Week 1: Success and failure in science

Question 1

Hvad er demarkationsproblemet?

Answer 1

Demarkationsproblem, i filosofien en betegnelse for problemet om, hvorledes man sondrer mellem egentlig videnskab og opfattelser, der ikke er videnskabelige, men påberåber sig denne status (pseudovidenskaber, religiøse, metafyiske).

Question 2

Hvilke svar giver hhv. positivisterne og Popper på demarkationsproblemet?

Answer 2

Positivists: empirical method & measurable statements (also popper)

Verification criteria - all statements which are not based on observations and experiences are meaningless

Role of philosophy is to produce inductive logic which explains relation between world and scientific statements

Popper: science is about conjecture and failures

Statements or systems of statements (in order to be ranked as scientific) must be capable of conflicting with possible/conceivable observations

A sentence (or theory) is empirical-scientific if and only if it is falsifiable

Question 3

Poppers “falsificerbarhed”

Answer 3

Falsificerbarhed betyder at en teori/hypotese kan efterprøves ved hjælp af eksperimenter

Question 4

Poppers falsifikation

Answer 4

Falsifikation betyder at man vil påvise at ens hypotese/ teori er forkert gennem eksperimenter, på alle tænklige måder

Question 5

Hvilke kritikpunkter har været fremsat mod Poppers teori?

Answer 5

Quine-duhem thesis

Risky generalisations do not always mean good science

Falsifiability criteria too restrictive

Hypotheses about evolution/function of genes/proteins cannot be absolut generalisations

Reproductability not always necessary

Question 6

Hvilke elementer af positivisterne og poppers teorier har overlevet i moderne videnskab?

Answer 6

Critical, unbiased

No ad hoc

Question 7

Hvordan håndteres uventede resultater typisk? Er der eksempler på eksperimenter som har falsificeret hypotesen (a la Popper)? Eller hvor man kritisk har efterset andre elementer i eksperimentet (f.eks. hjælpehypoteser)?

Answer 7

If the unexpected results are few and differ a lot from a larger set of results, the unexpected result might be removed from the results all together

The hypothesis could be changed to fit the results

When flat earthers did an experiment to prove that the earth is flat. The results actually ended up proving that earth was indeed not flat. Therefore their experiment falsified their hypothesis.

When scientists based their hypothesis on the knowledge of the way proteins are folded. They later discovered that not all proteins are folded and that their results therefore wouldn’t back the hypothesis.

Question 8

Hvordan vil I beskrive jeres eksempel ud fra Güttinger og Loves ”taxonomy of failure”? Kan I give eksempler på alle fejltyper, teksten nævner?

Answer 8

We expect that our results were because of a failure of precedent. It was most likely our materials or technique that were not proper and led to the unexpected results.

If you look at the experiment involving the folding of proteins, the failure to observe expected outcomes was because of failure of theorizing. They based their hypothesis on prior knowledge that turned out to not be true.

Fraud - pressure to publish. Removing or adding results. Changing results. P-hacking

Question 9

Hvilke tiltag kan man gøre for at mindske de forskellige typer fejl?

Answer 9

Lessen the pressure to publish

Publish their hypothesis and P-values before doing the experiment.

Very detailed description of experiment and protocol

Better funding and more background research

Question 10

Diskussion: Er reproducerbarhed lige vigtigt som videnskabeligt ideal i alle typer forsøg og felter? Skal/bør det være det? Inddrag gerne teksten af Leonelli i diskussionen.

Answer 10

Based on the text from Leonelli, the answer is no.

The need to be able to reproduce the same results, can lead to important results being pushed aside if they cannot be reproduced.

There are some situations where it’s impossible to recreate the exact same conditions and therefore impossible to receive the same results. This can once again lead to important findings being ignored or devalued, as they don’t live up to the condition of being “reproducerbar”.

important to weigh strengths and limitations - ecological validity vs reproducibility

Question 11

Karl Popper (1902-1994)

Answer 11

His philosophy was inspired by the success of physicists and by his skepticism of phycologists.

Question 12

Logical Empiricism

Answer 12

Developed by an influential group of philosophers and scientists who where a part of the Vienna Circle.

The logical empiricists were pro-science. and surprisingly anti philosophy.

They strongly rejected most of the German philosophy that was popular at the time.
- Philosophers that used terms and concepts that were not rooted in observation and experience were talking nonsense
- The proper role of the philosopher was to analyse the logical structure of scientific statements in their relation to each other and to the world

Question 13

The logical empiricists picture of scientific progress

Answer 13

Observation reports –> hypothesis –> Prediction/explanation –> confirmation –> and back to hypothesis.

The hypothesis is formulated as a scientific law or generalization

The explanation can be derived from the hypothesis.

Hypothesis and confirmation relies on induction

Hypothesis to explanation relies on deduction.

Question 14

Induction

Answer 14

Inductive is used to describe reasoning that involves using specific observations, such as observed patterns, to make a general conclusion. This method is sometimes called induction. Induction starts with a set of premises, based mainly on experience or experimental evidence. It uses those premises to generalize a conclusion.

Can be flawed, since An inductive argument may be highly probable, but even if all the observations are accurate, it can lead to incorrect conclusions.

Question 15

Deduction

Answer 15

Deductive reasoning (also called deduction) involves starting from a set of general premises and then drawing a specific conclusion that contains no more information than the premises themselves. Deductive reasoning is sometimes called deduction (note that deduction has other meanings in the contexts of mathematics and accounting).

Here’s an example of deductive reasoning: chickens are birds; all birds lay eggs; therefore, chickens lay eggs. Another way to think of it: if something is true of a general class (birds), then it is true of the members of the class (chickens).

Deduction can go wrong if you start with wrong premises.

Question 16

Hume’s opinion on induction

Answer 16

“we use induction all the time but can’t logically justify it”

Question 17

Popper’s picture of scientific progress

Answer 17

Problem statement –> Hypothesis/conjecture –> Prediction/explanation –> falsification/refutation –> NEW hypothesis –> prediction/explanation –> falsification/refutation

ad hoc modification is not permitted. New empirical content is necessary when it comes to the new hypothesis.

Question 18

Marxism and ad hoc modification

Answer 18

According to Marxism, the ruling class is not interested in ensuring decent living and working conditions for the poor. Marxism predicts that the suppressed working class will at some point start a revolution against the capitalists.
* Many have argued that this prediction was falsified by the rise of the welfare state, which ensures that the poor are looked after.
* Marxists responded with an ad-hoc modification to their view. They argued that the welfare state shows that the ruling class is aware of the imminent communist revolution and tries to delay it by pleasing the working class.

it made predictions that could be tested. However, Marxism was falsified: what it predicted did not happen.
Since Marxists tried to hold on to their theory without increasing
*
its empirical content, they effectively rendered their theory
‣
unfalsifiable. The wasn’t tested critically anymore, because any observation became compatible with it.
Marxism turned from a scientific theory into a pseudoscientific when it was no longer (treated as) falsifiable.

Question 19

Falsifiable theory

Answer 19

Makes predictions that can be tested and possibly disproven

Question 20

Falsified theory

Answer 20

Tests proved the theory/predictions to be wrong

Question 21

Pseudoscientific + examples

Answer 21

Looks for proof that support a biased result. You have presumptions of how something is and you try to get it proved (for any price).
Pseudoscience can’t be falsified, it does not live up to the falsifiability criteria.

Freud’s theories. They can explain everything with a patient.

Astrology/horoscopes. Too general to disprove

Marxism and the ad hoc adjustments.

Question 22

Freud and Adler. Pseudoscience examples.

Answer 22

Example: Imagine a man pushes a child into the water, intending to drown it. Another man dives into the water, intending to save the child.

Freud would say that the first man suffers from ‘repression’, and that the second man has reached ‘sublimation’.

Adler would say that the first man suffers from feelings of inferiority, and so does the second man

Popper argued that you could not think of any human behavior which could not be interpreted in terms of either theory. It was bc of this fact that they always fitted, were always confirmed. It seemed their strongest argument, but is in fact their strongest weakness.

Question 23

Falsifiable statements are risky

Answer 23

Falsifiable statements are ‘risky’ statements: they “stick their neck out” – exposing themselves to the guillotine of scientific testing.

Consider a classic example of swans:
It is clear what we need to look for to falsify this statement: a non-white swan.
If we find a non-white swan, we have refuted / falsified the statement “all swans are white”.
This is progress! Science advances by refuting false hypotheses.

Question 24

Logiske positivister

Answer 24

De var pro-science. De så deres opgave som det at støtte en “videnskabelig opfattelse af verden”, og for positivisterne udmærker god videnskab sig ved sin empiriske metode. (popper enig)

Antimetafysisk retning: De afviste meget af den eksisterende filosofi som værende ren spekulation, særligt tysk filosofi. Videnskabelige påstande skulle kunne gøres målbare. (popper enig)

Verificerbarhedskriteriet: positivisterne mente at alle udsagn som ikke kunne baseres på observationer og erfaringer var meningsløse (popper uenig)

Filosofiens rolle er at udvikle en induktiv logik som begrunder relationen mellem verden og videnskabelige udsagn. (popper

Question 25

Poppers kritik af de logiske positivister

Answer 25

1. Han mente at deres optagethed af meningskriterium er ufrugtbar (et “verbalt problem, et typisk pseudoproblem”)
2. Induktion (og verifikationen via induktion) kan ikke retfærdiggøres logisk pga induktionsproblemet.

Question 26

Induktionsproblemet

Answer 26

Vi kan ikke logisk slutte fra et endelig antal observationer til noget generelt. Vi kan ikke med sikkerhed basere vores viden om fremtiden på vores hidtidige erfaring.

Alle talte svaner er hvide, så man stopper og antager at alle svaner er hvide. Den næste talte svane kunne muligvis være sort.

Question 27

Poppers demarkationsproblem

Answer 27

Hvad er forskellen på videnskab og pseudovidenskab.

Utilfredshed med “altomfattende” teorier, som er så vagt og generelt formulerede at de bliver empirisk tomme og uden mulige modsigelser.

Question 28

Poppers billede af videnskaben

Answer 28

Den enkelte forsker og forskersamfund skal både udarbejde dristige hypoteser og søge at gendrive dem.

Brugen af ad hoc hypoteser er ikke acceptable i videnskabelig praksis

Falsifikationisme

Question 29

Falsifikationisme

Answer 29

En teori skal forbyde nogle mulige observationer, hvis teorien skal være videnskabelig

Question 30

Ad hoc hypoteser

Answer 30

Hypoteser som udelukkende opstilles for at redde en teori

Flat earthers

Question 31

Udfordringer for Poppers teori: Quine-Duhem tesen

Answer 31

Meningsholisme: Vi tester aldrig en hypotese særskilt. En hypotese står altid i forhold til en række andre udsagn, teorier, og forudsætninger i et netværk af antagelser og hjælpehypoteser.
En modstridende observation vil oftest fore til en revision af en eller flere af hjælpehypoteserne og ikke af teorien selv.

Hypotesen er ikke entydigt fastlagt af en given mængde data eller empiriske iagttagelser. - Flere forskellige hypoteser/teorier vil kunne redegøre for de samme data.

Question 32

Udfordringer for Poppers teori

Answer 32

”Enhver ”god” videnskabelig teori er et forbud: den forbyder, at visse ting sker. ”Jo mere en teori forbyder, jo bedre er den” (s. 43)

Men hænger ”risikable” generaliseringer altid sammen med god videnskab? Eller falsificerbarhedskriteriet for restriktivt?

Hvad gør vi når vores teorier ikke er absolutte generaliseringer? F.eks. hypoteser om evolution eller funktionen af bestemte gener eller proteiner?

Popper: Videnskabelige teorier skal generelt være “dristige”. Når der ikke er tale om observationer der fuldstændig afviser teorien kan man tale om meget usandsynlige hændelser.

“Non-reproducible single occurrences are of no significance to science” (Popper)

Question 33

Grader af (og variation i) reproducerbarhed som videnskabeligt ideal

Answer 33

Computational Reproducibility

Direct Experimental Reproducibility: Standardized Experiments

Scoping, Indirect, and Hypothetical Reproducibility: Semistandardized Experiments

Reproducible Expertise: Nonstandard Experiments and Research on Rare Materials

Irreproducible Research: Participant Observation

Question 34

Positivistiske ethos

Answer 34

-Politiske ideologier bør ikke få indflydelse på videnskabelige resultater (krav om deklaration af interessekonflikter i publikationer)

-Kravet om at underbygges videnskabelige teorier empirisk gennem præcise og reproducerbare observationer og eksperimenter.

-Videnskaben skal undgå ren metafysisk spekulation og henvisning til ikke-målebare og overnaturlige fænomener (f.eks. en intelligent designer i biologien).

Question 35

Hvad har vi lært af popper

Answer 35

Vigtigheden i at være kritisk og ikke bare lede efter bekræftende evidens. Fremsættelse af “ad hoc” hypoteser og andre forsøg på at redde en teori fra falsification er uvidenskabeligt.

Question 36

Nogle problemer med reproducerbarhed skal tages seriøst.

Answer 36

P-hacking og publication bias

Question 37

P-hacking

Answer 37

P-hacking:
* p-value: grænsen for statistisk signifikans.
* p = 0,05 betyder mindre end 5% chance for at resultatet skyldes tilfældigheder.
* P-hacking er når forskere efter udførelsen af et forsøg ændrer deres forskningsspørgsmål og/eller forsøgsdesign for at ”nudge” p-værdien over grænsen på de 0,05.

Question 38

Publication bias

Answer 38

Hvis mange forskningsresultater ikke publiceres, giver de publicerede resultater et forkert indtryk og kan være falsk positive resultater.

Question 39

Popperianske løsninger på P-hacking og publication bias

Answer 39

• Til at undgå p-hacking: forskere kunne evt. registrere deres hypotese og forsøgsdesign før de starter med at teste (som man skal i nogle kliniske studier)
• Mod publication bias: Tidsskrifter bør også udgive negative resultater.