Induction

OHear: Chap. 2

Baconian Induction

On from Aristotle

Aristotelean Science and Methodology

classificatory methodology: biology the model science
aim is correct description & classification: intellectual comtemplation of the world
leading to intellectual apprehension of the essence or true nature of things
four causes:

material: what it's made of
formal: the plan according to which it's constructed
efficient: the preceding occurences that brought the thing into existence
final: that for the sake of which the thing was done, the end at which it

Aristotelean science, in admitting & even laying special stress on the formal & final causation

qualitative & contemplative: aims at intellectual apprehension and contemplation of the true natures or essences or forms of things.
teleological: seeks to understand the purposes behind things

Modern Science commencing with Bacon's and Descartes reflections on methodology

experimental methodology: physics the model science
aims at prediction and control nature
leading to technological applications
emphasis on material & efficient causation

material emphasized
efficient emphasized
formal materialized & disempowered: not a kind of inner directedness or teleology, as for Aristotle
final eliminated

Modern science, in disallowing final and disempowering formal causation became

quantitative &
mechanistic

Baconian Method: Outlined in his Novum Organum (New Instrument)

allusion to Aristotle's logical & methodological treatises titled the Organon (or instrument)
four "idols": biases which cause us to turn away from knowledge and truth

"idols of the tribe" or social biases leading us to view things in relation to their uses for society (not as the are in themselves)
"idols of the cave" or personal biases leading one to view things in relation to themselves (not as the things are in themselves)
"idols of the market": verbal bias which can lead us to assume that words must refer to realties: "conceptions arising through the use of words standing for nothing" such as

fortune
prime mover
element of fire
planetory orbits

Positive Doctrine: "If we would command nature we must learn to obey her"

research should begin with presuppositionless observation
followed by tabulation of data gathered by such observation
followed by generalization or based on the tabulated data

Example: heat

collect or observe positive instances: all kinds of cases of (perceptions of) heat listed

sunbeams
boiling water
due to friction
due to strong drink
due to application of certain aromatic herbs

collect or observe negative instances: cases near to positive instances but lacking heat

moonbeams
cold water
things touching but not rubbing against each other
water
applications of nonaromatic herbs

factor analysis

list feature(s) present in positive instances: light (sun, fire), bubbling (water), intoxication (strong drink)
eliminate those absent in other positive instances: bubbling not present in sunbeams
eliminate those present in any negative instances: e.g., light present in moonbeams
remainder

feature(s) present in all positive but no negative instances
fire = "motion which acts on the smaller particles of bodies by exciting a dilating or expanding motion in any natural body"

Discussion

the chief virtue of Bacon's proposed method: stress on negative as well as positive instances

anticipating Popper's concern with falsification rather than confirmation
anticipating modern emphasis on crucial experiments

where competing hypotheses make conflicting predictions
this is the test case we seek

chief criticism: the ineliminability of "idols"

human subjectivity colors all our observation with

predispositions due to the very nature of our sensory and cognitive equipment: our nature
predispositions due to our upbringing, language, and culture: our nurture

guesswork (preliminary hypothesis formation) is indispensible

must have hypotheses already in mind to know what to look for
or what experiments to perform

Presuppositionless Observation

can't start with a blank slate
picking out types of cases of heat (e.g.) requires some prior conception of what it is

presupposition that the phenomenon in question is a single natural kind with a common underlying nature
probably not fulfilled in Bacon's cases of heat: no heat, really, in whiskey or aromatic herbs
Bacon himself was working under a faulty preconception (idol of the marketplace?)

since similarities are everywhere picking which ones are worth noticing requires having some background theory

inspired example

falling apples, the tides, the planetary orbits, are all gravatationalphenomena
similarity of extreme theoretical importance

uninspired example:

this lecturn, the flickering of the lites, the noises from the hall, the view from the window, are all SAC106 phenomena.
similarity of no theoretical importance

dubious examples: having your moon in Cancer; being a water sign; being a Saggitarian; etc.

you can't record (or even observe) everything: it's against some (pre)theoretical background assumptions that we pick and choose what to observe and record

example: record everything that happens in this room in the next five minutes
positive facts: the chalkboard remained black, the lights were on,
negative facts: no one died; nothing exploded.

presuppositionless observation may be too localized: things distant in place & time may be important factors that theory would lead us to appreciate by Baconian neutrality would cause us to miss

the moon and the tides
Semmelweis connection between

bloodpoisoning suffered by medical student who cut himself in cadaver lab
women contracting childbed fever

high incidence in ward visited by students on their rounds after cadavaer lab
try washing up after with strong (disinfecting) soap
resulted in reduced incidence of childbed fever
step toward a micro-organism theory of infection

The Role of Imagination in Scientific Thinking

Kepler: hypothesized elliptical orbits for the planets

Brahe had used virtually the same data to support a

modified geocentric hypothesis
with circular orbits

Kepler's heliocentric hypothesis not forced on him by the data but by the data in conjunction with the a priori theoretical desideratum of simplicity:

"the Pythagorean assumption that the world is organized on principles of mathematical simplicity and harmony"
"and attempting to read these harmonies in the simplist way into the data"

Newton's "revision" of Kepler's laws

do not simply build on Kepler's discovery (as pure Baconianism might lead you to suppose)
they actually contradict and correct them at some points, e.g.,

the paths of planets according to Newton are not exactly elliptical
due to their mutual gravitational attractions

Significant scientific advances

rarely come just from piling up more data of the sort we already have
but from "the inspired guess and the wild insight where the categories . . . that currently constrain our thoughts" and even our observations "are set aside" (p. 25)

Scientific pluralism:

The openness of scientific communities to people with different temperaments, presuppositions, and backgrounds
helps cancel out the possible negative influences of personal, social, and theoretical biases.

Inductive Proof

Four stage portrayal of inductive method

data collection: observation
tabulation of observed data according to its features

universal features: had by all positive instances of the phenomenon under investigation
commensurate features: had by no negative instances of the phenomenon

generalization from observed data: that commensurate & universal features are the cause of the phenomena
further testing: applying the generalization under new circumstances

Review

preceding criticisms were of Bacon's descriptions of the first two stages
Kepler/Newton v. Baconian takes on generalization mechanism (context of discovery or psychology)

Bacon: painstaking factor analysis
Kepler/Newton: bold conjectural insight

Inductive Proof: concerns the evidentiary force of observed or successfully predicted instances for confirming the general hypothesis or law: context of justification issue.

Inductive Proof Thesis:

collection of a large quantity of evidence consistent with a hypothesis
with no evidence against it
gives us reason to think the hypothesis true

Inconclusiveness of the evidence

Contrast deduction

All men are mortal; Socrates is a man; so, Socrates is mortal.
The inference is conclusive.
it's impossible for the premises to be true and the conclusion false

Induction is inconclusive due to its ampliative quality

since laws -- water boils at 100c at one a.p. or the law of Universal Gravitation -- cover many more instances than those observed instances which comprise the supporting evidence

it's always at least possible that the premises (supporting observations) are true
and the conclusion . . . generalizing universally from these (relatively few) instances ... is false

Hume's Difficulty

warrantedness of the inference from observed to unobserved cases requires tacit uniformity of nature assumption

the future will resemble the past or
unobserved cases will be similar to cases already observed

so the inference is no more warranted that this tacit assumption
but what warrants this assumption

cetainly not true a priori:

it's possible that the future might not resemble the past . . . all might change in the twinkling of an eye
in many particulars the future actually does fail to resemble the past

can't be warranted a posteriori (empirically or by observation) on pain of circularity

you are arguing, in effect, the future will resemble the past because it always has in the past
but this inference is warranted only if the uniformity of nature principle is warranted
so the inference cannot warrant the principle -- that'd be circular: one washerwoman taking in the other's laundry

appeal to probability is in vain

given the virtual infinitude of unobserved cases however many observed cases you have will be an insufficient sample
unless you somehow know it's a representative sample
but this is just where the uniformity of nature principle comes in
which Hume has shown to be insupportable.

Goodman's New Riddle of Induction

assume the future does resemble the past: you're absolved from Hume's difficulty
still doesn't tell us in what respects the future will resemble the past
Goodman's point: what similarities we feel warranted in projecting from past to future depend crucially on our conceptual categories

all emeralds so far observed by us have been what we call 'green': they all instance greeness: they're all similar in this regard.
suppose we also have the word `grue': grue means "green-and-examined-before-20[1]0 or blue": thus all emeralds so far examined would also be what we call `grue': they all instance grueness: they're all similar also in this regard.
question now becomes what warrants our projecting greeness into the future (expecting emeralds to remain green after 2010) rather than projecting grueness (expecting emeralds first examined after 2010 to be blue).

Worries about the fishiness of grueness miss the fundamental point: you can generalize from any given body of evidence in indefinitely many ways and we've no principled grounds for saying which generalization should be priviledged.

A hopeful observation: there is conclusive falsification

Examples

The generalization "All swans are white" is conclusively disconfirmed by the observation of one black swan.
The conservation of matter law was conclusively disconfirmed by the first atomic explosion.

General point

can never observe more that an insignificant proporition of ALL the instances (forever & everywhere) covered by a scientific law
to conclusively prove a universal generalization we would have to examine every case
but to conclusively disprove a universal generalization we only have to find a single counterexample

to prove "All Alma students are over 4'6" tall" we would have to examine every single student
but it's disproved as soon as we find one student shorter than 4'6".

Hempel's Paradox of Confirmation: skip