L3 - Philosophy of science, p.1 Flashcards
What is philosophy of science?
Branch of philosophy that studies the foundations of scientific research, to better understand the position of scientific research relative to other forms of information acquisition and generation
What was the initial view in Acient Greece about how is knowledge aquired?
Before the scientific revolution it was generally accepted that only deductive reasoning led to necessary truth (Plato and Aristotle, although Aristotle differentiated between inductive and deductive reasoning)
How did the work of Galilei and Newton infleunced the shift from deductive reasoning to inductive reasoning?
- Galilei and early Newton at first tried to convince their audience that the new way of thinking was very close to traditional deductive reasoning and demonstration
- They both pointed out to the improtance of observations and experimentation
- Later, Newton shifted from deductive approach to integrating induction in his work
How did the work of Bacon, Herschel, Comte and Whewell contribute to the shift to induction?
- Gradually natural philosophers started to argue that inductive reasoning could lead to conclusions as probable as truth, when facts were collected in large numbers and without prejudice, when effects could be replicated, and when theories led to new verifiable predictions
- Bacon: Among the first to promote the use of observation and experimental histories as the basis of knowledge acquisition, rather than reliance on tradition and authority
- Whewell, Comte and Herschel further pointed out that there is no clear distinction between observation and idea, between fact and theory. They are closely interconnected and influenced by the other so they cannot be separated
- As a result of the successes of science, most of the initial doubts about whether inductive reasoning could lead to true conclusions were swept under the carpet towards the end of the nineteenth century
What are two big ideas that built philosophy of science in the 20th century and to this day influence what we view as scientific? Why is this influence problematic?
- Verification proposed by logical positivists
- Falsification proposed by Popper
Even though they are not largerly supported in current philosophy of science, they still exert significant effect in psychological science (through their effect on research and methodological practices). This is problematic because it puts psychological science at risk of stagnation due to its implicit strict research norms that we blindly follow (operationalisation, hypothesis testing…)
What is logical positivism?
Philosophical movement in the first half of the twentieth century, claiming that philosophy should stop thinking about metaphysics, and instead try to understand the essence of the scientific approach; central tenet was the verification principle
What is the Vienna Circle (wiener Kreis)?
A discussion group set up in the 1920s in Vienna comprising scientists, mathematicians, and philosophers aiming to base philosophy on science and logic
- The basis of logical positivists
- They are plotting an attack on the traditional ‘‘vague’’ philosophy (metaphysics - part of philosophy that is about the essence of things rather than something we can observe)
So how did 20th century philosophy of science come to exist?
It came into being by the light of the Vienna Circle of logical positivist philosophers, scientists and mathematicians, who thought it was time to stop metaphysics and its boundless speculative discussions, and ‘to set philosophy upon the sure path of a science’
- logical positivists wanted to create a rule for philosophy that is based on ‘sure path of science’
Who and what influenced the logical positivists?
the Tractatus, a book of Ludwig Wittgenstein, in which he gives a philosophy of meaning & language
- Wittgenstein believed that paradoxes in logic were due to problems in not distinguishing between those propositions that are meaningful and those that are not
- Additionally, he claimed that language was a faithful depiction of physical reality which could be known by analysing the logical structure of language
What weapon did logical positivst intend to use to object to the traditional vague philosopy?
Meaning
‘If someone asserts “there is a God”, “the primary basis of the world is the unconsciousness” (refers to work of Freud which is also vague), we do not say to him: “what you say is false”; but we ask him: “what do you mean by these statements.” These statements reveal themselves as empty of meaning if one takes them in the way that metaphysicians intend.’ - Otto Neurath (logical positivist, member of Vienna Circle)
- They won’t argue against these people, they will just keep asking them for the meaning of their statements until it clarifies itself to that person that they actually have no meaning
What purpose was philosophy given thanks to logical positivism?
The clarification of language and the assessment of which sentences are meaningful and hence should be studied by empirical sciences - the linguistic turn
- This is revolution in philosophy: ‘‘philosophical questions are questions of language’’
How is linguistic philosophy defined?
“I shall mean by ‘linguistic philosophy’ the view that philosophical problems are problems which may be solved (or dissolved) either by reforming language, or by understanding more about the language we presently use.” –Richard Rorty
- That which remains after the language has been clarified and stripped of meaningless claims, becomes the subject of empirical science
What was the linguistic philosophy sometimes called and what other philosophy did it oppose?
It was called analytic philosophy and opposed continental philosophy which emphasizes historical context, human experience and critical analysis of culture, society and existence
What did we see in the past that was similar to the thinking of logical positivists?
- Hume’s rejection of the notion of causality and behaviourists ideas
- Both based on similar considerations as the logical positivists: Things that we cannot get from observations don’t belong in science
What was The Manifesto of logical positivism?
“The scientific conception of the world: the Vienna Circle”
- Starts philosophy of science as a separate philosophical discipline
- It deeply marks thinking about science, especially in psychology
- One of the most influential pieces in history
What were the starting points of the manifesto?
- Meaningful claims are either empirical or logical in nature
- Logical claims are verifiable by looking at their form (e.g, logic and mathematics) and help represent the world
- Empirical claims are about the world and can be verified by observation
- Only claims that are verifiable are meaningful!
What is demarcation of science?
- setting and marking the boundaries of a concept; used, for instance, in the philosophy of science to denote attempts to define the specificity of science
- Give a criterion that it has to fulfill and if it doesn’t fulfill it then it’s not scientific
- Demarcation criterion distinguishes science from pseudoscience
What demarcation criterion did logical positivists suggest?
The verification criterion
What is the verification criterion in logical positivism?
A principle stating that a statement is only meaningful if it can be verified through observation
- If you cannot find out the truth of a sentence by lookinh at reality, then the sentence becomes meaningless (e.g. meaningless: The soul is immortal; meaningful: The scale indicates 34kg)
- True is different from meaningful - meaningful sentence just has to have a possibility of verifying it
How do logical positivists propose to find the meaning of a proposition?
By transforming the proposition into successive definitions until finally only such words occur in it as can no longer be defined, but whose meanings can only be pointed out
What is sense data according to logical positivists?
Experiences gained through sensory perception that are objective and neutral
- They serve as a foundation for science
How do sense data and verification relate?
- Verification is the comparison of descriptions of observations (“observation sentences”) with these sense data
- Theoretical statements are verified through observation sentences
How does structure of science look like?
- We start at the bottom with the sense data since that’s the foundation that we are very certain of
- The observational sentences represent the sense data (reality) and they are directly observable
- These observation sentences can be linked to the theoretical vocabulary so this level indicates the summary of the observational sentences
- We need correspondance rules (e.g. measurement theory) which link the theory with the observation sentences, hence reducin the theoretical concepts into observation
- From this we can build overarching laws but again it’s nothing more than just compression of the observational sentences
Example from the structure of psychological science
What are the 5 problems with verifiability that ultimately lead to its failure?
- Theoretical concepts are richer
- Theory-ladnness of observation
- Underdetermination of theory by data
- Induction
- Unobservable entities
Problem with verification
Theoretical concepts are richer
- Theoretical terms, such as ‘force’ and ‘mass’, ‘depression’ or ‘extraversion’ do not seem to be reducible to observations
- Especially in psychology many open concepts (e.g. ‘intelligent’, ‘vain’), which cannot be defined exhaustively
- Theoretical statements are thus essentially ‘richer’ than summaries of observational statements
Problem with verification
What does ‘theory-ladenness of observation’ mean?
The concept that observations are influenced by the theories and frameworks that the observer holds.
Problem with verification
Why is theory-ladness of observation problematic and where does it stem from?
- The logical positivists make a strict separation between observation and theory
- They assume that observations (sense data) are “neutral”
- However, there is a big difference between observing X and observing that X has property Y
- The latter type of observation is important but relies on a theory that defines property Y
- Moreover, scientific observations are often based on instruments - and their accuracy is itself based on theory (e.g. observing the brain from an fmri - we have to make lot of assumptions about how the instrument works, which is based on a theory, before we can observe something from it)
- Important is also knowing how to observe - filter out the information that is actually relevant and specific, not just looking around the room and not knowing what we’re supposed to be observing (Popper)
Problem with verification
Underdetermination of theory by data
Theory #1: “This is a duck”
Theory # 2: “This is a rabbit”
- Observation does not uniquely determine theory so how do you determine which theory is correct according to the empiricists?
- Theories are sometimes equivalent in their empirical consequences = underdetermination of theories by empirical data
- To choose between theories, scientists then use other criteria (e.g. sparsity, elegance)
- But those criteria are themselves theoretical!
Problem with verification
Induction
From specific to general
- Induction is not logically valid and the whole framework of logical positivism is based on induction so verification is logically impossible
- !Inductive reasoning doesn’t lead to conclusions that are guaranteed to be true according to the rules of logic!
- General statements are not verifiable (so technically they are not part of science) since we will never observe all of the combinations of observations that lead to these general statements (white swan example)
- Specifically, statements about infinite sets, for example: continua in science
- So, general laws like F=ma are not verifiable…
- Back to Hume - logical positivists also want to reduce causal relationships to observations but it’s impossible becasue with the verification criterion, causal relationships can’t be part of science
Problem with verifiability
Unobservable entities
- In the 20th century, science fills up with unobservable entities - elementary particles, atoms, photons, etc.
- Statements about unobservable entities are neither verifiable nor reducible to observations
- In addition, new techniques make some entities observable
- But were statements about microbes “meaningless” (unscientific) until the invention of the microscope?
How did logical positivists argue to solve the problem of non-observable variables?
Operational definitions, however this creates another problem - how do we define these variables and which definition is correct?
What was the last drop that burried logical positivism?
Karl Popper introducing an alternative - falsificationism
What is falsification?
It says that statements that cannot be falsified because they make no clear predictions (all possible outcomes are in line with them) are not scientific
What inspired Popper to get his ideas?
- Popper worked in a daycare center with Alfred Adler where they had children with different problems
- Adler was a believer in psychoanalysis and explained lot of the cases by the theory of inferiority complex
- Popper presented Adler with a case that he felt didn’t fit into Adler’s theory
- Adler explained the case by modifying his theory, showing great certainty
- “Slightly shocked, I asked him how he could be so sure. ‘Because of my thousandfold experience’, he replied; whereupon I could not help saying: ‘And with this new case, I suppose, your experience has become thousand-and-one-fold.’” (Popper, 1978)
- The idea is that he has all those experiences that confirm his theory but if he observes something that was not in line, he just changes the theory so that it’s confirming it again = if a theory can explain all the observations and facts, it becomes weak and meaningless
What else was going on in the background when Popper was building his ideas?
- In the same period Einstein posits the theory of relativity
- This theory predicts that light deflects along a planet
- The solar eclipse of 1919 enables Arthur Eddington to test the prediction
- If the prediction didn’t come true, this would proof Einstein’s theory wrong
- So Einstein’s theory excludes certain events which if they happened they would falsify the theory
- This is a strong theory exactly because it can not explain everything
- Karl Popper recognizes that this marks the essence of science
- After all, if you’re wrong, you can find out!
- With a theory that can explain everything this is impossible, such a theory is uninformative
What conclusions did Popper reach based on Adler and Einstein’s theories?
- Theories are bold conjectures whose predictions can be tested against observations
- Theory-free observation is impossible, but is also not needed because you have your theory and then you test that theory against observations
- Induction is impossible: you cannot induce theories from observations
- But that is ok, because instead we can do deduction: theories can be used to derive predictions about observations!
How is the Black Swan a representation of Popper’s theory and his criticism of logical positivism?
- Europeans believed that all swans are white until black swans were discovered in Australia
- Logical positivism would deem “all swans are white” as increasingly valid with every white swan observed but seeing a thousand white swans does not guarantee that the next swan won’t be black
- The Black Swan illustrates this flaw in induction
- Popper’s falsification, however, shows that a single counterexample (a black swan) is more powerful than any amount of confirming evidence
- The Black Swan demonstrates that scientific knowledge progresses not by confirming hypotheses but by critically testing them and discarding those that are falsified
Besides falsification, what did Popper point out as important to do in science?
For scientists to be committed to put their ideas/theories to the falsification test and not just try to find evidence in favour of their statements
What did Popper argue about the development of scientific theories?
He argued that theories arise from creative acts of the mind rather than from observations and then predictions are derived from it and those can be tested on the basis of observation
What does Popper consider as the only scientific act and how can it be achieved?
The only scientific act is refutation and it can be done by deducing predictions from the theory that can then be tested
AND deduction is logically valid!
What is the hypothetico-deductive model?
A scientific method where theories are proposed on the basis of observation, induction and educated guesswork. From this theory, testable predictions (hypotheses) are deduced, and those predictions are tested against observations
- If these predictions don’t come true: falsify that theory
- If they do come true: corraboration (is not verification)
- A corraborated theory is one that is strong because it survived risky tests so far (but it’s not accepted since there is still the possibility that it might be falsified)
What does falsificationism emphasize in scientific theories?
Theories should be structured so that they can be potentially falsified by observations
What are the two phases that Popper identifies?
Theory developing phase and testing phase which are very separate and distinctive
- The theory developement phase is completely free = context of disovery → there is no logic
- The testing phase = context of justification → strict rules for the logic of testing: modus tolens
Logical positivists build on the tradition of empiricism. In which philosophical tradition is falsificationism rooted in?
Rationalism because we start with theories that spring from the imagination
- This means that part of the content of our theories does not come from observation but from the ratio
How does Popper’s view of rationalism differ from that of Plato or Descartes?
Popper considers the ratio fallible since we test and can falsify the theories that come to mind that’s why his version of rationalism is called critical rationalism
↪ This is still a popular movement in the scientific community (but not as much in the philosophy of science community)
So what is the demarcation criterion that Popper came up with?
The falsification criterion
What is the falsification criterion?
A criterion that states a scientific theory must be able to be proven false through observations
Statements or theories are scientific only if they are in conflict with possible observations
- A theory is scientific only if the theory can be proven false through observations (through falsification)
- No problems with induction, infinite sets, or unobservable entities
Based on what does Popper differentiate theories?
Degrees of falsifiability
What does greater falsifiability imply about a theory?
Theory excludes more = more falsifiable = greater informative content
What are two criteria that Popper identifies that increase a theory’s degree of falsifiability?
Precision of what it predicts and generality of what it pertains to
- A more precise theory and a more general theory excludes more
What does it mean when theory is more falsifiable if it excludes more?
“excludes more” possibilities = it makes specific and precise predictions that can be proven wrong by evidence, it leaves less room for interpretation or adjustment to fit any outcome, making it easier to test against reality
Example:
Theory A: “All swans are white.” (Excludes the existence of non-white swans.)
Theory B: “Swans can be white, black, or any other color.” (Includes almost all possibilities.)
Theory A is more falsifiable because discovering a single black swan disproves it, whereas Theory B can accommodate any color and is harder to falsify
Example: which statement is more falsifiable and why?
A) All women wear a red t-shirt
B) All women with blue jeans and brown hair, wear a red t-shirt
A is more general (applies to more instances) because it concerns all women
B applies to limited number of instances, namely the women who wear blue jeans and have brown hair
Every person who disconfirms B will also disconfirm A
So A is more falsifiable since it is more general
We are looking at general here
When can we make comparison of degrees of falsifiability between theories based solely on logic?
When the two theoretical claims are subsets - reasoning with sets
We could also compare two theories that are not subsets but we would have to make direct observations and count for example
Example: Which statement has greater informative content and why?
A) All women wear a red t-shirt
B) All women with blue jeans and brown hair, wear a red t-shirt
According to Popper’s theory, those theories which exclude more are more falsifiable so they have more informative content automatically
So again A would have more informative content
Example: which statement is more falsifiable and why?
A) People who are primed with Einstein, score higher on an IQ test.
B) People who are primed with Einstein and are not from Amsterdam and had a good night of sleep, score higher on an IQ test
In B, we are making the statement more and more conditional on specific events by adding information before we can look at our prediction so it makes it harder to falsify
B = more conditional = less general = less falsifiable
A is still specific on people who are primed with Einstein but it is more general
Example: Which statement has greater
informative content and why?
A) All women wear a red t-shirt
B) All women wear a red t-shirt and blue jeans and have brown hair
B is more informative because it is more precise
Now we are looking at how precise the prediction is
Which statement has greater informative content and why?
A) People who are primed with Einstein, score higher on an IQ test.
B) People who are primed with Einstein, score 6 points higher on an IQ test.
B because it makes more precise predictions so it’s easier to falsify it
How do these two examples differ?
1.
A) All women wear a red t-shirt
B) All women with blue jeans and brown hair, wear a red t-shirt
2.
A) All women wear a red t-shirt
B) All women wear a red t-shirt and blue jeans and have brown hair
First example looks at general vs conditional → What or who does the hypothesis pertain to? (A is more falsifiable)
Second example looks at precise vs imprecise → How precise is the hypothesis in what it predicts? (B is more falsifiable)
What are problems with falsification?
- No clear distinction between better or less supported theory
- The Duhem-Quine thesis
Problem with falsification
No clear distinction between better or less supported theory
Popper cannot make a clear distinction between better supported and less supported theories. At best, he can only say that theories are ‘not yet refuted’
- This feels weak to people because all of your knowledge is provisional, conjectural and hypothetical so the universal theories of sciences can never be conclusively established
- But Popper doesn’t consider this a big problem because that’s just how it is
Riet also said that she doesn’t consider this a big problem
Problem with falsification
What is the problem with hard falsification
Popper also admitted that hard falsification is impossible because of the Duhem-Quine thesis
Problem with falsification
What does the Duhem-Quine thesis suggest about testing theories?
A theory is never tested in isolation, they are always tested together with additional background assumptions
- So if a prediction doesn’t come true, it could be because of the theory
- But, it could also be because something else is wrong (e.g. wrong measurements, tests…)
Problem of falsification
Why is the Duhem-Quine thesis a problem for falsification?
- If a theory cannot be tested in isolation, it will also not be rejected in isolation
- Hard falsification is therefore impossible
- After all, you never know for sure whether your theory should be falsified or that one or more of the background assumptions are wrong
- The Duhem-Quine thesis is a structural problem for Popper
- Similar to how induction is a structural problem for the logical positivists
What are implications of Popper’s proposals for science’s status?
- Science proceeds by trial and error
- Confirmation bias
What did Popper argue about the progress of science in terms of trial and error?
- Theories can be proven false, but cannot be proven true; they are just the best available, the ones that (thus far) have passed the tests
- Because there is no guarantee that scientific explanations are correct, progress in science is best seen as process of trial and error, in which many possibilities are ventured and only the fittest survive
- Hence, scietists are less likely to stick to wrong opinions - wrong theory will hit on a falsification test and will have to be replaced = science learns from its mistakes
What is the negative side of trial and error when used as a progress for science?
Falsificationism implies that scientists can give wrong advice if they base their advice on wrong theories that have not yet been falsified
How does falsification help with avoiding the confirmation bias?
- Humans naturally tend to look for evidence which confirms their predictions
- Falsification goes against this human intuition (and disproves verification which results in confirmation bias)
How did science regard falsification?
Popper’s approach was soon perceived by the philosophers of science as more fruitful to the understanding of scientific progress than logical positivism. However, it had its problems too in the light of how science operated. Science does not reject theories as soon as they are falsified
Why did science not like that falsification pointed towards rejecting theories as soon as they are falsified?
In some instances, it proved that rejecting a theory immediately after data conradicted it, isn’t the best idea. When Newton’s laws were contradicted by Uranus’ deviations from how Newton predicted it should move in space based on his laws, it should have been rejected since it was falsified. However, instead of that, other scientists revisited the theory and suggested that the theory would still hold if another planet was present in Uranus’ orbit. This turned out to be true and let to the discovery of Neptune.
If they immediately rejected Newton’s theory, they would never reach such an important discover. This is true for many other instances in science that’s why it’s important to investigate whether the data contradicting the theory are sound and if they are then they should consider amending it in the light of the new finding.
Now that theories are allowed to be modified, what did Popper suggest to make distinction between theories that should be modified and whcih should be rejected?
Modifications that were not testable (not based on empirical evidence) or that made a theory less falsifiable were unacceptable to Popper = ad hoc modifications
Why do researchers not like to give up theories?
- No so much that they have to admit that they are wrong; rather that rejecting a theory means they have to search for a new, plausible theory that explains everything the previous theory explained plus the novel, contradictory finding. This, arguably, is one of the hardest challenges in science
- Additionally, falsification by itself doesn’t lead to the correct interpretation, it only eliminates the wrong ideas. So until someone comes up with a correct interpretation, falsification is a dead end. That’s why they try to revise the theories
