Notes on Subjectivity, Qualia and the Like


Ben Goertzel

July 21, 2005



This informal document is a continuation of a blog entry I wrote on July 19, 2005, in which I talked about the “circular containment” relationship between subjective and objective reality.  This was supposed to be a follow-up blog entry, but it grew beyond the size of a reasonable blog entry (though it still has the informality and speculative nature of a blog entry!).


In these notes, I will present a number of related ideas about subjective and objective reality, moving further into the domain of consciousness, qualia, and the like.  These ideas are somewhat diverse, but they lead in a very interesting direction – toward, for example, a possible way of articulating novel mathematical laws governing the behavior of qualia in subjective reality.


All this material is quite speculative and partially-baked, and is posted here in the spirit of open sharing and exploratory dialogue.


I have not definitely tried to answer every possible skeptical objection to the ideas proposed here.   I have thought of a number of arguments people would have with these ideas, and conceived various counterarguments, but I haven’t taken the time to type in this content. 


More on the Interpenetration of Subjectivity and Objectivity


As I discussed in the above-mentioned blog entry, starting from the assumption of subjective reality, one can derive the possibility of some kind of objective reality emerging – via arguments from developmental psychology.  The root of objective reality lies in the infant’s realization of “object permanence” – that mommy and daddy still exist when they leave the room, and their hand still has visual properties when they close their eyes. 


We don’t yet know any way to derive the detailed structure and dynamics of our physical universe from facts about subjective reality, but my suspicion is that this may be possible.  This gets into deep issues of “grand unification physics”, specifically John Archibald Wheeler’s idea of “law without law,” which hypothesizes that the laws of our physical universe are in some sense optimal, so that if one has an objective or physical world with unformed, indefinite laws, eventually the laws will settle into the optimal-law configuration (being the laws of our universe).   Wheeler’s idea was to assume some kind of pre-geometric, pre-lawful universe, and derive the laws of our universe from this.  But the idea makes just as much (or little ;-) sense if one places subjective reality in the place of Wheeler’s primordial pre-geometric world.


On the other hand, starting the assumption of objective reality – and furthermore assuming a complex, appropriately structured objective reality like our own, with the capability of the formation of complex, self-organizing systems -- one can derive the possibility of intelligent systems like humans and AI’s, that will describe themselves as having subjective realities, and whose internal dynamics will be conveniently describable using the notion of “subjective reality.” 


What these arguments suggest is that subjective reality may be derived as a pattern in objective reality (in that the assumption of minds having subjective realities may help one to explain the dynamics of intelligent systems within objective reality); and, objective reality may be derived as a pattern in subjective reality (in that the concept of objective reality arises in a mind as a way of organizing its subjective percepts/concepts).


Propositional Reality


An alternate way to think about these issues is, I suggest, to introduce a third kind of reality: propositional reality.  By this I simply mean the universe of logical statements, or more properly: logical statements using a vocabulary of logical atoms that includes a way to indicate “this” which is assumed understood by both speaker and listener.  One can think about both objective and subjective reality by associating them with subsets of propositional reality.


Propositions may include, for example, “1+1=2” or “This [indicating a ball] is red” or “This [indicating the speaker] feels sad” or abstract statements like “Red things make people and donkeys feel happy”, etc.


One can look at propositional reality as an abstraction of social reality and linguistic interaction.   All these discussions about subjective and objective reality that we’re having are basically an exchange of logical propositions among human minds.  This discourse only makes sense if one assumes that there is some reality to the space of logical propositions.


We can exchange propositions about things we see around us – so objective reality, in this context, boils down to a set of propositions that all or most of the participants in the dialogue can agree on.


We can also exchange propositions about our internal, subjective worlds – for instance, I can string words together evocatively describing a subjective feeling I’ve had, and you may read my words and map this approximately into some feeling you’ve had.  So subjective content also boils down to a set of propositions that multiple participants can understand.  And in some cases, it boils down to propositions that multiple participants can agree on, such as “Parting is such sweet sorrow.” 


We can’t say that “parting is such sweet sorrow” is objectively real in the standard sense, but we can say that it’s a proposition that is mutually comprehensible to members of a proposition-exchanging community, and that appears to very many of these individuals to be true.


This example shows that one of the main things distinguishing objective from subjective reality is that the former consists of propositions that nearly everyone in the dialogic community agrees to be true.  Nothing in subjective reality elicits such widespread agreement.  “Parting is such sweet sorrow” may be widely agreed on, but not quite as widely as “This rock is hard.”


In this context, the circular-containment relationship between objective and subjective reality boils down basically to mutual logical derivability.  If we take a large enough set of propositions about an individual’s subjective world, then through some complex derivations we should be able to derive important propositions about objective reality – and vice versa.  However, these derivations may be long and complex.  We may view subjective and objective reality as “islands” in proposition-space, separated by long and difficult chains of derivation.


The propositional point of view certainly has its shortcomings – one may argue that projecting into the space of propositions loses something of the essence of both objective and subjective reality.  However, this is exactly the kind of projection we make when we discuss these matters via intellectual discourse like this one.  (Though of course, sharing propositions is not the only way for humans to share knowledge: there is wordless transmission of feeling and intuition, and there is purely physical interaction in the context of the objective world.)


One reason I like the propositional reality idea is that it places the idea of scientific validation in a proper (i.e. sociological) perspective.  Some people like to say that objective reality is somehow more real than subjective reality because statements about objective reality can be scientifically validated.  But what does the scientific method really come down to, in practice?  It’s a complex thing (which I’ve addressed in an extended essay on the philosophy of science, written last year and posted on, but part of it has to do with the replication of experimental results by different experimenters.  In other words, it has to do with the formulation of propositions that multiple individuals will agree are true.  In this sense scientific validation is not really so different from the creation of agreeable abstractions regarding subjective experience.


The Problem of Qualia


Related to the above issues, I've been thinking hard about how to get around the "problem of qualia" in the philosophy of mind.


Let me first explain what I think the problem is.


First of all, a "quale" is defined (by as "a property, such as whiteness, considered independently from things having the property."  In discussions of consciousness, the term is normally used to refer to something that is perceived by a conscious, feeling, aware mind – as opposed to something that is conceived as having an absolute existence in an objective world.  We experience the whiteness of the wall (this is a quale); but we, as subjectively experiencing minds, infer the "objective" existence of the wall, via recognizing patterns of relationship among multiple qualia.


So what is the problem of qualia?


Sociologically, the problem is that there are two very distinct camps out there.


Some of us think "qualia" is a meaningful concept, and that it makes sense to think in terms of things like "Ben's subjective experience of whiteness."


Others believe that "qualia" is a nonsensical conceptual construction.  They make arguments like "The whole idea of qualia is meaningless, because I can never measure a brain and tell if there are qualia in there or not."  Or: "Not being George Bush myself, how could I ever tell the difference between George Bush, and a version of George Bush that had no qualia?"


How can we reconcile these two perspectives?


This sociological distinction reflects a basic conceptual problem, which may be cast as a linguistic problem: How can we meaningfully connect the language of subjective experiences (which we all use on an everyday basis) with the language of empirical observations?


Clearly there is a connection, because when I say "I experience that tree as tall", one can then go measure the tree and see if it's tall or not.  So, from a subjective point of view, I can detect correlations between my qualia and scientific measurements.  But, from a scientific point of view, I can never detect such correlations because there is no such thing as a quale-ometer.  I can detect correlations like "There is a link between a tree being tall and people saying they have the experience of perceiving the tree as being tall."  But this kind of correlation introduces an unsatisfying level of indirection: one is then dealing with talking rather than with experiencing directly -- one is dodging the qualia problem rather than confronting it.


I think the quale-skeptics are correct that qualia are not measurable.  However, I don’t think this means “qualia” are a meaningless concept.   I will give two examples now to explain the sense in which I think non-measurable entities can be very meaningful.


The first example is one that I’ll pursue a little later in these notes:  quantum phenomena.  Quantum reality contains many phenomena that are known to be unmeasurable, but that scientists still find very useful to talk about.  Because positing and analyzing these particular unmeasurables proves very useful for explaining various measurable results.


The second example is simpler: time.  How can we prove time exists?  We can’t.  The statement that time exists is not a falsifiable hypothesis, because the very concept of doing a scientific experiment involves the notion of time (how can we talk about “replication” of experiments without assuming the existence of time?).  Similarly, I suggest, the statement that qualia exist is not a falsifiable hypothesis, but the very concept of doing a scientific experiment involves qualia: we only accept the results of an experiment if we experience them ourselves.  “Seeing is believing” basically means that we want to observe the outcome of an experiment on a measuring instrument, in order to definitively accept it.  In short, individual experience, like time, is part of the language that is used to define scientific experimentation, rather than something that can be measured via scientific experimentation.  Asking “can the existence of qualia be scientifically verified?”, like asking “can the existence of time be scientifically verified?”, is making a category error.


But as with time and quantum reality, the concept of qualia can be judged in a human sense via its usefulness.  Is the concept good at producing interesting, surprising, valuable, ideas?  (And, a careful analysis of the history and philosophy of science reveals that this is ultimately the way scientific research programmes are validated anyway.) 


I’d say the jury is still out on qualia.  They are very useful as a way of explaining various aspects of human experience – for instance, I wouldn’t want to try to describe an LSD trip or a meditative experience or a love affair without referring to various aspects of qualia.  On the other hand, they’ve proved less useful so far for explaining aspects of human cognition.  But I tend to suspect that is because the theory of qualia (in contrast to e.g. the theory of quantum phenomena) has been extremely poorly developed.  With this in mind, in the second half of these notes, I will sketch some ideas indicating what I think some aspects of a theory of qualia might look like.  Of course, this is a very big topic and I’ll barely scratch the surface here – in fact I have a lot more thoughts in this direction, which I don’t have time to write down at the moment.


I note in passing that Buddhist psychology has a lot to say about the nature of -- but its vocabulary is obscure and tends to tangle up descriptive ideas with normative ones.   I have been conceptually inspired by Buddhist psychology in my thinking on these topics, but I prefer to take a fresh start and introduce a new vocabulary and set of concepts more closely tied to cognitive science than Eastern religion.


Qualia and their Properties


OK, so … suppose “qualia” exist in some meaningful sense, then what else can we say about them?  What general propositions can be made about this “subjectivity-reflecting propositional content”?


First, a comment about networks of qualia.  When thinking about qualia like "whiteness", it's hard to see how it could ever be possible to infer things like walls, trees, electrons and people from qualia.  But the trick is that qualia like "whiteness" are not the only kind -- there are also more abstract properties.  There are properties of relationship, such as "on-ness", "beside-ness" and so forth.  One can experience beside-ness, independently from the things that are beside each other.  This is the quale of besideness.  But then one can piece the beside-ness together with the things that are beside each other, which is a matter of building networks of relationship among qualia. 


This -- networks of relationships among qualia -- is what the subjective world is made of.


As noted above, from a subjective perspective, there is no such thing as a "quale versus a non-quale."  Every subjective entity has a quale aspect -- it's "being-in-and-of-itself" -- but then subjective entities may also have relational aspects as well (e.g. whiteness vs. whiteness-of-the-table, where "table" itself is a network of relationships among qualia; or besideness vs. besideness of whiteness and blackness, or besideness of the white wall and the black table).


Qualia, which are properties, can themselves have properties.  For instance, some important properties of qualia are the ones I call arity, centrality, intensity, solidity and historicity.  


Arity has to do with whether a quale is applicable to one thing ("whiteness") or two things ("besideness") or more ("give-ness" relates three things).  Qualia with different arities have different subjective feels to them.  Qualia of arity one may be called "elementary"; those of higher arity may be called "relational."


Centrality has to do with how much focus is on a given quale.  Some things seem to be at the fringe of awareness -- say, a vague sense of unease or confusion, or an idea that one can sense forming but doesn't quite grasp yet.  Other things are right at the center of awareness.


Intensity has to do with how vivid a quale is, how much attention it demands.  This is different from centrality -- because, for instance, sometimes the center of one's awareness can be occupied with something quite pale and calm, other times by something exciting and demanding-of-attention.


Solidity has to do with, for example, the difference between qualia that appear to be perceived and those that appear to be imagined.  A tree in the outside world has a different "feel" to it than a tree imagined inside the mind.  This is not a matter of intensity or centrality; it's a different dimension.  Normally non-solid qualia have less detail to them than solid qualia but this is not a hard-and-fast rule.


Relationship qualia can connect qualia with different degrees of solidity.  For instance, if I believe there is (in objective reality) a fly behind my computer monitor, but I can't see the fly (only the monitor), then I can experience the (somewhat solid) relationship between the non-solid quale of the imagined fly and the solid quale of the perceived monitor. 


Finally, historicity has to do with time -- with, basically, with whether the quale has ever been Present or not.  This quality gives us our innate sense of whether a quale is a memory or not.


There are empirical laws relating qualities of qualia; for instance, central qualia tend to be more intense than peripheral ones, but this is not a universal law.  One may distinguish and analyze those contexts in which peripheral qualia become unusually intense.


Building objective reality from subjective has to do (in a “proposition space” framework) with the formation of hypothetical relationships.  The blackness of the fly behind the monitor is not directly experienced as a solid quale, but the hypothesis is made that if I were to get up and look in back of the monitor, I would then experience it as a solid quale.  Now, getting up to look in back of the monitor itself involves a bunch of different qualia, including plenty of relational ones -- so the hypothesis of the fly behind the monitor is basically a set of implications of the form "If these qualia, then those qualia."  Once the ability for this kind of abstract implication emerges in a mind, then the capability to construct a working concept of "objective reality" is there.  Specifically, objective reality has to do with abstract implications whose conclusions involve solid, elementary qualia.


These various descriptors of qualia are somewhat useful for discussing the standard “paradoxes” of conscious experience.  For instance, Tennessee Leuwenberg, on the SL4 list, recently told the parable:



An intelligent scientist in the future is born on, and living in a spaceship. The inside of the spaceship is not devoid of light, but the colouring of all the internal surfaces happens to be black-and-white in appearance. However, she has a huge amount of information about physics. In this experiment, she is not capable of reproducing anything that is coloured for her to see, but she is able intellectually to fully understand the nature of light, its effects on the human eyeball, brain, nervous system etc.


One day she lands on Earth at the end of her mission. Upon opening the hatch, she casts her eyes first on an enormous bunch of red roses which have been given to her.


‘Oh’, she says, ‘so that's what it's like’.


(Similar parables have been told by others before, and exist in the philosophy literature; I just mention this one because I recently heard it.) 


Tennessee then asked: “Has she learnt anything new about colour? If you accept that she has, then qualia must be real, because she already knew everything that science could inform her about the world and about colour. There must, therefore, be something real about colour which is not addressed by science.”


The difference here seems to be between elementary and abstract qualia, and between solid and non-solid ones.  The scientist in the spaceship could understand roses using abstract, non-solid qualia.  Once on Earth, she could understand them using elementary, solid qualia.  The key point of this story, then, in terms of the current approach, is that qualia are differentiated by (among other qualities) arity and solidity.


Quantum Reality, Subjective Reality, and Exotic Probabilities


Next – going much further out on a limb, but also generating more original and interesting speculations -- I will argue that there are fairly interesting analogies to be built as follows:


Analogy 1 [objective-reality-centric]:

Objective reality is to subjective reality, roughly as classical-physics reality is to quantum-physics reality


In both cases, the latter contain phenomena that are more “nebulous” than the measurable phenomena in the former, but are still useful for explaining phenomena in the former.


Analogy 2 [subjective-reality-centric]:

Objective reality is to subjective reality, roughly as quantum reality is to objective reality


Objective reality is built up from subjective reality, and contains things that are not subjectively real but are useful for explaining things that are subjectively real.  Similarly, quantum reality is built up from objective reality, and contains things that are not objectively real but are useful for explaining things that are objectively real.


I will use these analogies to motivate a novel and interesting hypothesis regarding the nature of entities within subjective reality.


Please note, I am not claiming that subjective reality is quantum reality.  That would be silly.  I'm just making analogies.  These analogies are conceptually evocative, and also seems to lead in some interesting mathematical directions.


First I need to say something about the relationship between quantum physics and objective reality.  This is a distinction that I fudged past in my earlier discussion of objective vs. subjective reality.  Objective reality, as I discussed it there, is about things that we can imagine hypothetically observing if we were in a different position, a different situation etc.  However, the reality portrayed by quantum physics is a bit different, because it consists of things that we could never, in principle, observe, no matter what.  I don’t think that it makes sense to consider in-principle-unmeasurable aspects of the hypothetical quantum universe as part of “objective reality,” exactly.  Rather, these things exist in a different domain – they “exist” in a different sense.  They exist in the sense that postulating their existence is useful for explaining things that exist in objective, “classical” reality.  This is similar to the sense in which postulating the existence of things in “objective reality” is useful for explaining things that exist in subjective reality.  From a subjectivist perspective, objective reality is a useful hypothesis because it helps us explain observed patterns in our qualia, and quantum reality is a useful hypothesis because it assists objective reality in this job.  This is Analogy 2 listed above..


Explaining quantum reality properly seems to involve qualities like "probability wave" that are meaningless in the classical world.  These qualities can never be measured because as soon as they come into contact with a measuring instrument, they "vanish" -- quantum laws only work for unmeasured entities!  So we can, if we like, say that all these unmeasurable quantum properties are not "real."  That's fine, but they're still very useful for explaining the results of measuring devices.  Just as the ball that’s rolled behind the chair is not “real” in the sense of immediate subjective reality, but is still useful for explaining the results of immediate subjective observations at other points in time. 


Similarly, subjective reality seems to be reasonably well explicable using qualities like arity, solidity, centrality, intensity and historicity (and of course, other qualities that I haven’t explicated here).  These qualities can be measured in various ways (e.g. by asking people questions, or by doing brain measurements and correlating the results with peoples' verbal responses) but they also (so I suggest, and I realize this will be obvious to some people yet controversial to others) seem to have aspects that can never be measured -- that are not fully captured via verbalization and are hence, so to speak, beyond the domain of measurement. 


This is Analogy 1 listed above, and it leads up to a very interesting question: What is the right way to measure and manipulate uncertainty within subjective reality?   In other words, if one has propositions about subjective reality, and wants to quantify their uncertainty, what’s the right way to do it?


Cox’s Theorem shows nicely that “probability” is the right way to handle uncertainty – but Saul Youssef showed that ordinary real-number probability is not the only kind of probability there is.  One can construct probabilities obeying all the rest of Cox’s axioms besides real-number-ness, by using probabilities drawn from any of three other algebras: the complex numbers, the quaternions or the octonions.  Youssef showed that quantum theory can be derived from the assumption that one should use complex numbers rather than real numbers to measure uncertainty, together with some other simple assumptions.


Which leads up to the very interesting hypothesis that, maybe, subjective reality also calls for a non-real-number type of probability?  


There is significant intuitive support for this, in the form of the intuition that subjective reality contains things that are “inexpressible in words” – i.e. that vanish when crisply expressed in language , in a similar sense to how quantum uncertainty vanishes when measured by a classical instrument.  This kind of intuition is why Amit Goswami and others have posited that the human unconscious is a quantum system.  Goswami ties in this subjectively experienced uncertainty of the human mind with the hypothesis that the human brain is a quantum system, but I don’t make this same connection.  I think it’s possible the human brain is a  “macroscopic quantum object” in important senses, but the notion I’m putting forth here is quite different from that.  I’m suggesting that subjective experience has some of the same mathematical/conceptual properties as quantum reality, regardless of whether or not the physical system associated with that subjective experience relies on macroscopic quantum dynamics for its intelligent functionality.


Basically, the concept here is that in the subjective world, the dichotomy between X and not-X has a different sense than in objective reality.  In objective reality (by which I mean “measurable reality”, a la classical reality) we can’t have two alternative possibilities both occur – at any particular point in time, the ball is either behind the chair or not, and the rock is either hard or soft.  On the other hand, in quantum reality the electron can pass through both slit 1 and slit 2 (in the classic double-slit experiment).  In subjective reality can for instance, one really be both happy and unhappy at the same time, in the sense of mathematical superposition?


There is a trivial sense in which one can be both happy and unhappy at the same time: the mind/brain is a large, complex system, and one part of it can be happy while the other part is unhappy.  But this is not what I’m talking about; I’m proposing something more radically.  The question I’m asking is whether it makes sense to think of subjective states as unresolved superpositions among multiple possibilities.  If the answer is yes, then this means one wants to use non-real probabilities to reason about entities within subjective reality.  Which would be quite a major and exciting conclusion.


Note also that merely concluding one should use non-real probabilities to model subjective states doesn’t resolve the issue of what kinds of non-real probabilities to use: complex, quaternionic, or octonionic.  In the case of quantum reality the math only works out so as to agree with experiment if one uses complex probabilities.  The situation with subjective reality is far less clear to me at the moment.


My intuition screams out at me here: Octonions!   Octonionic probabilities to measure the uncertainties of qualia!   But I have nothing to substantiate this at present, except for vague intuitions.  This is a train of thought I will definitely keep percolating in the back of my mind, waiting for a breakthrough.  If one wants to show a role for octonions here, the key is to come up with an argument why probabilities of subjective events would be nonassociative … this would mean they have to be octonions rather than any of the other possible algebraic entities.


Finally, let me retreat from this particular train of thought for a moment and return to the big picture.  Whatever the fate of this particular speculation about probabilities and subjectivity, it is an example of the kind of possibility that arises when one takes subjective reality seriously as a domain of being.  I hope I have made a good argument that this is worth doing, at least on a provisional basis to see what comes out of it.