Académique Documents
Professionnel Documents
Culture Documents
TYPE
EVALUATION
1. Deductive:
o valid/invalid (no middle ground)
o T premise, false conc are the only actual truth values that determine validity,
others don't
o Sound/unsound
o To be unsound the false premise(s) must actually be needed to support conc.
(False superfluous premises are no problem)
o Sound arg with necessarily true conc doesn't require premises to be related
2. Inductive:
o Strong/weak
o Based on uniformity of nature (future tends to replicate the past/regularities
prevail accross spatial regions
o Assuming truthfulness of premises, see if conclusion probably true by linking up
premises with regularities in experiential background AND probability of
conclusion must be based on the assumption that premises are true (if probably
true independently of premises then its weak)
o Allows for degrees of strength. Strong must he more probable than improbable
(more than 50% probability of truthfulness of conclusion) adding extra premises
may weaken or strengthen argument
o True premises and probably false conclusion are the only actual truth values that
determine strength UNLESS total evidence requirement (TER) not met.
o TER: premises must not exclude or overlook some crucial piece of evidence that
undermines the stated premises and requires a different conclusion
o Cogent is strong, all true premises and total evidence requirement met i.e. total
evidence taken into account for actual truth of premises and probability of
conclusion
**superfluous false premise wont make an originally cogent argument uncogent
COUNTER EXAMPLE
- Argument form is an arrangement of letters and words such that the uniform substitution
of words and phrases in place of the letters results in an argument
- Any substitution instance of a valid argument form is a valid argument
- Substitution instance of invalid form is an invalid argument IF it is not also a substitution
instance of a valid form or conclusion is a tautology
- Counter example method consists of isolating the form of an argument which it is relying
on and constructing a substitution instance that has true premises and a false conclusion
(useful to start with conc)
- Identify premises and conc, form words and content words, replace content words with
letter
- Only useful for deductive arguments
- For categorical syllogisms remember: cats, dogs, mammals, fish, animals
- For hypothetical syllogisms: better if substitution for conditional premise expresses
necessary relation. For conditional conclusion use true antecedent and false consequent
LANGUAGE
INFORMAL FALLACY
- Relevance: Appeal to force, pity, people (fear, bandwagon, vanity, snobbery, tradition),
accident strawman, red herring, ad hominem (direct, circumstantial, tu quoque), missing
the point
- Weak induction: hasty Generalization (converse accident), unqualified authority, false
cause (post hoc ergo proper hoc, non causa pro causa, multiple causes, gamblers fallacy),
weak analogy, appeal to ignorance, slippery slope
- Presumption (insufficient support): begging the question (shaky premise missing, shaky
premise, restated in conclusion, circular reasoning. For the latter two, conclusion is
usually presented first. Illusion of support is necessary to count as fallacy and not
redundancy. See if arguer trying to hide something) complex question, false dichotomy
(disjunction of what you want + some unacceptable proposition presented as only
alternative. Reject unacceptable proposition), suppressed evidence
- Ambiguity: equivocation (meaning of word/phrase), amphiboly (syntactical ambiguity)
- Illicit transference: composition (conclusion depends on erroneous transference of
attribute from parts of something onto the whole), division (opposite of composition).
These differ from generalization and accident based on collective vs distributive
predication. Also for these, you need general knowledge of situation and nature of
attribute
CATEGORICAL PROPS
- Relates two categories (part or all of the class denoted by subject term is included or
excluded from the class denoted by predicate term)
- Standard categorical form (quantifier, subject, copula, predicate)
- Quality, quantity for propositions. Distribution for terms
- Existential import (all or some of the things denoted by subject term of universal
categorical prop actually exist).
o Boole: Venn, modern square, immediate inferences, boolean existential fallacy.
o Aristotle: traditional square. Conditionally valid immediate inferences: Contrary
(A and E, both can't be true but both can be false), Subcontrary (I and O, both
can't be false but both can be true), Subalteration (truth goes down, falsity goes
up). Existential fallacy: when contrary, sub contrary, subalteration used for
nonexistent things (different from "illicits" which is when conditions are present
but inference is wrongly applied). Venn diagram (subject and predicate term must
be same as those of premise. Use circled x for universal premise, not conclusion)
- Logically equivalent (same truth value and meaning) and unconditionally valid:
Conversion (Switch subject term with predicate term/ EI), Obversion (switch quality,
not quantity and predicate term complement), contraposition (switch subject and
predicate, and complement both/ AO)
- Unconditionally valid: Contradictory (diagonals)
- Not all cases of universal to particular fallacies are existential eg illicit contradictory
(wrong application of contradictory)
TRANSLATING TO CATEGORICAL
CATEGORICAL SYLLOGISMS
NATURAL DEDUCTION
- PHILOSOPHY
o LNC: -(P and -P), LEM: (P or -P), identity: P = P
o Tautological invalidity and Q invalidity imply actual invalidity iff a valuation that gives it
true premises and a false conclusion corresponds to a real world situation
o In both propositional and predicate logic interpretations, it is up to us to determine under
what conditions atoms say something true and it is up to the world whether the relevant
conditions actually obtain. We ignore arguments with mixed interpretations (neither
definitely true nor definitely false), incomplete propositions (doesn't denote anyone), liar
sentences and certain interpretations of mathematical truths.
o Don't confuse valuations (assignment of truth value)) with interpretations (conditions
under which statements are true)
o QL= syntax, semantics, existential import, countermodels
o Existential quantification isn't true if and only if one of its substitution instances is true as
there may not be enough named things. Universal quantification isn't true if and only if
all of its substitution instances are true as some unnamed things may not have the
property
o Extensional/intensional contexts
o Leibniz’s law and co referential designators
o Theory of descriptions (Russellians and anti Russellians)
o Existence claims: QL= can do it for predicative experssions like Fs do/don't exist,
definite descriptions (translate to “there is one and only one A” or “it is not the case that
there is one and only one A). QL= can't do it for existence claims involving names. Eg Ex
x=x is always true (problem for real names as we didn't imply necessary existence) and
it's negation always false (problem with empty names as they don't refer to anything).
This is because QL= assigns objects to all names.
- PRED LOGIC
o Nothing is B: -EvBv / Av-Bv
o All B are C: Av(Bv⊃Cv) [needs existential import of A for "some B are C" to be true]
o No B is C: Av(Bv⊃-Cv)/-Ev(Bv∧Cv)
o Some B are C: Ev(Bv∧Cv)
o Some B are not C: Ev(Bv∧-Cv)
o Not all ...: -Ax
o Equivalence between
-Ax; E-x
-Ex; A-x
-Ex-Fx; AxFx
-Ax-Fx; ExFx
Ax(Fx ∧ Gx); (AxFx ∧ AxGx)
Ex(Fx ∨ Gx); (ExFx ∨ ExGx)
(AxFx ∨ AxGx) q entails but isn't equivalent to Ax(Fx ∨ Gx)
ExAy(Dxy) isn't logically equivalent to but entails AxEy(Dxy)
o For presence of variable in just A or just B universal or existential quantification can be
extracted from both "ands" and "ors", to give equivalent wffs
o For conditionals, above rule applies if variable only in consequent, but if only in
antecedent, a universal or existential quantification in antecedent extracts into the other.
Extracting any type of quantification doesn't work if variable in both antecedent and
consequent EXCEPT: Ex(Fx ⊃ Gx); (AxFx ⊃ ExGx). (These switches are all equivalent
btw)
- TRANSLATIONS (see chap 29 for practice)
o Identify DoD, set predicates. If all objects in the argument have same property then you
could make things easier by restricting DoD to them.
o When translating, don't forget assertions about names, whether an object actually exists,
and proper translation of "ands" in english.
o Consider:
▪ All A are B
▪ All A are every/”any” B
▪ All A are any/some B
▪ No A are B
▪ No A are every/“any” B
▪ No A are any/some B
▪ Some A are B
▪ Some A are Every/”any” B
▪ Some A are some/any B
▪ Some A are not B
▪ Some A are not every/”any” B
▪ Some A are not any/some B
o No A (relation) every B (or emphasis "any" B)
¬Ex(Gx ∧ Ay(Fy ⊃ Lxy))
or Ax(Gx ⊃ ¬Ay(Fy ⊃ Lxy))
o No A (relation) any B
¬Ex(Gx ∧ Ey(Fy ∧ Lxy))
or Ax(Gx ⊃ ¬Ey(Fy ∧ Lxy))
or Ax(Fx ⊃ Ay(Hy ⊃ ¬Rxy)) (from chap 29: no philosophy student admires any rotten
lecturer)
o For "all A relation any B" take into consideration if B exists or not (chap 29: all wise
people admire any good philosopher example)
o If everyone loves Nerys then Owen does. (Or Anyone loves Nerys, and if anyone loves
Nerys, then Owen does) (AxLxn ⊃ Lon)
o If anyone loves Nerys then Owen does (ExLxn ⊃ Lon)
- IDENTITY
o Qualitative and strict numerical identity are both transitive, symmetric and reflexive
relations, i.e. are both equivalence relations. But strict numerical identity is the ‘smallest’
equivalence relation – it relates an object to nothing other than itself
o Ref: Ax x=x OR ∀xRxx OR ∀x∃yRxy
Sym: AxAy(Rxy ⊃ Ryx) OR AxAy(x=y ⊃ y=x)
Tran: AxAyAz((x=y ∧ y=z) ⊃ x = z) OR AxAyAz((Rxy ∧ Ryz) ⊃ Rxz)
Euclidean: AxAyAz((Rxy ∧ Rxz) ⊃ Ryz)
You can instantiate multiple universal quantifiers with the same names in a wff. (Useful
to close tree if there is a reflexive relation wff present anywhere)
o Only, the only, no... except (followed by individual): the general form of such
statements is that a designated individual has a stated attribute and anything having that
attribute is identical to the designated individual. (if the only are >1 then join by
disjunctions when expressing equivalence to all people who have the property). (For all
x, If x is not A then x is not B)
o All except: the general form of such statements is that a designated individual lacks a
stated attribute and that anything not identical to the designated individual has the stated
attribute. (If all except >1 things, then join them by conjunctions when stating their non-
equivalence to everything else). (For all x, if x is not A then x is B)
o Superlatives: These are statements asserting that, of all the members of a class,
something is the largest, tallest, smallest, heaviest, lightest, and so on. To translate these
statements, first give the designated item the class attribute, and then say that, if anything
else has that attribute, it is somehow exceeded by the designated item
o Definite description: an item of a certain sort exists, there is only one such item, and that
item has the attribute assigned to it by the statement. (Scope matters here when negating a
definite description)
o Only a relation b: (Lab ∧ Ax(¬x = a ⊃ ¬Lxb))
o Everyone except a relation b: Ax(¬x = a ⊃ Lxb)
o a relation everyone but b: Ax(¬x = b ⊃ Lax)
o Some predicate only relation b: Ex((Gx ∧ Lxb) ∧ Ay(Lxy ⊃ y = b))
o Whoever wrote the Iliad wrote the Odyssey: Ax(Ix ⊃ Ox)
o Whoever is a present King of France is sexy: Ax(Fx ⊃ Hx)
o Angharad loves only Bryn and Caradoc, who are different people: {((Fb ∧ Fc) ∧ ¬b = c)
∧ Ax(Fx ⊃ (x = b ∨ x = c))}
o the tallest girl: {Gy ∧ Az((Gz ∧ ¬z = y) ⊃ Myz)}
o Existence claims involving predicative expressions and definite descriptions get rendered
into QL = using the existential quantifier. Existence claims involving names don't e.g. Ex
x=n is always true and its negation always false. This is a problem with real names and
empty names (as QL= assigns objects to all names)
- PRELIMINARIES
o When in doubt, use indirect proof
o Tautologies can be proved by conditional or indirect proof. Tautologies expressed as
equivalences are usually proved using two conditional sequences, one after the other.
- ND RULES OF IMPLICATION
Applicable to whole line only
Note that whenever we use this strategy of working backward from the conclusion, the rules of
replacement are the only rules we may use. We may not use the rules of implication, because
these rules are one-way rules.
o Standard rules
▪ Modus ponens
▪ Modus tollens
▪ Pure hypothetical syllogism
▪ Disjunctive syllogism
▪ Constructive dilemma
▪ Simplification (& E)
▪ Conjunction
▪ Addition (or I)
OR
▪ And introduction/elimination
▪ Or introduction/elimination
▪ Conditional elimination (ponens, tollens)/ conditional intro (conditional proof,
weakening, Pure hypothetical syllogism (look for diagonals))
▪ negation introduction (indirect proof)
▪ Constructive/ destructive dilemma
▪ Argument by cases
o Quant rules
▪ Since these are rules of implication, they apply to whole lines only
▪ As long as quantifier attached, rules of inference can't be applied
▪ When dealing with multiple quantifiers, with each successive
instantiation, the outermost quantifier drops off generalization restores
the quantifiers in reverse order
▪ Instantiation is an operation that consists in deleting a quantifier and
replacing every variable bound by that quantifier with the same instantial
letter
▪ EI: introduce only name (do before UI). The name introduced must be a new
name that has not occurred in any previous line, including the conclusion to be
derived.
▪ UI: introduce name (old name or new name {if it doesn't appear before}) or
variable (that ends up free and not captured in the process by another quantifier)
depending on context (variable if intending to universally generalize later over
some part of the statement, name if intending to match some part of universal
statement to another statement).
▪ Generalization is an operation that consists in introducing a quantifier
immediately prior to a statement, statement function or another
quantifier AND replacing one or more occurances of a certain instantial
letter in the statement or statement function with the same variable that
appears in the quantifier. The instantial letter may be the same as the
variable in the quantifier.
▪ EG: quantify over name or variable and replace atleast one of their occurrences
with quantifier variable. It is important that the instantial letter be replaced by a
variable that is captured by no previously introduced quantifier and that no other
variables be captured by the newly introduced quantifier
▪ UG: quantify only over variables and replace all their occurrences with quantifier
variable.
a. In conditional and indirect proof, UG must not be used within the scope
of an indented sequence if the instantial variable y is free in the first line
of that sequence.
b. To keep AxEy(Mxy) from entailing ExAy(Mxy), UG must not be used if
the instantial variable y is free in any preceding line obtained by EI.
c. It is important that the instantial letter be replaced by a variable that is
captured by no previously introduced quantifier and that no other
variables be captured by the newly introduced quantifier
The instantial letter is one over which quantification occurs or which is an
instance of the quantification.
For universals you can generalise and instantiate with the same variables as given
before.
When translating these statements, the point to remember is simply this: The
subject of the original statement is represented by a capital letter in the
antecedent, and the predicate by a capital letter in the consequent.
- ND RULES OF REPLACEMENT
Applicable to parts or whole even when quantifier attached
o Demorgans
o Commutativity
o Associativity
o Distribution
o Double negation (--A⋁B and -A⊃B are equivalent). See result of 8.3, 1, 16
o Transposition
o Material implication
o Material equivalence
o Exportation: [(p ^ q) ⊃ r] :: [ p ⊃ (q ⊃ r)]
o Tautology
o 4 quantifier negation rules
ANALOGY
PROPORTIONAL SYLLOGISM
- z% (<100%) of all As are Bs
x is an A
There is a z% probability that x is a B
INDUCTIVE GENERALIZATION
- Things and instances
x no of As have been observed
All or z% of observed As are B
All or z% of all As are B
Use interval, random sampling with replacement and large sample (or increase interval
which will reduce accuracy). Interval size doesn’t change much unless observed relative
frequency less than 0.1 or greater than 0.9.
INDUCTION BY CONFIRMATION
- hypothesis is a description or explanation of an object (too small, too far away), event
(too long ago) or a process (too long) that cannot be observed
- Based on data and goes beyond it (not mere description of data), though it can be
completely novel as well.
- data is information that comes from observation of the world, not simulations
- prediction should be a testable deduction that is preferably only likely true if hypothesis
is true, not otherwise. P(E|-H) must be low. Make sure it is precise enough so that you
can tell if it matches data
- data for building hypothesis and testing should be separate (though testing data can be
older data not used to build hypothesis)
- If data doesn’t match prediction then hypothesis false by Modus Tollens (given you can
rule out ambiguities in the extent to which data matches prediction; and any errors in
observation of data)
- After confirmation check for any other reasonable explanations using criteria for best
hypothesis (Hypothesis indeterminate if other explanations just as good. Hypothesis
supported if this is the best explanation)
- Sometimes prediction isn’t a deduction (which can compromise Modus Tollens) and
maybe prediction is also likely if hypothesis false. In that case, use Bayes Rule, which
can quantify degree of support hypothesis gets from E or lack of support from –E. You
can also find posterior odds of H against –H
- Procedure
o State hypothesis and its important features
o Make prediction (If … Then …). Assign probabilities to hypothesis.
o Describe data then collect
o See if match
o If doesn’t match use MT to falsify hypothesis or Bayes rule to quantify probable
falsity. If match then look for alternate explanations that provide explanations
just as good (if yes then hypothesis indeterminate, if no then use Bayes rule to
quantify evidential support for hypothesis).
o Make conclusion
NECESSARY CONDITION
A is a necessary condition for B if consequent of conditional
OR
Not -> Not
Yes <- Yes
Rest are indeterminate
SUFFICIENT CONDITION
A is a sufficient condition for B if antecedent of conditional
OR
Yes -> Yes
Not <- Not
MILLS METHODS
- Reason from phenomenon to probable cause
- Mills MoA for necessary condition: antecedent condition always present with
phenomenon, but may be present in absence of phenomenon which rules it out as
sufficient cause
- Modified Mills MoD for sufficient condition: Antecedent condition present in atleast one
case with phenomenon and absent in all cases where phenomenon is absent, but may be
absent when phenomenon is present which rules it out as necessary cause.
- Mills JMoAD for necessary and sufficient condition: One to one correspondence between
antecedent condition and phenomenon
- Mills MoR: rule out possible causes. The one that remains is the probable cause.
(probable because not all causes may be known)
- Mills MoCV: primitive correlation
- Start by looking at patterns. If you see mostly cause and effect present together, go for
MoA unless one to one correspondance (JMoAD). If you see mostly absence of cause
and effect then try MoD.
ANALYSIS PROCEDURE
- identify nature of statements
- Find conc and reasons
- Find hidden premises and separate useless stuff - Lay out pattern: vertical, horizontal
(independent), conjoint (dependent), multiple conc.
- Deductive or inductive evaluation
- Language issues + fallacies
- If argument defective, give counter example or RAA
- If you need to pick hypothesis is given time frame, try to falsify as many as possible.
From the remaining, use induction by confirmation or crucial experiment unless
predictions not possible. Pick inference to best explanation then.