Why I Am A Dualist
Marlowe C. Embree, Ph.D.
Assistant Professor of Psychology, UWMC
Revised as of August 27, 2008
Some students have been curious about why I am a dualist, not an epiphenomenalist. I don't care which view you personally adopt; I encourage students to think critically for themselves. But as a matter of intellectual honesty and as "food for thought", I offer the following rather lengthy and technical essay. It represents the state of my thinking at this time, which is, of course, subject to change and further refinement without notice. I might most accurately describe myself as a "moderately reluctant" dualist since I recognize the problems inherent in this position. But my view of dualism is like the famous take on democracy - "the worst possible form of government, except for all the others".
Note to fans of F.F. Centore: I am using the term "dualism" to refer to what Centore would properly call "psychosomaticism", though it may (see below) also encompass what he would call "vitalism". In contrast, "epiphenomenalism" in this paper means what Centore calls "materialism", whether reductionistic or not. This comment can safely be ignored by readers who have no clue what I am saying here, but if you're interested in an overview of Centore's ideas, click here and scroll about two-thirds of the way down that Web page.
Preliminary considerations
Even the most extreme epiphenomenalist will concede that the brain is not a computer. Not only is it many orders of magnitude more complex than the most sophisticated computer in existence today, but (thinking purely of the infrastructure or "hardware") it operates in a fashion very different from that of a computer. However, for the purposes of this essay, it will often be convenient to think of neural connections and firings as if they were analogous to the "memory" and "operating system" of a silicon-based computer. This is only an analogy for purposes of argument and should not be taken literally, as it is completely inaccurate in many important details. However, the arguments presented in this essay will not depend on the assumption of any substantive similarity between how computers work and how the biology of the brain works.
Why is the question of dualism vs. epiphenomenalism a philosophical, not a scientific, one in my view, unresolvable by means of observation or experiment? (At least, that is my view, though for the record, some epiphenomenalists disagree vociferiously -- without, however, it seems to be, any empirically well grounded basis for so doing.) One major reason is the belief, now shared by many in both psychology and philosophy, that it is inherently and logically impossible for any finite conscious entity ever to fully understand itself or the basis of its own self-awareness. This is not due to any existing limit in scientific knowledge; a tenfold, hundredfold, or millionfold increase in scientific knowledge would not resolve the problem. Let's take a look at the rather convoluted argument.
Imagine (here's the computer analogy) that any idea, concept, or experience (element of consciousness) can be represented by some combination of neural connections and/or firing patterns (that's the assumption of psychophysical parallelism, which -- it is important to keep in mind -- both epiphenomenalists and dualists share). To keep things manageable, let's develop some symbols. Let small letters represent an idea, concept, or element of conscious experience; for instance:
x = "2 + 2 = 4"
Let capital letters represent the neurological representation of that idea, concept, or element in terms of some pattern of neural connections and firings, which will be depicted here (by analogy) as a pattern of zeros and ones, though this is of course a vast oversimplification. For instance, we might find (by experiment) that
X = 0101110
Now let the term e(x) mean "an explanation of why x is represented neurologically by X". (That is, an answer to the question, "Why is "2 + 2 = 4" represented psychophysically as 0101110?") Whatever the explanation (which is likely to be incredibly complex) might be, it is itself an idea or concept that (by the assumption of psychophysical parallelism) must be able to be represented by some neural pattern E(X). By definition, E(X) must include X:
E(X) = 100100101011100010101
Since E(X) contains but is larger than X by definition, it must be true that just X is a subset of E(X), so E(X) is a subset of E(E(X)), and E(E(X)) is a subset of E(E(E(X))), and so on. In other words, we have an infinite regress. For any neural pattern (which must, in general, correspond to an element of cognition or conscious experience), we can ask, "why is that experience represented physiologically by that neural pattern," and so on. There is no upper limit to this process.
Yet, the human brain, as a physical system, is finite. Even though no living human being has ever "exhausted the capacity" (speaking analogically, for again, the brain is not a computer) of the brain, in theory there must be a finite limit to the number of neural patterns that could ever possibly be generated, for there is a less than infinite number of neurons in the brain, the brain is limited by the size of the skull, and so on. Thus, at some point in the infinite chain X, E(X), E(E(X)), E(E(E(X)))... the brain must be unable to make a further step. At that point, it is logically impossible to avoid the conclusion that the brain is having an experience it cannot explain (because it lacks the capacity to contain the explanation). In one way, this idea can be seen as a logical extension of Godel's Theorem in mathematics.
An infinite being could understand human cognition fully, of course. In fact, even a sufficiently great but finite superhuman being could do so. All that is required is, in essence, a data table containing three columns and a huge (but finite) number of rows. In each row, column 1 would list a specific pattern of neural firings (with all possible patterns listed in the table); column 2 would list the corresponding conscious experience or idea (some cells would be blank, for instance, when a part of the brain not associated with conscious awareness was implicated); column 3 would list a comprehensive theoretical explanation of the connection between the two. Since the number of possible neural patterns is finite, the data table is finite (though it far exceeds the capacity of the human brain, as argued above, and hence could never be encompassed in its entirety by a human thinker).
As humans, therefore, we presumably are incapable even in theory of understanding the exact links between neural patterning and conscious experience. At present, as noted below, we have literally not even a clue of how this connection might work. At best, we have (a) a good understanding of the physical workings of the brain and (b) the rudiments, at a very gross anatomical level, of the parallelism between neural patterning and variations in conscious experience. To call this a working model of consciousness is like saying that a Neanderthal who knew that a dropped rock would fall understood Einstein's general theory of relativity.
Note that the argument above, and in fact all the arguments in this essay, do not depend upon a "spiritual" or "religious" view of the universe. Although historically and statistically most dualists have believed in God (as I do), and many if not most epiphenomenalists have not, this is more of a historical accident than a logical necessity. Dualism requires the belief that materialistic explanations of consciousness are insufficient; it does not require any specific assumptions about what that might mean or why that might be possible. Conversely, theistic faith does not rest on a foundation of dualism as such, though it is more easily reconciled with dualism than epiphenomenalism. (Some staunch theists have been epiphenomenalists, though they are currently in the minority.)
Now to the arguments themselves.
Argument #1: Epiphenomenalism inherently undermines the validity of thought
This argument was framed most briefly and cogently by C.S. Lewis, as follows: "If naturalism is true, we can know no truths." (By "naturalism", Lewis meant, or for our purposes can be interpreted as having meant, epiphenomenalism.)
Let's see what this means. To regard our thoughts as valid and defensible is to argue that we have freely applied rules of abstract logic to them. The word "freely" is important and an essential part of what rational thought means. But the epiphenomenalist must, by defintion, regard freedom as an illusion pure and simple. To him or her, we do not "choose" our thoughts; they are inevitable products of our brain states, which are deterministic products of external causes outside our control. Why, then, should we assign any importance or validity to them, or presume to think that they correspond in any sense to objective truth?
We can only have confidence in our thoughts (or in some rationally evaluated subset of our thoughts) if our thoughts are free, which means dualism. While deterministic systems (like computers) can be rule-governed, we can only have confidence in the meaningfulness or validity (as opposed to the mere consistency) of those rules if they are put there by a logic-governed being (the human programmer). Unless we believe that human beings are deterministic systems governed by rules that are meaningful because they were put there by a free and logical Creator (a conclusion repellent to many epiphenomenalists), we can only regard humans as logical if we view them as free (hence dualism).
Modern epiphenomenalists attempt to avoid the force of this argument in two major ways, both of which, I believe ultimately fail of their purpose.
Counterargument #1: The Darwinian explanation
This argument suggests that we can presume a general (though not complete) correspondence between our conclusions and external reality because survival is enhanced by the ability to analyze and solve problems logically. There is some force to this argument, but as outlined below, I think that it is insufficient.
While evolutionary mechanisms doubtless exist, natural selection only "cares" about reproductive fitness or survival-to-reproduce. Thus, only traits that contribute to the individual's ability to pass on genetic material to future generations (or that, by virtue of how the genetic code works, are directly correlated with traits that do this) can be selected.
Does the ability to reason abstractly about purely philosophical or theoretical matters qualify? It's hard to see how. (To argue that this "must" be so because we now have that ability is purely solipsistic, arguing in a circle.) There is certainly no evidence that mate selection is driven by a desire for an abstractly logical partner (versus, say, a physically attractive one). Nor is abstract logic necessarily at all correlated with the kind of concrete, pragmatic problem solving ability that might contribute to ongoing physical survival. (In fact, research indicates that "absent-minded professor" personality types are anywhere between two and five times more likely than others to experience serious accidents that might diminish survival-to-reproduce.)
Counterargument #2: Quantum mechanics
A second argument (which tends to be more vague and unfocused but nonetheless worthy of consideration) is that we no longer view the world as a rigidly deterministic system. As Newtonian mechanics has given way to a post-Einsteinian universe, we now know that (at least at a micro level) physical entities behave stochastically: the behavior of, say, a particular quark can never be specified exactly, but only in terms of probabilities. The concept of randomness may, these thinkers say, provide a way that a purely material system could be regarded as "free".
However, there are two major problems with this idea. First, quantum behaviors only occur at the micro level. Something as large as a neuron (let alone an entire brain region) is not subject to these kinds of stochastic processes. (Or, more technically, the probabilities cancel each other out at the micro level, leading to the same outcomes that an old-style deterministic model would have generated.) Second, randomness is not the same as freedom. Freedom implies purposive or intentional activity, which neither a purely random nor a purely determined system can be said to have.
Argument #2: Epiphenomenalism provides no explanation for the subjective elements ("qualia") of conscious experience
A word about the term "mechanism" as it will be appearing from time to time, like currants in a bun, throughout this essay.
The assumption of metaphysical naturalism (the notion that all events in the universe can be explained solely and sufficiently through chains of impersonal cause and effect) absolutely requires that a scientific mechanism be found to account for any empirical phenomenon. Hence, an epiphenomenalist is duty-bound to assert that consciousness can be explained mechanistically. A dualist is (no pun intended) free to assert either that mechanisms of this sort are involved or not (see the section on counterarguments to dualism below, however).
Of course, it is perfectly legitimate for the epiphenomenalist to state (as s/he must do, if s/he is being honest about things, given the current state of scientific knowledge and progress) that we do not yet have such a mechanism, but that s/he believes that one will eventually be found. Nothing illogical about that: but note that the second part of this statement is nothing more or less than an article of faith. It is no different from stating, "No evidence of twelve-foot-tall green-skinned space beings with six eyes and a remarkable ability to play the saxophone has yet been found, but I believe that eventually it will." The second half of both statements is outside the universe of discourse of science in that they are not falsifiable. If the necessary data is found, the belief is confirmed; but in the absence of such data, one can always claim that further investigation "is believed likely to turn it up". While the statement may be entirely correct, it is not a scientifically grounded statement as such, merely a wish or hope. (It may be defensible on nonscientific grounds, of course.)
At the present time, there is not even a ghost of a hope of a hint of how any material system, no matter how complex, might come to "exude" consciousness. The notion that consciousness is a "necessary emergent" of certain kinds of physical systems is nothing more or less than a pipe dream at this time. No relevant mechanisms have been identified. Note that the existence of psychophysical parallelism, which no one disputes, is not such a mechanism but is merely a correlation. (The fact that, when the temperature drops, people feel cold does not mean that a drop in temperature is the same as the subjective experience of being cold. The same can be said of the link between brain states and conscious experience.)
Again, this argument is of a different nature than argument #1 above. My contention in section #1 is that belief in the truth of epiphenomenalism is logically inconsistent, whereas here in section #2 I am simply arguing that there is no concrete evidence to justify that belief at the present time. Further research could disprove my second argument (though, again, there are no specific grounds to believe that this would ever happen; it is just a hope on the part of those who wish to view consciousness in emergent terms); no amount of empirical evidence could vitiate the first argument as such.
Argument #3: Epiphenomenalism requires acceptance of deterministic assumptions about human nature and behavior
The argument here is still weaker in that it does not represent either an inherently logical flaw in epiphenomenalism (as does argument #1) nor an evidentiary flaw (like argument #2), but simply a necessary consequence of accepting epiphenomenalism as true -- a consequence that happens to run absolutely contrary to the foundational concepts of Western civilization and, particularly, our legal and ethical systems. This fact does not disprove epiphenomenalism; but it shows that a logically consistent epiphenomenalist would have to argue (as does B.F. Skinner) that we must scrap the current foundations of our culture and rebuild it on a completely new basis. This most people are unwilling to countenance; hence, it can be argued, belief in epiphenomenalism is an existential impossibility. It can be imagined, but not lived out in practice. A worldview that cannot be lived out in the real world is unlikely, all things being equal, to be correct; though it is not inherently impossible that it is correct.
If epiphenomenalism is correct, we are all "puppets on a string" (our behaviors are strictly and solely determined by forces outside our control). If this is true, it makes no sense to think of holding others morally accountable for their actions. Social and legal sanctions lose the only ethical or philosophical justification that they might have, namely, that people are being subjected to outcomes that could have been anticipated and avoided had they chosen differently (i.e., more responsibly). We can no longer speak of justice (just as, based on argument #1, we can no longer speak of truth) if epiphenomenalism is correct, but only of brute force, of "might makes right", of the lex talionis disguised (through a sort of social deception or sleight of hand) as true moral justice.
If this is true, I'm going to have to stop watching reruns of Law and Order, for while the universe (on these assumptions) does contain a deterministic order, it doesn't contain any ethical laws. But, wait, if it's true, I can't help but watch; my behaviors are determined. That's a relief; I like the show.
Counterarguments against dualism
It's only fair-minded for me to mention some of the major counterarguments that epiphenomenalists have historically made against dualism as I close out this essay. Obviously, the good arguments are not all on one side or the other; were this the case, no controversy would remain, as people would have long ago resolved the question. No philosophical worldview is watertight; all contain seeming paradoxes, all are challenged by anomalous data. The goal of critical thought is to weigh the arguments and make a rational, though necessarily tentative (for the jury can never fully be in, at least not in this life) choice among competing alternatives, as I have tried to do in this essay. Weigh my arguments above against the best counterarguments from the epiphenomenalist side, which I will attempt to present fairly; then make up your own mind.
Counterargument 1: The absence of a mechanism/means by which the material and immaterial worlds could influence one another
The word "mechanism" may not be precisely the right one here. But if (as dualism must in some way postulate) consciousness is a nonmaterial entity by definition (however we might speculate about what that might mean -- and note that, as indicated earlier, this does not necessarily imply a Cartesian concept of the immaterial soul as such, though it does not exclude that possibility), we have to ask, "How can these two 'realms' or 'worlds' or 'types of reality' influence one another?"
After all, even the dualist must admit that there are strict limits to the ways in which these two realms can interact or intersect. I cannot directly alter the physical world merely by thinking about it (if I could, I would add some recently materialized dollars to my bank account). As the old joke goes, "all who believe in telekinesis, raise my hand". My soul or spirit or mind or self or whatever we might call the conscious part of me can only control my own body; it can only do so by means of the mediation of physical brain processes; it can only do so as long as the physical mechanisms of brain and body remain alive and intact, and so forth.
Thus, the dualist must postulate some sort of special connection or "psychosomatic unity" between mind and brain, but it is impossible to elucidate precisely why this is so or how it might work. While this is perhaps a "wash" (since the epiphenomenalist cannot specify how pure matter no matter how constitute might become conscious or exude consciousness: in fact neither side has any "mechanism" about which to speak), it does seem a weak link in the dualist position.
Note, however, that some of the red herring arguments commonly advanced against traditional Cartesian dualism are sheer nonsense. We need not in any sense specify a "part of the brain" or "region of the brain" that serves as the "mind-brain link". We don't know enough about what consciousness is to draw any such conclusions. All we need say is, with some humility and reverent agnosticism, that we don't know how this works, and that (if my argument under "preliminary considerations" is correct) we can never know. If the dualist has a hard time explaining why my mind is linked to just one body in just one limited way, the epiphenomenalist has an equally challenging puzzle on her/his side of the divide.
Counterargument 2: The presumed unscientific nature of dualism
Those who define science in a certain way tend to argue that dualism is "prescientific" or "unscientific", in that it introduces concepts that are inherently unobservable and untestable (the existence of an immaterial self, soul, mind, or spirit). The fact that these are concepts that have been important within the religious history of the West adds fuel to the fire for those who see science and religion as inherently opposed or oppositional (a view I strenuously dispute; the motto of the University of Denver, where I obtained my undergraduate degree, is pro scientia et religione, and I see no reason not to support both enterprises wholeheartedly). Besides, in the fundamental sense of holding that there is some kind of foundational, non-contingent reality on which all else depends, everyone in the world is religious, and therefore philosophically neutral thought is technically an impossibility since all religious conclusions (including those of atheism or agnosticism) are "pre-theoretical" (see Roy Clouser, The myth of religious neutrality, for a wonderfully cogent exposition of that idea).
In my view, however, this rests on a philosophically biased view of what science is. It is true that dualism is not reductionistic But to define science as necessarily reductionistic or materialistic is to stack the deck unfairly against dualism. It is not a logical requirement of science to make that assumption. Science is about observables, but it is not necessary to conclude that only observables are real. A scientist is free to be a materialist, but cannot correctly claim that science has proven that view. At most, materialism is an operationally convenient viewpoint for scientists to take; convenience is a weak test of truth at best! The fact that naturalistic science can explain many empirical phenomena well does not prove that it can explain all of them, just as the fact that many drinks contain alcohol does not prove that all drinks are alcoholic by definition.
Some individuals like to cite Occam's razor (it was once said by a nameless wit that Occam bled to death when he cut himself with his own razor) as evidence that epiphenomenalism is "more parsimonious" than dualism. I think this doubtful (we could be sure of this only if a coherent mechanism to explain consciousness in material terms were known; as noted above, we do not have this essential point of comparison), but even so, Occam's razor correctly states that the simplest sufficient (that is, complete or coherent) model is the one that is presumed correct, and we do not know if epiphenomenalism qualifies as a sufficient model. In my view, for reasons outlined above, it does not. (And, in any event, the principle of parsimony remains only an assumption. It does not constitute a logically coherent proof of the notion that simpler models are actually more likely to be true.)
Counterargument #3: Is dualism vitalistic?
This may or may not be an offshoot of the first argument. Some see dualism as opening the door to belief in the ability of nonmaterial agents to influence physical reality in a host of other ways that dualism itself does not postulate. If we believe that the spirit (or spirits) can change physical events in even one way, what prevents us from "regressing" (so the argument would run) to a "prescientific" belief in a host of other such events, such as the notion that the spirits of my ancestors can visit my home in the middle of the night and rearrange the order of cans on the pantry shelf, or the notion that my soul can leave my body and temporarily inhabit the body of a stray cat?
Clearly, phenomena of that nature, if they can be postulated to exist at all, are completely outside the universe of discourse of science. (I can observe that the cans on the pantry shelf have a different arrangement than formerly, but can draw no scientific conclusions about the causes involved, since they are inherently unobservable and outside the scope of my own introspective awareness.) Consciousness is not like that; all of us know from direct experience that we are conscious and self-aware; that phenomenon is not in doubt (except by radical behaviorists like Skinner, who appear to use a mind they do not have to doubt the existence of mind in the abstract, a paradox if ever there was one).
It is true that dualism requires the belief that the material cosmos (or, at least, the three-dimensionally-bounded spacetime continuum we physically inhabit) is not all there is. But our choice among ways to conceptualize that reality must be bounded by the empirical phenomena we observe. Most phenomena that do not involve consciousness seem well explained by the kinds of scientific models that view the universe as a closed system. This does not mean that all of them are; it means that, in doing normal science, we can operate under the presumption that this is generally (not necessarily universally) true. This is one reason why science and religion are not inherently at loggerheads; they usually operate at different levels of explanation. ("The lot is cast into the lap, but the decision is wholly from the LORD"; both parts of this proposition can be objectively true and valid even though the scientist can only draw empirical conclusions about the first part. The scientist, speaking as a scientist, cannot tell us whether the second part is true or false, though as a human being s/he is likely to have drawn conclusions one way or the other since the existential demands of existence practically require all thinking adults to adopt some kind of philosophical worldview.) In other words, dualism need not lead to the wholescale abandonment of empirical science, which (contrary to popular opinion) is historically and philosophically dependent on theism, not atheism, as the logical ground of its emergence. (See Alfred North Whitehead for one exposition of this idea.) Nor does dualism require the uncritical acceptance of any and all religious ideas; those in direct conflict with clear empirical evidence must be viewed with extreme caution at a minimum, if we are to regard religious truth as propositional or objective in any sense.
Closing comments
The notion that dualism and epiphenomenalism are the only possible positions is overly simplistic. John Polkinghorne, in his intriguing book The Faith of a Physicist (just barely missed making the cut of the fifty books that have most influenced me in my life) proposes the intriguing alternative of "dual-aspect monism", which is an entirely different position that is neither dualistic nor epiphenomenalistic. In short, his idea is that matter and consciousness are flip sides of the same undifferentiated "stuff", so that there is only one kind of reality in the cosmos, matter/mind or mind/matter. There are problems with this (such as the question of why humans can think but protons, to our knowledge, can't), but he makes an interesting case for his ideas. I don't yet find them fully convincing, but am keeping an open mind (or matter/mind) about it. His position does offer a way out of the epiphenomenalism-dualism polarity.
This essay was not meant as a polemic. My goals are entirely irenic (and perhaps a bit eristic) but not deliberately confrontational. As noted previously, thinking persons have always disagreed about these issues. If anything, my purposes were to promote further critical thinking and analysis on the part of you, the reader, and to encourage productive dialogue. Disagreement is not only expected, but useful, as long as it occurs within a context of mutual respect. "The essence of dialogue is not that we relativize our convictions, but that we agree to accept one another as persons" (Martin Buber).