As a psychologist doing philosophy, I offer this essay on the centrality of one’s subjective experience, focusing on its implications as to qualia and dualism.

My conclusions as to dualism are difficult to classify.  The world in itself is monistic (wholly material), but lacks agents to have knowledge of it, and so is unknown.  As soon as anything models itself as an agent with beliefs, that model includes an implicit, necessary assumption of dualism.  One might say that there is no inherent dualism, but that any adequate model of the world must be dualistic, in order to include the possibility of knowledge.

We will see that this approach to dualism leaves the basic implications of materialism intact.  Materialism denies the reality of immaterial mental entities such as your red quale when you fantasize a red Ferrari.  Further,it suggests that everything that happens either has a physical cause or is random -- that neural processing always follows the laws of physics,leaving no room for us to be initiators of causes.  Ignoring randomness, which limits predictability but is simply a further limitation on free will, the implication is determinism:  all human behavior, overt and covert (thought, dreams, intentions), could in principle be predicted, so that science is adequate to understanding human behavior.  (See the article, “Agency” for how all of this is compatible with our having free will).

For the most part, I believe, philosophers fail to take the implications of materialism far enough.  There is, I propose, a key sense in which you and I are automata:  rule-based systems with no slightest awareness – no beliefs, no hopes, no awareness of pain.  In that view, there are no meanings, no knowledge or intentionality.

Evolution tends toward rule-based systems that mimic intentionality.  Amebas and jellyfish, although lacking central nervous systems, find food and flee from danger, as though they knew and cared.  In the natural course of things, there is increasingly sophisticated mimicry of intentionality, and one might expect the evolution of an organism which, like a chess computer, produces possible future scenarios and has automatic systems for choosing the most adaptive.  For all you know, other humans than yourself might be such systems, utterly lacking in intentionality, but mimicking beliefs, hopes and awareness of their surroundings.

Of course, evolution doesn’t intentionally mimic intentionality.  More accurately, sophisticated systems are more likely survive and reproduce.  The simplest way to model such systems, as Dennett says, is by taking the intentional stance with them.

Such a position works wonderfully to explain all that is, with one glaring exception:  it ignores one’s own subjective experience.  I propose that we should take this automaton position as one adequate explanation of reality with a fatal flaw.  Beginning with that viewpoint we can then look at the origins of subjective experience.

It is difficult even to conceive that intentionality might not be real, in that we deny its reality intentionally and by conceiving.  Here is a science fiction illustration of how there might be a complete model of the world without conception, intentionality or actors.

                                                                   Temnons
 
  A Temnon lives for a billion years or more.  Its body is a thousand light years in diameter.  Most of its body is of unimaginably low density, but a small portion is extremely dense, and records sub-nuclear activity.  A Temnon travels rapidly through space, with whole galaxies passing through it, doing the Temnon no harm.  It records only whatever interacts with its body.

   We might imagine that Temnons evolved conventionally into highly advanced processing systems, never achieving intentionality.  If so, they eventually reached the point where all their needs were automatically supplied, such that their sole remaining behavior is an aimless processing of data.  A Temnon doesn’t act, think or predict, but merely amasses and processes its data as 0s and 1s.  Still, we might describe the sort of processing it does in intentional terms, as we do with artificial intelligence.  It processes information selectively, organizing it into a model of the world, utterly without intention.

   Over a billion years, one Temnon developed a model of the world that is generally superior to our own.  It modeled human civilization thoroughly, without recourse to the notion of intentionality, tending to model at the microphysical level.  In its radical translation of human terms such as ‘pain', 'hope', etc., it modeled humans as mere links in the causal chain, without feelings or aims.

   It was entirely successful in modeling the debates of philosophers of mind, including detailed conversations about qualia, consciousness and mind.  But, not having personally experienced such things, it neither used nor needed any semblance of what we mean by the subjective aspects of such terms.  For example, it “interpreted” the discussion as to what it might be like to be a bat in a way that modeled both sides of the argument functionally, missing what we consider to be the point.  Not having experienced subjective entities such as qualia, it failed to recognize the subjective aspects of the issue.  Yet it succeeded in modeling the full functional content of each side of the argument.

** ** **

I believe that the Temnon’s model of people could be fully adequate, such that adding intentional terms (pain, meanings, belief, purpose, etc.) would in no way increase its predictive power.  This suggests that such terms point at entities that are either non-existent or epiphenomenal (without causal influence).

Notice that the Temnon could not possibly notice our intentionality.  Our sole basis for supposing that there might even be awareness, meanings or aims is each person’s subjective experience.  There is no hint for a non-intentional system that there might be intentionality.  Writing operas and producing philosophy can be interpreted as rule-based behaviors of wonderfully sophisticated automata, evolved as the result of social interaction being adaptive.

Such a position is, of course, unsavory, counterintuitive and apparently ridiculous.  It implies that you and I don’t know that we are alive, that we have no interest in affecting the future, that death and suffering are no worse than life and comfort.  It implies that we simulate having aims in precisely the way a chess computer does:  running possible scenarios and choosing one by built-in rules, while having no idea that we are doing so.

Just what-all is missing from this materialist view?  I don’t think that anything physical is missing.  Only the thinnest veneer of personal experience is left out of the Temnon model: that related to the intentional aspects of perception, belief, desire and choice.  In particular, the Temnon model is missing intentionality – the quality of aboutness in a mental entity, such as my meaning apple being about that apple.  Let’s try to characterize this veneer by distinguishing mind from brain.

Mind and Brain

The brain has been described as a syntactic engine, with its neural processing fully controlled by proximate causes, distal causes (such as operant conditioning and the evolution of the brain), and some randomness.  ‘Syntax’, here, is “the rules governing construction of an orderly system of information, whether in sentence structure, logic or elsewhere”.  A syntactic engine, then, is a mere rule-following device such as a chess computer.  The brain is an input-output system, structured through evolution and learning to make certain sorts of responses to stimuli, without aim or awareness.   The implication is that the brain is not a true meaning-user that comprehends and chooses, but is a mere rule-based processor.

The mind has been described as a semantic engine (a user of meanings), intentional but supervenient on the brain – a flow of epiphenomenal mental states that have no causal power – have no impact on succeeding mental or physical states. 

This view creates a basic problem for the thin veneer of intentionality, making it epiphenomenal (utterly outside the causal chain).  As an apple comes into my field of vision, my brain activates processes for identification; the only epiphenomenal part is my personally meaning anything by that neural process.  I fantasize a red Ferrari.  The mental redness (the quale) is epiphenomenal.  You stick a needle in my arm.  The wince and squeal are physical; only the mental quality of the pain is epiphenomenal.  One ought to conclude that the intentional aspects, being without causal influence, are superfluous.

However, the claim that they are epiphenomenal is thought to imply that mental states are both real and immaterial.  I’m going to try to show that mental entities are real but not immaterial.  They are merely special ways of describing neural entities.

Agency

I find it useful to divide reality into two levels:  the physical and agency, where agent is simply a necessary construct that creates the possibility of personal experience, belief, desire and free choice. (I more fully describe necessary constructs in the article, “Knowledge”).  Any complete model of reality (including the Temnon’s), requires necessary constructs.  Examples of necessary constructs include the perfect circle, the axioms of geometry, and the surface of a star.  They are invented entities, and are modifiable to varying degrees.  Different models of reality will make use of different necessary constructs.  In my own terminology, a necessary construct is valid/real:  not physical, but an integral part of reality within the given model.

However, the necessary construct of agent is unique, in that it is the sine qua non for the world that we know.  We can say that it is no more than a concept, but it is also the very ground of there being persons with minds that can conceptualize.

I propose that our society unknowingly uses (and conflates) two competing models of reality.  The radical scientific model (which I will call the automaton explanation) is consistent with a microphysical explanation, and ought to treat people as automata.  The other, the agent explanation, treats people as agents with minds and free will.  I will try to show that it differs from the automaton explanation primarily in adding the single necessary construct of agents, and then including functional relations to that construct.  Brain terms and mind terms tend to explain the same phenomena, and need not conflict if we keep these two explanations separate.

I suggest, then, that there are no minds (and no intentional phenomena) without agents.  Once we employ the agent construct, an intentional relation is between an agent and an object.  It is a functional relation to the degree that we accept the agent as real.  Note that the Temnon lacks agency, even in the agent explanation.  It has accidentally modeled the world, but its model is not about anything; it mirrors the world in some ways, but doesn’t assert any such relation.

Experiencing the redness of an apple is a functional relation between agent and apple, and feeling pain in your leg is a functional relation to your leg injury.

My premise, then, is that the automaton explanation can be expanded to be fully adequate for all but one’s own subjective experience.  The agent explanation is fully mappable onto the automaton explanation, so that there is no multiplication of entities, but cannot properly be subsumed by it, because there is no appropriate way for the automaton explanation to deal with intentionality.  (The automaton explanation already uses necessary constructs, such as infinity; we need only add agency.)

If you fully accept that the automaton explanation is accurate and complete for the Temnon, you are left with just two alternatives (see the article, “Agency”).  In one alternative, mind, consciousness and qualia, although epiphenomenal, have genuine existence.  They are part of reality itself, to be accounted for in any complete understanding of the world.  (Qualia are genuine entities, independent of the model used).  As I argue below, there is no evidence of such entities.  Agents and automata would report the same qualia, so that this alternative is empty of content. 

In the second alternative, the very adoption of a subjective stance by automata creates the context for personally experiencing one’s surroundings and developing personal beliefs about the world.  And that context necessarily takes account of the agent as a necessary construct, with meanings and other intentional entities as functional relations to agents, all valid/real in the only sort of model of reality in which truth is possible.  There is dualism, but only as a necessary invention of the model of reality taken when using the subjective stance.

Qualia

Perhaps the most compelling argument for dualism is the incontrovertible subjective experience of qualia.  I propose that a red quale is the semantic relation between an agent and an intended (seen, recalled or fantasized) red surface.  The quale is not, in addition, a red mental entity.

We might consider three supposed levels to qualia:  the syntactic underpinnings, the semantic relation of making a claim about the apple, and the quale itself as a mental entity. 

Let’s start with the syntactic underpinnings -- firings in a neural network, designed to facilitate appropriate responses to redness in a wide variety of ways.  I will try to describe them while relating them to their semantic counterparts that produce qualia.  Imagine a schematic of the physical structure of a neural network, with each meaning represented by a simple on/off (yes/no) switch.  When the switch for apple turns on, there is no content to it other than “yes”.  The content is merely implied by the position of the switch in the syntactic equivalent of semantic space (ie, in the neural network with its primings of other neurons).  For example, the redness of an apple is represented in a space for color, defined in relation to blue, yellow, brown, black, etc.  There is, then, another syntactic/semantic space for sound, somewhat similar in structure.

But how can a mere position in syntactic/semantic space imply genuine content?  The content is in the associations, supported by the priming of connected switches.  We might say that my meaning apple has its content in what I tend to do after that switch turns on – take a bite, recall what apples taste like, categorize it as a fruit, etc.

It might feel to me like I’m holding the full meaning of apple in my mind at a given moment.  The reason is that this content is immediately available to me, with no delay when I look to see that the content includes its typical shape, color, taste, etc.

We can model a thought process syntactically as a succession of switches being turned to the “on” position (or being slightly activated with priming that affects which switches turn on next).  It’s a succession of covert and overt behaviors, such as priming the William Tell story, or reaching out for the apple.  The covert can be explained not only as neural behaviors but also as a thought process -- a path through semantic space, where the active content is always thin and the richness is in the full complexity of directions the mind can turn, relating my current situation to my whole history of experience and knowledge.

Said differently, my mind has a whole model of the world.  My current situation (seeing this apple) makes sense because I position it within my model.  I have a sense of knowing, because of the availability of the various potential directions, but I don’t have the full content as a single mental entity.

Mirroring the syntactic underpinnings of qualia is one’s semantic relation to the red surface:  a personal assertion of redness (utterly different from perceptions of saltiness or stinging), or an identification by placing this sensory quality in the semantic space for color.  The content of this assertion is its placement in my model of the world, suggesting all the appropriate ways I might want to respond.  A perception is the sort of assertion that even the most primitive sentient creature makes.  The semantic relation is the syntactic process described from a personal perspective.

The personal relation to an elemental sensible quality such as color must mirror the mere on/off switch underlying it, as a simple assertion:  that’s red, or that’s salty, or that’s stinging.  It isn’t yet a red quality in the mind, but is a yes, an assertion of that redness being found out there.  The relation is one of intending a unique location in syntactic/semantic space to be about the real color, identified within the context of one’s model of the world.

Suppose for a moment that you have no immaterial qualia, but only on/off switches that position visual wavelengths in a syntactic/semantic space for color.  As you scan a horizon, your system instantly recognizes hundreds of colors and shades, knowing their inter-relations, and activating connotations such as liking certain colors.  Given your intimate familiarity with each color and the ability to recognize it instantly, it would surely present to you as an inherent perceptual quality of some real surface – the barest recognition, a simple that is unique; that apple is red with the slightest purplish tinge.  When you later recalled it, it would still hold that quality.  Red would seem like a simple, like one of the elements that make up the visual world.  I propose that discussion of the qualia of direct perception is about exactly that.

The third supposed level to the qualia discussion is the epiphenomenal, immediate content – its mental redness, in addition to the real redness of the apple that I am seeing or recalling, and in addition to my assertion of familiarity with it (my assertion of where it fits within my model).

There is the tendency, then, to want to identify the qualities of the quale:  boy, this mental redness is really red.  But that can only be done by asserting about the quale, and that assertion must be mediated by locating the color of the quale in semantic space.  Whether or not that can be done introspectively, we must recognize that it would be a new assertion – “perceiving” the quale in an I-meaning-quale relation.  That is, an on/off switch for red would mediate my relation to the quale.  The implication is that the quale is not epiphenomenal, but is noticed (generated by the appropriate neural processes for noticing, such as electrical connections from the quale to the appropriate part of the neural network).

We are capable of introspection, and we take an interest in qualia as entities in themselves.  But it is easy to show that, if qualia are epiphenomenal (if our relation to a quale is not mediated by a new meaning through neural connections), we have no way of knowing whether we are really introspecting qualia.  The argument goes as follows.

Suppose that there is this third level of qualia – immediate, conscious mental contents without added processing – in addition to the semantic relation of asserting.  Suppose further that there is a defect in my brain that causes my qualia to disappear for minutes at a time.  Let’s say that I happen to be introspecting during my fantasy about a red Maserati when my red quale suddenly disappears.  My neural processes would, of course, be unaffected by the absence of something epiphenomenal.  Thus, I’d think, “yep, there it is, red as ever”.  If, soon after my qualia returned, I recalled that little attempt at introspection.  I would be able to recall the redness of the previously introspected quale nicely, underlain by the same neural processes that would have occurred had there been no temporary absence.
 
Whatever epiphenomenal qualia are, even the most careful introspection and retrospection would miss their blinking into and out of existence.  Introspection of your epiphenomenal qualia, then, provides you with no evidence of whether they occur.

The semantic relation to the redness of an apple is an assertion with genuine mental content (that apple is red), limited to a spot in semantic space and all the connotations thereto.  That spot in semantic space can only be a simple, a declaration of semantic location (a yes with its content being implicit).  The same, then, can be said of a pain, if we separate the pain itself from its entire connotation, such as the attendant suffering.  The pain itself is neither unpleasant nor pleasant, but is a simple.

Consider the pain of a wasp stinging me.  Underlying it is neural processing, tending to lead to appropriate responses such as wincing.  In addition, there is a semantic relation, asserting a certain physical location and a unique simple, somewhat related to an ache or to tactile pressure, but far away from redness.  Is there, in addition, a pain quale in consciousness?

If there is such a pain quale, am I in an intentional relation to it, mediated by another meaning, as in, I-characterize-quale?  Or is the pain quale a mere content of consciousness, not noticed by the agent, but mere consciousness-in-itself?  It can only be the latter if its disappearance would fail to alter the neural flow.  And in that case, we have no private evidence of whether or not we have such qualia.

But why does red have just the “feel” that it does, and green such a different “feel”?  As an agent, I directly experience the world, including elemental experiences in which the assertion is no more than a “yes” – empty of content but in a precise location in my model of the world.  This gives rise to the sense of immediate recognition.  The question of whether we could switch the mental experiences of red and green for each other falsely Implies that they have content.  One must take a stand.  If my qualia for red and green were switched, would I notice?  If so, then they are not epiphenomenal.  Neurally, I am thus in an I-assert-greenness relation to the quale.  If, instead, they are epiphenomenal, I could not notice the switching in any meaningful way.

When I say that the elemental assertion has no content, I am ignoring its connotation (primings), which includes the complexity of my history of seeing this color.

I experience pain and imagine red dragons in ways that are inherently private.  There is no question that these are real relations, but they are not relations to me as automaton.  They are relations that are dependent on my necessary construct that I am an agent.   Given that necessary construct, I make use of a true explanation that I experience the world directly, have beliefs and desires, and influence the course of things.  None of that is true in the automaton explanation, but all of it is a different way of describing the physical events that can also be described neurally. 

My leg is injured and my automaton reacts accordingly, neither knowing nor caring that it is injured.  In the agent explanation of the same event, I personally experience the injury by means of a system that gives it a place in a semantic space for tactile sensation and/or pain.  The pain has instant familiarity, and is nothing like a sound or a color.  That is, it’s an ineffable simple of a perceptual sort.  Besides its location in semantic space, it has all sorts of connotations.  Typically (but not always) pain is accompanied by suffering, which provides most of its content.

Science might one day be able to read my thoughts precisely, finding the precise neural activity underlying my claims of redness and pain, and being able to decipher my thoughts in far greater detail than I can.  Science will have unraveled my automaton, but will miss the intentional relations – what it means to be an agent experiencing a red apple.  It is not missing an additional, immaterial entity.  Rather, it is missing the perspective:  being an agent who is thinking and experiencing.  This is just a different explanation of the same things, but the perspective is fundamental to our design, and will never be replaceable by automaton explanations.
 
Consider Mary, a scientific investigator of color, locked for all her life in a black-and-white room, so that she has full objective knowledge of the color red, but no subjective experience of it.  Now she comes out of the room and sees a red rose.  At the syntactic level there is a neural activation in a previously unused part of her brain that is reserved for the semantic space for color.  We need not even consider the question of qualia to understand that this is radically new information to her.

I proposed early on that the agent is a necessary construct, and that all intentional terms are functional as relations to the agent.  The implication is that we can map the agent explanation onto the Temnon explanation by adding a single necessary construct (and working through all its implications). 

More obviously, belief and desire are genuine functional relations, although they imply the agent as a necessary construct as one of the terms of the relation.  (I explore these sorts of functional relations in “Desire, Fear and William James”).

The term ‘supervenient’ is commonly used to describe the relation of brain and mind in a dualist model.  I like the term for describing the relation between the agent and automaton explanations.  The two are arguably not combinable (see “Knowledge”), but attempt to describe identical events.  In either model, the competing explanation can be seen as supervenient on it, not having causal power.  From the viewpoint of the automaton explanation, meanings are epiphenomenal, and are merely supervenient on neural events.  In the agent explanation, my meaning is intentional with causal power, and the neural event underlying it is supervenient – not causing it, but parallel with it.  That is, my intending that meaning will always be underlain by that precise neural event as a mere co-occurrence (rather than as a subsequent effect, caused by my meaning). 

The agent explanation fully acknowledges the neural.  Whereas the automaton explanation can deny all agency, the reverse is not true.  But causally, the agent explanation assigns real causal power to the agent, with the neural as the necessary collaborator, neither causing nor caused by the meaning, but the meaning itself in its physical aspect.  That is, a meaning is a neural event described with a model that includes agency. 

The two explanations can be mapped onto each other, with neither one adding to the functionality of the other.  I mean no more than that when I say that they are epiphenomenal on each other.

Independent Reality

We are mistaken to think that there is a single true version of independent reality (see the article, “Knowledge”).  Truth is inherently relative to the inter-relation between independent reality and the model in which it occurs.  What is true about human agency is a function of which model is used.

Dennett treats the intentional stance as a mere stance.  I can imagine accepting his view for all cases except for oneself.  My view of myself as an agent is not a convenient fiction, nor an error of folk science.  It is true.  And since I believe that you and I are fundamentally alike, I can hardly help but believe that you, too, are truly an agent.

The agent is a necessary construct, only as real as entities such as the laws of physics and the axioms of geometry.  But that is an entirely different thing from calling it a mere stance, because it isn’t optional.  Dennett wants to say that the microphysical model is the more accurate, with the intentional stance merely a shortcut.  In doing so, he is conflating the automaton and agent explanations.

Which of the two models of reality is primary, or is more true?  The automaton explanation contains a paradox:  it is a set of concepts and beliefs proposed by automata that are incapable of conceiving or believing.  Properly, the automaton model should be approached as the Temnon model.  The Temnon has no beliefs.  And Its model makes no assertions until agents come along to try it out.

But the agent explanation has an inherent weakness in that its implied dualism is wholly a function of the model chosen, rather than a feature of independent reality.
 
I conclude that the agent explanation is the superior model of the world for our use.  The automaton explanation can be precisely mapped onto the agent explanation, and is superior for some purposes, but is inadequate for describing knowledge, personal experience, desire, being free to choose and anything mattering.

Interestingly, I think that the ultimate version of the Temnon model would take the lack of initiating causes to the extreme.  It would model the world as a unity, with no separation, no distinctions, everything so inextricably connected that all is cause and effect of all else.  A star has no surface, but just lesser and lesser density, with its photons and gravitons extending across the universe.  This is a view that plays a prominent role in Eastern philosophy.  In some versions, they propose that the unity is broken up only by concepts.

I conclude that the key move in history was the development of a self-concept as agent (see “Agency”).  All that matters in life is a function of the subjective stance.  The automaton explanation will continue to be valuable as a tool of science, but, in its pure form, it leaves out all that matters.

Many of the problems addressed by philosophy are due to the conflation of two parallel explanations.