Week 2: Reasoning & Philosophy of Science Flashcards
Reasoning
our conscious ability to make sense of the world around us to modify our understanding of the world around us by evaluating and using facts
Humans and animals do this all the time
Science history
historically knowledge was derived from authority, you have been a sponge, and the idea of systematically testing “truths” is often attributed to Galileo in the 17th century (though it was probably more common) e.g; Aristotle, Galileo and falling masses
Galileo’s Leaning Tower of Pisa experiment
a thought experiment in order to prove that all objects fall at the same rate, whatever their mass.
What are the 3 assumptions we can make about facts?
- Careful observation will lead the observer to collect an accurate fact
- Facts are independent of any theory (they lead to theory)
- Facts are firm and reliable (repeatability)
1 Careful observation will lead the observer to collect an accurate fact - how can this be influenced?
natural tendency to use our experience to make sense of what we observe. We see what our previous experience tell us/expects us to see. There is nothing wrong with this - this has guided some of the greatest scientific discoveries. You see what previous experiences tells you to see.
Search image
two observers do not necessarily see the same thing, search image is getting good at seeing something:
X-ray interpretation: Doctors practice looking at X-ray and to be able, to interoperate what others cannot.
Fish Eye colour: Dr Clare practiced looking at the fish eye colour until her ‘search image’ was able to identify eye colour because she got good at looking for something.
Observers accounts in criminology, medicine, and in the legal profession are notoriously bad.
2 Facts are independent of any theory
stair-case example, optical illusion doesn’t work on someone who has never seen stairs, they will only see lines on a page.
Having a concept of what stairs look like guides your ability to see and interpret the illusion.
3 Facts are firm and reliable
Facts change, our judgement about the validity of an observation is guided by what we know or assume. This makes facts just as fallible as our assumptions.
Scientific knowledge is based on facts established by observations. What is the role of experiments?
The art of an experiment is continual refinement. to improve the conditions or design/technology and see if the results are consistent.
Attempt to create objective observations rather than subjective ones, minimizing subjective observation and maximizing amount of independence from what you observe.
Reasoning is the way we make sense of facts.
What are the the most common forms of reasoning?
Inductive and deductive
Logic is very different to reasoning, although science is based around logic. What is logic?
When we use logic in common everyday ‘speak’ we are using it inappropriately. Logic is a set of rules, if something is logical it does not make it true. Logic is the basis of how we study “reasoning” but also math, economics, computer science etc.
Logic is not right or wrong, it is just a set of rule - nothing to do with correctness
Logical arguments:
- Philosophy is dull
- This lecture is on philosophy
Thus: - This lecture is dull
Is this logical?
Yes, but it does not make it true.
- A bird is a mammal This is also logical
- A crow is a bird
Thus:
3. A crow is a mammal
Is this a logical argument? is it correct?
This is logical, however it is not correct because premise 1 is incorrect.
We have drawn a false, but logical conclusion
Monty Python example:
All witches burn
All things that burn are made of wood.
Therefore, all witches are made of wood
All things made of wood have the same weight
All things that float have the same weight, all wood floats, a
All ducks float
Therefore if she weights the same as a duck she’s made of wood
Entirely logical argument, they have constructed logical argument yet it is false
What is inductive reasoning?
Build up a series of observations, which support a particular fact.
It is possible the fact is still wrong, we use probability to measure uncertainty.
All life forms are composed of carbon, hydrogen, oxygen, nitrogen, sulphur and phosphate.
All life has phosphate based DNA
This is alive
Therefore it is probable that it has phosphate based DNA.
Is this inductive? What is the probability that the next life form follows this rule?
This is inductive reasoning.
It is possible to find something that does not follow the rules - but highly improbable.
1.6 million species have been named, every-single one of them are phosphate based,
Probability of not being phosphate based in 1.6 million: 1
The Arsenic story
Phosphate and Aresenic have similar properties: atomic radius, electronegativity…
Biological form of P is PO33- which is similar to AsO43- if it gets into your blood will kill you
Mono Lake, eastern CA – Alkaline, this is a toxic soup of salt, with high concentration of arsenic.
Bacteria GFAJ-1, bacteria that lives in this lake, scientist found this and found that it could live off an arsenic base rather than phosphate. They argued that the case of life’s composition is proven false.
Is science mostly inductive or deductive?
Most science is inductive because we use a finite amount of available evidence to generalize about how all cases will respond.
Induction can never really demonstrate something completely.
How do we make good inductive conclusion?
*Make a large number of observations:
This can get ridiculous, should I burn my hand 100 times to demonstrate that fire burns?
True under a wide variety of conditions –
This can also get ridiculous; should I do it with different coloured socks?
- No contradictory cases – is this fair? There are very few rules that do not have exceptions
What are the problems with inductive reasoning?
- Humans see patterns: we have a perception of the world
* Humans are biased: we seek evidence to confirm our hypotheses.
Is inductive reason logical?
No, even if it may be correct your conclusion is based on observations
Observation ===|> Conclusion (NOT logical)
Copper example:
- Copper expands when heated
- Iron expands when heated It is however correct
- Gold expands when heated
thus
- All metals expand when heated.
This is not a logically valid argument
It is correct - but it is also possible to find a metal that does not expand
Animals can do inductive reasoning as well example, give example:
Two Monkeys Paid Unequally’
Monkeys understand that grape is unfair relative to cucumber, monkey able to reason inductively
What is the opposite of inductive reasoning?
Deductive reasoning
What is deductive reasoning?
You proceed from a general statement to a specific conclusion, this conclusion must be true given the previous statements – it uses logic! It is based on the principle of logic.
Draw the diagram
(page 5)
bottom up (inductive reasoning) series of observations -->
top down (deductive reasoning) laws and theories --. predictions and explanations
Why is deductive reasoning more powerful than inductive, from a logical perspective?
In logical terms, deduction is more powerful than induction. We deduce conclusion, but if facts are true then you are arriving at correct explanation.
What two criteria is deductive reasoning evaluated by?
Validity & Soundness
What is validity?
does it meet the criteria of logic
What is soundness?
are the assumptions correct
- A bird is a mammal
- A crow is a bird
Thus:
3. A crow is a mammal
Which criteria has it failed?
This is also logical but incorrect statement,
It has failed the criteria of soundness.
Do statisticians use induction or deduction?
Classical statistics, induction is used because we use samples, to try to know what most of it is going to be like.
Drug X – works on sample (clinical trial) and reduces the disease. this is inductive.
What branch of statistics is deductive?
Bayesian
Which fictional character is famous for his deductive reasoning? is it actually deductive?
Sherlock Holmes, but he doesn’t use deductive reasoning, he uses inductive reasoning because he goes from a general fact, series of observations, inductive reasoning, that leads to theories.
What is the Philosophy of Science?
Philosophy of science is just how we formulised our way in science through history
Philosophy of Science
Francis Bacon “new method”
Before Francis Bacon we were suing logical appraoch proposed by Aristotle.
New method is essentially inductivism, the process of building up evidence.
What are the four important ‘idols of the mind’, proposed by Bacon?
- Tendency to see patterns - evolution has led us to this as it is important to predict what is going to happen, but has the potential to overlook the observation and have false positives in research
- Personal preferences - that guide how you see the world
- Linguistic confusion – “theory” in science has completely different understanding in science then in public. Public expects us to have ‘laws’. General public things theory MAY not be true.
- Failure to question authority, assume what you hear or read is correct without questioning the source.
Philosophy of Science
Karl Popper and Falsificationism
He introduced a concept called falsification, what is this?
falsification advocates that all scientific theories should make testable predictions.
You should be able to extrapolate them down, set up to be rigorously tested, and to embrace the fact that they can be found false.
‘The point of Science is to prove the theory wrong and if you can’t find a way to falsify it then the theory is more powerful.’
Falsification allows you to test theories and rank hypothesis
Falsification
Hypothesis 1; All substances expand when heated
Hypothesis 2; You may find success today if you pursue a risky venture
Which can be falsified?
Hypothesis 1 can be tested! But the second cannot, it can be accommodated!
Hypothesis 2 can not be tested, very hard to prove they don’t because they are worded so vaguely that you can always find a way to fit it in
Falsificationists argue:
The best theory makes many specific predictions and runs the highest risk of being found false. Any new hypothesis should be more falsifiable than the one it replaces. Hypotheses cannot be “true” but can be more widely supported than a previous one. Allowing us to compare the relative strength of competing hypotheses. If you fail to contradict it then you have a powerful hypothesis
- Mars moves around the sun
- All planets move around the sun in an elliptical orbit
- All planets move around the sun in closed loops
Which hypothesis is the most sophisticated;
- because it makes three specific predictions! – most sophisticated!
What is the goal of science from a perspective of;
Inductivism
Falsification
Inductivism: the goal of science is to increase the observations which support an hypothesis
Falsification: the goal of science is to test the falsifiability of an hypothesis or to replace it with a more sophisticated hypothesis
Ad hoc modifications,
anti-aristotelians
Galileo’s telescope
Problem with ad hoc modifications is that they are untestable
Beware of ad hoc modifications…
Anti-Aristotelians advocated that all celestial bodies were perfect spheres, Galileo’s invented telescope and showed the moon had craters and mountains. His opponent modified his hypothesis to say that the moon was covered in an invisible substance which filled in the craters making it smooth.
Problem with this modification is that it was difficult to prove, hard to falsify. It was untestable (at the time) and thus could not be falsified.
Galileo responded by arguing that he was prepared to accept that the invisible substance existed, but that it was actually built up on the mountains making them even higher!
Ad hoc modification is only acceptable to Falsificationists if it generates a new set of predications and is thus itself falsifiable
What is the main problem with falsification?
Is it worth it to throw out hypothesis based on one piece of evidence!
A general hypothesis is usually based on a series of interrelated hypotheses, testing a prediction normally relies on a complicated experiment.
If the hypothesis is falsified, it may be impossible to decide why… a minor sub hypothesis. Technological problem?
You need to carefully consider what went wrong
What did philosophers argue about falsification?
Falsification was far too strict, it would be wrong to through out an entire hypothesis over a single piece of evidence.
The 1960’s = Theory based methods
The response in 1960s was the emergence of the idea that science should be based on an underlying theory, not the acquisition of facts
What are Kuhn’s Paradigms?
Pre science paradigm (the world is like this)
Normal science (paradigm not really working)
Crisis (not working at all any more)
Paradigm revolution (somebody proposes something better)
New-normal science (new paradigm)
New crisis
(Cycle repeats)
What is the problem with Kuhn’s Paradigm?
No way to evaluate which paradigm was better, he relied on the consensus of the community, until enough scientists agree with him.
By looking at the history of science one could see this is how everything shifted.
Lots of Lamarckism which then shifted to Darwinism
Lakatos’ research programmes
How did Lakatos attempt to fix falsification?
Strongly influenced by Popper, he attempted to “fix” falsification, not all theories are equal if you falsify a theory, you can modify “lesser” components and thus maintain the “research programme”.
Programmes can be replaced by ones that makes better predictions you can change them.
The core is often hard to find, in reality we do change them and essentially revolutions happened
This was in attempt to merge falsification with paradigms
What is Feyerbend’s anarchistic theory?
Anything goes, Feyerbend argued that scientists rarely follow a specific method, and to do so would constrain creativity. He did not think science should be given any special status.
He argued that scientists should pick any theory they want to follow and just pick something that works for you!
Against method
“The idea that science can, and should, be run according to fixed and universal rules, is both unrealistic and pernicious. It is unrealistic for it takes too simple a view of the talents of man and of the circumstances which encourage or cause their development. And it is pernicious for the attempt to enforce the rules is bound to increase our professional qualifications at the expnse of our humanity. In addition the idea is detrimental to science, for it neglects the complex physical and historical conditions which influence scientific change”
The Bayesians
Thomas Bayes, theory is mathematical, Bayesians rediscovered, by assigning probabilities to various hypotheses based on prior evidence. New types of evidence provide more support than additional evidence of the same kind.
You can use the probabilities to give you support with quantify previous evidence!
The branch of statistic that might be deductive
Problem with the Bayesians?
probabilities can be subjective and thus make them do what we want!
Where else does Bayesians theory come up?
Phylogenetics, We can predict probability of gene changes, we can accurately predict them
Mayo’s experimentalism
What is experimentalism?
NOT THE SAME THING AS CONDUCTING AN EXPERIMENT
Experimentalists have practical methods of measuring effects without the need for theories. It is just a measure of effects, not theory. Progress in science is the accumulation of experimental data. This likely encompasses what we call applied science and the science of error statistics.
How do we reduce error?
What is New experimentalism ?
New Experimentalists advocated;
Experiments can be independent of confirming a theory, they can focus on ruling out factors, accuracy of estimates, detecting artefacts or error in evidence’s
Evidence is basis of argument, thus we should suspect observations, observations and knowledge can be retained even if theory changes.
Testing the parts and performance, trying to improve the outcome.
Mayo’s Version
Concentrate on experimental design, modelling, data analysis, statistical methods.
Beyond standard statistics we need to focus on how these increase reliability of observations.
We need to design analyses with a high chance of detecting error
What experiment can you plan an detect error.
Bacon –
inductivism with an idea of bias
Falsification –
theory depended theories are those that survive test
Kuhn’s Paradigms –
consensus of the community on what paradigm to use
Lakatos Research Programmes –
protect core ideas and modify peripheral hypotheses
Feyerbend –
rejects “method” in favour of theory
Bayesian’s –
use probability to evaluate competing theories
Experimentalists –
it is possible to use method to evaluate data independent of theory
Occam’s razor vs. Sherlock Holms
What is Occam’s razor?
Occam’s razor; when you have multiple hypotheses, the one, which has the fewest assumptions, is the more likely.
In science ‘parsimony’; simplest explanations are best – more “falsifiable”
Goals of science
- To gather an understanding of cause an effect relationships
- To predict future event based on past observations
- To explain mechanisms
Methods of her science
- Make observations
- Explain observations in light of theory
- Make predictions which test theory
- Be willing to acknowledge sources of error
