What Is This Thing Called Science?
My notes
What do I want to get out of the book? (Shopping list):
- [ ] What is a fact that forms a science?
- [ ] What distinguishes science from pseudoscience
- [ ] How to distinguish scam from intuition
- [ ] What does science say about believing
- [ ] Where is a place for believing
- [ ] What is theory?
- [ ] What is a statement?
- [ ] What does make a good experiment?
- [ ] What is inductivism?
- [ ] How do we derive law from facts?
- [ ] What is falsification?
- [ ] How to tell if a theory is falsifiable.
- [ ] How do induction and falsification play together?
- [ ] What does it mean for Theory to be a structure?
- [ ] What is normal science?
- [ ] What is paradigm
- [ ] What does the Bayesian approach have to do with science?
- [ ] What is Subjective Bayesianism
- [ ] Is what we are looking for laws?
- [ ] What is realism?
- [ ] What is anti-realism?
- [ ] What is Scientific realism and conjectural realism?
- [ ] Where is the place for philosophy in science?
- [ ] What is axiom?
- [ ] What is empiricism?
- [ ] What is positivism?
- [ ] What are the differences?
- [ ] How does science solve the issue of the Bayesian Brain, which means that there is no objective perception as everyone has different beliefs?
Induction lets us derive principles and laws from facts obtained directly through perception.
Those principles and laws (also called scientific knowledge) can be then used to estimate and predict other facts. As well as, to obtain new scientific knowledge via deduction.
perception(world) -> facts
induction(facts) -> scientific knowledge
deduction(scientific knowledge, initial conditions) -> scientific knowledge
Facts are claims about the world that can be directly established by senses.
Scientific knowledge can be used to predictions prediction(scientific knowledge) -> predictions
.
Example
- Observing thermometer and state of water.
- Gathering a large number of observations under a variety of conditions, air pressures, etc. in such a way that no one conflict with the desired law. Inducting a law that wather change it's state at 0 and 100 degrees.
- Getting the laws of state of water, together with initial conditions like, the water, air pressure, and drawing prediction that at 100 degree the water will boil.
Inductive reasoning is neat but naive. Even if we gather large amount of facts (singular statements) that are true, then we can not state universal statement based on them, as we can not be sure that the next fact will not contradict with the conclusion.
Even if all observations are true, there is no logical guarantee that the next observation will also be. If all observations are true, then the conclusion that not all observations are true, is valid. Because we can not observe all instances.
Deduction does not work in stohastic world.
[[Inductive arguments can not be justified]]
Any finite number of observational statements, can not justify general law of infinite possible cases.
Even if we use probability, a large (but finite) number of observations, divided by (infinite) possible cases, results in zero probability.
Falsification is based on the fact that specific observations can falsify universal statements by finding counter arguments.
Hypothesis is falsifiable if there can exist logical observational statement that falsify the universal statement.
Examples of statements that can not be falsified are:
- It is raining or not.
- Each point in the circle is of the same distance from the center.
- You may be lucky in tomorrow lottery. The first sentence can not be falsified as is true no matter of the actually weather. The second sentence is always true because the definition of a circle is like that. The third sentence is used in astrology and is also always true no matter what.
Falsification require sentence to be falsified because is says something about world by excluding some logical observation. If the senstence is in falsifiable then the world can act randomly and it won't falsify our sentence.
In similar way Adler's feeling of inferiority and Freud's psychoanalysis are infalsiable.
Adler claims that every human action is driven by [[Feeling of inferiority]].
In scenario when child get into river he may jump to save it or not jump. In case he jump, Adler says that is supports his theory and says "he jumped because he felt inferior and wanted to overcome his feeling by being brave." If he don't jump it also suppers Adler theory because "he overcome his feeling of inferiority and didn't jumped into water. Showing his courage to be indifferent to other people needs."
[[Theories that want to explain everything does not explain anything]]
If theory wants to be scientific it has to be falsified.
Levels of falsification
Two sentences:
- Mars moves around the orbit of Sun.
- All planets moves around the orbit of a sun.
The second one is of higher level of universality and hence there are more observations that can falsify it this it's of higher level of falsification, thus better.
Set of observations which lead to falsification are called potential falsifications.
And hence, Kepler theory explaining movement of planets was replaced by Newton's theory which explained all movements and gravity.
Newton theory had larger set of falsifications yet, it survived the time challenge.
[[Scientific work is about proposing hypothesis and then working hard on falsifying it]]
This is hard truth as it often wants some working theory than proving that no theory survived.
Yet it's the only way to move forward, we are learning on our mistakes only.
[[Science is moving forward by trials and mistakes]]
Falsification encourages to state brave hypothesis, because if they fail (and will) then we learned. If they don't fail then we moved forward. There is no danger of popularising flawed theories as they will be ruled out by falsifications.
For falsifications, science progresses by falsifying theories and proposing new ones that can not be falsified by those examinations. No theory is true,, but succeeding ones are better than the previous ones.
New radical theories are encouraged because if they are wrong they will quickly be rejected via examination.
For falsificationist, science comes from problems. Problems in context of the old theory that can not explain new observations.
The difference between naive inductivenest and falsificationist is that the first derive science from observations and the second one from problems. Inductionist gather many similar singular observations and state universal statement. Falsificationist on the other hand observe nuances that can not be explained using old theory and thus propose new theory that can explain it.
How to use falsification in research?
Let say we face a problem, some nuance that the theory can not explain.
Bats can fly at night and dodge collisions and hunt insects, however they have weak eyes.
Current theory says that all animals including human, use eyes to coordinate in reality.
Proving that it's false for bats would falsify the theory.
The falsificationist pose a hypotheses "although bats have weak eyes, it's enough to see the obstacles"
They cover the bats eyes and see what happens. They set the dependent variable to 0 and see what happens.
Before experiment we conduct the deductive reasoning:
- all animals including bats can dodge collisions because of their eyes and without eyes they would not be able to do it.
- The sample of 🦇 were closed 👀
Deductive conclusion:
- The sample of bats will not be able to dodge collisions
It turns out that nothing happens, bats are dodging collisions with and without eyes. The theory is falsified. But let's find out why and propose a better one.
To propose better hypothesis we have to use imagination. Someone suggest that maybe it's the 🦇's👂 that let them navigate.
They close their 👂 and observe that 🦇 can no longer avoid collisions.
Hypothesis supported, but now, falsificationist has to make it more precise. It says "Bats are using their mounth and ears to navigate".
Reserachers close their mounth and observe that once again, bats can not navigate well.
He is close to formulate the thesis, but his reserach does not prove how exactly the bat navigate. Maybe they don't use ears, but some device close to ears that was also coverted. Maybe the sample is not representative and reserachers used some specific set of bats.
Progress from Aristotele to Newton to Einstein was very similar. Aristotele physics was correct to some degree. It explained why things go down (they seek their natural place in the universe), it explained how siphon and pumps work etc.
But the theory didn't explain many things:
- Why rock thrown from the high mast felt down close to the mast even though the ship was moving very fast.
- Why Jupiter's moons round around Jupiter, not Earth.
In 17th century Galileo and Newton proposed new theory that explained everything that Aristotele physics explained and many more, it lead to discovery of Neptune, and Moon's influence on sea.
Newton's theory survived 200 years, no one was able to falsify it.
But the theory didn't explain:
- Details of Mercury's oribit
- Chaning mass of electron.
Einstein was the one that proposed new relativity theory. His specific relativity theory explained mass as a function of speed m(v) and that mass can be transformed into energy and reverse. His universal relativity theory expected bending light close to high mass objects. Currently people try to falsify Einstein theory.
This is how the progress in science
In falsificationism, confirmation are valuable because they support new theory that replaced the old one.
The value of a confirmation of a theory depends on the state of the art. Confirmation of something radically new is valuable, but as the time and number of confirmations passes it's significance decreases.
For falsification theory can not be proven, it can be only falsified. It can be supported by many observations, but in context of infinite number of observations they are always of probability of zero. [[Acceptance of a theory is always temporary, falsification of theory may be permanent]]
Empirical base of objective observations has nothing absolute. [[Science is not build on solid ground. It's build on stilts put into a quagmire]]. They are high enough to stay still but fundamentally does not touch any concrete fundaments.
Theory fanboys will always try to defend a theory by stating that the anomalies, nuances were correct but the theory can explain them also, they state adhoc hypothesis which explains the nuances and try to prove them. If they fail, they state different hypothesis and repeat and repeat until no one care about them. And that's how the new theories surpass old ones. Because no one is able to protect the old theory.
There are some numbers that are special, and may describe reality better than others. Like pi, golden ratio, or four, the number of proteins.
Copernicus theory needed houndreds of years to be accepted not because scientists were stupid, stubborn, but because it was not obvious. The telescopes were so poorly made that it was not clear whether the observations are correct. There were still some problems with Copernicus theory which could not be answered, why would this new theory be better than the current Aristotle's one? We needed Galileo and Kepler and Newton work to support this theory.
Both inductionists and falsificationist are not able to explain great historical breakthroughs and progressions.
Copernicus Revolution was not created via observations and then inducing universal statement. Not it was made by proposing radical hypothesis which was not falsified.
The biggest breakthroughs have structural character, which can not be explained by any theory.
Why is that? Observatory sentences has to be formulated in a language of some theory. As a result they are precise and meaningful as the theory which provides the language to express them.
Theory which is based on word "democracy" is as precise or vague as the word democracy.
One of the ways of giving terms a meaning is via definitions.
Terms can be defined using different terms which meanings are given. If the meanings of those others terms are given by definition, then we have a problem of infinite regression.
Dictionary is useless if you don't know many words.
Newton could not define mass using terms that already existed. He had to surpass the old system and create a new one.
We can not define a mass by observation. Likewise colour.
Terms get a meaning by their role in a theory.
Terms can be defined either by a definition or by observation.
Galileo did most of his experiments in his mind. He did many thought experiments.
Empirists would argue that only experiences can lead to theory.
He was experimenting mostly with analogies, visual metaphors, thought experiments rather than concrete experiments.
[[Creativity workbench]] is exactly what the scientist were doing.
The strict experiments were possible only when the theory was formulated.
[[Strict experiments are possible only when there exist strict theory]]
History shows that each new term start with vague definition and vague theory. With time it's definition gets more precise and so the theory.
Theory can be developed if they are hints on how to develop it.
Scientific Research Programmes
Imre Lekantos proposed that theory is a set of organised structures in "Falsification and methodology of scientific research programmes"
It's theory is an answer to Popper falsificationism.
Negative heuristic says that the fundamental assumptions can not be rejected nor modified. They are protected from falsification by the safe belt of supportive hypothesis.
Positive heuristic says how the program can be extended and developed.
Scientific programmes can be progressive or degenerative, depending on how well they are doing in explaining what happened and predicting what will happen.
The immutable core of the theory are assumptions that can not be changed.
For example Copernicus Theory assumed that sun is still and planets are going around the sun. Earth is spinning once per day.
Newton's they assumed the laws and principles of force and gravity.
Marx's historical materialism assumed that social changes can be explained in terms of war of classes, and that economic base of latest instance determine the importance of class, and elements of the war.
What is economic base? What are elements of war?
The immutable core of the theory is unfalsifiable because of the methodological decision of its supporters. If the observatory data, and facts, do not match with the theory, it's not because of the immutable core but because of the other supportive thesis.
Ad-hoc hypotheses are those that says that there is an exception which can not be explained using the theory or "those rules does not apply to this one instance".
Kunh Paradigms
Kunh proposed new theory of philosophy of science. It was close to Lacaston's theory. The difference was that Kunh put much emphasis on sociological aspects. The so called "sociological relativism" is what is emphasised in his theory.
According to Kunh, the development of science goes through following steps:
- Pre-science, [[non-organised and differentiated activities that lack paradigm are not science]] (e.g., sociology can not be science as it lacks paradigm)
- Normal science, happens when scientific association agree on one common [[Paradigm consist of general theoretical assumptions, principles and techniques of using those principles to measure and predict reality|paradigm]]. When scientists research reality using paradigm, they are doing normal science. Scientists move science by formulating and developing paradigms by trying to explain more and more phenomenons in reality.
- Crisis, when they can't, crisis arises. It is resolved when new paradigm that solves all the issues arise.
- Revolution, when new paradigm offers solutions to all the unsolved problems lure scientific circle.
- New normal science, when the new paradigm becomes the normal one;
- New crisis, and the cycle repeat.
Science is ruled by single paradigm which coordinates and rules problem solving.
What makes science scientific is the existence of paradigm. Lack of paradigm means lack of science.
The nature of paradigm is that it can not be strictly defined. However all of them have common properties:
- clear principles and theoretical assumptions (e.g., max speed of light, Maxwell's equations, Newton's rules, everting everything in physical reality can be explained as a mechanical system).
- typical examples of how to use their principles (e.g., prediction of the movement of planets)
- General methodological suggestions like: you should try to match your paradigm to nature, lack of match is a result of error in theory.
The goal of normal science is to formulate detailed paradigms which aligns with nature.
Paradigm is always imprecise and open, so that there is always room for improvement.
The goal of normal science is to solve puzzles using the paradigm.
Puzzles can be both theoretical or experimental.
For example in Newton paradigm
- the theoretical puzzles were to invent mathematical techniques that explain movement of planets on which multiple forces apply and to. Or developing assumptions that would allow for explaining fluid physics using the paradigm.
- the practical puzzles where to invent telescopes and measurement's techniques that would allow for measuring.
Scientists must assume that the paradigm can solve all those problems.
Inability to solve the problem is assumed to be the failure of scientists nor the weaknesses of a paradigm. Puzzles that can not be solved are considered anomalies not paradigm falsifications.
Every paradigm consists of some anomalies. In Kopernik theory the anomaly was wrong size of Venus. In Newton paradigm Mercury's orbit was an anomaly.
The difference between immature pre-science and mature normal science is that in the former scientists should be constantly arguing on its correctness while in latter scientists are not doubting in its fundamental correctness.
In the lack of normal science there is the same number of theories as scientists in the discipline. Each one has to explain their statements.
Example is optical science which— before Newton proposed corpuscular dualism—could not reach an agreement.
Paradigm is hard to define precisely. Similar to how hard it is to call something "game". Wittgenstein stated that it's impossible to define what game is using common rules and definitions. However similar to how games are hard to define, nonetheless does exists and the term is understandable in common sense, same with paradigms.
Even though the paradigm is hard to define, scientists can understand paradigm in the process of careful education.
[[Even though something can not be strictly defined doesn't mean it's useless]]
Significant amount of knowledge in normal science scientists is implicit or hidden in common sense.
Polanyi's Personal knowledge explain it.
Scientists should not blame the theory if they can not solve some problems. They should doubt their skills and try the best they can. Same as craftsman should not abuse his tools because it can not create a masterpiece.
The existence of unsolved problems doesn't not lead to crisis. There are always anomalies.
The problem is however when the anomalies touches paradigm's fundaments, and constantly can not be solved by the community.
The significance of anomalies is influenced by the social demands. People need some answers more than others. Therefore if the anomalies are in the area of social demand, it can lead to crisis. An example may be the need for calendar.
Significance of the anomaly is also influenced be the period can not be solved.
When the anomalies are significant, scientists start philosophical and metaphysical disputes. From the paradigm point of view they may have low value.
Scientists start to argue and express their insecurity, impatience, and instability. The governance of the paradigm collapse. We have a revolution.
The Revolution leads to deep contemplations, contemplations which leads to completely new paradigm or just a new idea significant enough to allow for formalisation.
New paradigm sees a world s as an object made by different things.
Maxwell's Electromechanical theory assumed the existence of ether, while Einstein completely got rid of it.
Different paradigms leads to different questions. Questions about the weight of the planets were useless for Arestoteles paradigm, while for Newton's they were crutial.
Paradigms drive the way scientists look at the world.
Scientists using different paradigms live in different worlds. We observe things that are governed by paradigm. Of paradigm does not allow for some things we can see them.
That's why no one saw anomalies before Copernicus posted his theory.
We need the lens, glasses of a theory too see it.
There are no rational reasons to change paradigm.
[[leafs are as valuable as roots]]
Some change it because of the simplicity, fine because of social needs, usefulness. Someone will reject the new paradigm because of troublesome implications, religion,
Rationalism vs relativism
Rationalism states that there exists only one timeless and universal criterium which can be used to judge theories.
For instance,
- Inductionist beliefs that such criterium is degree to which a theory has been supported by facts.
- Falsifications beliefs that such criterium is the degree of falsifiability non falsified theories.
Rationalist beliefs at scientists use this criterium to judge theories, and will choose the one that is better.
Inductions beliefs that astrology is not a science because it can not be induced by facts. Falsifications beliefs that Marxism is not a science because it can not be falsified.
Relativism states that there is no universal and ahistorical criteria. Decision whether a theory is scientific or not is based on what the society values and sees as important.
According too relativism there is no rational criteria to judge theory. And so we can not say that one is better than the other. [[Better and worst]]
[[Relativism says there is no universal criteria to judge some theory better than other, rather better is what is more valuable and important to community]]
For instance, western capitalistic societies values control over nature. Other cultures thinks that knowledge and science should bring well being, piece, calmness.
Pitagoras once said "Man is the measure of all things"
Not good, not moral law. But human is the ultimate measure. Which makes sense to what we know now with the [[Bayesian brain or Perception = Sensory + Beliefs]]
It is relativism to human beings. Truth is relative to human being.
Kunth said "there is no higher standard than agreement of the community of interest". It's relativism to human communities.
Relativism - Truth is relative to X.
Same applies to what is science and what is just pseudo-science.
If there is no way to measure a theory other than approximate the number of supporters, then the truth is power. [[Truth is power]] [[Power is the truth]]
Change of the theory depends on the psychology of mass.
Progress of science is a result of propaganda.
Change of theory is similar to change of religion faith.
We can not distinguish heresy from truth.
Kuhn is relativist. He believes that the way to compare theories is to measure:
- correct quantitative predictions,
- balance between esoteric and everyday problems,
- number of different solvable problems.
- Simplicity, range, interoperability with other specialties.
Values of those measurements are considered by scientific community. Measures are based on psychological and sociological analysis. So it's based on value hierarchy and ideology.
There is no higher standard than agreement of some community
[[Truth is what scientific community agrees on]]
Kuhn theory is relativistic.
The distinguishing between science and not-science is level to which the activity can withstand the normal science.
It's extremely hard to distinguish science from non-science.
There are questions that we have to answer:
- can one theory be closer to truth?
- Can one theory be better at solving problems?
- Is a better instrument of predicting?
Objectivism
Objectivism is a position that knowledge (simple sentences, to complex theories) surpasses human believes.
Objectivism is the opposite of individualism; believe that says that knowledge is constrained and expressed in a form of human believes.
We know only as much as our believes allows us to know.
Individualism
From individualistic point of view, knowledge is a set of believes that changes over time.
Knowledge is a set of correct believes that are documented or something similar.
Knowledge must be based either on correct sentences (deductions) or correct facts (inductions). Both are flawed. Sentences are based on other sentences. Facts are based on observations but who says if they are correct?
Knowledge must be based on fundamental knowledge that we assume is true.
For rationalists they are axioms used to derive other sentences.
For empirist they are sensory information. Which are then used to build theories by induction. // TODO: check what original says and create a note.
[[From Blast point of view, knowledge is a set of correct thesis and theories; allowing for predictions and stacking up]]
[[From Consume point of view, knowledge is a set of correct information]]
[[Experiment is planned, theory-based interaction with nature, with the intent to investigate and test a theory]]
[[Objective knowledge is knowledge without the subject who knows]]
[[Subjective knowledge is knowledge with the subject who knows and holds the knowledge]]
[[What distinguishes a respected thinker from a crazy man is not his initial theory, but the level of details, awareness of the hardness, familiarity with the state of the art, awareness of the raised allegations]]
There is a separation of government and religion but there is no split between government and science.
[[Realism says that theory describes reality. Instrumentalism says that theory gives instruments to associate theory with observable things and make predictions]]
[[Theories can not be compared for the same reason why sentences in different languages can not be translated. They describe phenomenons that does not exist in other cultures, and ecological niches]]
The only way to compare theories is via confronting them with observable situations and registering degree in which they overlap with what they categorise.
Another way is comparing theories is to rate whether they are linear or nonlinear, coherent or not coherent, dare or approximate.
At the end, when we want to choose theory, we are left with subjective choice. It's because criteria lacking content makes them decision non rational and non objective, becomes discussion, includes conflicting preferences, and most of the work plays the propaganda.
Then we are also left with rating aesthetics, taste, metaphysical biases, religious needs, beliefs or subjective preferences.