor know what X is. Putnam is then able to derive his anti-skeptical argument.

Putnam’s argument against premise C2 of the canonical argument can formally be expressed thus:

P1. Either I am a brain in a vat, or I am not a brain in a vat
P2. If I am not a brain in a vat, then the thought I express by uttering ‘I am a brain
in a vat’ is false.
P3. If I am a brain in a vat, then the thought I express by uttering ‘I am a brain in
a vat’ is false.___________________________________________________
P4. The thought I express by uttering ‘I am not a brain in a vat’ is false.7

The law of the excluded middle justifies premise P1. Premise P2 seems correct on reflection; if I’m a person sitting in front of a computer, then it is false that I’m a brain in a vat, and any thought whose content expresses that I am a brain in a vat is false. Premise P3 is the most controversial premise of the argument, and the one most motivated by Putnam’s thought experiences. Premise P3 is justified because the thought ‘I am a brain in a vat’, thought by a brain in a vat, expresses a different content than the thought ‘I am a brain in a vat’ thought be the non-envated person. The thought expressed by the brain in a vat refers to the configuration of electronic impulses used by the vat system to simulate brains and vats. The thought expressed by the non-envated person refers to actual fleshy brains glassy, wiry, chemically vats. Since Putnam thinks that the words used by the brain in a vat and the non-envated person refer contingently, the brain in a vat and the non-envated person could use the same words in expressing the thought “I am a brain in a vat’, yet the sentence will have a different meaning depending on the speaker. In P3 the thought expressed is false, since the content of the thought uttered by the brain in a vat does not refers to actual fleshy brains and glassy, wiry, chemically vats, but to vat system stimulations. The conclusion P4 implies that whenever I assert that I am a brain in a vat, I am speaking a false statement. Thus, I can realize that it is true of me to think that I am not a brain in a vat. Thus I can know that premise C2 is false for any skeptical scenario in which I am a Putnamian brain in a vat.

There are some problems with the theory of content externalism, and this has lead to criticism of content externalism as a response to the canonical argument. One non-useful but strongly made criticism that underlies some criticism is that content externalism seems to simple, or is a semantic trick. For example, McGinn sarcastically writes “I can achieve the anti-skeptical result DesCartes needed God to vouchsafe by exploiting considerations about what determines content. Ah the wonders of analytical philosophy! It is an optimistic man who believes that Cartesian skepticism may be refuted in this way.”8One possible motive for this type of reply could be its target, premise C2. Some philosophers doubt that any a-priori argument cannot challenge premise C2, though the reasons for this doubt rely more on the failure of a-priori arguments thus far to produce a result rather than an explanation about the capacities of a-priori argumentation. If these philosopher’s intuitions are correct, then it might be useful in considering the canonical argument to know why such a-priori arguments must fail.

A more serious problem Putnam faces is that his conclusion as stands is somewhat limited in its scope. While his argument may refute the particular scenario he considers, in which everyone is a brain in a vat hooked up to a machine the simulates the world, some scenarios escape the argument. For example, the skeptic could argue that up until yesterday, I was in the real world perceiving objects normally. I thus would acquire all my concepts correctly; my thoughts about glasses of water that a glass would refer to the object that we treat as real. The skeptic will then argue that yesterday, governmental scientists kidnapped me while I was sleeping, removed my brain, and placed it in a vat filled with the necessary electrical equipment to reproduce all the experiences I have had since then. Thus, I might be deceived in thinking that I see my concept of water when I’m actually seeing a simulated version. This problem is potentially solved by investigating into more depth on how transferring concepts of one language to another really works; if proponents of content externalism are to respond to this criticism, they need to show that their is some noticeable change when a person’s concepts change.

Finally, the strongest criticism against Putnam’s argument is that it actually makes thinks worse, as a revised skeptical scenario could show that we don’t know the content of our own language. The skeptic could argue that I don’t know whether the thought I express when I utter ‘I am not a brain in a vat’ refers to the language of the brain in a vat, or the non-envated person. Formally, this results because in P2 and P3, Putnam has the thought ‘I’m not a brain in a vat’ refer to the language of the non-envated speaker. In evaluating this response, we should note that this reply does not refute Putnam’s view, but shows that the skeptical scenario is different if he is correct. In other words, the truth or falsity of this response does not alter the truth of Putnam’s argument. Some part of his theory of reference will need to be shown to be wrong if it is to be challenged. If Putnam’s theory holds though, he has changed the type of skepticism that we are responding to. As such, we should try to resolve skepticism with this revised argument in mind.

9One question we might ask about this new scenario is whether it needs to be resolved a priori; do responses beg the question in using sensory experiences. I think surprisingly, the answer might be no, we can use our sensory experiences in responding to this scenario. In the revised skeptical scenario, I don’t know whether I am a brain in a vat, and my language is vat language or if I am a non-envated being and my language is non-vat language. However, whatever I refer to in my language represents what it is I have experienced. For example, if I am a brain in a vat, and I think ‘there is a glass of water in front of me’, this thought refers to the vat system stimuli that I experience every time I have water. Likewise, if I’m not en-vated, my thought refers to the glass of water. In either case, my thought necessarily refers to whatever system caused it; I could not have the same thought in a different language without causally interacting with the sensory causing mechanisms in the right way. Then, I could know that a glass of water is in front of me in the same way that I think that I can know that a glass of water is in front of me without knowing what the glass is made of. The negative result of this thought is that does not do much to resolve the question of whether or not we are brains in vats. The positive result is that in a Putnam type scenario, my thoughts refer to the things that I think they do, though I still can learn about the thing’s nature. Thus, we might be justified in using empirical methods to investigate the nature of our words, though it’s not apparent that the investigation will be helpful in determining whether were are brains in vats or not as it would presumably come up with results using the same words in both languages. The main positive result of this is that we avoid the switching of realities the skeptic uses in undermining knowledge. In response to the skeptic who says that I am a brain in a vat, I could concede that the glass in front of me, in addition to having the perceptual properties I think it does, could also have the property of being produced by a vat system. This doesn’t mean that the glass is something other than what I have always thought it is; it just means that there could be more to learn about the glass.


Notes

1. I’m using all four of my extension days on this paper even though I only have three days to use them on, so imagine what one extra day would do in improving this paper when reading.
2. Taken From Class Notes
3. My apologies if there is something obviously wrong with this response that we covered, but I don’t recall anything like it, and I thought I’d throw it out there.
4. Or at least three responses in dealing with the argument’s logic.
5. Brains in Vats, pg.38
6. From Class Notes
7. Ibid
8. A Priori Knowledge of the World: Knowing the World by Knowing Our Minds, pg.76
9. I’m not convinced that this response works, but I thought it might be interesting to work out the details of this sort of reply to the revised skeptical scenario. Take this as explorative hypothesizing and not my definitive view.

Works Cited

Putnam, Hilary. Brains in Vats. In Skepticism A Contemporary Reader. Edited by Keith DeRose and Ted Warfield, Oxford University Press, 1999.

Bruekner, Anthony. Semantic Answers to Skepticism. In Skepticism A Contemporary Reader. Edited by Keith DeRose and Ted Warfield, Oxford University Press, 1999.

Forbes, Graeme. Realism and Skepticism: Brains in a Vat Revisited. In Skepticism A Contemporary Reader. Edited by Keith DeRose and Ted Warfield, Oxford University Press, 1999.

Warfield, Ted. A Priori Knowledge of the World: Knowing the World by Knowing Our Minds. In Skepticism A Contemporary Reader. Edited by Keith DeRose and Ted Warfield, Oxford University Press, 1999.