Destroying the dichotomies
Introduction
In a previous post, I talked about the ultimate importance of destroying certain dichotomies, but I didn’t really present their destruction in detail. In this post, I'll do just that.
Each dichotomy I’ll discuss is obviously a huge philosophical topic in its own, and I won’t be covering the entire literature on the subject. I’ll mostly offer a taste of how one might attack the dichotomies, to give a sense of what losing it might be, using arguments I personally like.
You’ll see that I talk a lot about predictive processing (PP), but the destruction of these dichotomies doesn’t depend on it. I just like predictive processing, and it gives a bit of a concrete scientific feel. But these dichotomies have already been extensively challenged philosophically, and often the idea can be recovered just through phenomenology.
Analytic Statement vs. Synthetic Statement
An analytic statement is roughly supposed to be true or false by virtue of the concepts it uses, while a synthetic statement is supposed to be true or false by virtue of “how the world is”, or “how the things it talks about are”. For example, “bachelors are unmarried” is said to be analytic, while “bachelors are short” is supposed to be synthetic.
I’d like to introduce here the way predictive processing (PP, a cognitive neuroscience framework I’ve discussed several times on this blog) accounts for concepts, as seen in the work of Christian Michel.
In this model, concepts are “prediction units”; they are the entities in terms of which predictions are made. A given concept is instantiated by a prediction unit, taken as the root node of an extended tree of other prediction units. The extended network of a given concept corresponds to the sub-network in the PP model that consists of the concept root node and all of its child nodes (each child node is itself a concept root node). The idea is that the activation of a concept’s root node makes available a body of information, namely the subnetwork depending on that root-node. The body of knowledge represented in a PP model is a web of interconnected concepts.
I highly recommend reading Christian Michel’s work on PP and concepts, it’s fascinating, and it allows unifying various theories of concepts (theory-theory, prototype, exemplar, etc.) and phenomena (like context sensitivity).
So a concept is roughly made up of a root node, a cluster of nodes densely connected to it, and a tree that extends into the network in such a way that the boundary between what is part of the concept and what is not is fuzzy.
What we see with this plausible account of concepts is that there's no real difference of profound nature between statements considered analytic and those considered synthetic. A connection between two concepts is “more or less analytic” depending on how densely they are connected in clusters.
We also see that there’s no such thing as an “inseparable connection” between two nodes. Analytic statements are generally thought to be non-revisable, unlike synthetic ones, but in PP, any connection can be modified, whether intracluster or intercluster. Everything is “theory”. The distinction between “a component of a theory” and “a theory” fades in favor of a fuzzy distinction between “big theory” and “mini-theory”, where, for example, the concept of APPLE is seen as a kind of mini-theory that can be revised in itself.
I think the analytic/synthetic distinction is more of a socio-linguistic phenomenon than a deep doxastic one. For communication, we need contact points between our theories. We attach names to mini-theories (like “apple”), and if someone’s mini-theory diverges too much, we refuse to grant them the name, saying they don’t “really” have the concept of APPLE or aren’t “really” talking about apples.
Concepts are dynamic and fluid entities. There’s no connection in our minds that is non-revisable, not even within a concept; it’s just that when we revise an intraconcept connection, we sometimes say, as part of our socio-linguistic practices, that it’s no longer “really the same concept”.
With this, the analytic/synthetic dichotomy dissolves somewhat. Analyticity becomes a gradual, socio-linguistic feature, all beliefs are revisable, etc.
It might seem that what I’m saying doesn’t destroy the dichotomy but merely destroys its analytic pole. It’s true that the analytic has been more directly attacked, but that’s not entirely accurate.
On one hand, both analytic and synthetic statements are reinterpreted and absorbed into a single framework (PP) where the dichotomy doesn’t exist.
On the other hand, even if it seems we’ve mainly “added synthetic to the analytic”, we’ve also added some analytic to the synthetic.
This account of concepts illustrates the theory-ladenness of every statement. When we say “there is an apple on the table”, it’s not “just” an innocent bite-sized statement; it carries with it the mini-theory of apples, the mini-theory of tables, etc. So, the truth of this statement doesn’t depend just on an apple and a table; it depends on a whole theoretical framework that postulates apples and tables as theoretical entities.
This may be hard to notice for everyday concepts, but it’s obvious with outdated scientific theories or conceptual frameworks. For example, saying “there are miasmas in this house” is not just an innocent bite-sized statement; it brings with it the entire (now outdated) theoretical framework of miasmas. Words are not just innocent “pointers” to external entities; they are theories themselves.
So we can say that there’s a bit of analyticity in statements considered synthetic because their truth depends on the truth of mini-theories (the concepts they use), and these mini-theories are/contain statements considered analytic. A synthetic statement contains concepts, and a concept is a cluster of statements considered analytic.
Some may say this stretches the point too far, but still.
In short, all this should lead us to think the analytic/synthetic dichotomy is quite wrong, and we should absorb it into a framework where it doesn’t appear. This impacts our relationship to doxastic matters; we realize that an apparently innocent statement brings a whole theoretical machinery with it, and that there’s no doxastic connection that is non-revisable outside of artificial socio-linguistic considerations.
Component vs. Compound
As we saw earlier, a concept is somewhat of a mini-theory in itself. In that sense, we lose the distinction between “a component of a theory” and “a theory”. I think this is quite interesting to reflect on. It means that when we build a theory in the classical sense, what we’re actually doing is connecting mini-theories together to form a larger theory composed of mini-theories. We lose the distinction between “what the theory says” and “things about which the theory says something”.
We might realize that the way we treat differently mini-theories and large theories may be arbitrary. Why do we say a concept refers to objects, while a statement/theory refers to a state of affairs? They’re (concepts and theories) kind of the same thing. Why do we say a statement/theory has a truth value, but not a concept, which is instead said to have an extension, to be more or less “appropriate”, etc.? They’re kind of the same thing.
It seems we may need to abandon the theory/concept dichotomy, the object/state-of-affairs dichotomy, and perhaps even the truth value/extension dichotomy. But what would that leave us with?
For the object/state-of-affairs dichotomy, perhaps a structural, relational, or other exotic ontology could be suitable.
For the loss of the truth value/extension dichotomy, it seems we’d need a kind of holism where, instead of talking about concepts that refer to objects and statements/theories that correspond or not to facts, it’s the total doxastic network (or large parts of it) that corresponds or not to the whole of reality (or large parts of it). Reference, correspondence, and truth become fused, just as concepts and theories do, and just as objects and states of affairs do.
(This probably has affinities with slingshot arguments.)
Phenomenal Experience vs. Beliefs
Phenomenal experience in perception is generally thought of as a kind of neutral or quasi-neutral source of information we use to form beliefs about the world, a kind of medium between beliefs and the external world. In this tripartite model, we have, relatively separately, beliefs, the external world, and phenomenal experience as a relatively neutral intermediary.
In PP, phenomenal experience of perception is not a neutral medium between the doxastics and the world; it is the result of what the brain believes is tickling its sensory organs. With a bit of phenomenology, introspection, and effort, one can quite literally notice that our theories are in our phenomenal experience of perception. I emphasize this; you can really see your theories in your perceptual experience.
When you look at an apple, you don’t just believe that it’s a 3D object or that it’s persistent; you see it as “having” the quality of “3Dness” and “persistentness”. And as explained in previous posts, advanced meditators can in a certain sense gradually “deactivate” the theories in their brain, and this is accompanied by a transformation of their phenomenal experience; it becomes less and less fabricated.
And it’s not just that the brain “overlays” or “superimposes” things onto a “raw”, “neutral” phenomenal experience; every phenomenal experience is the result of theories. A sufficiently advanced meditator can deactivate all the theories in his brain, producing what’s called a “cessation”; a loss of consciousness. Even the most minimal phenomenal experience is still the result of our theories.
So what we have here is a kind of destruction of the phenomenal/belief dichotomy. Phenomenal experience is quite literally woven from our beliefs.
This raises a serious problem for traditional epistemology. How can we know whether a belief is good? By testing it against perception. But not everyone perceives the same thing. So we need to find “the good perceptions.” But perception depends on beliefs. So we need to find “the good beliefs.” But how do we know if a belief is good? By testing it against perception…
This loop might seem dumb, but it’s actually extremely hard to escape in traditional epistemology. With the destruction of the phenomenal/belief dichotomy, we lose that “neutral medium” between the doxastics and the world, and we find ourselves immersed in a world woven from our beliefs, in which we can’t really separate what comes from the world and what comes from us because there is no such distinction.
We can try triangulation, amplify sensory data, open brains to see beliefs, etc., but in the end, we remain immersed in something that, rather than being a neutral medium between world and belief, is woven from our beliefs.
Another interesting consequence is that, if our doxastic/conceptual structure is deeply relational as we saw earlier, and our phenomenal experience is woven from that structure, then our phenomenal experience is also probably deeply relational. When advanced meditators say that “everything is interconnected”, it’s not just a New Age metaphor; it’s a deep phenomenological insight.
Non-Cognitive vs. Cognitive
I’ve discussed and challenged this dichotomy in another post, but here are the key points.
We generally have a very different attitude toward the non-cognitive (desires, emotions, sensory sensations, etc.) and the cognitive (beliefs, thoughts, understanding, reasoning, memory, judgments, etc.). To use the analogy from that post:
Let's imagine a tiny island in the middle of the ocean, an island that is constantly being battered by the waves. Here, the island represents our self, and the waves represent our sensations. But it doesn't seem to us that our cognitive is waves; it seems to us that waves include sensory sensations, emotions, etc., but we imagine our cognitive instead as a lighthouse standing in the middle of the island. We see the cognitive as something that transcends sensations, that sees beyond; we associate the cognitive with activity (not passivity), solidity, penetration, beyond, transcendence, light, the self (thought as activity, as movement, as a manifestation of our self), and so on.
But there’s no deep difference in nature between the cognitive and the non-cognitive:
“It feels like something to have a non-cognitive attitude” : The cognitive also has a phenomenology.
“The cognitive controls behavior” : So does anger.
“Cognitive states can be dispositional; we keep believing something even when not thinking about it” : So can anger (we can be “hot-tempered” without being currently angry).
“The cognitive influences what we perceive” : So does anger.
And so on.
The more we reflect, the more the difference between cognitive and non-cognitive seems superficial. Sure, the cognitive feels different from the non-cognitive, but sound feels different from color too. The cognitive might be a very fine-grained phenomenological modality, but why not treat it like very fine-grained personality traits? The cognitive surely allows things the non-cognitive doesn’t, but why not think of the cognitive as a whirlpool, or a big wave that synchronizes others, rather than a metaphysically separate lighthouse?
(Also, with a bit of phenomenology, we realize that emotions are also woven from our beliefs, not just perception.)
Once we go beyond this dichotomy, the notion of intentionality starts to seem questionable. Why do we consider that cognitive attitudes “say things about the world”, but not non-cognitive ones? If we abandon the lighthouse image and bring the cognitive closer to the idea of a big wave that synchronizes others or to fine-grained personality traits, then the view of concepts that “point” to objects, and of propositions that “point” to states of affairs, with doxastic attitudes having “truth values”, may begin to fade away.
Subjective statement vs. Factual statement
Here, I want to ask a question. Why do we say that aesthetic judgments like "this painting is beautiful" are subjective, in contrast to "factual judgments" like "this painting is rectangular"? Of course, some argue that aesthetic judgments are objective, but let’s ignore those mad philosophers.
What I’m suggesting is that the judgments considered “subjective” often are so because they concern a topic where members of the community have a kind of direct, non-inferential, and fundamentally irresolvable disagreement, and because that disagreement isn’t particularly serious (/because it’s tolerated by the community). If we look at things this way, it could explain why some philosophers are anti-realists about colors; because a significant part of the population doesn’t perceive colors in the same way.
Obviously, philosophers will offer all kinds of philosophical arguments for their anti-realism about color, but still, could it not be that the differing treatment of color, due to the difficulty in achieving intersubjective agreement, is what originally motivates those philosophical arguments?
When you think about it, many concepts ultimately seem to be grounded in our subjective experience, not just concepts of color or aesthetics. I’m not trying to defend a form of empiricism here, but take for example the concept of SOLIDITY. Isn’t it at least somewhat grounded in or linked to the feeling of resistance to penetration when we press on an object with our bodies? If that’s the case, is the concept of SOLIDITY really so different from aesthetic concepts, which are said to be grounded in subjective experience? And if people had a radically different experience of the world and failed to achieve intersubjective agreement on the solidity of objects, couldn’t some philosophers become anti-realists about solidity?
In the Predictive Processing (PP) framework, our hierarchical generative model isn’t there to mirror an external reality, it’s there to predict incoming sensory data in ways relevant to our needs. And in active inference (an extension of PP), our hierarchical generative model handles action, desire, and prediction/belief all at once. In fact, in active inference, the belief/desire and action/prediction dichotomies are quite literally destroyed; everything is modeled the same way by a single system. It’s not just that it does all these things simultaneously; it does them the same way. There’s no distinction.
So in a way, we are a system managing a “thing” (free energy) that is a fusion or blend of sensory predictions, sensory interpretations, beliefs, desires, and actions. Faced with that, we might conclude that color and aesthetic judgments being “subjective” isn’t unique at all. Probably all concepts relate, more or less closely, in one way or another, to our subjectivity.
But we’re social animals with language, so we have to align our subjectivities a bit, and that intersubjective (idealized) alignment is what we call “the factual.”
(By the way, just as with a bit of phenomenology or meditation, we can notice that our beliefs are quite literally part of our phenomenal experience, we can also notice that “possibilities for action” or “interaction potentials” are quite literally part of our phenomenal experience too (this makes sense in active inference); see for instance the idea of “perception of affordance” in enactivism).
There are many topics where the general population has no trouble saying there’s “no objective answer”, that it’s “a matter of opinion”. In my view, philosophers dislike the idea of propositions that “have no objective answer”. So they’ll create complex and convoluted analyses to reduce propositions that seem not to have objective answers into ones that do: “What you REALLY mean by ‘this painting is beautiful’ is ‘this painting elicits a pleasant aesthetic qualia in me’”.
I find that a bit ridiculous. No, I insist; what I mean by “this painting is beautiful” is “this painting is beautiful”, and no, I’m not going to adopt an error theory, a non-cognitivism, a subjectivist cognitivism, or an aesthetic property realism.
Many seem to think that either objective aesthetic properties exist, or aesthetic statements refer to the speaker’s feelings. But there is a third option: destroy the subjective/objective dichotomy.
I think some philosophers could benefit from studying other cultures’ or eras’ worldviews. That sense that “everything is objective, and if a statement seems subjective, it’s because it’s covertly about objective facts concerning the speaker” is not at all universal. For example, in Amerindian perspectivism, there’s not really a “common world”; each individual’s perspective defines their world, and that world can change. Interactions between individuals are seen as indirect and not really taking place in an “objective reality”. Each person has their own perspective => each person has their own reality.
That’s just one example, but anyone who’s studied anthropology or history knows how strange “exotic” worldviews can be from a contemporary Western perspective.
So we’re arriving at the destruction of a very important dichotomy: the subjective/objective or subjective/factual dichotomy. Likely all concepts relate, more or less, to our subjectivity, and the notion of “factual” applies to judgments that can ideally achieve, or that the community wants to achieve, some form of intersubjective agreement.
Normative vs. Factual
I’ve already talked about this in another post, so I won’t repeat everything, but the idea is that the concept of TRUTH is, in my view, constitutively normative, just like the concepts of BEAUTY or MORALITY.
An ugly painting is “incorrect,” a beautiful painting is “better/correct”.
An immoral action is “incorrect,” a moral action is “better/correct”.
A false belief is “incorrect,” a true belief is “better/correct”.
I won’t repeat my arguments, but what I think is that, since we’re social animals, we have a “feeling of correctness” in certain domains. In the domain of actions, this feeling manifests as morality; in the domain of contemplative objects, it manifests as beauty; in the domain of beliefs, it manifests as truth; in the domain of beliefs' origins, it manifests as justification; etc.
With this view, “aesthetic facts” are kind of an outgrowth of the feeling of correctness in the contemplative domain, “moral facts” are an outgrowth of the feeling of correctness in the domain of action, and “facts” simpliciter are an outgrowth of the feeling of correctness in the domain of belief. Metaphysical realism is an outgrowth of a normative feeling.
This collapses the normative/factual dichotomy, in the sense that the very concepts of TRUTH, FACTS, or OBJECTIVE REALITY are all in a way originally and constitutively normative; they’re ways of seeing things that stem from the social nature of humans and their norm-governed practices.
Representation vs. Referent
Representationalism holds that assertions, beliefs, words, concepts, propositions, etc., are “representational tools.” These linguistic and mental entities are seen as having some semantic relation to a mind-independent external reality. Concepts have an extension, a reference, an intentionality; they “stand for” elements of an external world, they have “aboutness.” Propositions and assertions “represent” an external reality; they can, in a certain way, be “compared” to that reality, and they’re “true” if they “mirror” or “correspond” to it. Language is seen as saying things about that reality, and when it fails to correspond, or cannot correspond, there is an error, either of truth or of language use.
The problem is that the notion of “representation” just doesn’t work. I’ve written an article on this so I won’t repeat everything, but here’s the big idea, inspired by Putnam’s model-theoretic arguments.
Basically; language is not “pre-connected” to the external world. The connection between language and the external world must be made by humans. Since humans have no direct contact with the external world, they can only try to connect language to the external world with language. Obviously, this is impossible. Language cannot connect itself to the external world. If we want to explain under what conditions which concept refers to which piece of the external world, we're immediately confronted with the problem that language must be pre-connected to the external world if the conditions we're stating are to capture something, or capture the right thing, about the external world.
To reuse an analogy from another post:
On a paint palette, you can put colors and mix them to create new ones. Now, imagine you've been asked to create not a color, but a sound on your palette. The best you could do would be to create a color that evokes or is associated with that sound. That's not what you were asked to do, but it's not that bad. But now imagine that no sound has ever been associated with a color, and that therefore, for humanity, no color evokes any sound. In this situation, you're asked to create a given sound on your palette (using only your palette). In this situation, you realize that not only can you not create the sound on your palette, but you can't create a color that evokes that sound either. You could try to create in humanity this association between a color and this sound yourself, but you've only been allowed to use your palette and mix paint; you can't create an association between a color and a sound by just mixing paint on your palette!
The point of this analogy is that our thought is supposed to be of one kind, and reality of another. Our thought cannot “contain” reality. Some people remain indifferent to this; “obviously our thought cannot contain reality, our thought refers to reality”. The problem is that for our thought to refer to reality, there needs to be a connection between our thought and reality, but we can only use our thought, and thought cannot connect itself to reality.
The loss of representationalism, for me, really completes the destruction of this kind of view with thoughts on one side, the world on the other, phenomenal experience as medium, and thoughts “pointing” to the world.
Some may want to think that the fall of representationalism means that “reference slides freely over the surface of noumenal waters”, but in fact it means there’s no fundamental difference between “representation” and “referent”; or rather, that the very notion of “representation” is flawed.
The idea of “referents” likely comes from the fact that we’re constantly “improving” our “descriptions”, and so we imagine some kind of “ultimate description” that we reify as an external objective thing, especially in a social context. But upon reflection, there’s no real fundamental difference between “description” and “object of description”:
In the non-dualizing mode of discourse the term ‘object of description’ is therefore replaced by ‘description so far’ and the ‘description of the object’ by ‘description from now on’
Source
(This quote is truly magnificent.)
World for us vs. World in itself
Many people say things like “okay yeah, we can’t grasp the external reality, the world in itself, in our minds,” but they don’t really internalize that. They kind of have the impression that, okay, the world isn't made of chairs, trees, etc., because that stuff comes from our concepts that are “world-for-us-kind-of-things”, but still, they have the impression that the world is made of structures, or maybe causation, etc., when that stuff too comes from our concepts that are just as much “world-for-us-kind-of-things”.
What needs to be realized is that we are fully immersed in the world for us. The idea that we can, in any way, come into contact with a “world in itself” is incoherent. The very notion of the “world in itself” is incoherent. To reuse a correlationist analogy:
Looking through an (empty) beer bottle, you'll see the world as brown and radially distorted. Obviously, you know that the world is not brown and radially distorted. Even if you'd been wearing beer bottles in front of your eyes all your life, you'd probably know that the world isn't brown and radial (it might perhaps be difficult to discover this, or hard to grasp what a non-brown and non-radial world means, but this difficulty doesn't seem insurmountable).
But now imagine that pieces of beer bottles are in every one of your concepts. Now, the problem is deeper; the problem is not “only” at the level of an “obstructed interaction” with the world, the problem is embedded in thought itself, such that a thought that the world is not brown and radial is incoherent, impossible, unintelligible, erroneous, illusory.
The idea is that we're totally immersed in the world for us, so we think through “world-for-us-kind-of-things”. Some try to salvage things by distinguishing between “grasping the world in itself” and “referring to the world in itself”, but that doesn’t escape the correlationist problem (since that distinction itself is a “world-for-us-kind-of-thing”), and as we’ve seen, representationalism makes no sense.
Once again, we shouldn’t try to destroy one pole of the dichotomy. We shouldn’t destroy “world-in-itself” and keep “world-for-us”. We must destroy the whole dichotomy. Just as there’s no true distinction between perspective/reality or description/object of description, there’s none between world-in-itself and world-for-us.
Conclusion
In my previous post, I talked about the destruction of the “ultimate dichotomy” which must be done after the destruction of the others. But I think that having destroyed all these dichotomies, the ultimate dichotomy should have destroyed itself.
This “ultimate dichotomy” is the one between perspective and reality, or fiction and reality, or description and reality.
By losing this dichotomy, we realize that everything said in this article is just a perspective/description no more or less valid than the conventional one that preserves all these dichotomies and remains metaphysically realist, representationalist, etc.
This transforms our relationship to the doxastics. Everything said here isn’t a set of deep or ultimate truths; there is no such thing. It’s just useful descriptions. This transformation loosens our attachment to “views”, and we acquire a relationship to belief that is one of lightness.
We have to climb all the rungs of anti-representationalist ultimate truths that reveal the falsity of cognitive/non-cognitive, normative/factual dichotomies, etc., to finally reach the top and realize that the “ultimate truths” of anti-representationalism are in fact conventional truths.