That 'knowledge' is structural is fairly uncontroversial. Although there are certainly different types, I would like for sake of this question to characterize (what I see as) a generalization of the concept: Knowledge are familiarity or understanding of particular items of experience. It can be communicated, though not always perfectly, and thus are sometimes "fuzzy" (people may have slightly different ideas about the facts). The more structured, the less fuzzy knowledge becomes, but then less facts are accommodated by such a structure.

When we think of "Confirmation bias" it is usually in terms of an error of induction, or a psychological propensity to misconstrue. However I would like to likewise expand and generalize the idea: In the case of psychological propensity it manifests at many different levels, from the interpretation of language to sense perception. Confirmation bias is (loosely) fitting knowables into some structure established by prior "knowledge". (If one is fanciful one may even make a case for probability distributions and other quantum phenomena)

“In visual perception, you have a process that suppresses ambiguity, so that a single interpretation is chosen, and you’re not aware of the ambiguity.” - Daniel Kahneman

Now without getting into a very long tirade, this is the hypothesis I got to:

Knowledge structures are formed by Confirmation bias.

Edit: From some of the references I've come to realize that my use of "comfirmation bias" need to be better elucidated. There is a phenomenon/process, recognizable as the same, and most apparent in the guise of confirmation bias in the areas of sociology and psychology, but also in the physical and metaphysical that for the sake of this question I will call Existential Normalization (EN). EN is then a reification of this phenomenon as distinct from knowledge, and knowledge structures, and operating on all levels from social constructs down to the aggregations of matter.

Note that now "confirmation bias" has changed from a somewhat vexing phenomenon into a sort of ontological entity. Knowledge is structured fact which is formed by EN, which is informed by Knowledge. A triad of concepts that allows for what IS to become MORE.

Question: Has anybody done something along these lines?

Why is this interesting? Most creatures aren't aware of their own biases, an awareness of one's own confirmation bias are likely a first step to Self-awareness. Also think of a DNA sequence as "knowledge"...

  • Again this sounds very Kantian, the patterns we get out of experience are the ones we ourselves put into it. Making those patterns relative to the priors makes it neo-Kantian, but "confirmation bias" is not used this way. You can look at Friedman's relativized a priori, and fallibilistic apriorism generally.
    – Conifold
    Nov 1, 2019 at 23:29
  • @Conifold Fallibilism certainly shares a core conception of "knowledge" with the premise of this question. Your linked answer mentions how the notion of a priori has changed; what I'm asking about here could be said to be a principle or a mechanism that would act to limit the rate of change, if knowledge had mass this would be gravity.
    – christo183
    Nov 5, 2019 at 9:28
  • The evolution of a priori is a side issue. What I meant is that relativized a priori are exactly the "confirmation bias" and the "gravity" you are looking for. They are the structuring patterns/principles used to shape new knowledge, and change much slower than that which they shape. The stuff paradigms are made of, if you will.
    – Conifold
    Nov 5, 2019 at 9:35
  • @Conifold Thanks, I see. So Reichenbach and Friedman, any others? Did they take knowledge to be only declarative? Did anyone try to make a case for relativized a priori, or more likely a precursor concept, being somehow metaphysical, pre-linguistic or observer independent?
    – christo183
    Nov 5, 2019 at 10:42
  • 1
    Look at Mormann's recent survey Toward a Theory of the Pragmatic A Priori: From Carnap to Lewis and Beyond. Of course, Kuhn also comes to mind, and the elaboration on him by Lakatos.
    – Conifold
    Nov 6, 2019 at 6:36

3 Answers 3


Well, yeah.

A version of this problem has been tacked by many people for millennia. In terms of philosophical school, the idea can be found in the ancient school of skepticism.

In terms of examples, we can probably start with Heraclitus's famous fragment that a man cannot step into the same river twice. There are many questions regarding this fragment, but it is clear that he is pointing a confirmation bias - seeing a familiar-looking object leads us to believe that this is the same object that we have seen before, although it is actually not so simple.

About the same time, at the other end of the world Buddha was teaching his disciples to free themselves from Upādāna, translated as attachment, clinging which included attachment to any kind of doctrines and worldviews. He recognized that this attachment to ideas is a major cause for confirmation bias which in turn caused suffering that the world doesn't meet our expectations.

Later, Hume attacked the concept of causality, on which all knowledge is based, saying that causality is a matter of custom or habit. So, he basically takes the extreme view that all knowledge is subjective and biased.

But if knowledge structures are formed by confirmation bias, then how does knowledge begin i.e. how are these structures formed in the first place? Kant gave this problem a long treatment in his first Critique.

Later treatments of the topic are more technical, for example we can review Wittgenstein's attempt to separate bias from objectivity, which can be summarized by the following quote:

So too the fact that it can be described by Newtonian mechanics asserts nothing about the world; but this asserts something, namely, that it can be described in that particular way in which it is described, as is indeed the case. The fact, too, that it can be described more simply by one system of mechanics than by another says something about the world.

Source: Tractatus 6.342

  • Illuminating view of Heraclitus and Buddha...
    – christo183
    Dec 6, 2021 at 13:27
  • Added some more context, hit "Accept" if you like ;)
    – Jencel
    Dec 6, 2021 at 14:53
  • +1 Me like. I love me a good LW quote.
    – J D
    Dec 6, 2021 at 17:06

The basic concept as outlined is well-represented in philosophy, though no one I know of uses this particular terminology. 'Confirmation bias' is a modern term — mid-20th century — and is mostly used in psychology, with a few ports to the philosophy of science. But the idea that knowledge is built from imperfect structures is implicit at least back to Hegel (where the notion of a dialectic works with it explicitly), and arguably all the way back to Plato's theory of forms (where it is the philosopher's job to see through the imperfect manifest structures to understand the perfect form behind it).

Certainly critical theorists are constantly working with biases and skewed knowledge structures. I'd guess the epitome of that work would be Gadamer's "Truth and Method", where he argues that all knowledge structures are bounded and localized: that we come to understand the world as though we were standing in a valley surrounded by mountains. Not only do we not know what lies beyond the horizon of those mountains, we may have no common understandings at all with someone raised in a different valley.



I'm going to reduce your statements presuming that the reformulation is accurate; please let me know if it's not.


Knowledge is structured by propositions articulated by people who from experience construct them with "biases", cognitive or otherwise. Since the "bias" is built into the process, somehow, all knowledge is itself "biased". Thus, where knowledge exists, it is necessary that "bias" exists.

Short Answer

Yes, this is a prevalent theme in a number of Analytical and Continental philosophies. Critical theory, for instance, presumes that not only is this the case, but it is used for one group of people to maintain a power imbalance over another. Deconstruction is a philosophy that challenges the notion that a text is easily understood, if understood at all, given differences in the biases of the author and the reader. Quine's arguments about the indeterminancy of translation might also be seen as problems regarding a coherence between structures of knowledge. Lastly, in the philosophy of mind, embodied realism can be understood as all knowledge being built on a particular bias rooted in the nature of physical being and experience.

Long Answer

Since my interest is the the analytical tradition with a strong emphasis on the philosophy of language, I'll respond from regarding the theory of embodied cognition, which is a philosophical theory built on the findings of the linguistic turn and the naturalized epistemology that includes cognitive science.

In philosophy, meaning is the lynchpin. No meaning, no knowledge, plain and simple. Thus, let's lay bare that semantics has some light to shed on philosophy, which is uncontroversial in analytical philosophy. First, you need to wrap your mind around the fact there are multiple theories that go into meaning. Certainly, Gottlob Frege's paper Über Sinn und Bedeutung which posits the distinction of sense and reference is important. But, the philosophy of language has grown since. From Szabó and Thomason's Philosophy of Language, page 196:

[The] conservative approach to pragmatics has the advantage of being congenial to linguistic theorizing...[but it] excludes many important aspects of language use... Even simple utterances are ambiguous... "natural language understanding" draws... heavily on world knowledge and common sense.

From artificial intelligence, we know, that common sense is a tremendous difficulty to overcome in computational linguistics and improving computational intelligence exactly because common sense is a series of propositions of knowledge that are essentially biases in the structure of knowledge. That's to say, that common sense is viewed as a structure and method of knowledge, thus both ontological and epistemological knowledge, which is highly normative; unless that normativity is explicitly specified by codified ontology, computers simply don't have the ability to decide what are axiological conditions.

The question that has been looming since Turing's test, is, how to deal with this bias in propositions that arise from intuition such as common sense, folk psychology, or naive physics. It's a deep philosophical dilemma that lies at the basis of the question in computational linguistics that asks, for instance, 'How should a computer know when 'tall' is used as an adjective when using different nouns in different contexts?'. And in the philosophy of language, it is a universal consensus that there is no solution yet to the problem.

Now, from the philosophical theory of embodied realism, there is a broad solution of sorts. One has to abandon truth-conditional semantics and find a theory of semantics that somehow captures where this meaning comes from in a way that goes beyond the notions of truth and falsity. There's a school of thinking called cognitive semantics that seeks to pick up where later Wittgensteinian thought leaves off with ideas like language-games and family resemblance.

In short, cognitive semantics posits where the "bias" in language comes from. It comes from the neurons in our brain which makes the philosophical theory a both a psychologist and conceptualist theory of mind since it quite literally addresses how the brain and the basic categories relate to the "bias" you are referring to. This is why as a philosophical theory, one has to accept Quine's advocacy of using scientific, and thus psychological propositions, in philosophical reasoning. While the article on embodied cognition covers the various proposed structures, three important ideas are that of prototype theory of definition, conceptual metaphor of meaning, and ontological neural computation, all of which inhabit niches of classical philosophical problems.


So, there are philosophies that certainly recognize the tendency of biases and preferences in meaning, and explore exactly what they are. I've tried to review what current embodied realism entails, though a much better start would be Johnson and Lakoff's Philosophy in the Flesh. In it, the main thesis is that knowledge is never objective and disembodied, but rather biased on account of certain dispositions (SEP) of biochemicals and teleological functions of biological processes (SEP) that inevitably lead to a disposition to normativity including informal fallacy, cognitive biases, defeasibility in argumentation, as well as explaining the very nature of semantics itself. In this way, propoponents absolutely accept that where knowledge exists, it is necessary that "bias" exists.

  • 1
    What interest me is the ontological implication of bias as a given.
    – christo183
    Dec 14, 2021 at 8:13
  • 1
    @Christo183 Right? I think it's rather Kantian to believe that all thought is phenomenological and that noumenological in the absolute sense is beyond our grasp. At best we settle for a transcendental knowledge that is a bias of our subjectivity.
    – J D
    Dec 14, 2021 at 10:14
  • 1
    CPR Section III.10 pure understanding in the synthetic a priori starts with Aristotelian categories which might be thought of as the biases of the mind.
    – J D
    Dec 14, 2021 at 10:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .