Qualia epiphenomenalism is the view that qualitative properties of events, such as the raw feel of itchiness or painfulness, lack causal efficacy. One common objection to qualia epiphenomenalism is the epistemic argument, which states that this loss of causal efficacy undermines our capacity to know about these epiphenomenal qualitative properties (Sterjnberg, 1999; Watkins, 1989). A number of rejoinders have been offered up to insulate qualia epiphenomenalism from the epistemic argument. In this paper I consider and ultimately reject two such replies, namely, the common underlying cause reply and the appeal to physical-effect epiphenomenalism.
This paper is divided into five sections. After introducing and motivating qualia epiphenomenalism (§1), I outline the epistemic argument against qualia epiphenomenalism (§2). I then address two common responses to the epistemic argument. First, the Common Underlying Cause Reply (§3) states that we have knowledge of our epiphenomenal qualitative properties by virtue of the fact that the physical properties that strongly subvene epiphenomenal qualitative properties cause the beliefs/memories that are appropriate to the epiphenomenal qualitative properties. I argue (§4) that this tactic fails because this relation between the qualitative properties and the beliefs/memories about the qualitative properties is indirect, and hence it is possible to render the counterfactual “Had my pain not occurred, my belief that I was in pain would not have occurred” false. If the beliefs/memories about the qualitative properties can occur without the qualitative properties having occurred, these beliefs/memories are unjustified and possibly false, so are not considered knowledge. I then consider Physical-Effect Epiphenomenalism (§5), according to which qualitative properties are epiphenomenal vis-à-vis physical effects but are efficacious vis-à-vis mental effects, so beliefs/memories about prior qualitative properties are justified. I conclude that if beliefs/memories are irreducibly mental they are epiphenomenal with respect to the physical domain themselves. On the other hand, if beliefs/memories are reduced to the physical, qualitative properties once again lack the ability to cause them to occur.
1. Qualia Epiphenomenalism
Here is a (seemingly) trustworthy saying: our desires and raw feelings influence our behaviour. We desire a hamburger, and not surprisingly, we eat one. We feel pain and so our hand moves away from the hot stove top. Can this common sense assumption that qualitative properties of events play a causal role in human behaviour be misguided? Plenty of philosophers argue that it cannot. Terence Horgan, for example, argues that mental inefficacy is an “outrageous offense against common sense” (Horgan, 1987, 504), while Jerry Fodor suggests that “if it isn’t literally true that my wanting is causally responsible for my reaching ... then practically everything I believe about anything is false and it’s the end of the world” (Fodor, 1989, 77).
On the other hand, there are some weighty reasons to take qualia epiphenomenalism seriously. According to two leading qualia epiphenomenalists (Gadenne, 2006; Robinson, 2004), qualia epiphenomenalism follows naturally from the following three principles: (1) physical causal completeness, which states that every event has a sufficient physical cause; (2) irreducibility, which states that qualitative properties of events are distinct from physical properties of events; and (3) the principle of causal exclusion, which states there can be no more than a single sufficient cause for any given effect. In support of physical causal completeness, William Robinson provides a thorough account of the physical transactions from muscle contraction back to the chemical activity surrounding neurons before noting:
The picture that these facts support is one in which all our actions are effects of extremely complex, but completely physical, interactions of the cells in our brains and the neurons that lead into and away from our brains. (Robinson, 2004, 158)
In other words, all physical effects, including behavioural effects, have a sufficient physical cause.
There is also ample reason to endorse (2) irreducibility.Footnote 1 Volker Gadenne alludes to several famous thought experiments—Thomas Nagel’s bat argument, Frank Jackson’s Mary argument, and David Chalmers’ zombie argument—that indicate there is a distinction between qualitative properties of events and physical properties of events. He then concludes that these thought experiments are not even necessary, since:
It is sufficient to point out that there are undeniable events, such as pains or afterimages, which we identify by features that are not physical. We do not find these features among properties like shape, volume, mass, or electrical charge ... Hence [qualitative mental properties] cannot be identical with (identity understood as sameness) or realized as physical properties. (Gadenne, 2004, 103)
Qualitative mental properties are discernibly distinct from physical properties—we cannot weigh out a pound of love, or measure out the positive/negative charge of hope (though, some may speculate that hope would have a positive charge!). Given this distinction, qualitative mental properties must be distinct from physical properties.
Finally, the qualia epiphenomenalist also argues that (3), the principle of causal exclusion, is intuitively obvious. Overdetermination is plausible in rare cases, such as when two assassins pierce the target’s heart at the same time, but Robinson argues that it is implausible in the case of mental causation:
Jones might have died of fright from the noise of a gunshot even if the bullet had not entered his heart and killed him. We can understand an alternative causal route in this case, but there seems to be no analogue in the case of qualitative events and neural causes, i.e., no ground for saying that some neuron(s) would have had raised activation in virtue of a qualitative event even if there had been insufficient pre-synaptic activation to bring about this result. (Robinson, 2006, 86)
For Robinson, and for other qualia epiphenomenalists, countenancing overdetermination in the context of mental causation is implausible.
These three principles, when combined together, lead to qualia epiphenomenalism. If there can be no more than a single sufficient physical cause for an effect, then it follows that distinct qualitative properties of events must be excluded from causal efficacy. Qualia epiphenomenalism is the doctrine that qualitative properties of events lack causal efficacy, so qualia epiphenomenalism is justified by these three propositions taken together (Robinson, 2004, 158; Tammelleo, 2008, 42-43; Gadenne, 2006, 105).
Before moving on, it is worthwhile distinguishing several different versions of epiphenomenalism. In its broadest form, epiphenomenalism is the doctrine that mental properties lack causal efficacy tout court, meaning that mental properties of events cannot have physical or mental effects. Following Watkins (1989, 159), call this Classical Epiphenomenalism, as expounded by C. D. Broad (1925, 118; 1959, 791). On the other hand, many consider epiphenomenalism to only be the doctrine that mental properties of events cannot causally influence any physical effects (Jackson, 1982; Nagasawa, 2010). Call this Physical-Effect Epiphenomenalism.
There is a second set of worthwhile distinctions: the events under discussion in this paper have, at least, four different properties. Events have physical properties, such as synchronous neural firing; they have functional properties, such as the causal relations they bear to other events; they have intentional properties, such as what these events intend for, or direct at; and they have qualitative properties, such as what this event feels like for the subject. For example, I am starving in the desert when a stranger walks up to me and tells me that there is a chocolate bar lying in the sand on the other side of the dune. This news causes an event to occur in me. This event has qualitative properties (feelings of relief and excitement), it has intentional properties (my excitement is about, or specifically directed at, the one chocolate bar out of the set of all chocolate bars), it has functional properties (the presence of chocolate bars typically cause mouth preparation behaviour, and chocolate bar pursuit behaviour), and it has physical properties (neural/chemical reactions cascade throughout my brain). Given these differing properties of the event, it is possible to only be an epiphenomenalist with respect to one, or perhaps two, of these properties. Thus, following the taxonomy of Frank Jackson (1982), the Qualia Epiphenomenalist argues that qualitative properties of events lack causal efficacy, while other properties may be causally efficacious. Similarly, it is possible to only maintain the irreducibility thesis with respect to one of the properties. Thus, it is possible to believe that qualitative properties fail to reduce to physical properties, but intentional and/or functional properties do reduce to physical properties. In what follows, I focus on qualia epiphenomenalism—the doctrine that qualitative properties are both irreducible and causally inefficacious—while I only briefly discuss the possibility that intentional properties are irreducible and epiphenomenal.
2. The Epistemic Argument
Epiphenomenalism has its share of problems. Indeed, the presumption in favour of mental causation is so strong that Tyler Burge argues, “There is no real hope that epiphenomenalism could become a credible view” (Burge, 1993, 112). Or again, Brian McLaughlin calls epiphenomenalism “truly incredible” (McLaughlin, 1994, 284). Perhaps the most pernicious argument against epiphenomenalism is the so-called Epistemic Argument. While there are several versions of the epistemic argument, this paper focuses on unjustified intentional properties (i.e., beliefs/memories) about epiphenomenal qualitative properties.Footnote 2 According to this version of the epistemic argument, since qualitative properties are not causally efficacious they play no role in bringing about the instantiation of intentional properties in the subject, such as beliefs/memories about those qualitative properties. Since these beliefs/memories about epiphenomenal qualitative properties do not arise because of the epiphenomenal qualitative properties, these beliefs/memories about the qualitative properties may occur without the occurrence of the corresponding qualitative properties. Therefore, beliefs/memories about epiphenomenal qualitative properties are unjustified and possibly false. Since beliefs/memories must be both true and justified in order to count as knowledge, these beliefs/memories about epiphenomenal qualitative properties cannot be counted as knowledge, and these epiphenomenal qualitative properties cannot be known. Michael Watkins raises this concern:
If qualia are not causally efficacious, then my beliefs and memories would be just as they are whether there were qualia or not. Beliefs about qualia cannot be justified on the basis of qualitative experiences since those experiences do not cause those beliefs (Watkins, 1989, 160; see also Stjernberg, 1999).Footnote 3
Epiphenomenal qualitative properties are not relevant to the occurrence of beliefs/memories about those experiences; in fact, those same beliefs/memories may be in place in the absence of those experiences. But if I can believe that pineapples are sweet, and have a memory that pineapples are sweet, even if the actual taste experience was sour, or there was no taste experience whatsoever, the belief that pineapples are sweet is unjustified, and possibly false, and does not count as knowledge, while the quality of the experience is unknown.
Frank Jackson calls this epistemic argument “the most powerful reply” (Jackson, 1982, 134) to his knowledge argument which endorses qualia epiphenomenalism. Similarly, Alexander Staudacher calls this epistemic argument “the most challenging” (Staudacher, 2006, 154) objection to qualia epiphenomenalism, while William Robinson calls it “the most powerful reason for rejecting epiphenomenalism” (Robinson, 2011). Clearly, this epistemic argument requires a response if qualia epiphenomenalism is to remain a viable option. In what remains of this paper I consider two prominent responses to this epistemic argument, and I find them both wanting.
3. The Common Underlying Cause Reply
According to the Common Underlying Cause Reply, epiphenomenal qualitative properties supervene (with nomological or metaphysical force) upon the physical properties that cause the effect. This being the case, it is not correct to suggest that epiphenomenal qualitative properties have nothing to do with the occurrence of the effect, or are irrelevant to the occurrence of the effect. Rather, qualitative properties, though causally unrelated, are still intimately related with the effect in virtue of their supervenience on the cause of the effect, and this close relation suffices to dispel any epistemic worries that qualia epiphenomenalism faces.
I want to briefly consider three different articulations of this reply. Frank Jackson hints at the Common Underlying Cause Reply in his discussion on qualia epiphenomenalism. According to Jackson, epiphenomenal qualitative properties are related to behavioural effects as the movements on a theatre screen at one moment are related to the movements on a theatre screen a moment later. The first movement (a punching, for example) seems to cause the second movement (a falling), but the first movement is caused by the projector, which also causes the second movement. Jackson summarizes, “No matter how often B follows A, and no matter how initially obvious the causality of the connection seems, the hypothesis that A causes B can be overturned by an overarching theory which shows the two as distinct effects of a common underlying causal process” (Jackson, 1982, 133; see also Nagasawa, 2010, 47). Since the qualia epiphenomenalist maintains that epiphenomenal qualitative properties are the products of the same physical cause that brings about the behavioural effect, the epiphenomenalist can appeal to these underlying causal processes to explain how epiphenomenal qualitative properties are related to these effects.
Alexander Staudacher makes a similar point, but makes some significant expansions. Staudacher distinguishes between direct and indirect causes: direct causes are either immediate causes of the effect, or parts of the causal chain that leads to the effect, whereas indirect causes mean that “the cause of A is also the cause of B” (Staudacher, 2006, 156). Thus, beliefs/memories about past qualitative experiencesFootnote 4 will have the same cause as the actual qualitative experiences, or, in Staudacher’s words, the qualitative experiences will bear an “indirect causal relationship” (Staudacher, 2006, 156) with the beliefs/memories about the experiences.Footnote 5 This means that the beliefs/memories are counterfactually dependent on the qualitative properties: “the belief would not have arisen if the fact hadn’t obtained” (Staudacher, 2006, 159; see also Robinson, 2004, 165-167). This counterfactual dependency is sufficient to dispel the epistemic argument outlined above. After all, we wouldn’t believe we were in pain if the cause of the pain had not occurred as well, so we are justified in believing we were in pain.
Of course, qualitative properties will only be indirect causes of the memories/beliefs about those qualitative properties if there is a sufficiently stringent necessity relation from the physical cause to the qualitative properties. Volker Gadenne adds this crucial component to the picture. He argues that qualitative properties are properties of physical systems, so it is reasonable to expect qualitative properties to be determined by the physical properties of the physical system. Thus, the relation between qualitative properties and their underlying base is one of “strong supervenience, or dependence” (Gadenne, 2006, 103). This supervenience relation grounds the psychophysical law P 1→M 1, stating that whenever the physical cause occurs, the qualitative properties necessarily occur as well (Gadenne, 2006, 104). Thus, the counterfactual “If Q [the qualitative property] had not occurred, P2 [the effect] would not have occurred in the given situation” (Gadenne, 2006, 109; see also Robinson, 2004, 161) is necessarily true. As Gadenne summarizes:
Is it really possible, on epiphenomenalist assumptions, that you experienced pain yesterday (t1), when the dentist treated your teeth, but now (t2), for some unknown reason, you believe you experienced pleasure in that situation? It is improbable, even if Q-states should be epiphenomena. The brain states that produced pain at t1 are causally connected with the brain states at t2, so that the latter are inclined to produce memories that correspond to what was actually experienced at t1. Causation on the physical level together with the supervenience of Qs leads to the same result as would causation on the mental level. (Gadenne, 2006, 112)
The strong supervenience relation ensures that the belief/memory nomologically depends upon the past qualitative property. This nomological dependency ensures that we believe we were in pain only when we actually were in pain, which dissolves any epistemic concerns.
4. Problems with the Common Underlying Cause Reply
This Common Underlying Cause Reply is endorsed by a number of qualia epiphenomenalists. Indeed, Hans Muller, a critic of epiphenomenalism, acknowledges that he was “convinced by it for many years” (Muller, 2008, 86). There are, however, significant problems with this solution, all of which stem from the gap between, or the indirectness of the relation between, the belief I was in pain (P 2) and the qualitative property of pain (M 1). If qualitative properties are causally efficacious, the causal relation from the qualitative property to the effect is direct and immediate (M 1→P 2), so the appearance of P 2 guarantees that M 1 occurred, which justifies and proves the belief P 2 true. If qualitative properties lack causal efficacy, there is no causal relation from the epiphenomenal qualitative property to the belief about the qualitative property, so there is a gap between the occurrence of the pain and the belief about the pain—the belief about the pain follows from a physical cause, which in turn determines that the pain will be present as well (M 1←P 1→P 2). This gap enables the critic to render false the requisite counterfactual “Had the epiphenomenal mental property not occurred, the effect would not have occurred” by demonstrating certain instances where the mental property does not occur, but the effect still does. If this counterfactual fails, then the belief that I was in pain can occur without any actual pain, so the belief that I was in pain cannot be counted as knowledge.
But is it possible that for a belief about an epiphenomenal qualitative property to occur without the occurrence of the corresponding qualitative property? The most rudimentary demonstration that the qualitative property can be absent while the effect still occurs is found in Daniel Dennett’s argument that epiphenomenal mental properties, since they make no difference, should be able to cease to appear, or run ten years behind, while everything else should remain the same (Dennett, 1978, 252). This means that my experience of pain may have occurred ten years ago, and yet I now believe that I just experienced pain a moment ago—a dubious epistemic predicament, to say the least.
Volker Gadenne provides the characteristic reply to this expression of the gap problem. As highlighted above, the qualia epiphenomenalist has at her disposal the claim that the appearance of the physical cause necessitates the appearance of the supervening qualitative properties, so qualitative properties cannot be “arbitrarily distributed” (Gadenne, 2006, 111) in the manner that Dennett’s thought experiment suggests. Rather, if the qualitative property is absent, or lags ten years behind, then there will be some alteration to the physical cause as well, which will in turn result in some alteration to what effect occurs. The strong supervenience relation guarantees that the effect will only occur when the epiphenomenal qualitative property does, so the epistemic credentials of epiphenomenalism are re-established.
Gadenne’s response raises an important question: how strong is the necessity relation from the physical cause to the qualitative property that supervene upon the cause? Typically, supervenience relations are either metaphysically necessary or nomologically necessary. The epiphenomenalist rejects a metaphysically necessary supervenience relation. After all, if the physical cause metaphysically necessitates the qualitative properties that supervene upon the cause then there is no possible world where the physical cause appears but the supervening qualitative properties do not. Many epiphenomenalists, however, are persuaded by zombie arguments, or inverted qualia arguments, which presuppose that the physical cause can appear without the supervening qualitative properties (Chalmers, 1996, 123; Robinson, 2004, 168; Jackson, 1982). The epiphile, therefore, rejects the metaphysical necessity of the mental properties. This stands to reason, as a number of authors suggest that a metaphysically necessary supervenience relation is hardly distinguishable from reductionism (Kim, 2005, 49; Pauen 2006, 134). The supervenience relation, therefore, is likely to be one of nomological necessity. But now the epiphobe can insist that missing and/or inverted qualitative properties are metaphysically possible again, which reintroduces the aforementioned problem.
William Robinson gets around this problem by granting that there are nomologically distinct possible worlds where the belief/memory that I was in pain occurs without the actual pain, but in this world, under these laws, the belief/memory that I was in pain cannot possibly occur without the actual pain. As William Robinson concludes: “the possibility of a world in which laws are different and we do not know that our qualitative events occur does not show that in this world we do not know that they occur” (Robinson, 2004, 168-169). And if our belief/memory that we were in pain is necessarily preceded by the pain in this world, this re-establishes the epistemic merit of the belief/memory.
There are two reasons to think that Robinson’s reply will not work, the first of which arises from a consideration of the problem of epistemic luck. Imagine that Sally spins around until dizzy, points in some random direction, and says “Montreal is in that direction.” Sally’s belief could by chance be true, but because this true belief came about due to luck, it is hardly considered knowledge. To avoid epistemic luck, it is typically assumed that true beliefs must be justified as well. For example, if my true belief that Montreal is in the specified direction is rooted in an analysis of the local geography and a familiarity with the relevant maps, this true belief is counted as knowledge. Not surprisingly, Duncan Pritchard suggests that epistemic luck’s “elimination is an adequacy condition on any theory of knowledge” (Pritchard, 2005, 151) while Jonathan Dancy concludes that “knowledge must somehow not depend on coincidence or luck” (Dancy, 1985, 134).
As it turns out, while Robinson’s solution may demonstrate that the belief that I was in pain is true in this world, the belief arises from an unreliable epistemic mechanism, so the belief is still not justified. Rather, the belief is only true due to epistemic luck, which prevents it from being considered knowledge. Here is the epistemic mechanism that I claim is unreliable: an effect (i.e., belief that I was in pain) is justified because it is caused by the same physical property instance that determines that a qualitative property (i.e., pain) will be instantiated, which in turn supports the relevant counterfactual. But imagine a nomologically distinct world where P 1 causes the belief that I was in pain and also subvenes itchiness (M x). Imagine also that there is a psychophysical law in that world such that P 1→M x so that whenever I have the belief that I was in pain I will have had an itchy experience. In this case the cause of the effect (i.e., the belief that I was in pain) was the cause of the qualitative property (i.e., itchiness). It is also the case that the counterfactual “Had itchiness not occurred the belief that I was in pain would not have occurred” is necessarily true. These are the same criteria that the epiphenomenalist appeals to in this world to justify the belief that I was in pain. Does this mean that in that distant world the belief that I was in pain is justified as well? The same mechanism brought it about, so it should be justified. However, it seems plain that my counterpart’s belief that he was in pain cannot be considered justified knowledge, as there never was any pain, so the epistemic mechanism is unreliable.Footnote 6 And, if the principle “the cause of the belief is the cause of the qualitative property” does not deliver knowledge in that world, why should it be considered a reliable epistemic mechanism in this world? In fact, assuming that there are many metaphysically possible psychophysical laws (i.e., P 1→M r, P 1→M s, P 1→M t, etc...), there will be many possible worlds where the cause of my counterpart’s belief that he was in pain subvened tinglyness, anger, jealousy, etc... One of these many possible worlds will hold a person whose belief he was in pain was caused by the physical property instance P 1 that subvenes a painful quale. But now it looks like this unreliable epistemic mechanism has simply gotten lucky on one occasion. In the same way that Sally’s pointing finger will correctly point to Montreal on occasion, or a broken clock will be correct once a day, so the cause of the belief that I was in pain will subvene a pain quale in some one possible world. And, in the same way that Sally does not know where Montreal is, or a broken clock does not know the time, because of the epistemic luck involved (even though it was correct in this case), so Robinson does not know that he was in pain (even though he may be correct in this world).
Not only does Robinson’s model fail due to the epistemic luck involved, but his model also appears to be question begging. Robinson’s response relies upon the view that in this world the belief that I was in pain is preceded by a pain—otherwise the belief that I was in pain would not be true, and would not be knowledge. In order for the belief that I was in pain to be preceded by a pain, however, the psychophysical law P 1→M 1 must hold. But how can we be sure that the appropriate law P 1→ M 1 holds in this world?Footnote 7 Humans lack the ability to introspect their own neural activity, so they cannot correlate the introspected qualitative state M 1 with specific neural activity P 1. Even if humans gained this ability (perhaps through being hooked up to a real-time fMRI machine), abstraction to law requires more than a single confirmed instance, so we will have to confirm the law through memory of prior instances. According to epiphenomenalism, however, memory lacks causal efficacy, so the epiphenomenalist cannot believe in the psychophysical law P 1→M 1 because she remembers prior instances of this law occurring.Footnote 8 Thus, the epiphenomenalist cannot be certain of the psychophysical law P 1→M 1. And, if there is no certainty that this law holds, then it is not possible to appeal to this law holding in this world as a foundation for knowing that M 1 occurred.Footnote 9
To summarize, if mental causation is true, then my pain causes my belief that I was in pain. The relation between the mental cause and the effect is immediate, so it renders the counterfactual “Had my pain not occurred, my belief that I was in pain would not have occurred” true. As a result, my belief that I was in pain is true and justified knowledge. In contrast, on qualia epiphenomenalism, my pain is caused by the cause of my belief that I was in pain. The relation between the pain and the effect is indirect or mediated, and hence it is possible to render the counterfactual “Had my pain not occurred, my belief that I was in pain would not have occurred” false. So the belief that I was in pain is an unjustified and possibly false belief. The epiphenomenalist responds to this charge by supposing that there is a nomologically necessary psychophysical law holding between the qualitative property and the appropriate belief. But, for so long as the psychophysical law is only nomologically necessary, it is possible that this law fails, which falsifies the relevant counterfactual and casts our knowledge of our qualitative states in doubt. Robinson replies by emphasizing that in this world, under the proper psychophysical law, the belief that I was in pain is true, and hence is counted as knowledge. But, the unreliable justificatory mechanism implies that the epistemic luck involved prevents this belief from being counted as knowledge, even if it is true. Moreover, the presumption that the appropriate psychophysical law holds in this world is question begging, for the epiphenomenalist cannot demonstrate that this is a world where the appropriate psychophysical law holds. For these reasons, appeal to a common underlying cause will not overcome the epistemic argument against qualia epiphenomenalism. Another strategy is needed to salvage qualia epiphenomenalism from the epistemic argument—and to this I now turn.
5. Physical-Effect Epiphenomenalism
Physical causal completeness combined with causal exclusion implies that every physical effect has a sufficient physical cause, to the exclusion of mental causes. But this does not self-evidently imply that qualitative properties cannot cause mental effects (Jackson, 1982, 133; Cavendar, 2009, 105-106; Nagasawa, 2010, 42). Thus, as noted above, there are those who endorse physical-effect epiphenomenalism, which supposes that qualitative properties are epiphenomenal vis-à-vis physical effects, but causally efficacious vis-à-vis mental effects. If qualitative properties retain causal efficacy within their own domain, then qualitative mental properties (i.e., pain) may cause mental effects (i.e., the belief that I was in pain). If the belief that I was in pain was caused by the pain, the belief that I was in pain is justified knowledge, and the epistemic argument against qualia epiphenomenalism dissolves. As Nagasawa explains:
If epiphenomenalists accept ... that events in the physical domain, but not necessarily in the mental domain, are causally inefficacious in virtue of their falling under a mental type, then they can reject the inconsistency objection by claiming that Mary’s colour experience, which is an event in the mental domain, does cause Mary’s acquisition of knowledge about the experience, which is also an event in the mental domain (Nagasawa, 2010, 42).
According to Nagasawa, Mary’s colour experience is efficacious vis-à-vis mental effects, and knowledge acquisition is a mental effect of the colour experience, so it is knowledge.
Assuming that qualitative properties causally influence the instantiation of mental intentional properties (beliefs/memories), a key question that arises is whether these intentional properties are reducible to functional/physical properties or not. According to some philosophers in the debate, not only are qualitative properties irreducible but intentional properties are irreducible as well (Chalmers, 1996, 82; Chalmers, 2003, 229; Searle, 2003, 155ff). David Chalmers, for example, argues that beliefs about qualitative properties are partly constituted by those qualitative properties, so a functional/physical duplicate lacking the qualitative properties cannot have the same belief content. Hence:
Phenomenal realists often hold that while the phenomenal is conceptually irreducible to the physical and functional, the intentional can be analyzed in functional terms. But if what I have said here is correct, then this irreducibility cannot be quarantined in this way (Chalmers, 2003, 230).Footnote 10
If intentional properties are irreducible to the functional/physical, as Chalmers gestures at, then certain problems arise. First, if qualitative properties can only cause intentional properties because intentional properties are irreducibly mental, intentional properties lack the ability to cause behavioural effects for the same reason. Behavioural effects, such as hand movements and utterances, are physical effects. These physical effects have a sufficient physical cause, which, assuming the qualia epiphenomenalist continues to adhere to physical causal completeness and causal exclusion, excludes these intentional properties from causing this physical behaviour (cp. Nagasawa, 2010, 42). This leads to the unusual consequence that my unreduced belief that I was in pain is considered knowledge, but my utterance “I was in pain” is not knowledge, because it was caused by neither the belief that I was in pain nor the pain itself. It also leads to the view that the belief that I was in pain was caused by the pain itself, but the physical utterance “ouch!” was not. And, as Daniel Dennett suggests, this move amounts to, “retreating into a solipsistic world where there is only himself, his beliefs, and his qualitative properties, cut off from all effects in the world” (Dennett, 1991, 403). These counterintuitive consequences suggest that there is something wrong with the model.
Perhaps the epiphenomenalist can swallow these counterintuitive consequences and still maintain that intentional properties, though they do not cause or justify utterances, are still justified themselves since they are caused by the relevant prior experience. As it turns out, even this concession is not available to the epiphenomenalist. The causal exclusion problem, as originally formulated by Jaegwon Kim, implies not only that the qualitative property (M 1) cannot cause a physical effect (P 2), but also that the qualitative property M 1 cannot cause the intentional property M 2. To see why, notice that according to strong supervenience, which we have seen that most qualia epiphenomenalists accept, M 2 is completely determined to occur by its subvenience base P 2, which, according to the causal exclusion principle accepted by the qualia epiphenomenalist, excludes the qualitative property M 1 from being counted as a cause of M 2 (Kim, 2005, 39ff; see also Bailey, 2006, 495). Thus, the qualia epiphenomenalist is not able to suppose that M 1 can causally influence M 2. Interestingly, the classical epiphenomenalist position that Broad articulates excludes the possibility that mental properties can causally contribute to the instantiation of mental properties for precisely this reason:
(2) The complete immediate cause of any experience is a simultaneous bodily event in the brain or nervous system of some one living organism. (3) No experience is a cause-factor in the total cause of any bodily event. (It is unnecessary to add that no experience is a cause-factor in the total cause of any mental event, for that follows immediately from Proposition (2) above). (Broad, 1959, 791)
The epiphenomenalist agrees that mental properties are dependent upon, and determined by, physical properties, which means that the intentional property M 2 is determined by physical properties, so there is no work left for the qualitative property M 1 to do.
Given these difficulties associated with the view that qualitative properties causally contribute to the instantiation of unreduced intentional properties, a number of epiphenomenalists suggest that qualitative properties causally contribute to the instantiation of reduced intentional properties. As Alexander Staudacher points out:
The motivation for taking qualia-epiphenomenalism into consideration at all is mainly the uneasiness felt by many in accepting one of the available reductive accounts of qualia. These reservations against a reductive account of qualia may well be compatible with a reductive account of other mental states and properties as beliefs or intentional properties. (Staudacher, 2006, 155)Footnote 11
It is possible that intentional properties reduce to the functional/physical, and if this is the case then qualitative properties may causally influence these mental intentional properties. This solves both of the problems that plagued the nonreductive account of intentional properties. It guarantees that the belief that I was in pain can cause the assertion “I was in pain”, since the belief is physical so it can influence physical behaviour. It also prevents the pain (M 1) from being pre-empted as a cause of M 2 by the subvening physical base P 2, since M 2 is P 2. Since pain caused the belief that I was in pain, this belief is justified and can be counted as knowledge.
This move does not work. After all, the epiphenomenalist, by definition, argues that qualitative properties do not causally influence physical effects. If intentional properties are physical, as this solution presumes, then qualitative properties will not be able to causally contribute to the occurrence of these physical intentional properties. Hans Muller expresses the difficulty:
... It seems clear that beliefs must count as physical according to the qualia epiphenomenalist. But this means that beliefs cannot be caused by qualia on pain of violating the following restriction: ‘the instantiation of qualitative properties makes a difference to other mental states though not to anything physical.’ (Muller, 2009, 110)
Qualitative properties do not influence physical effects, so, to the extent that intentional properties are physical, these qualitative properties will not causally contribute to their occurrence. And, if qualitative properties contribute nothing to the occurrence of beliefs/memories about qualitative properties, then those beliefs once again lack proper justification and cannot be counted as knowledge.
To summarize, on one version of epiphenomenalism, physical-effect epiphenomenalism, qualitative properties are inefficacious vis-à-vis physical effects, but causally efficacious vis-à-vis mental effects. My pain may therefore cause my belief that I was in pain if this belief is a mental effect. Since the belief that I was in pain is caused by the pain itself, it is a justified, true belief, so this belief is considered knowledge and this qualitative property can be known about. If this belief is irreducibly mental, however, it cannot cause or justify utterances or behaviours appropriate to the belief. Moreover, since this belief itself is determined to occur by a physical base, the qualitative property is excluded as a cause. Reducing the belief to the physical hardly solves the problem, for then the qualitative property will not be able to cause or justify the belief. Since the belief is not caused by the qualitative property, it is an unjustified belief that cannot be counted as knowledge.
In conclusion, two popular responses to the epistemic argument against qualia epiphenomenalism have been blocked off. Beliefs/memories about qualitative properties are not justified by physical causes that strongly subvene correlating qualitative properties, for this indirect relation between the belief and the qualitative property allows for the epistemically precarious scenario whereby the belief appears without the corresponding mental property. Nor are unreduced beliefs/memories about qualitative properties justified by endorsing physical-effect epiphenomenalism, for this move either leads to the view that unreduced beliefs/memories cannot cause their appropriate behavioural effects, or the view that reduced beliefs/memories cannot be caused by qualitative properties. Having blocked off these responses to the epistemic argument, the epistemic argument still stands as a powerful objection to qualia epiphenomenalism. The qualia epiphenomenalist is left with several options. She can attempt some other method of resisting the epistemic argument,Footnote 12 or she can consider some other solution to the mental causation problem that does not involve rejecting the principle of mental causation. What should be avoided, however, are the persisting appeals to common underlying causes or physical-effect epiphenomenalism. These methods of response simply will not work.Footnote 13
