What is knowledge?

It might not be what you think it is

We tend to understand instinctively what is meant by knowledge, even though we might not be able define it. 

The Oxford English Dictionary defines it as:

the understanding and information one gains through education or experience, encompassing both theoretical understanding and practical skills.

This seems fair enough and tends to match our experiences of gaining knowledge. There are, of course, caveats, such as can we have knowledge of something but not understand it? Or know something that later turns out to be false?

Knowledge from a philosophical perspective

Philosophy generally posits that knowledge is ‘justified true belief.’ I know something if I believe it, that it’s true, and that I have good reason for believing it (that my belief is justified). This then presents us with several other questions, such as what counts as justification, as well as whether we can ever be certain that something is true. How, then, should we respond to scepticism? How are we capable of questioning knowledge if knowledge is self-evident? 

Philosophical discussions of knowledge are normative, that is, they relate to how we ought to think, not necessarily how we actually think.

Knowledge from a Cognitive Psychology perspective

Cognitive psychology explains knowledge it term of how information is represented and used. Knowledge then becomes: ‘all the information in one’s memory’ (McCarthy & McNamara, 2021), or as, ‘an individual’s personal stock of information, skills, experiences, beliefs, and memories’ (Alexander et al., 1991). 

Knowledge, therefore, is seen as mental representations, such as schemas, semantic networks, or procedural rules. Cognitive psychology and the broader cognitive sciences (including cognitive neuroscience) investigate how knowledge is acquired, stored, processed, and used in the brain or in artificial systems. This includes declarative (semantic and episodic) and procedural (acquired skills). Knowledge we can test empirically.

While philosophy looks at how we ought to think, cognitive science investigates how we actually think.

This means that mental representations can be justified but wrong, perhaps because of faulty assumptions and coincidences. This challenges the justified true belief view - knowledge can be wrong. 

We may, then, know that Canberra is the capital of Australia, that goldfish have a memory span of around 3 seconds, or that the Great Wall of China is visible from space. That the capital of Australia in Canberra might come as a surprise for some people because there’s a tendency for non-Australians to believe it’s Sydney (after all, who goes to Australia and visits Canberra?).

We might ‘know’ that goldfish have a 3 second memory span - only they don’t. They have pretty good memories and can be trained to perform simple tricks. What you might have ‘known’ about the memory span of a Goldfish is wrong. Also, the Great Wall of China isn’t visible from space - just ask an astronaut. 

Cognitive science doesn’t necessarily care if the information contained in long-term memory is wrong because that’s not really the point. However, this doesn’t stop cognitive scientists from studying why we believe something that isn’t true (the Earth is flat, the moon landings were faked), or false memories (events we believe to have happened but didn’t).

Being right for the wrong reasons

Philosopher Edmund Gettier suggested that justified true belief isn’t enough to constitute knowledge - there must be something else. We can illustrate this through what’s become known as a Gettier problem.

Say I’m waiting for a train and there is a large clock on the platform. I check the clock and it reads 3pm. So far, so good. But what I don’t know is that the clock in broken - it’s been saying 3pm for days, if not weeks. But here’s the punchline: it really is 3pm and the fact that the clock is accurate is simply a coincidence. I believe it’s 3pm and I have good reason to. I’m right, but for the wrong reasons.

Similarly, I might know that the United States entered the Second World War in 1941. But my belief might be that this resulted from them being attacked by Germany. My conclusion is correct (The US entered the war in 1941), but my belief pertaining to why is wrong - it was Japan who attacked the US, not Germany.

Gettier would argue that this doesn’t represent knowledge, because it’s not grounded in evidence - it fails the criteria for justified true belief.

Cognitive science, therefore, assumes that knowledge can be wrong, but that these errors can be corrected in the light of new evidence. This is the major premise of Bayesian models of cognition. Bayesian models attempt to explain how the mind works using probability - Bayes’ Theorem is a mathematical rule for figuring out how likely something is, given what we already know and what new evidence we have discovered. This also explains Gettier-type problems, in that the correct belief can be the result of a low-probability event - when predictions match outcomes by accident, the brain reinforces a false belief.

Our minds aren’t perfect logic machines, but they are good at guessing. What we know and how we learn is influenced by what we already know and what we have learned. We use what we already know to make sense of the world, so imagination and prediction are important to how we learn and think.

Cognition can, therefore, be viewed as our prior beliefs plus any new evidence we encounter, resulting in updated beliefs. This can be seen in perception, for example, recognising objects in poor lighting; language, such as predicting what someone is about to say next; and learning, adjusting our beliefs in response to new learning.

The brain doesn’t wait for the information before making a guess based on what is likely. Often the guess is wrong and based on cognitive faculties such as object recognition (and is why we often see faces in slices of toast). Our brains make pre-emptive guesses based on limited information.

We use schemas to generate these guesses, so schema theory fits well with Bayesian models of cognition. Our prior beliefs about something are stored in existing schemas, but when new evidence comes to light, this results in schema adjustment. In terms of knowledge, you might have a belief that goldfish have a 3 second memory span, but when presented with evidence to the contrary, you will (hopefully) consolidate this new information into a pre-stored schema. This isn’t always the case, of course, and several cognitive biases (such as confirmation bias) may prevent full integration.