NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/sci/philosop/tech/4120

Text File | 1992-11-17 | 6.9 KB | 133 lines

Newsgroups: sci.philosophy.tech
Path: sparky!uunet!secapl!Cookie!frank
From: [e-mail] (Frank Adams)
Subject: Re: Hypotheses (was: Re: Assumptions vs. assertions)
Message-ID: <[e-mail]>
Date: Tue, 17 Nov 1992 00:44:04 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Organization: Security APL, Inc.
Lines: 122

In article <[e-mail]> [e-mail] writes:
>In article <[e-mail]>, [e-mail] (Frank Adams) writes:
>> In article <[e-mail]> [e-mail] writes:
>>>In article <[e-mail]>, [e-mail] (Frank Adams) writes:
>>>> It has since that discussion occurred to me that simultaneous interpretation
>>>> on multiple levels is not strictly necessary, and may in fact be false (that
>>>> is, very unusual) psychologically.  It suffices that when the truth of
>>>> statement is falsified at the accepted level of precision, there is a
>>>> fallback to a less precise interpretation; and that this fallback is
>>>> understood by both speaker and listener.  This, I hope you will agree, is
>>>> fairly common.
>>>
>>>We are in *partial* agreement here.  It is assumed in pragmatics that a
>>>hearer normally takes an utterance to be meaningful and tries
>>>various strategies in order to find an interpretation.  However, in
>>>contrast to your formulation, I would claim along with other
>>>pragmaticists that the *speaker* knows what level of precision he
>>>intends to convey at the time of utterance;
>>
>> We are in complete agreement to this point.
>>
>>> when you say "this
>>>fallback is understood by both speaker and hearer" I feel you are
>>>describing something special that happens or is supposed to happen
>>>in scientific discourse, specifically, when a scientific hypothesis is
>>>stated.
>>
>> I am talking about a different phenomenon.  What I am saying is that,
>> assuming that the speaker and hearer have reached the same interpretation of
>> the original statement, when that interpretation is later falsified, they do
>> not thereafter always discard the statement entirely.
>
>        OK, but there has to be a way for the hearer to know
>        that this is what is supposed to happen.  I assume
>        it is due to the institutionalization of the
>        procedure;  both participants know the special rules
>        that are in effect.  No problem here, then.

I don't think institutionalization is necessary at this point, though it can
help.

>> Instead, they change
>> the level of precision at which the statement is interpreted.  Neither need
>> have considered this alternative interpretation at the time the statement
>> was made, but they will fall back to approximately the same interpretation.
>>
>> I think this occurs with at least two kinds of statements: hyphotheses and
>> observations.  Not only scientific hypotheses have this kind of property; the
>> hypotheses of a detective investigating a crime will behave the same way.

And, I might have added, hypotheses encountered in ordinary life.  "Jim
isn't here.  He must be waiting for me in front of the station."  The second
of these sentences is a hypothesis; a fallback to "... inside the station."
or "... near the station." would be normal if it was falsified.  Failure to
adopt the same fallback is a real possibility, of course.

>          The approach seems to presuppose that all hypotheses are
>          formulated in such a way as to admit of degrees of precision.
>          But are there not hypotheses that do not seem to admit of
>          degrees of precision, so that upon finding them falsified,
>          there could be no fallback?  For example "Air has weight."
>          Suppose this were falsified (i.e., suppose enough evidence
>          accumulated for us to declare it false); there would seem to
>          be no less precise way of interpreting it to fall back on;
>          either air has weight or it doesn't.  (Again I know I'm being
>          conservative here.)

(There is a real problem here, which you glossed over: "air has weight" is
not falsifiable.)

However, insofar as we *do* reject it, there appears to be no fallback.

>          Moreover the greater the explicit precision with which a
>          hypothesis is stated, the less susceptible it would seem to be
>          to fallback.  For example if I hypothesize the value of
>          something to five decimal places, that's even more of a yes-
>          or-no proposition than "Air has weight."  *Yes* of course we
>          can revise the figure later, but that constitutes rejection,
>          not fine-tuning, of the original hypothesis *as stated*.

Perhaps so; but that isn't the way it really works.  A calculation of the
value of something to five places is based on a theory of how that something
works/is consistituted/etc.  In a real sense, it isn't a hypothesis, but a
conclusion drawn from a hypothesis.  A slightly incorrect result suggests
that the hypothesis is approximately right, but fails to take some
relatively minor factor into account.

>          I may be naive about scientific method, but I see a conflict
>          here between two principles, both having to do with the
>          statement of hypotheses:
>
>          Principle A.  State the strongest, i.e., most precise and
>          therefore most easily falsifiable hypothesis possible (so that
>          the search for truth is advanced);
>
>          Principle B.  State the weakest, i.e., least precise and
>          therefore least easily falsifiable hypothesis possible (so
>          that it can be fine-tuned via fallback rather than be rejected
>          outright).

I think the equation of strongest with most precise is in error here.  The
strongest hypothesis is the one which has the fewest "free variables"; the
one which explains the most.  Hypotheses are, for the most part,
essentially qualitative, not quantitative, so "precision" is not directly
applicable to them.

Stating a hypothesis which has a free parameter with a very specific
arbitrarily chosen value for that parameter doesn't strengthen it; it just
constitutes a statement of the hypothesis with an arbitrary value for the
parameter.  The strength of the hypothesis is in the hypothesis, not in its
statement.

>          I have a feeling though that what you're really after is
>          a sort of linguistically-oriented version of the simple notion
>          that we don't abandon a good hypothesis (or, especially, an only
>          hypothesis) until we're absolutely up to our ears in counter-
>          evidence.  But that ought to be stateable without bringing in
>          the notions of precision and fallback to less precision.
>      If we agree that all this is institutionalized anyway, then
>      we're already allowing for special rules.