File spoon-archives/bhaskar.archive/bhaskar_1997/bhaskar.9710, message 3


From: "Marshall Feldman" <marsh-AT-URIACC.URI.EDU>
To: <bhaskar-AT-jefferson.village.Virginia.EDU>
Subject: Re: BHA: Re: Foundations of Probability
Date: Thu, 2 Oct 1997 05:57:24 -0400


 Louis,

I think the two -- statistical and physical properties -- are equivalent.
The statistical  properties are caused by the physical, and we know (but
usually do not mention) the causal mechanisms by which systems with certain
physical properties attain
certain statistical characteristics.

Perhaps using urn models will be a bit less misleading than the fair coin
(since
it possibly could land on its side). E.g., we have an urn with 50 white
balls and 50 black ones, what's the probability of pulling out a white one?

Probability is a MEASUREMENT. As such, it's no different than weight, mass,
valence, or any other measurement. It's a theoretical construct designed to
measure some characteristic of a system. Just as a square with area 4 must
have sides equal to 2, an urn with 50 B & 50 W balls must have probability
.5 of choosing a B or W (providing the chooser does not intentionally try to
pick one kind of ball). We give measurements ontological standing just as we
give any other theoretical construct ontological standing: through argument
involving empirical corroboration (see Fig. 0.1 RTS). I don't see
probability as posing any unique problems here.

As for the future, probability is a measurement of certainty. As such, we
can use it to characterize our knowledge of the past ("there's a 50% chance
I left my sweater at your house"). We can also use it to characterize the
past ("at the rate the plague was spreading, there was a 70% that towns in
southern Germany would lose their entire population"). We can use it to
characterize the future ("the chances of getting
the grant are 25%"). We can use it as an ontological property (as in quantum
mechanics) or as a tool for corroboration (as in inferential statistics).
But in all cases, it's just  measurement based on a reasoned account of the
system under discussion. Seen this way, we readily see that much of the
debate tends to conflate open and closed systems (e.g. the reason the
"future" sometimes seems ontologically different than past is that we are
tacitly assuming an open system coupled with a set of actual, and perhaps
observed empirical, events) or really a special case of the more general
issue of empirical corroboration and its validity.

-----Original Message-----
From: Louis Irwin <lirwin1-AT-ix.netcom.com>
To: bhaskar-AT-jefferson.village.Virginia.EDU
<bhaskar-AT-jefferson.village.Virginia.EDU>
Date: Wednesday, October 01, 1997 5:30 PM
Subject: Re: BHA: Re: Foundations of Probability



>Carroll,
>
>I think you are missing an important aspect of probability as it relates to
>the future.  Your example refers to things that have already happened and
>are determinate (at least if we are realists).  But probability statements
>often, and perhaps usually, refer to the future, and we cannot extrapolate
>the view that you express unless we take the future as closed.  What is the
>probability that that the die I am about to throw will come up heads?  If
>we do not take the future as determinate, your view prevents us from giving
>any answer at all.  In fact, even if we do take the future as determinate,
>if no one knows the future then your view still prevents us from giving any
>kind of answer.
>
>Don't object that you do not know enough about my coin to make a
>probability prediction.  It's true you do not; maybe my coin is two-headed
>or not 'fair' (tends to land heads more often than not).  But now you are
>getting onto the messy terrain of the foundations of probability.  For
>example, why is my coin fair (if it is) - because it has certain regular
>physical features, or because statistically it lands heads half the time?
>If the latter sounds circular (a coin that lands heads half the time will
>land heads half the time), the former is a bi less obviusly so (a fair coin
>is one which is flip-symmetric; flip-symmetric is relevant because coins
>with that property tend to land heads half the time).  I'm not getting into
>this, so get your own coin.  I only want to make the point that probability
>predictions do seem to reflect properties of the world, and not just our
>knowledge.
>
>Louis Irwin
>
>P.S. Oh - the flip came up heads.  You owe me a hundred bucks! :)
>
>At 10:59 AM 9/30/97 -0500, you wrote:
>>Folks,
>>
>> I don't get this problem. What (for you reading this) is the
>>probability of [put in your own guess] resting on my desk under the
>>front edge of my monitor?
>>
>> I would guess that whatever you used to fill in the [], the
>>probability of it being the case would be a decimal followed by
>>at least several billion zeroes. But for me the "probability" is
>>1.0 that two slightly used pencils and a mostly eaten sugar cookie
>>are resting on the back of *Achilles in Vietnam*, the cookie covering
>>about 3/4 of each of the lines of the first jacket blurb, the pencils
>>covering the blank line between the 2d and 3d blurbs and the first
>>two lines of the 3d blurb.
>>
>> The probability of what is is 1.0. The probability of what
>>is not is 0.0. Probability belongs strictly in the realm of knowledge
>>of the unknown, but does not affect the content of the unknown.
>>
>> What?
>>
>>Carrol
>>>
>>>  Hans,
>>>
>>> I do not know of any formal theories, but when I learned probablity it
was
>>> taught as a formal branch of mathematics. As such, emphasis was placed
on
>>> its status as an axiomatic-deductive system. Now the question arises as
to
>>> how such systems can apply to external reality, particularly varied
>>> ontological realms (quantum mechanics, evolution, economics, etc.).
>>>
>>> I think a CR answer would stress the transitive-intransitive distinction
>>> and, borrowing a bit from pragmatism, treat probability theory as an
>>> instrument. Following Fig. 0.1 on p. 15 of RTS, "empirical testing"
>>> eventually warrants treating probabilistic mechanisms as real. This last
>>> move is, I think, the weakest in RTS. It's especially problematic here
>>> because probability comes in as part of the empirical testing as well as
in
>>> the probabilistic (stochastic) phenomenon we're claiming is real.
>>>
>>> Please share your thoughts and others' comments, as this issue is very
>>> important and interesting.
>>>
>>> -----Original Message-----
>>> From: Hans Ehrbar <ehrbar-AT-marx.econ.utah.edu>
>>> To: bhaskar-AT-jefferson.village.Virginia.EDU
>>> <bhaskar-AT-jefferson.village.Virginia.EDU>
>>> Date: Tuesday, September 30, 1997 8:59 AM
>>> Subject: BHA: Foundations of Probability
>>>
>>>
>>>
>>> >
>>> >Do you know of any treatments of the foundations of probability
>>> >which do not commit the obvious errors of epistemic fallacy
>>> >(such as many subjective theories of probability) or
>>> >a flat empirical realism (such as those theories which try
>>> >to reduce probability to observed frequencies) or regularity
>>> >determinism (such as those who define probability by the
>>> >absence of certainty)?
>>> >
>>> >Hans Ehrbar.
>>> >
>>> >
>>> >     --- from list bhaskar-AT-lists.village.virginia.edu ---
>>> >
>>>
>>>
>>>
>>>      --- from list bhaskar-AT-lists.village.virginia.edu ---
>>>
>>
>>
>>
>>     --- from list bhaskar-AT-lists.village.virginia.edu ---
>>
>>
>
>
>     --- from list bhaskar-AT-lists.village.virginia.edu ---



     --- from list bhaskar-AT-lists.village.virginia.edu ---

   

Driftline Main Page

 

Display software: ArchTracker © Malgosia Askanas, 2000-2005