Overcoming Quine's objection to the analytic / synthetic distinction

Known unknowns and unknown unknowns!

Moderators: AMod, iMod

Skepdick
Posts: 5027
Joined: Fri Jun 14, 2019 11:16 am

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by Skepdick »

PeteOlcott wrote: Wed Apr 08, 2020 5:58 pm Comprehension means successful queries.
OK, can you encode the notion of a {QUERY} in your system?

What language would you use for querying this thing?

PeteOlcott wrote: Wed Apr 08, 2020 5:58 pm I am not focusing on the rat's nest of complexity
of translating between natural language and the internal structure, I am only focusing on
the internal structure.
But you are deriving the internal structure from the English meaning, are you not?
PeteOlcott
Posts: 970
Joined: Mon Jul 25, 2016 6:55 pm

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by PeteOlcott »

Skepdick wrote: Wed Apr 08, 2020 6:30 pm
PeteOlcott wrote: Wed Apr 08, 2020 5:58 pm Comprehension means successful queries.
OK, can you encode the notion of a {QUERY} in your system?

What language would you use for querying this thing?

PeteOlcott wrote: Wed Apr 08, 2020 5:58 pm I am not focusing on the rat's nest of complexity
of translating between natural language and the internal structure, I am only focusing on
the internal structure.
But you are deriving the internal structure from the English meaning, are you not?
I updated the root post of this sequence possibly totally
refuting all of Quine's objections with three lines of Prolog.
Skepdick
Posts: 5027
Joined: Fri Jun 14, 2019 11:16 am

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by Skepdick »

PeteOlcott wrote: Wed Apr 08, 2020 6:43 pm I updated the root post of this sequence possibly totally
refuting all of Quine's objections with three lines of Prolog.
*sigh*

You haven't refuted Quine. You have affirmed Quine.

You haven't defined the meaning of {MARRIED} anywhere.
If you go and define a married() predicate, you will discover that

for all X, if bachelor(X) is false, then married(X) is true.

bachelor(X) <-> not Married(X)

But you already said that. Here:

Code: Select all

bachelor(X) :- \+ marital_status(X, married).
Skepdick
Posts: 5027
Joined: Fri Jun 14, 2019 11:16 am

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by Skepdick »

The crux of your problem is that you don't understand what logic is, and why logic is (inevitably) a reductionist endeavour, whereas the notion of meaning (as humans use it) is holistic. Logic is strict - natural meaning in natural language is not.

https://www.metalevel.at/prolog/logic

And this is a quote I came across recently

"Cyc failed to understand a story about a person named Fred shaving in the morning... Its inference engine detected an inconsistency in the story: it knew people do not have electrical parts, but because Fred was holding an electric razor, it believed the entity 'FredWhileShaving' contained electrical parts. It therefore asked whether Fred was still a person while shaving. ("Deep Learning" by Goodfellow, Bengio and Courville)

From: https://bit.ly/2y4w00Q
PeteOlcott
Posts: 970
Joined: Mon Jul 25, 2016 6:55 pm

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by PeteOlcott »

Skepdick wrote: Thu Apr 09, 2020 10:02 am
PeteOlcott wrote: Wed Apr 08, 2020 6:43 pm I updated the root post of this sequence possibly totally
refuting all of Quine's objections with three lines of Prolog.
*sigh*

You haven't refuted Quine. You have affirmed Quine.

You haven't defined the meaning of {MARRIED} anywhere.
If you go and define a married() predicate, you will discover that

for all X, if bachelor(X) is false, then married(X) is true.

bachelor(X) <-> not Married(X)

But you already said that. Here:

Code: Select all

bachelor(X) :- \+ marital_status(X, married).
http://www.liarparadox.org/Meaning_Post ... p_1952.pdf

The way that meaning postulates work is that each one of them defines one atomic
unit of meaning (a single relation) between two otherwise meaningless finite strings.

We could add meaning postulates to {Married} up to the point where the system
would know divorce law of the state of California well enough to win a divorce case.
had_a(wedding, X) :- marital_status(X, married).
has_a(spouse, X) :- marital_status(X, married).

Quine goes on and on using some variation of the word synonymous 93 times claiming
that the synonymy between bachelor and unmarried cannot be defined. Since I did
define the synonymy between the otherwise totally meaningless finite strings: "bachelor"
and "unmarried", I proved Quine wrong.
Last edited by PeteOlcott on Thu Apr 09, 2020 6:23 pm, edited 1 time in total.
PeteOlcott
Posts: 970
Joined: Mon Jul 25, 2016 6:55 pm

Re: Overcoming Quine's objection to the analytic / synthetic distinction

Post by PeteOlcott »

Skepdick wrote: Thu Apr 09, 2020 10:23 am The crux of your problem is that you don't understand what logic is, and why logic is (inevitably) a reductionist endeavour, whereas the notion of meaning (as humans use it) is holistic. Logic is strict - natural meaning in natural language is not.

https://www.metalevel.at/prolog/logic

And this is a quote I came across recently

"Cyc failed to understand a story about a person named Fred shaving in the morning... Its inference engine detected an inconsistency in the story: it knew people do not have electrical parts, but because Fred was holding an electric razor, it believed the entity 'FredWhileShaving' contained electrical parts. It therefore asked whether Fred was still a person while shaving. ("Deep Learning" by Goodfellow, Bengio and Courville)

From: https://bit.ly/2y4w00Q
All that shows is that you need more meaning postulates. A human degree of comprehension
of any subject (or all subjects) can be composed as soon as enough meaning postulates are specified.

That story about Fred was at the very early Cyc system back in 1992. Cyc has only been touted as
commercial ready in the last four years.

I am only proposing to use two aspects of Prolog:
(1) Its syntax (may need augmentation)
(2) It model. Provable(X) means True(X), ~Provable(X) means ~True(X) and Provable(~X) means False(X).

It took the Cyc project 700 labor years to manually construct meaning postulates representing far
less than one quadrillionth of one percent of all human knowledge.
Post Reply