Philosophy of Science
Chapter 1: Some problems of induction
Science sometimes seems inaccessible due to the complexity of it. Meanwhile, scientists
merely use our cognitive abilities in a structured way and draw conclusions from that.
However, that means that scientists also inherit the ‘problem of induction’: Using induction is
fallible since it still leaves open the possibility of the opposing thing happening (if you claim
all donuts are tasty because you have eaten 1000, the 1001th one might still be disgusting),
and to solve this a chain of inductions must lead to a deductive proof or will end into an
infinite regress. This is a general problem we experience with our knowledge: We don’t know
everything.
Though not really being aware of this himself, Newton kept himself busy with considering the
various ways he might go wrong in drawing conclusions from induction. He held his own
experiments regarding the retractability of sunlight, and concluded that some rays of light are
more refractible than others. However, Newton only did some tests on particular rays of
sunlight, yet he claims that his conclusion applied to much more; sunlight and starlight. Here,
Newton is not concerned with the problem of induction: maybe the light works differently in
another time/location, maybe this phenomenon can only be represented correctly if all light is
present, and what is he really claiming? Is Newton stating things about light in general?
Hooke challenged Newton on this, but Newton responded that he doesn’t want to do that.
To solve his problem, Newton would have to show he could eliminate all possibilities of error,
or show how he is justified without falling into infinite regress.
What Newton did in testing different possible errors is called ‘active error probing’, which
represents experiments really well. But since we can’t truly probe everything, which sources
of error are relevant?
Lecture notes:
This course tries to apply philosophy to the discipline of science; it uses the philosophical
tools for clarification. Studying science can be done in a descriptive, or normative way:
descriptive questions are questions examining …, and normative questions regard ….
There have been historical differences in the various scientific methods we had throughout
the ages. Before the scientific revolution, people believed there is a natural order: rules must
be based on solid, undoubtable first principles since the visible world is not really the best
yardstick.
First indubitable principles were:
● The sun, planets and moon are perfect spheres.
● Heavenly bodies move in perfect circles.
● There must be 7 metals, 7 virtues and 7 planets.
Plato was one of the thinkers who thought there was a more fundamental world behind ours:
the realm of ideas, such as ‘The Good’, ‘The Just’, ‘The Chair’, ‘The Circle’, etc. Things in
our world were merely approximately derived from these ideas (acting on justice often also
includes other possible motives).
,Aristotle is a good example of scientific differences from modern science. He distinguished
different types of causes of something: material cause (bronx), formal cause (, efficient
cause and .... He applied this last cause in his search for something’s goal/purpose (telos).
For example: a seed has the goal to become a plant, an object with a lot of ‘the element
earth’ will tend to fall closer to the center of the earth due to some inherent goal to go
towards the universe, rain serves to let plants grow, plants serve to feed animals, and
animals serve to be food or to plow fields for humans, and humans serve to flourish.
Christian thinkers incorporated this in Christianity as ‘God’s plan’. This idea of natural order
stayed dominant until the 16th, 17th century.
This is all derived from certain first-principles. For Descartes, this was ‘I think, therefore I
exist’, and from there, he could create a whole structure of knowledge. This makes field work
unnecessary; all conclusions follow by logic. This is based on Euclides’ manner of deriving
mathematical axioms from each other. It ended up being the ideal structure, since it would
give absolute truth.
Here, there is a difference between realists and instrumentalists: realists think scientific
theory gives the description of the world and has the goal to explain phenomena by giving a
description which is true, while instrumentalists think theories only serve as an instrument to
correspond to observations.
For example: Ptolemy made the model where the earth is at the center with everything
spinning around it. This was never interpreted as reality, but merely an instrument to predict
things (instrumentalist). During the scientific revolution, Copernicus discovered the opposite
by using Ptolomy’s method; planets had to move in perfect circles, but a heliocentric model
was used as another hypothetical instrument. Kepler then went further by claiming that orbits
actually were ellipses, based on a large set of data (realist). Then Galileo Galilei innovated
by experimenting with self-made devices in order to test phenomena under controlled
conditions (‘how can it be that something flies, without being pushed?’). From the way inertia
works, he then derived the planets’ elliptical orbits.
Repetition of chapter 1:
Then, the first normative theory happened: scientific knowledge is built on
observations (empiricism), through connecting the knowledge we get from
observation by means of induction. Another is Logical Positivism, which believes that
philosophy does not exist.
There are induction and deduction. Induction can be enumerative: a proposition gets
more justified the more cases of justification have happened.
Active Error Probing is what you do when you try to eliminate any possible errors out
of your induction-based scientific theory by repeated testing under different
circumstances.
Some might say that induction works, since we use it all the time to make small rules that we
base our behavior around. From this, you can sum up all the times induction has worked,
and then you have an enumerative induction chain which would prove induction works. This
is circular and doesn’t really work in practice. We really only use induction after we have
given it a deductive base so that it is already a kind of ‘mini theory’. Only after that do you
include induction so that you don’t end up in infinite regress or a mere enumerative chain.
Induction also doesn’t always represent the truth because the correlations it finds often do
not mean anything at all (drowning victims in pools / vs amount of nic cage movies).
, Regarding the empirical cycle: it does not end up back into itself like an actual circle; it
works more like a spiral, with different theories eventually being deduced from each other.
Chapter 2: Falsificationism; science without induction:
Popper wanted to get rid of all forms of induction from science. Induction might seem
necessary, but anything it finds can always be wrong, leading Popper to falsificationism. This
view doesn’t think theories can be proven, only disproven. If you rewrite the LDD from the
previous chapter as ‘if it is [a], it is [b]’, you can fill in the variables with whatever you would
like them to be. If a ever happens to be not-b, then the new LDD can be refuted in this way:
If LDD’ is true, then it is not the case that if it is [a] it is not [b]. It is the case that it is [a] and
not [b], therefore LDD’ is not true. Here you falsify a proposition without appealing to
induction (modus tollens).
This extends to Newton: He succeeded in falsifying a claim about light by denying a
proposition he thought was true, but failed to falsify his own hypothesis because his
experiment failed to falsify what he could have falsified. However, you might feel like each
observation is then compatible with infinite amounts of falsifications, but these do not
corroborate.
Popper distinguishes something as being growth to knowledge if it corroborates, which
means that something is free of inductive confirmation, support or verification and is thus
more falsifiable. Something is more falsifiable if it affords more opportunities to be falsified.
This means that simpler theories that describe the world are more falsifiable since it can be
easily recreated. This means that Popper endorses methods which are as falsifiable as
possible.
Popper applied this to various things. He saw how Einstein’s theory was very corroborative,
but how Freud’s psychoanalysis was everything but. Here he found the ‘demarcation
problem’ of distinguishing science from pseudoscience, and eventually concluded that
something is science if it is falsifiable. For Popper, engaging in science is then to only
consider falsifiable theories. Freud’s theory would be mere conventionalism.
These beliefs seem to rest on the claim that it is better to engage in inquiry by attempting to
falsify them than by attempting to confirm or defend them. However, the falsificationist
cannot accept this since it is based on induction and unfalsifiable. Popper responds by
saying that he considers scientific methods to be mere sets of rules which he uses for their
perceived effectiveness. This makes it necessary that the rules about those methods are
conventional, since they then wouldn’t apply themselves to themselves. However, he
perceives falsificationism to be susceptible to criticism but he himself thinks it is a good way
to pursue the use of modus tollens.
But when do you accept falsification? You can also decide there is something wrong with
whatever you are testing, or add a condition to it. Also, we are justified in our own
experiences, but not justified in generalizing. Accepting a statement then requires us to
make the right decision based on the given experience. Here, a balance must be found
between pursuing falsification and pursuing scientific growth. Popper doesn’t see science as
absolute and thus thinks we inevitably need to choose something.
