Once, the story goes, in 4th century BC China, the philosopher Zhuang Zhou1 dreamed he was a butterfly, flitting and fluttering about, happy with himself and doing as he pleased. He didn’t know that he was Zhuang Zhou.
Suddenly he woke up, and there he was, solid and unmistakably himself. But he didn’t know if he was Zhuang Zhou who had dreamt he was a butterfly, or a butterfly who was now dreaming that he was Zhuang Zhou.
This is one of the most famous tales from the Zhuangzi, the ancient self-titled writings of Zhuang Zhou. The apparent impossibility of determining when we are dreaming was clearly profound to him. It is a shame, then, that he had to endure his confusion, not knowing about this simple trick: hold your nose, close your mouth, and try to inhale. If you can’t, then you are awake and in reality. If you can… then you are currently in a dream. 👁
When we devise some explanation or claim about the world; whether it’s about dreams and butterflies, or science, or politics; we are stating a belief about the reality that we live in. Saying the sky is blue is a statement of belief that we live in one of the realities where the sky is blue, as opposed to a reality where it is purple or green. Whether or not such other realities actually exist in a greater sense is irrelevant—before we go outside and look, we could be in any of them.
The nature of knowledge, then, is understanding more specifically which realities might be ours, and which are certainly not. It is to escape a world where “false” and “true” are indistinguishable, unnecessary concepts into one where they have meaning. Zhuang Zhou knew we were in one of two realities, but he couldn’t identify which; the terms “dreaming” and “awake” had no meaning to him, since he knew no sentence which could tell them apart. All he knew was that the answer to “do I feel awake?” would always be yes.
We, on the other hand, have a check: a claim or question whose truth depends on something specific about our reality. A check is more than just something that is true to us: it has to be false in some other reality, where things are different. Without such a check, a claim is meaningless. If you can account for any evidence, even that which is fabricated or false, you might as well be able to explain nothing. The baseline for a claim is that it is falsifiable. Whether or not a lone tree falling in a forest makes a sound is a pointless issue, because there’s no way either claim can be checked to be wrong.
Falsifiability is a requirement for a claim or belief to be taken seriously, but is it sufficient? If a fervent prophet runs up to us in the street and proclaims the sun will go out next year, then this is falsifiable: our check is “wait a year and see if the sun is still there.” However, we don’t have good reason to believe them. The check is not very practical, because we are restricted in when we can do the check and get results from it. It would be better if we or someone we trust could do some checks now that suggested whether the sun would go out. Better claims are testable: they have checks which can be performed at arbitrary times by independent people, so we don’t have to wait or trust the word of a few people who can do the test.
That said, there are some interesting examples of non-testable predictions like “we should discover X” in history. The theory of general relativity predicted black holes should exist decades before they were first observed. Similarly, when the comet orchid was discovered, a flower whose nectar is at the bottom of a 30cm tube, Charles Darwin used his theory of evolution to predict there should be a insect with a 30cm proboscis which pollinates it. Sure enough, such an insect, a type of sphinx moth, was discovered 21 years later. If after lots of searching none of these things were found, we’d have reason to doubt the theories involved.
Finally, really great claims are ones whose checks correspond with all of its details.
The way the Ancient Greeks explained the seasons was that Hades, god of the underworld, abducts Persephone and tricks her into an agreement where she must spend 6 months of the year with him in the underworld. The despair of her mother Demeter, goddess of the harvest and agriculture, at losing her daughter causes the vegetation of the world to wither, causing the autumn and winter seasons. When Persephone’s time is over and she is reunited with her mother, Demeter’s joyousness causes the vegetation of the earth to bloom and blossom, which causes the spring and summer seasons.
The problem with this explanation is that there are so many details which are entirely irrelevant to temperature or crop yields. Maybe Persephone actually likes Hades, and flees her over-protective mother for half a year. Or maybe it’s Persephone, not Demeter, who represents agriculture, and her powers don’t work when she’s in the underworld.
Our current explanation for seasons is that the Earth is slightly tilted, and as it travels around the Sun, our part of the planet periodically tilts towards and away from it. Every detail of this plays a part in evidence we are trying to explain, and so we can’t alter any part of it without affecting what we’d expect to see. Really good claims are difficult to vary, because every change we can make while still explaining the same thing is another possible reality that the explanation doesn’t rule out.
The importance of meaningful checks, and these three aspects of falsifiability, testability, and difficulty in varying are some of the closest things we have to a formula for escaping the false realities that we often find ourselves in.
- My other post How do we really know what’s true?
- A New Way to Explain Explanation - David Deutsch
- Making Beliefs Pay Rent (In Anticipated Experiences) - Eliezer Yudkowsky
- Karl Popper - Stanford Encyclopedia of Philosophy
- Zhuangzi translation by Nina Correa
P.S. I am oversimplifying the dream argument, since we might argue that we don’t know the reality where the holding-nose check fails is the waking one, just that we can now tell the difference between the two. I didn’t go into this problem because telling the difference is enough when empirical experience is involved.
Said “juang jo.” ↩