[Return to X200 Home Page]

Outline for Lecture #3: Three Phases of Scientific Reasoning

Overview: Scientific Reasoning in a Nutshell

Phase 1: Why P?
- where P is some puzzling phenomenon

Phase 2: Maybe H!
- where H is an hypothesis such that
if H were true, then it would explain P

Phase 3: But is H true?
- do independent tests of H
- weigh the evidence for and against H

Phase 1: Why P? Curiosity Phase

The Choice of Scientific Problem is influenced by:

Intellectual Puzzlement
- violated expectations
- deep explanation of a regularity

Practical Significance
- instantiate sufficient conditions for desired
outcomes
- find a removable necessary condition for
something we want to prevent

Strategic Factors
- availability of equipment, know-how, funding

Phase 2: Maybe H! Creativity Phase

Philosophers often call this stage in the development of scientific explanations the "Context of Discovery". It is structured in the following ways:

The Typical Explanatory Problem: to find an H (explanans) that would give us reason to expect P (explanandum)
- H often describes a causal process that
results in P and there is a valid deductive argument from H (perhaps in conjunction with more or less trivial auiliary premises) to P

"Hard" Constraints on H:
- it should be consistent with everything else we
know about the world

"Soft" Constraints on H:
- it should be "plausible" given what else we
know

It is difficult to describe exactly where bold new Hypotheses come from. Sometimes analogical reasoning is involved.

Phase 3: But is H true? Critical Phase

Philosophers sometimes call the evaluation process the "Context of Justification".
Here we can give a much more precise account of the reasoning involved:

Basic Logic of Testing:
- derive a new prediction from H (call it P*)
- do an experiment to check the truth of the
prediction

If the prediction is false then we use the following valid deductive argument which is called "Modus Tollens":
- Suppose H implies P*
- But we observe not-P* (i.e., P* is false)
- Then we must conclude using deductive logic that not-H (i.e., H is false)
(We say that H is refuted.)

If the new prediction is true, then we can sometimes - but not always - 
use inductive reasoning to conclude that H might very well be true:
- Suppose as before that H implies P*
- In this case we observe P* (P* is true) 
- No valid deductive argument leads from P* to H 
(after all their may be other possible explanations of both P and P*
So we can conclude nothing deductively. However, in certain cases
 (for example, if there is no other plausible explanation available)
we may say that P* confirms or gives inductive support to H.