So Occam's Razor is one philosophical principle that the average Joe Q. ScienceLover goes absolutely gaga over, so maybe I can stimulate some interesting discussion! I have a couple questions about it, which, depending on the answers of all you Joe Q. ScienceLovers, may form the basis of a problem which will result in either a) a revision of Occam's razor, b) a revision of some other common views of Joe Q. ScienceLovers.

Question #1: What does Occam mean when he talks about "the number of entities" (you know, the thing that we are not supposed to be unnecessarily multiplying). What sorts of things constitute "entities"???

Question #2: What does Occam mean by "necessary". That is, how do we adjudicate how many entities are necessary?

:science: ANSWER AWAY! :science:

Well, I had an answer somewhere, but for the moment, entities can be specific causal actions and/or governing principles, depending on what we're examining.
So if we know that a neuron is stimulated by a neurotransmitter, and that this transmitter is released, we don't need anything further to explain the stimulation of the neuron (even though many neurons are stimulated by multiple neurotransmitters). We have a cause, and to suppose anything further is unnecessary, since from the facts given, our theory expects stimulation, which is what we observe.

These definitions are relative, IMO.

Yea Philosophy 101, or maybe 50.5.

First Wiki cite!!! :king:
http://en.wikipedia.org/wiki/Occam's_Razor


Basically, entities are the components of an explanation or philosophy.

And 'least is best' in Occam's viewpoint. Meaning the least amount of description is best for arguement's sake.

For instance in formulating an explanation or hypothesis, the least complex formula is best, in Occam's view. Basically a Reductionist stance, that maximumally simplifying arguements are best.

If applied to modern science, I believe it's usually taken as the simplest hypothesis is easiest to prove/disprove.

Historically it had something to do with Nominalists at the time arguing that nature is made of everything. I think skimming some original works by Aristotle is a good example of Nominalism gone wild.

Well, I had an answer somewhere, but for the moment, entities can be specific causal actions and/or governing principles, depending on what we're examining.

So, say we have a theory, that is equal in explanatory power whether we say that it implies the existence of objects A, B, and C, or all those objects plus object D. You're saying we ought to just forget about object D?

These definitions are relative, IMO.

What's that mean!?

I dunno what Occam meant (and all my uses are post-Occam), but I do know of a couple of interesting uses of Occam's razor.

The first is in statistical learning theory, where entities are parameters of a certain type (think coefficients of a polynomial curve), it is often the case that running a least squares regression with greater number of coefficients, a method that will actually improve the curve's explination of current data will result in a curve that is less adept at predicting new data! Here, entities are parameters, and neccesity would be determined by certain statistical techniques (like cross validation)


Another one is medicine. Where entities would be causes of symptoms and neccesity would be explaning all the symptoms. If we treat all symptoms of a patient as seperate causes we could be missing a single cause which would continue to cause patients problems, so it is very important for doctors to try to see if there is an underlying cause to symptoms.

So, say we have a theory, that is equal in explanatory power whether we say that it implies the existence of objects A, B, and C, or all those objects plus object D. You're saying we ought to just forget about object D?

Well, I wouldn't recommend forgetting it, but I'd certainly suggest that if an important decision required us to decide at that moment whether D existed or not, we should say not.
If the evidence doesn't imply it, and a theory already explains the evidence, then it shouldn't be part of our accepted explanation.

Well, I wouldn't recommend forgetting it, but I'd certainly suggest that if an important decision required us to decide at that moment whether D existed or not, we should say not.
If the evidence doesn't imply it, and a theory already explains the evidence, then it shouldn't be part of our accepted explanation.

So suppose it is stipulated that D will never ever figure into the explanatory power of any theory for all time. Then do you think its safe to just let go of D?



Oh and here's what ol' Occam had to say about what's necessary:

“For nothing ought to be posited without a reason given, unless it is self-evident (literally, known through itself) or known by experience or proved by the authority of Sacred Scripture.”

So go look in your friggen Bibles for what's necessary!

Another one is medicine. Where entities would be causes of symptoms and neccesity would be explaning all the symptoms. If we treat all symptoms of a patient as seperate causes we could be missing a single cause which would continue to cause patients problems, so it is very important for doctors to try to see if there is an underlying cause to symptoms.

Is that like the whole thing they do with trying to figure out which symptoms are epiphenomenal?

epiphenomenal properties pose some problems with certain popular theories of causation!

Which sense of epiphenomenal are you using here? Medicine seems to me to have its own mangled version based on my 2 minute wikipedia/wiktionary research on the question of "WTF does 'epiphenomenal' mean?"

Which sense of epiphenomenal are you using here? Medicine seems to me to have its own mangled version based on my 2 minute wikipedia/wiktionary research on the question of "WTF does 'epiphenomenal' mean?"

Well, maybe an analogy would work:

Assuming that you're trying to figure out what causes some thunderstorm. You have two bits of data: that barometric pressure dropped right before the storm, and that the needle on your barometer dropped right before the storm.

The barometric pressure dropping is the cause, and the needle on your barometer dropping is a bit of epiphenomena, because although it is constantly correlated with storms of that sort just as much as the drop in barometric pressure, it is clear that it is the pressure drop that caused the storm and not the movement of the needle on you barometer.

So suppose it is stipulated that D will never ever figure into the explanatory power of any theory for all time. Then do you think its safe to just let go of D?


If D cannot explain anything then D is not a scientific concept and should play no role in science.
If D could be found, but for some reason won't explain any other observation then 'D exists' is a valid scientific theory to test, although a pointless one.

What's that mean!?

Well, I don't think you can really define 'entities' in a general sense.. or maybe you can, but it doesn't really matter.

All you're interested in is comparing the complexity between two theories that attempt to explain the same phenomenon. That should be relatively easy to do, even without a proper definition of 'necessary' or 'entity'.

So I guess what I'm trying to say is that you don't need such definitions.. as they might differ from theory to theory.

Occam's Razor isn't supposed to be a well-defined methodology, anyway. It's meant to be a guide, so you don't need to define everything.

Mr. Fiftyson,

I suppose then yes, such a distinction is quite important in medicine. Of course, in the case of a common cause, it could be that the phenomenon that is causing it is not a symptom in itself.

It's pretty much just common sense. "If you hear hooves think horses not zebras."

I'm not sure how that's Occam's Razor...

Please 'splain.

Well, there's not many zebras in the UK, so for the hooves to have come from zebras, the zebras must have, perhaps, escaped from a zoo, shipped over for a circus, long lost survivors of Noah's Ark, etc etc. Horses, OTOH, were fairly commonplace around the time when the phrase was invented. You have to assume fewer things to arrive at the conclusion that the hooves were horses than zebras.

I mean, these days, the whole Occam's Razor thing is more about Einstein's "as simple as possible, but no simpler than that" thing, than the original "less assumptions are better" anyway...

Are the hooves shod? If so you would also be assuming the hitherto unknown domestication of Zebras and therefore a whole team of people solving a previously insoluable difficulty. OTOH if these hooves are not shod then they are very unlikly to be horses - assuming if we may we are not in the tiny minority of the world where there are still wild horses. Can the listener to the possible hoof dicern from sound alone if the hoof is cloven? If not large deer would therefore be the most probable explination, or whatever geographic equivalent.

Occam's Razor sez that the hooves are shod.