I started reading this paper about a year ago, I don’t know how i found it or why i read it. Recently when looking through my notes I found that I had written extensive notes and I figured the three of you might like to read them and so I whipped them into readable shape.
Full text source: https://www.dan.sperber.fr/?p=35
SPERBER, D. (1997), Intuitive and Reflective Beliefs. Mind & Language, 12: 67-83. https://doi.org/10.1111/j.1468-0017.1997.tb00062.x
Do we have beliefs that are “truthier” than others? Does the truthiness of a belief live in the brain, or somewhere else? Are all beliefs created equal? Are some “beliefs” really just facts, and some “facts” really just beliefs?
These are the kinds of questions some philosophers like to ask, Dan Sperber being one of them.
What follows is my understanding of his paper, in which he argues that we have two main kinds of beliefs.
We know that organisms act, choose, and survive, which requires interacting with the world in a reasonably successful way. That, in turn, means there must be some way for the organism to represent what is going on around it, and to update those representations as reality changes. In humans, Sperber thinks there is a data-base where representations are automatically treated as describing how things actually are, and are freely used as premises in inference and action. These are intuitive beliefs. They feel like “how the world is”. 1
On top of this, humans can also think about representations themselves. We can believe that someone said something, that a claim was proven, that a text is sacred, that a teacher is reliable, and so on. These are meta-representations. When you accept something like “The scientist has proven that X” or “Scripture says that Y and that is authoritative,” you are holding a belief whose content is a representation embedded in a context that you treat as validating. That means, you believe in this representation Y because you have an intuitive belief X that tells you you can trust Y.
Sperber calls these beliefs reflective beliefs. These are embedded in some internal validating context.
Not all beliefs that come from teaching, religion, politics, or expert authority automatically become intuitive. Some of them remain tied to their validating context: “The physicist says that spacetime is curved,” “The Church teaches the Trinity,” “The doctor says this drug works,” and so on. In those cases, your grip on the content can be partial or hazy, yet you believe it because you treat the source or argument as good enough. Sperber’s point is that this is not just a temporary derivation step on the way to an intuitive belief; many such contents stay reflectively held.
Some people claim that whenever we accept something inside a validating context, we immediately strip away the context, like “my teacher said that” or “it has been proven that” and drop the bare facts into the intuition data-base. Sperber calls this disquotational incontinence, and he argues against it. If that view were right, all our real beliefs would be intuitive beliefs, and reflective belief would be a mere moment in the process of learning. But Sperber points out that this would be a bad cognitive design for cases where we only half understand the content: dogmas like the Trinity, or scientific claims, or technical uses of concepts we can repeat but not fully master. Treating such half-understood contents as if they were fully intuitive, free to drive inference everywhere, would be risky and could easily generate contradictions or confusion2.
Instead, he thinks our minds can keep some contents in a more cautious, reflective format. A child can believe that “there are millions of suns in the universe” because a parent says so, even though the child cannot yet represent “sun” and “universe” with the same concreteness or mastery as an astronomer. A believer can affirm “The Father, the Son and the Holy Ghost are one” as a religious truth, even if they cannot make that idea cohere with all their other intuitive representations. These are cases where we can meta-represent and endorse more contents than we can hold as fully intuitive data.
This also shows that intuitive beliefs are not automatically “truer” or more solid than reflective ones. There can be intuitive beliefs that are wrong (“the Earth is roughly flat”) and reflective beliefs that are well supported (for example, quantum mechanics, which most of us hold true largely on trust). The intuitive/reflective split is about how the belief is stored and used in cognition, not about its epistemic status.
In everyday life, many of our intuitive beliefs grow out of perception and simple inference. You see your friend in the smoking area, notice her frown, and spontaneously take it that she is uncomfortable; that can feed into further, possibly more reflective thoughts like “she’s worried about something” or “it’s the cigarette smoke.” Likewise, seeing a bird shiver might ground an intuitive belief that it is cold, which then supports a reflective judgment that giving it warmth would help. These intuitive beliefs are not “written down” anywhere, but they act as direct premises in your ongoing inferences.
By contrast, a lot of what we learn, mathematics, theology, history etc. start off as reflective beliefs. We learn procedures or theorems from teachers and texts we treat as authoritative. We may be able to follow rules (like an algorithm for addition) long before the relevant principles become intuitive. Over time, with practice, some of these domains can partially “drop into” the intuitive box: certain moves in arithmetic or geometry can come to feel as obvious and effortless as perceptual judgments. But Sperber’s claim is that there will always remain many beliefs that we hold reflectively: they are anchored in validating contexts and meta-representations, and never fully migrate into pure, freely inferential intuition.
One consequence is that humans can, thanks to language and other symbolic systems, carry around far more than could ever be represented in a purely intuitive way. Our meta-representational abilities let us believe contents that exceed our intuitive representational capacity. That is part of what allows for abstraction, science, and religious or ideological systems: we can stack and scaffold beliefs about claims, proofs, authorities, and texts, without needing every last bit of that structure to be intuitively available.
Disclaimer: I probably got a lot of details wrong. I am not qualified to read most of the stuff I seem to read. I do not know how to fix this.