, Volume 76, Issue 1, pp 137–145

Reply to Nagasawa on the Inconsistency Objection to the Knowledge Argument


    • Department of PhilosophyWilfrid Laurier University
Critical Discussion

DOI: 10.1007/s10670-011-9284-0

Cite this article as:
Campbell, N. Erkenn (2012) 76: 137. doi:10.1007/s10670-011-9284-0


Yujin Nagasawa has recently defended Frank Jackson’s knowledge argument from the “inconsistency objection.” The objection claims that the premises of the knowledge argument are inconsistent with qualia epiphenomenalism. Nagasawa defends Jackson by showing that the objection mistakenly assumes a causal theory of phenomenal knowledge. I argue that although this defense might succeed against two versions of the inconsistency objection, mine is unaffected by Nagasawa’s argument, in which case the inconsistency in the knowledge argument remains.

In a recent paper Nagasawa (2010) defends Jackson’s (1982, 1986) knowledge argument from what he calls the “inconsistency objection.” Although the objection has taken several different forms (Stjernberg 1999, Watkins 1989), including my own (Campbell 2003), Nagasawa maintains they share the claim that Jackson’s hypothetical neurophysiologist (Mary) cannot learn anything new when she first sees colours if, as Jackson supposes, qualia are epiphenomenal. Drawing on Chalmers (2003), Nagasawa appeals to an acquaintance model of phenomenal knowledge to falsify this claim and thereby restore Jackson’s argument. In this reply I argue that although such an approach might succeed at undermining other versions of the inconsistency objection, it leaves mine untouched because I offer a slightly different version of the objection than Stjernberg and Watkins.

I’ll begin with a brief description of Jackson’s knowledge argument. According to the argument Mary is a brilliant scientist who, despite being cloistered in a black and white environment her entire life, learns all the physical facts about colour perception. When she is finally exposed to colours Jackson asks whether or not she learns something new about the colour experiences of others. He claims it is clear that she does, since she then learns what it is like to see colours. But if we grant this we are saying that Mary learns new facts about colour perception. This means her previous knowledge was incomplete, in which case there are more facts about colour experience than the physical facts, so physicalism is false.

Nagasawa works with Jackson’s (1986) version of the argument which I will reconstruct for convenience since I will refer to it later:
  1. 1.

    Mary (before her release) knows everything physical there is to know about other people.

  2. 2.

    Mary (before her release) does not know everything there is to know about other people (because she learns something about them on her release). Therefore,

  3. 3.

    There are truths about other people (and herself) that escape the physicalist story.


Jackson concludes from the knowledge argument that qualia are not only nonphysical, but are also epiphenomenal. Although he is not very clear about this, it seems obvious enough that Jackson’s reason for thinking qualia are epiphenomenal is that Mary notices no causal gaps or mysterious causal interventions in the physical processes involved in colour perception. If qualia were causally efficacious but nonphysical, and hence unknown to Mary, there would (barring a form of systematic overdetermination) presumably be mysterious causal contributions to physical processes that she could not understand or explain.

Nagasawa rightly claims that the main intuition of the inconsistency objection is that there is a tension between the knowledge argument and the claim that qualia are epiphenomenal. Although he acknowledges that there are some differences between the various articulations of the objection, Nagasawa claims they all share the following structure:
  1. 1.

    If epiphenomenalism is true, then qualia are causally inefficacious in virtue of their falling under a mental type.

  2. 2.

    If qualia are causally inefficacious in virtue of their falling under a mental type, then Mary cannot acquire new knowledge about qualia upon her release. Therefore,

  3. 3.

    If epiphenomenalism is true, then Mary cannot acquire new knowledge about qualia upon her release.

  4. 4.

    If the knowledge argument is sound, then Mary acquires new knowledge about qualia upon her release. Therefore,

  5. 5.

    If epiphenomenalism is true, then the knowledge argument is unsound, and vice versa (Nagasawa 2010, p. 42).


Nagasawa proposes that premise (2) is the key to the argument, and indeed, Watkins and Stjernberg are both quite explicit in their defenses of this claim, stating clearly that knowledge of qualia would require their causal efficacy. Nagasawa sees this as the weak link in the argument and sets to work showing it is false. He does this in two stages: first, by providing reasons to be suspicious of causal theories of knowledge generally; second and more interestingly, by proposing that Mary’s knowledge of colour qualia should be understood in terms of the special relation of acquaintance which is captured in Chalmers’ (2003) constitution thesis about phenomenal knowledge.

The constitution thesis claims that for a subject S and quale q, “S’s phenomenal knowledge about qualia [sic.] q is partly constituted by q” (Nagasawa 2010, p. 52). According to Nagasawa’s interpretation of this idea, “since there is an intimate constitutive relationship between phenomenal knowledge and its object, there is no causal gap between them.” He adds,

The constitution thesis is hypothetical but nevertheless plausible. In particular, it correctly characterises the idea that qualia are directly accessible by a knowing subject unlike objects in the external environment, without committing the problematic idea that qualia are numerically identical to phenomenal knowledge (Ibid.).

Chalmers uses a version of this thesis to support the claim that there is a special epistemic relationship between conscious subjects and their qualia, which he calls the justification thesis:

When a subject forms a direct phenomenal belief based on a phenomenal quality, then that belief is prima facie justified by virtue of the subject’s acquaintance with that quality (Chalmers 2003, p. 249).

Nagasawa makes use of this special form of acquaintance to undermine premise (2) of the inconsistency objection. The premise assumes Mary can have justified beliefs about her qualia only if there is a causal connection between her qualia and her beliefs about them. Chalmers’ justification thesis provides an alternative account of phenomenal knowledge that does not require this assumption. Hence, premise (2) is false and the inconsistency objection is unsound.

I am sympathetic with the justification thesis and the related idea that qualia are self-intimating.1 I also think that an appeal to the justification thesis, if defensible, may very well undermine the version of the inconsistency objection that Nagasawa considers. However, in the paper to which Nagasawa replies I took myself to be offering a different version of the objection than Watkins and Stjernberg.

The main difference between my version of the inconsistency argument and the others Nagasawa considers is that mine is not necessarily committed to a causal account of phenomenal knowledge. Hence, I do not rely on the claim that Mary cannot have knowledge of her qualia as encapsulated in premise (2) of Nagasawa’s presentation of the inconsistency objection. It is certainly understandable why Nagasawa would think otherwise. Part of my original discussion involves the claim that what we imagine when Mary first sees colours involves expressions of surprise and astonishment, and utterances such as, “So that’s what a sunset looks like!” And, as Nagasawa notes, I did say that “All of these descriptions of Mary learning something new, of having realizations, and presumably making exclamations (instead of saying ‘ho, hum’), encourage and exploit the intuition that qualia have causal efficacy” (Campbell 2003, p. 263). However, my aim was not to suggest that if qualia were epiphenomenal Mary couldn’t have knowledge of them. My intention was to distinguish between two ways of imagining what would happen when Mary enters the coloured world.

According to the first way we assume that qualia are causally efficacious. After all, when we initially ponder the thought experiment involving Mary we have no reason (yet) to think that her qualia lack causal efficacy. What most of us imagine is Mary expressing shock and astonishment about her experiences. And if we assume that her qualia are efficacious (as I suspect we all do when we first consider the thought experiment) Mary’s behaviour serves as evidence in a straightforward way for her knowledge of new facts. It does seem “just obvious,” as Jackson says, that Mary learns something new about the experiences of others. This secures the plausibility of the crucial parenthetical remark in premise (2) of the knowledge argument, in which case Jackson’s conclusion goes through. This is not to assume that qualia have causal efficacy, but only that one way of imagining the thought experiment does—the way that makes its conclusion so appealing.

However, this is not the end of the matter. In the remainder of my original article I considered an alternative way of thinking about Mary’s case that was designed specifically to anticipate arguments like Nagasawa’s. I said:

Let’s explore the possible objection that the changes in Mary’s behaviour are consistent with epiphenomenalism. The epiphenomenalist would say that qualia are without any causal role, and hence, that Mary’s utterances such as, “So that’s what a sunset looks like” … are caused by the physical antecedents of the qualia in question. This means the reason Mary says she has learned something new has nothing to do with her new acquaintance with colour qualia, but is due entirely to the fact that her visual cortex now operates the way cortices do in normal observers (Ibid.).

That is, the alternative way of thinking about Jackson’s thought experiment that follows in my earlier article assumes the kind of relationship between Mary’s qualia and her knowledge of them that Nagasawa describes in the justification thesis. Hence, one of the aims in my original discussion was to suggest that there is an inconsistency in Jackson’s argument even if we assume something like the constitution thesis. My argument did not depend on the assumption that qualia are efficacious.

In what follows I wish to expand on this second way of imagining the thought experiment in order to show that qualia epiphenomenalism, together with the basic conditions of Jackson’s thought experiment, erode the intuition that Mary learns something new. This is a distinct claim from the one Nagasawa attacks on Jackson’s behalf, and hence, represents a distinct version of the inconsistency objection.

So how should we imagine Mary if we abide by the constraints of the knowledge argument and the claim that her qualia are epiphenomenal? Well, consider again the kind of knowledge Mary is assumed to possess:

She discovers, for example, just which wave-length combinations from the sky stimulate the retina, and exactly how this produces via the central nervous system the contraction of the vocal chords and expulsion of air from the lungs that results in the uttering of the sentence “The sky is blue” (Jackson 1982, p. 130).

Since Mary has complete physical knowledge of these processes and qualia are epiphenomenal, she has complete causal knowledge about colour experiences. So what does this suggest about how we ought to imagine Mary’s reaction when she enters the coloured world? Are things any different from the way we imagined them when we naively assumed her colour qualia were efficacious? I think so.

Dennett’s blue banana trick (1991, pp. 399–400) is illustrative of the kind of concerns I want to articulate. Dennett proposes that as a prank we present Mary with a bright blue banana for her first colour experience. What most of us would expect since Mary has never seen yellow or blue before is that she will think the experience she has of the banana is as of a yellow object. However, Dennett proposes something different. He claims that Mary isn’t fooled and that she says the following:

Hey! You tried to trick me! Bananas are yellow, but this one is blue! … You have to remember that I know everything—absolutely everything—that could ever be known about the physical causes and effects of color vision. So of course before you brought the banana in, I had already written down, in exquisite detail, exactly what physical impression a yellow object or a blue object … would make on my nervous system. So I already knew exactly what thoughts I would have … I was not in the slightest surprised by my experience of blue… (Ibid.).

If we assume that Mary is in possession not just of all the physical facts but all the causal facts, it is difficult to disagree with Dennett’s way of imagining Mary’s response to the blue banana trick. It does seem as though she would be able to predict her own behavioural reactions to coloured objects, including her utterance, “Bananas are yellow, but this one is blue!” If we add to this Churchland’s (1985) image of Mary having the ability to introspect her own brain states, Dennett’s portrayal of the blue banana trick is well nigh irresistible.2 How could she fail to identify the correct colour if she knows what effects light wavelengths of 450 nm have on normal brains (including her own) and exactly what behavioural effects this will produce in her, including utterances about colours?

Assuming that Mary’s causal knowledge is complete, this should lead us to think differently about Jackson’s thought experiment. Given Mary’s fantastic grasp of the effects of coloured objects on the nervous systems of normal observers it would appear as though she would not (contrary to the way most people initially imagine the thought experiment) express any shock or surprise when she first sees colours. It is entirely reasonable—in fact required by the parameters of the thought experiment—to assume that Mary would know in advance what she would say (and how she would act) when first stimulated by wavelengths of light that fall within the visual colour spectrum. If Mary knows how she will react to colours there doesn’t appear to be room for the idea that she would utter expressions of surprise and astonishment, for everything she would say and do in response to colours must have been an open book to her during her confinement in a black and white environment. If one imagines otherwise one is not following Jackson’s instructions and is assuming that there are behavioural effects of light stimulation Mary couldn’t know about in advance—one is denying she has complete causal knowledge and illicitly assuming that her nonphysical qualia are playing a causal role in her behaviour.

This has an important implication. We have to say in light of Mary’s lack of genuine surprise and astonishment that it is no longer “just obvious” that she learns something new about colour experience. Indeed, I think the blue banana trick does this sufficiently, but when we generalize this way of thinking about the thought experiment it is clear that there is room for doubt about Jackson’s central premise. If Mary can predict every reaction to colour it isn’t obvious that Mary learns something new. When we imagined things the first way, which involved the assumption that qualia are causally efficacious, it did seem obvious that Mary learned something new. We could take her reactions as evidence for her acquaintance with new facts, but it turns out now that we were imagining things incorrectly. When we treat qualia as epiphenomenal we eliminate the kind of evidence that made it seem so obvious that Mary learns new facts.

One might object to the above line of argument in the following way. Although Mary can predict her behavioural reactions with perfect accuracy, this doesn’t preclude the possibility that she is genuinely surprised. Why can’t she correctly predict that she will be (and act) surprised? If I tell my class that there will be a surprise quiz next week they can’t be surprised that there is a quiz, but they surely can be surprised by what’s on the quiz. So analogously, although Mary can know what her reactions to colour will be ahead of time, why can’t the character of her experience still astonish her?

This is somewhat strange territory. It is unclear what one should say about the possibility or extent of Mary’s surprise when she can predict her own behaviour. I see no internal contradiction in the claim that she could predict her being surprised, but I suspect many will have the intuition that something is amiss in such a situation. Related considerations do seem to tell against this defense of Jackson’s argument. If we generalize the implications of the blue banana trick it is difficult to see how there could be any reason to think Mary is astonished about anything. If we were to show her a child’s coloured building block (or some other object not associated with any particular colour in the way bananas are), we would expect her to fail to identify which side of the block is which primary colour. Yet, if Mary possesses all the relevant causal facts she would know which colour word she will (and should) utter when presented with each side of the block. She knows what wavelengths of light are being reflected off of each surface, she knows exactly how these stimulate the visual systems of normal perceivers, and she knows exactly how this produces assertions about the colour of each surface including which colour terms are uttered. There is no reason to think she would fail to identify the colour of each side of the block. Under these conditions does the character of her experience surprise her? Is it “just obvious” that she learns new facts, or do we begin to suspect that she might have known all along what these colours look like? It seems to me there is plenty of room for doubt about how to answer these questions, and that’s all that is needed to resist the knowledge argument.

But perhaps this is thinking about things in the wrong way. In imagining the thought experiment I have been approaching it from the standpoint of a bystander who observes Mary’s behaviour once she enters the coloured world. If we adopt Mary’s perspective, perhaps things are different. Indeed, one might think that all that is needed to support Jackson’s conclusion and defeat the inconsistency objection is the claim that Mary is undergoing a colour experience for the first time. When this is paired with Chalmers’ constitution thesis we know she is having phenomenal knowledge since such knowledge is partially constituted by the presence of a phenomenal property. How she behaves, or whether she expresses surprise or learning behaviour is beside the point.

In response, I think it bears pointing out that there is something illicit about adopting Mary’s perspective when we imagine the thought experiment. After all, what Mary is supposed to know is so far beyond what any of us can realistically imagine that we cannot successfully put ourselves in her shoes without running the risk of begging the question. That’s one reason for adopting the perspective of the bystander. A second is that even if we admit that Mary’s phenomenal beliefs are justified, the question remains whether she is then in the possession of new knowledge. Certainly, the constitution thesis tells us once she is in the coloured world Mary knows what it is like to see blue (or whatever), but the issue here is whether or not this is something she could have known before. I have been arguing that from the perspective of a bystander there is plenty of room for doubts about this issue, and in fact, that there is plenty of evidence to defeat the claim that what Mary knows upon her release into the coloured world is new knowledge. If Mary doesn’t fall for the blue banana trick and can correctly identify any colour presented to her it isn’t clear that she acquires new knowledge. Since the source of these doubts all stem from the completeness of Mary’s causal knowledge what’s getting in the way of Jackson’s crucial premise is the claim that qualia are epiphenomenal—hence the inconsistency in his argument.

If the above tactic seems excessively verificationist it bears noticing that I have not employed verificationism in a way that Jackson should find objectionable. After all, Jackson himself makes use of behavioural evidence to draw conclusions about another’s qualia. His original version of the thought experiment involving Fred depends on our having the ability to verify his unique ability to see two different shades of red we cannot. Fred’s claim is taken seriously only because he can pass certain behavioural tests, such as reliably sorting the same red objects into the same two groups (Jackson 1982, p. 182). Were he to fail such tests we would have no reason to think he has a richer or more complex phenomenal life than the rest of us. This suggests there is nothing illegitimate about drawing conclusions about an agent’s qualia on the basis of their behaviour.3 If Mary’s qualia are epiphenomenal we have plenty of behavioural evidence to suspect that she already knew what it was like to see colours, and so Jackson’s argument fails.

My overall concern then, is that if we carry out Jackson’s thought experiment in a way that does not assume qualia are efficacious, it is unclear whether or not Mary learns new facts when first exposed to colours. I have not said anything to imply that Mary’s epistemic relationship to her qualia requires their causal efficacy. That is the difference between my version of the inconsistency objection and Stjernberg’s and Watkins’. All three of us challenge the veracity of Jackson’s second premise—the second parenthetical remark in particular:

Mary (before her release) does not know everything there is to know about other people (because she learns something about them on her release).

Watkins and Stjernberg attack this claim by denying it is possible for Mary to learn new facts (about colour qualia) because this requires a causal relation between Mary’s beliefs and her qualia that is precluded by Jackson’s endorsement of epiphenomenalism. My strategy is different. I have not tried to show that if qualia are epiphenomenal Mary cannot have knowledge of them. Instead, I have distinguished two ways of imagining the thought experiment. According to the first way, if we assume that qualia are efficacious it does seem obvious that Mary acquires new knowledge. And, as I emphasized in my original paper, Jackson himself describes the thought experiment in ways that encourage precisely this way of imagining what happens when Mary first sees colours (1986, pp. 291, 292). However, if we imagine the thought experiment in the second way, according to which qualia are epiphenomenal, there are plenty of reasons to doubt that Mary learns new facts when first exposed to colours. So for premise (2) of the knowledge argument to seem plausible we need to imagine the thought experiment in the first way and deny epiphenomenalism, which renders Jackson’s position inconsistent. But if we assume qualia epiphenomenalism in an attempt to avoid the inconsistency our intuition that Mary learns something new fades and the knowledge argument collapses.

I applaud Nagasawa’s attempt to defend Jackson’s argument. If Chalmers’ justification thesis can be maintained it certainly seems to undermine two versions of the inconsistency objection. However, Nagasawa has failed to notice that there are important differences between my version of the objection and these others. There is still good reason to think the knowledge argument is inconsistent.


Though for a contrary treatment of qualia see (Rosenthal 2002). Those who follow Rosenthal think there can be unconscious qualia, which suggests that ones epistemic relationship to qualia must be more complex than the relation of constitution.


To deny Mary has such knowledge of the workings of her own brain seems to require the denial of the claim that she possesses all causal knowledge about vision.


Notice how the thought experiment about Fred also seems to encourage the idea that qualia are causally efficacious. Jackson seems to depend on the idea that Fred’s special discriminative abilities are caused by his unique qualia.



My thanks to two anonymous referees for their helpful comments on an earlier draft of this paper.

Copyright information

© Springer Science+Business Media B.V. 2011