Documente Academic
Documente Profesional
Documente Cultură
(University of Miskolc)
Preliminaries
Project: to explain how logical knowledge can be a priori.
Even though Ill provide material for arguing that logic is a
priori, I wont argue for that.
Logical knowledge consists of sentences ascribing logical
properties to arguments (valid, invalid), sentences (logically
true, contradiction), or sets of sentences (logically
equivalent, inconsistent).
We know something a priori, if its justification does not
depend on anything empirical.
The argument on slide 24 is invalid you need experience
to understand this claim, but once you understand it, you
need no further experience to justify it.
Knowledge of chess
Knowledge of chess consists of sentences ascribing
properties to moves (legal, illegal) and arrangements
(check, checmate).
Knowledge of chess is a priori.
Sentences in chess should not be construed realistically,
because the facts making these sentences true/ false are
not objective, they depend on human decisions.
Alternatively: they are objectively true and the
corresponding facts are independent of the mind only if the
decisions about rules are already assumed.
Knowledge of chess is knowledge of mind-dependent facts.
We have non-empirical knowledge of them because we
made them. It is a kind of makers knowledge.
Agenda
1.Argue that logic is an artifact, just as chess is. Deductive
logic is not extracted from lay thought, it is created.
2.Dispel intial worry about anti-realism: truth is objective, so
preservation of truth is objective, and validity is the
preservation of truth.
3.Argue for logical anti-realism. The argument in brief: there
are many systems of logic; the way we choose between
them does not indicate that there are system-independent
logical properties the various systems seek to capture.
4.Clarifications: why the artificiality thesis is necessary, why
Quines anti-conventionalist arguments do not apply, how
this account differs Boghossians.
Artifact 1.
(Subconscious cognitive mechanisms do not count.)
1. Deductive inference is extremely rare among laymen.
Lay inference is a large leap from unarticulated premises to
a distant conclusion, not a meticulous stepwise process.
Inferences can be later put into a deductive form.
2. Experimental studies of deductive reasoning show that
people often do not identify deductive tasks as deductive
ones or do not try to solve them by deductive means.
Wason selection task: which card(s) do you have to turn to
check the truth of if p is on the one side, then q is on the
other side?
p
~
p
~q
Artifact 2.
In tasks where syllogistic or conditional inferences have to
be evaluated there is an ineliminable belief bias.
No police dogs are vicious.
Some highly trained dogs are vicious.
Some police dogs are not highly trained.
10%
No addictive things are inexpensive.
Some cigarettes are inexpensive.
Some addictive things are not cigarettes.
71%
valid:
valid:
Artifact 3.
3. Deductive inference is of little use in practical situations
in which we only have partial information and ceteris
paribus generalizations. Deductive inference is monotonous
(remains valid if we add new premises), but reasoning in
practical situations is defeasible.
Evaluating candidates for the position of professor of logic.
Generalization: Anyone publishing in the best logic journals
should be shortlisted.
Information: X publishes in the best logic journals.
Conclusion: X should be shortlisted.
New information:
X was fired from his previous positions because of sexual
harassment.
Those were papers were jointly authored with his wife, a
noted logician, and X has no independent publications.
Artifact 4.
4. Logical properties presuppose logical form, logical form
presupposes logical constants. Hence, we cannot have
logical properties without logical constants. But the
meaning of logical constants is fairly different from that of
their natural language counterparts.
Misbehaving close relatives
Frau Merkel likes/does not like fuzzy logic.
If I have two heads, then I have three hands.
I ate the pizza and ordered it.
All politicians are liars, but there are exceptions.
More distant relatives of &
Time: after, before, when, while, since, until
Causal(?): so, because, since, for, now that, as, in order that
Opposition: although, even though, whereas, while, yet
Artifact 5.
5. Comparison with ethics. Morality is present in practice, it
is not invented by moral philosophers. Even among laymen,
we have:
(1)awareness of ethics ethical considerations show up in
deliberations, bad conscience, education
(2)basic ethical thought ethics distinguished from custom
and self-interest, ability to formulate rules
(3)terminology
Among laymen
(1)no awareness of logic: arguments and inferences are
evaluated, but that evaluation is global rather than
specifically logical, no attention to logical considerations, no
education
(2)no basic logical thought no distinction between
inductive and deductive argument, no distinction between
false moves and false premises, no ability to identify formal
rules
Anti-realism 1.
When you devise a system of rules for a certain task you
seek how to best satisfy the various requirements.
In choosing a logic you seek the best balance of scope,
simplicity, closeness to natural language (intuitiveness),
philosophical tenability (Resnik calls this cognitive
equilibrium.)
But we do the same in choosing between scientific theories,
we just balance different considerations (empirical
adequacy, simplicity, compatibility with other theories, ).
This does not threaten realism about science! Why should
the case of logic be different?
Anti-realism 2.
1. Why is this not a rule of logic?
For all x, y, z terms and + comparative adjectives, x+y,
y+z x+z
Pros
no exception
completely formal
instances are not justified empirically
Cons
complicates syntax (need a special class of 2-place
predicates);
unlike other rules: the logical form is created by a special
class of predicates
Would the cons entitle us to reject this rule if we were in the
business of describing mind-independent facts like in
science?
Anti-realism 3.
In science we are not allowed to sacrifice scope for
familiarity and simplicity unless special conditions obtain.
(1) The would-be law does not yield new predictions.
(2) It is reasonable to hope that a more elaborate version of
the current theory will be able to explain the facts the
would-be law describes
(1) does not apply, since logic does not seek predictions.
(2) does not apply, since there is no way to capture these
inferences by the customary rules.
If logic were meant to capture the objective property of
validity, this rule should not be rejected.
Anti-realism 4.
2. How can we resolve conflicts in logic? Consider some
(apparent?) problems with classical logic.
a. inferences involving empty names
Fa x(Fx)
Pegasus is a winged horse. So there is at least one winged
horse.
We may (C) stick to the rule and ban empty names, (M)
reanalyse names as predicates or (R) develop a free logic.
b. inferences with the material conditional
~(A B) A
Its not the case that if the greens win at the next election,
theyll attack the US. The greens will win at the next
election!
We may (C) stick to the material conditional but be careful
in translation to and from natural language or (R) develop
Anti-realism 5.
c. sorites
(antecedent) My daughter is not an adult at time t.
(conditional) If my daughter is not an adult at time t, she
wont be an adult one second after t.
Instantiating (antecedent) now and iterating modus ponens,
we may conclude that my daughter wont be an adult even
when she is 90.
We may (C) simply take care when dealing with vague
predicates, (M) adopt a metaphysical view which makes
(conditional) false at one point (epistemicism), (M) adopt a
semantic view that makes the (antecedent) not true for a
range of instantiations (supervaluation), or (R) adopt fuzzy
logic which does not support modus ponens.
d. ex contradictione quodlibet
We obviously do not want to infer to all propositions from a
contradiction. So we may (C) abstain from using logic when
Anti-realism 6.
We may choose between four types of solutions:
(Conservative) Stick to classical logic but apply with care.
(Moderate) Put the blame on metaphysical views, semantic
views, etc. and not on classical logic.
(Radical) Abandon classical logic in favour of another one.
(Opportunistic) Keep classical logic and use non-classical
logic in the problem cases.
Claim: the way we decide shows that logical properties are
not objective.
If you are an opportunist, you have abandoned realism.
Classical logic and non-classical logics are compatible only
if they are taken to make claims about system-specific
properties, but are not compatible if they are taken to
capture absolute logical properties.
Anti-realism 7.
The arguments between the various options are
inconclusive, which shows that they do not concern matters
of fact.
Irresolvable conflicts do not imply that debate is not about
objective facts. It might happen that we are just ignorant of
the facts knowledge of which would be sufficient for the
decision (e.g. debates about past events). In these conflicts,
however, it is hard to see how the case could be decided if
we knew further facts.
The disagreements concern which considerations (scope,
simplicity, intuitiveness, philosophical
considerations)should be given priority. Such
disagreements are not factual in nature.
Objection: isnt there a common core between various
systems of logic which yields an objective property of
validity?