When are thought experiments valid…and when not?
September 13, 2017 at 5:17 am #5026
I’m a structural engineer responsible for designing a bridge across a wide body of water. The bridge will have to bear unusually heavy loads up x tons. I have a sudden idea for a new support system. To test the bridge, I first imagine the bridge and then imagine a truck weighing x crossing it. No problem. I then imagine a truck weighing twice x and again the bridge holds.
I take it that’s an example of an absurd thought experiment.
What’s the difference between this thought experiment and a good one?September 13, 2017 at 5:54 am #5027
Well, if one’s mind is as expansive as Edward Wittens’, then scientifically thinking about concepts without writing them down is probably okay.
If not like Witten, and lack built-in evidence, then thought experiments must be substantiated by keen research/evidence, as much as possible.
September 13, 2017 at 7:39 pm #5057
- This reply was modified 3 years, 5 months ago by God.
Half my cats are at war with each other. A good thought experiment enabled me to work out where to put the cranky ones so they never meet yet still have access to all the house. I worked out how to inspire each of them to go into the room designated when the time came to revolve their presence without them encountering each other, and without having to rely on my memory on a daily basis.September 15, 2017 at 1:14 am #5096
Half my cats are at war with each other. A good thought experiment enabled me to work out where to put the cranky ones so they never meet yet still have access to all the house. I worked out how to inspire each of them to go into the room designated when the time came to revolve their presence without them encountering each other, and without having to rely on my memory on a daily basis.
Half your cats? How many dozen cats do you have, you cat lady?September 15, 2017 at 1:55 am #5099
Ha Unseen! What is the minimum number of cats I must have to make my above post grammatically and logically correct?October 17, 2017 at 11:56 am #5848
Defining “valid” in the sense of “strong”.
When are thought experiments strong? They aren’t, but conclusions resulting from them might be strong enough to survive criticism, or weak enough to not survive it.
For instance, one or more conclusions of Einstein’s thought experiments are weak because he assumed without evidence that nothing travels faster than light.
Unless gravitation’s attractive force travels faster than light, the earth is orbiting not where the sun is, but where it was just over eight minutes ago. Mercury and Venus are orbiting where the sun was less than eight and a fraction minutes ago; the other planets are orbiting where the sun was more than eight and a fraction minutes ago. The farther the planet, the longer its lag time.
Which of Einstein’s conclusions from that thought experiment are valid?October 17, 2017 at 1:15 pm #5849
I would say a thought experiment is “valid” when it produces a model that appears to be correct.
A thought experiment could maybe sometimes be thought of as a hypothesis: “if the world was like [hypothesis] X, what would happen?”.
If “what happens” matches up with observable reality, then it’s good supporting evidence that the world is in fact like X.October 17, 2017 at 8:47 pm #5850
The force of gravity does not “travel” in the sense that a photon of light travels. The light we get from the Sun is 8 minutes old but that is because of the distance between the 2 masses and not because of the “pull” the Sun exerts on the Earth. The speed of light is not determined by gravity even when a massive object bends it. Photons of light have no mass. Gravity itself is not really a “force”. That is the old Newtonian way of visualizing it. Gravity is just a description of the warping of Spacetime which happens when 2 massive objects attract each other. Light can have its momentum changed by those objects which will change its frequency (wavelength) but not its speed.
You are standing in the middle of a vast flat plain. You hold a gun at arm’s length (90 degrees) in one hand and a similar bullet for that gun in your other hand, also at the same outstretched position. As you fire the gun you open your other hand and allow the bullet to fall. Which bullet hits the ground first?October 17, 2017 at 8:52 pm #5851
Strega, I think you need to invest in some cat traps.October 17, 2017 at 10:28 pm #5852
Good topic, sorry I missed it when it was new…
There are two types of thought experiment: explanatory (used to explain a concept) and exploratory (used to explore a concept).
Schoedinger’s cat is explanatory… it explains the concept of superposition. Your example is exploratory: you build a model of the bridge in your mind, then test it in your mind. The validity of such an experiment is based on how accurately the model is constructed, and how accurately the model is tested. Once the model passes the thought experiment tests, it’s about as valid as it can get and should be implemented in simulation software (or scale model), to be tested in an environment with stricter adherance to the laws of physics.
I take it that’s an example of an absurd thought experiment.
What makes you say it’s an absurd thought experiment? It builds a mental model, then tests it…. or are we to believe you (the structural engineer) isn’t aware of the physical limitations of the intended materials, or structure of the bridge? Obviously, testing the model by imagining it holding up to extreme testing is not the same as doing rough mental simulations and finding that the model holds up.
What’s the difference between this thought experiment and a good one?
Appropriate testing of the mental model.October 18, 2017 at 1:41 am #5853
I would say a thought experiment is “valid” when it produces a model that appears to be correct. A thought experiment could maybe sometimes be thought of as a hypothesis: “if the world was like [hypothesis] X, what would happen?”. If “what happens” matches up with observable reality, then it’s good supporting evidence that the world is in fact like X.
Matt has already explained well what the two most common kinds of thought experiments are. They are usually informal ways of (as Matt said) exploring or explaining formal arguments. That means a thought experiment almost always accompanies a formal argument. Nearly always a thought experiments are secondary to a more formal text or a formal critique of a formal argument (such as Schrodinger’s critique of the Copenhagen solution). They may be (subjectively) more helpful than some formal arguments but they are not valid in the sense that an argument is valid (or sound).
You can say “that was a useful example” or “I think the thought experiment was constructed well” or on the negative side “that thought experiment doesn’t help me” or “I could have constructed it way better” or “I think this thought experiment has entered cuckoo-land”. But they cannot be “valid” in the way a formal argument is. As for your definition of “appearing to be correct”, you’ve just complicated matters. What is the difference between “being correct” and “appearing to be correct”? What is the difference between a model created from empirical observation vs. a model presented as a hypothetical thought experiment? How can you validate a hypothetical model in a way that “appears” to be correct?October 18, 2017 at 3:05 am #5856
Which bullet hits the ground first?
If the bullet approaches the speed of light, time contracts… nyuck nyuck …..otherwise the same time since gravity acts the same on each bullet.October 18, 2017 at 10:15 am #5859
@davis – “What is the difference between a model created from empirical observation vs. a model presented as a hypothetical thought experiment? How can you validate a hypothetical model in a way that “appears” to be correct?”
– you could say that empirical observation, factual knowledge, and model-building go together to make a complex explanatory picture of reality. A picture is not the same as the real thing, so it’s more of a structural system of knowledge that explains how elements are related together.
It’s a good question, how does someone test a hypothesis. I think we can say we’re doing well when our hypothesis elegantly explains a variety of related information better and more completely and neatly than any other, fills in the gaps in those other similar (but mistaken) hypotheses, and accommodates new information easily.
Also, it’s very important that it fits into existing well-established knowledge, since this is the way that reality itself is put together – it all fits together logically, as Judge Judy is fond of pointing out. When the model can actually predict how something will happen, it’s good evidence that it’s a good model.October 18, 2017 at 10:16 am #5860
@robert – what if you’re facing East-West? Then, the rotation of the Earth must have some kind of effect.October 18, 2017 at 10:30 am #5861
Robert, that is the correct answer.
You must be logged in to reply to this topic.