There’s a long history of debate about what the definition of knowledge is. For a long time it seemed to be settled and the accepted definition was “justified true belief”,until the Gettier Problems showed that that definition doesn’t always work. And that’s actually something I really enjoy about writing definitions, even if we’re not sure what the right definition is, we can still tell when one is incorrect.
This is why the Gettier Problems are so convincing, they showed examples that strictly matched the stated definition, but they were situations where we wouldn’t use the word “know” to describe what was happening. In those situations where most people wouldn’t use the word, but the definition says the word is appropriate, we say that the definition is wrong. Even in situations where we can’t identify a better definition, as long as we can use the word consistently, we actually don’t need a definition.
But that doesn’t mean a definition isn’t useful. If we can identify the characteristics necessary for a definition to match our actual usage it can reveal interesting facts about how we view the world, and help us think about and discuss these situations better. In fact, I think failing to do this is one of the biggest problems we have in coming up with good definitions. We start with a few things that we think it has to include and then add on or modify them until the definition works. But often we can be completely wrong about what’s really fundamental to the definition, and we end up with lots of extra baggage to make everything fit together.
It’s better to start with what seems right, but be willing to abandon any part of it. Often I’ve found that I end up getting rid of everything I thought should be in the definition and end up with a completely different set of characteristics. Let’s work though “justified true belief” and try to see if we can improve each of the terms, or if they need to be removed entirely?
- Belief – most people think their beliefs are true and justified already, so what’s left when we strip those qualifiers away? Maybe something like “acceptance” or “expectation”?
- True – this is extremely problematic, at first I wanted to say that we can’t be sure anything is true. But then I convinced myself there’s at least one true thing that we can always be sure of – that our experiences really exist. Maybe we’re in a simulation, or we’re hooked up to The Matrix, or maybe we’re being tricked by an infinitely powerful demon? But no matter what’s happening, whatever we experience we can be sure that at least the experience actually exists. In fact this is such a fundamental fact, that I almost want to say that we don’t “know” our experiences. Everything we “know” has some chance of being wrong or incorrect. So maybe a better term than “true” is “confident” or something like that?
- The definition of “justified” is having a good reason, and the definition of “reason” is a cause, explanation or justification. Of those three I think “cause” is only part that’s likely to be useful.
Getting rid of the problematic ideas or parts, and piecing together what’s left, here’s a definition that seems to work:
To Know: The expectation that a cause and effect relationship has been identified
It’s maybe a little more complicated than I would’ve liked, but let’s go through all the parts:
- Expectations – the other options to convey this idea were “confident” or “likely”, but those bring up questions about how confident/likely something needs to be to count. With expectations, we either have it or we don’t. I’ve been thinking about expectations a lot, for example, they’re critical to the definition of imagination, and I think that they’re just an automatic part of the way our brain works. In a sense they’re so common that we take them for granted and stop noticing them.
- Cause and effect – at first I was just going to say either “cause” or “effect” since one implies the other, bit I think this is a case where being explicit is worthwhile.
- Relationship – I added this because a relationship is between two or more items/concepts/etc. That, plus the idea of cause and effect implies that a chain of events would also count, if they were all dependent on each other. For example, the immediate cause of almost anything is subatomic particles bumping up against each other. But when we talk about cause and effect we’re usually talking about big groups of tiny little events that are all related.
- Identified – it’s not enough to just accept that a cause-and-effect exists, you have to be able to pick one specific one (or a related group) as the one causing the expectation
- What’s left out – truth, as I said earlier I think the only thing we can ever be sure is 100% true is that our experience exists. I also didn’t mention experience since everything we know, and also every mistake we make, comes from our experiences, so it like unnecessary extra description.
Truth is something that I’d think most people would’ve considered a required part of the definition. And maybe if we could find a better definition for truth, something that matches the way we actually use it better, it could be? But if we’re trying to include the perfect version of truth, of an ideal of being perfectly correct, I don’t think that’s useful or obtainable, so we have to remove it to make the definition of knowledge work.
Note: this definition was originally the result of a five part series of posts starting here. The above is a lightly edited version of the conclusion of the series.