Our heads are stuffed with beliefs. Where do they all come from?
Some beliefs are perceptual. That is, we believe some things because we perceived them to be true in the world. For instance, you might see a woman wearing a red dress and then believe that she is wearing a red dress.
Other beliefs are inferential. That is, we believe something because we figured it out. If you come home and say "Hello!" and nobody answers, you used reasoning to believe that nobody's home. You didn't perceive it directly, you inferred it from the fact that nobody answered you.
Other beliefs are testimonial. We believe many things because somebody told us it was true. This might be because you heard it spoken aloud, or it might be that you read something that a person wrote. It's easy to think that most of what we know is testimonial.
But is that true?
To get an answer we need to think about what it means to "know more." The simplest way to think about it is in terms of pure numbers. If the number of testimonial beliefs is greater than the number of non-testimonial beliefs, then can say we know more from testimony, based on the number claim.
Let's look at that for a moment, and compare the number of testimonial beliefs one has to, say, perceptual beliefs.
We'll look at vision. Cognitive scientists agree that your mind does not store pictures, as such, in long-term memory, but rather stores descriptions of them, in some kind of sentence-like format.* So when you look at some scene and remember it, like, for example, a birthday party, what you remember are facts about the visual and spatial structure of the scene. How many?
It's hard to know, because your mind does compression and other memory-saving tactics. But even with those in effect, it's reasonable to assume we encode, say, ten perceptual beliefs every second. This is probably a low estimate, as I'll show below. This happens just about every second that you are awake, and a person is awake about 21 million seconds per year. That's 210 million perceptual beliefs a year. A ten year old kid already has encoded 2100 million perceptual beliefs!
Here's another way to look at it. A team of scientists made a robot that could see and navigate around its environment (Laird & Derbinsky, 2009). They found that the robot needed to encode between 100 and 1000 elements every 50 milliseconds or so. If each "element" has its own belief, and 420 50ms episodes per year, the robot would have 210 million to 420,000 million beliefs per year. A twenty year old would have 8,400,000 million perceptions in her head.
Let's compare that to testimony. Reading or listening to a sentence takes a second or two. So even if you were reading all day, the testimonial beliefs would never match the perceptual ones, at least in pure numbers. Further, every time you get a testimonial belief, you also get perceptual beliefs, because you have a memory of the experience of reading or listening! (For example, the sound of the voice.)
There are other meanings of "knowing more," however, which I will address in another post. But it's clear, from the number claim anyway, that we don't have more testimonial beliefs than perceptual ones.
This essay is an expansion of a paper I recently published with Carleton University philosopher David Matheson (2012). You can read the full paper here:
Pictured: A reflection of the Eiffel Tower. From Wikimedia Commons.
* These representations are not in English or any other natural language. They are just sentence-like. That we don't store images in long-term memory is something everyone in the "mental imagery" debate in psychology agrees upon. What they don't agree on is whether or not people are able to create bitmap-like images in the mind from these descriptions (Kosslyn, 1994).
Davies, J. & Matheson, D. (2012). The cognitive importance of testimony. Principia: The International Journal of Epistemology. 16(2), 297-318.
Kosslyn, S. M. (1994). Image and Brain: The Resolution of the Imagery Debate. MIT Press. Cambridge, MA.
Laird, J. E. & Derbinsky, N. 2009. A year of episodic memory. Paper presented at the Workshop
on Grand Challenges for Reasoning from Experiences, 21st International Joint Conference
on Artiﬁcial Intelligence, Pasadena, CA. Paper retrieved from http://www.eecs.umich.edu/~nlderbin/ijcai09.pdf