Theory-ladennes of observation
what we observe is influenced by the theoretical beliefs, assumptions, or expectations we already hold
Verificationism
- Abduction: derive theory from data
- Deduction: deduce testable hypotheses
- Test hypotheses against the world
- Induction: observations that meet the expectations of a theory verify it
Induction in science
Reasoning from particular claims to general claims/ from known cases to unknown cases.
Principle of uniformity in nature
future will resemble the past, and that similar powers will be conjoined with similar sensible qualities
A Priori Justification
A statement is justified a priori if its truth can be established merely by thinking about it
Contradiction is Absent: We can easily conceive that things in the future will be different than they have been in the past
A Posteriori Justification
Justify PUN by showing that it is supported by experience
But would need to infer from experienve that the future will resemble the past
Justifying PUN by appeal to experience is circular
Hume’s Problem of Induction
If Hume is correct;
We lack a rational basis for assuming that observed patterns will continue to hold
Popper’s falsificationism
Science does not aim to confirm scientific theories through experience and observations → it aims to falsify them
Theory survives attempted refutations ⇒ corroborated
Define corroborated
- It has withstood detailed and severe tests.
- It has not been falsified by experience.
- It has not been superseded by another theory in the course of scientific progress
Empirical content
This is defined as the set of observable states of affairs that a theory precludes
Boldness
the bolder a theory the more empirical content → probability decreases → greater the predictive power
Theoretical superiority
A theory T₂ is superior to T₁ if T₂ is corroborated and T₂ has greater empirical content than T₁:
T2 must be riskier and more informative by ruling out more possible observations.
A good scientific theory is one which:
- is precisely formulated and easily testable
- has empirical content
- is bold
- has not been immunised on the basis of ad hoc assumptions
Holism about testing
- impossible to test, hence falsify, hypotheses in isolation (in conjunction with numerous auxiliary hypotheses)
- observation report contradicts a prediction → some assumption is false (not sure of which)
Describe Lakatos’ Research Programmes
The hard core: central assumption
Protective belt: auxiliary assumptions
Lakatos’ Heuristics
Negative heuristic:
- hard core of a theory must be protected
- falsifications directed towards the protective belt
Positive heuristic:
- progressive modifications of the protective belt (after results contradict predictions)
Progressive Modifications
Modifications are considered progressive if they increase the hard core’s predictive power
Degeneration
If a research programme fails to be progressive
Abandonment Criteria
Research programmes are not abandoned because of one falsifying observation, nor immediately upon degenerating
Research programmes are abandoned only if they degenerate for long enough AND can be replaced by new, progressive programmes
(4) Types of Inductive Reasoning
Statistical argument: infer from relevant statistics
Inductive generalisation: most observed cases have had a certain property, so future cases will, too
Analogical argument: two cases are similar in a certain way -> infer more similarities
Inference to the best explanation: have a body of facts, and aim to infer the best explanation of those facts
Criteria of Strength of the types of inductive arguments
Statistical argument: Is the result statistically significant
Inductive generalisation: similarity of past cases to current/ future cases? Properties of the sample?
Analogical argument: extent of the resemblance? Do they resemble each other in the most relevant ways?
Inference to the best explanation: Does the hypothesis explain the data well? Are there better explanations?
Practical guidance
Corroborated does not mean a theory is true, just that it hasn’t been falsified (yet)
Corroborated theory = untested theory
Choose corroborated theory, but then you’re using induction
