Making Theoretical Progress
Deduction vs. Induction
Deductive (top down) reasoning:
More narrow
Concerned with hypothesis testing
Inductive (bottom up) reasoning:
More open-ended
More exploratory
Every swan I see is white so all swans are white (bottom up) --> specific observations --> patterns --> develop theory
All swans are white so the next swan I see will be white -->
Theory --> hypothesis --> observation --> confirmation
Problems with Verification
20th Century Approach: What is Science?
Critical Questions:
- What's unique about science?
- What is science? What is non-science?
- Can the scientific method be identified & applied to other fields of knowledge?
A group of philosophers discussed these questions --> logical positivism
What are the demarcation criteria that define science?
The Vienna Circle: Logical Positivism - Verification, observation, induction --> demarcation of science (proceeds through cycle)
Translating individual observations into general conclusions --> demarcation
A statement was only regarded as scientific if it could be verifed as true or false through objective observation
Falsification (Popper)
Verification is logically impossible
Induction Problem: Just because you make a lot of observations coming to the same conclusions, it doesn't mean this will always happen
Unobservable facts: Overtime, unobservable things may become observable e.g looking at the atom
Observations may be wrong: e.g. illusion of the lines that look different but are the same
Observations not objective: Using verification for the demarcation of science is problematic - logical positivists were called naive & called into question
Falsification instead of Verification
Austrian born scholar
Member of the Vienna Circle
Initial interest in psychology - 'I can't account for the essence of scientific progress
Popper
Falsification instead of verification
Science constantly questions its explanations
Considerations the differences between physics & Freudian Psychoanalysis
Falsifiable or not?
Hypothetico-Deductive Method
Degrees of F.
Criteria for Choosing Theories
Kuhn's Theory of Scientific Progress
- It never rains on Wednesday - yes
- Kennedy died from a bullet - yes
- If he'd lived, Kennedy would have ended the Vietnam War - no
Physics: Progress through finding weaknesses of theories
Psychoanalysis: Progress through finding corroborating evidence
Falsification differentiates science from pseudo science
Verification is impossible to prove in an inductive conclusion - easy to disprove
Observation --> Interpretation --> Hypothesis -- > Test
Based on inductive reasoning
Make sure scientific explanations are scientifically grounded
The more falsifiable a theory, the higher it's scientific status
Example - Louis Pasteur - why do wine, beer & milk sour?
Due to organisms in the air --> test
Provide evidence for theory using filter to prevent organisms entering
E.g. 1. Wine sours from organisms
- Wine sours from bacteria in the air
- = more falsifiable because it has more specificity --> more specific = more prone to falsification = higher scientific status
Scope
A good theory is one that makes wide ranging claims with a wide scope
Example: 1. Mars moves in an eclipse around the sun
- All planets move around the sun
Precision
- has a higher status as it tells you what the first said and more
The more precise a theory the more falsifiable it is
Example: 1. Planets move in ellipses around the sun
- Planets move in closed loops around the sun
Oval orbits would falsify 1st not 2nd
ANY orbit falsifies 1st AND 2nd
Operational Definition - Describes exactly what the variables are & how they'll be measured --> e.g. are toddlers happier in warm or cold weather? Operationally define by testing how many smiles in each
Parsimony
Among competing theories, the one that makes the fewest assumptions should be favoured. Occam's Razor --> simple theories prefered
Example: Event - it's raining & I saw a bright flash
- Lightening
- Someone is taking a picture of me
- Is simple - so it's preferred
Increasing Falsifiability
The more falsifiable the theory the better. More content, more informative
A replacement theory should be more falsifiable than what it replaced
Example:
- All bread nourishes
- All bread, except a particular batch nourishes
- All bread nourishes except that batch due to a fungus
3 Is best --> modified theory is not ad-hoc
Fruitfulness
Has the theory --> new empirical discoveries --> scientific progress
Example: Theory = Bats have weak eyes but can see
Test = Blindfold bats & release
American psychiatrist - Science is a social activity
Scientific paradigm --> Set of common views of what the discipline is about & how problems should be investigated
Introduced the term scientific paradigm - any scientific achievements were part of a paradigm --> stages
Pre-science --> normal science --> crisis --> revolution --> new normal science --> new crisis
A paradigm determines:
- What is to be observed & scrutinised
- What kinds of questions should be asked? etc.
Kuhn- Structure of Scientific Revolution
Each discipline starts with facts to explain small scale phenomena
Researchers share paradigms - attempts made to falsify the theory - modifications can be made
Revolution - confidence in paradigm decreases, crisis, scientific revolution & paradigm shift
Does a revolution really mean progress or does it just mean change?
Puzzle solving NOT problem solving
Kuhn:
- Normal science = important
- Scientists should criticise occasionally
- Conformity & puzzle solving
Popper:
- Critical is best
- Criticism & non-conformity are essential
Problems
Too vague
How long should scientists stick with a theory in the face of trouble spots
Why would working with a theory for a long time improve the chance of identifying limitations
Both Popper vs. Kuhn are opposite extremists on the issues
Popper --> all scientists should follow same critical procedures
Kuhn --> should puzzle solve
Middle Ground - Constant coupling of ideas to observations by using verification & falsification is a guarantee that the ideas will not be completely in contradiction with the reality as it can be observed = Rule of Thumb - look at converging evidence