You can find further information on my thoughts on my website
www.logicalparadoxes.com/index.html
Solution of Hempel’s paradox.
Concisely. An evidence e confirms a hypothesis h iff
P(h/e)>P(h). Let us consider the hypothesis h=“all ravens are black” in the most current context; the value P of its
probability calculated on the observation of some ravens, all black, is by far
higher than the value P’ calculated on the observation of some non-black
individuals, all non-ravens, though h and h’=”all non-black individuals are
non-ravens” are logically equivalent. Now (Nicod’s postulate) the new evidence
of a non-black non-raven individual (a red apple, say) actually confirms h’,
increasing P’ up to P”; but even thus increased, P”<<P, therefore, by
definition, h is not at all confirmed by the red apple, because we can already
count on its by far higher value P.
Here is the
winning trace for the definitive solution of Hempel’s paradox (in the full
respect of classical logic and of Nicod’s postulate). I have similar traces for
Goodman’s riddle, McGee’s counter-examples of Modus Ponens, Bértrand’s polyvalent geometrical probabilities, up to
the general solution of logical paradoxes; in fact all these troubles arise from
a wrong approach to the various problems.
If some reader is interested in the matter, write to
me at italo@italogandolfi.com.
The
considerations below sketch an extremely concise exposition of the solution.
Reflexivity
is a kind of indexicality. Indexical expressions depend on the context because
they are affected by some lack of information: and the context is just the
integrative source of information which converts the indexical (open)
expression into an absolute (closed)
one. On this ground a basic distinction can be focused between linguistic
conversions (Socrates was a great man; he ...) and ostensive conversions
(“yesterday”, “down there”). Ostensive conversions can be neglected for the
present purpose.
The
aforesaid lack of information is symbolizable by a free variable whose
conversion is carried out by its substitution with a constant. Of course that
conversion cannot be effective if the free variable to substitute continues to
appear in the substitutor. Here I do not dwell on the improvement that the
introduction of a reflexive variable can bring
to a formalization of the entire theme.
The
current criterion for the classification of open sentences (Skolem) is the
number of different free variables occurring
in the sentence under examination. We
are interested in monovariable sentences. Skolem criterion does not account for
the basic difference between sentences where the free variable occurs either in
the subject or in the predicate, and sentences where the same free variable
occurs both in the subject and the predicate (reflexive sentences). Logical
paradoxes arise with self-conversions. Whatever sentence resulting from a
self-comversion cannot be closed (the
conversion cannot be effective), as the free variable to substitute continues
occurring in the substitutor. Paradoxical dilemmas are defective and no
defective dilemma can admit a well-founded answer; the usual argument (if it is
were so, then it ought to be the contrary) is nothing but the intrinsically
vain attempt to fill through an arbitrary hypothesis the lack of information
created by our same definitions.
This
general solution of logical paradoxes entails very wide consequences.
SOLUTION OF GOODMAN’S RIDDLE
ABSTRACT
Goodman’s riddle arises from a
flagrant misapplication of the truth table for a basic propositional
connective. I mean that the truth of a
binary inclusive disjuntion does not at all require the truth of both disjuncts;
so, provided that an examined emerald is green, it is grue regardless of its
future color.
The point seems to me manifest; yet the vast bibliography about the riddle instils
the doubt that the same point so manifest is not. Anyhow I assume that the next
few pages will make my opinion unquestionable.
CURRENT APPROACH
A premise. The occurrence of asterisks and inverted commas as
diacritical symbols is not a
typographical carelessness, but the consequence of theoretical reasons (here left out).
DEFINITION:
(i) something is grue iff
(it is examined before t° and is green) or (it is examined after t° and is
blue)
(superfluous brackets facilitate the analysis below).
Owing to the above definition, before t° for each evidence statement
asserting that a given emerald is green, there is a parallel statement
asserting that the very emerald is grue. Thus the two inductive hypotheses *all
emeralds are green* and *all emeralds are grue* result equally confirmed. But
as soon as we survey an emerald after t°, we knock against a contradiction, for
the two hypotheses lead to incompatible results.
SOLUTION
Goodman’s example is potentially misleading because, so to say, its
institutional (then unobjectionable) oddity contributes to mask the logical mistake
affecting the argument on which the riddle stands; then I prefer to propose another
example .
Our new universe is constituted by the twenty bear cubs (briefly: twenty
cubs) of a breeding. They are classified on the basis of two oppositions: in
fact the attitude of every cub depends on the moment it is examined (either day
or night) and can be either sulky or aimable.
For the sake of concision I will speak also directly of day
examinations, of aimable cubs and so on.
Furthermore I integrate the exposition by a representation R where our
universe is visualized in a circle whose
horizontal diameter opposes day examinations (upper semicircle) to night
examinations (lower semicircle) and whose vertical diameter opposes sulky
examinations (left semicircle) to
aimable ones (right semicircle). Thus the circle is partitioned in four quadrants:
Q1 for the day-aimable, Q2 for the night-aimable, Q3 for the night-sulky and Q4
for the day-sulky examinations. Each examination is represented by a dot (which
obviously falls within the respective quadrant).
Since the first sequence of n examinations has been performed during the day
and did not give any sulky outcome it tells us that
(ii) all
cubs are day-aimable
and is represented by the corresponding sequence of n dots falling within Q1. Indeed the greater is n the more reliable is the inductive
generalization, until, for n=20, (ii)
becomes a deductive statement etcetera; anyhow here I do not dwell upon this
theme because, once the reliability of (ii), inductively or deductively
supported, is achieved, the reasoning follows identically. In fact our very
problem is: can (ii) be the basis for predicting the night-attitude of a cub? Intuition
suggests us a negative response, anyhow, in order to strengthen it let me agree
that by definition
(iii) a
cub is domestic iff he is (day-aimable or night-aimable)
and
(iv) a
cub is tempered iff he is (day-aimable or night-sulky)
(where “or” continues meaning an
inclusive disjunction). Now, can the
contradictory conclusions of an argument like
as all cubs are day-aimable, by definition they all are domestic,
therefore night-aimable, but by definition they all are tempered too, therefore
night-sulky and not night-aimable
be seriously accepted? Claiming that a given evidence enables us to
predict something means simply that the information provided by that evidence
contains the information of which the prediction is the bearer. And, until we
have no information about the night attitudes of our cubs, we cannot reasonably
infer any specific information about such attitudes. In other words: since the information that all cubs are day-aimable
is compatible with both their possible night-attitudes, it does not allow us to
infer specifically one of them. All what we can reasonably infer is just that,
at night, a day-aimable cub will be either aimable or sulky. In R, while the dots
representing day examinations fell within the upper semicircle (more precisely,
within Q1), the dots representing night
examinations will fall within the lower semicircle. A cub is domestic iff AT LEAST
ONE of his two dots (less concisely: iff AT LEAST ONE of the two dots
representing the respective day-examination and night-examination) falls within
Q1 or (inclusive) within Q2: analogously a cub is tempered iff AT LEAST ONE of
his two dots falls within Q1 or within Q3. Thus, since the first dot both of
domestic and tempered cubs falls within Q1, both conditions are already
satisfied, within any quadrant of the lower semicircle the second dot will fall. That is: owing to the inclusive import of the
disjunctions occurring in (iii) and in (iv), the day-aimableness implies the
domesticity, but the the domesticity does not at all imply the
night-aimableness (analogously: the day-aimableness implies the temperedness
but the temperedness does not at all imply the night-sulkiness). In order to
impose these latter implications we should betray Goodman’s definition,
replacing the inclusive disjunction with a conjunction, as, for instance,
(v) a
cub is capricious iff he is (day-aimable AND night-sulky)
yet it would a sterile betrayal
because, under (v), the capriciousness is sufficient to imply the night-sulkiness,
but the day-aimableness is no more sufficient to imply the capriciousness: the
conclusion is that, in any case, the information obtained by the performed day examinations
is insufficient to infer any night-attitude of a cub. So, going back to the
original example, we see that if an emerald examined before t° is green, by
definition it is already grue with full rights, therefore its being grue does
not at all exact its being blue after t°.
At this point the riddle is overcome, because the argument according to
which we should predict a blue emerald has been confuted. Anyway, for the sake
of completeness, I wish also to confute the argument according to which we should
predict a green emerald. Thus the claim is achieved that Goodman’s procedure,
far from entailing two incompatible
expectations, does not entail either.
Concisely. As soon as a definition gives a discriminating value to a
predicative opposition concerning the universe we deal with, no generalization
can disregard this opposition. Now (i), besides the discriminating chromatic
factor, gives a discriminating value to a chronological factor too (actually
the brackets in (i) show that the two disjunct predicates are not “green”/“blue”
but “green-and-examined-before-t°”/ “blue-and-examined-after-t°”). Therefore, since
the green emeralds a, b, c etc. have
been examined before t°
(vi) all emeralds EXAMINED
BEFORE t° are green.
and not
(vii) all
emeralds are green
is the right inductive generalization. The difference between (vi) and
(vii) is indeed momentous since (vi) does not provide any information about the
color AFTER t° of an emeralds exactly as (ii), where the chronological factor si
meant by “day”, does not provide any information about the night-attitude of a
cub. In other words: the chronological specification occurring in (vi) (respectively: occurring in (ii)) reduces the
domain of “all”, leaving out of quantification
every emerald examined after-t° (respectively: every cub examined at night).
Frankly there is little need to point out that in our daily practice we
all are convinced that a green-before-t° emerald is also a green-after-t°
emerald; in fact this conviction does
not involve the analysis above because, rather than being derived from the
mentioned theoretical assumptions, it is derived from collateral and contingent
peculiarities as for instance the high implausibility of the hypothesis that an
emerald changes its color at t°. Here is the reason why I think that Goodman’s
example is misleading. Would the hypothesis that a day-aimable cub changes its
attitude at night be equally implausible? It is just the plausibility of both
the incompatible alternatives about his night attitude, that is the absence of some
collateral and contingent notion privileging one of them, which shows their
theoretical unpredictability.
To summarize: a definition is not an evidence , then, regardless of any
definition we propose, on the only ground of the evidence that all the emeralds
examined before t° are green neither we can predict that an emerald examined
after t° ought to be blue nor that it ought to be green.