More Details of 
Scientific Method
Chapter 16

Reaction to Failed Tests

If the desired results are not obtained when we follow a known recipe or procedure, we then look for

1. an incorrectly described or followed step in the recipe,
2. a malfunction in the equipment,
3. an incorrectly measured ingredient,
4. an ingredient polluted by a foreign substance, or
5. a factor not previously considered.

The foregoing may identify a remedy or leave a puzzle.

Cause and Effect 

When a reproducible operation is under way or running, nothing unusual happens. If we introduce a disturbance or do something to affect the process, we may see a departure from the ordinary or usual behavior. So we should strongly suspect that the departure is (most likely) due to the disturbance. This suspicion is tested and confirmed if the departure is repeated (several times) whenever we make the disturbance re-occur, or we may be fooling ourselves — always a possibility.

Through trial and error, we may look for a disturbance producing or causing a variation in behavior which we want to keep. This disturbance can then be made a normal part of the process. A new controlled situation results. Experimentation and fine-tuning of the process at hand can continue.

When introducing a variation in the operation of a mechanism, care needs to be taken to ensure the variation is the only one that is done. When two disturbances are introduced simultaneously, one unknown to us, we may think a variation in behavior is due to the disturbance we saw. But the variation could be due to the other disturbance or to the fact that both disturbances occurred simultaneously. Pattern recognition is not always straightforward.

Confidence

Confidence in old and new rule-based processes may be built through repeatable and reproducible experiments and observations. Experiments and their results are accepted if they can be reproduced and repeated by others besides their inventors. For this, the inventors or first discoverers of a phenomenon must carefully record the method, art or recipe used to get their results. With no such record lengthy or cryptic, results are subject to argument and doubt.

Making Theories and Predictions

Rules (and suggestions) can be linked together to get or suggest further rules. That is, combining a rule which says when a first situation A occurs so will a second B to another rule which says when the second B occurs so will a third C gives a new rule: when the first situation A occurs, so will the second B and the third C.

By combining reliable and not so reliable implication rules together, we can make predictions. The chains of reason by which the predictions are made are called theories. When a prediction fails, at least one part of the chain of reason leading to it is uncertain. Knowing which parts of the chain are the most certain and which are the least may suggest a new course of action: informed trial and error. In contrast, when a prediction succeeds, we may become confident or overconfident in the chain of reason which suggested it.

Proposed implication rules can be shown to be false. They can never be shown to be true. As a result, rule-based reason depends on implications rules and assumptions supposed or pretended to be true and reliable. 

Prediction versus Self-Delusion

In physics, amazingly long chains of implications and approximations are used for predictions. Because of approximations or suspect implications, the reasoning is unsure. The predictions need to be tested. A prediction gives a value for a number or quantity. The value could be one of many. If the observed value is far from a predicted value, a step or two in the creation or derivation of the prediction must be wrong. The theory needs repair. Confidence in the predictive method/theory is increased when the predicted value and the observed value are close.

In physics, confidence grows in a theory (one long chain of reason) if it correctly predicts one result, not seen before. In physics, those making the predictions are the theoreticians while those making the observations are the experimenters. [3]  

[3] A division of labor has occurred between scientists making predictions (the theoreticians) and those doing the experiments (the experimenters).

Implication rules and theories become more trusted (or useful) when predictions based on them are seen to be true or very close to the observed values. Successful theories are hard to find. But when a theory gives a good prediction, variations of the theory may be tried to get other predictions.

Avoiding Self-Delusion

A theory can predict a future event. If the event occurs, confidence in the theory grows. A theory can also suggest a value or values for a quantity which has already been measured. This suggestion is like a prediction of a future value of a quantity provided the measured values were not used in the construction of the theory.

In contrast, if past measurements are used in the construction of a theory, the agreement of the theory with the measurements on which it is built is no surprise. It is expected for the sake of consistency. A theory that did not agree with the values used in its construction would be inconsistent.

Confidence only grows in a theory when it fits and predicts observations that are not part of its construction. Here the prediction of future measurements instead of matching past measurements provides a more reliable test. It is less susceptible or prone to self-deception. Not all theory makers are logical and some illogical theory makers can be partially correct. Self-deception is to be avoided.

Chapter Sections: [ 16 Private Agreements ] [ 16 Public Laws ] [ 16 Physical Laws ] [ 16 Accidental Patterns ] [ 16 Reliable(?) Patterns ] [ 16 Scientific Method ] [ 16 Reaction to Failed Tests ] [ 16 Chaos ] [ 16 Statistical Inference ] [ 16 End Notes ]

www.whyslopes.com
Volume 1A, Pattern Based Reason

 Chapters 1 to 24

FOREWORD
Three Remarks
1 Introduction
2 Communication
3. Elements of Reason
4 Implication Rules
5. Deception
6 Chains of Reason
7 Longer Chains
For & From Consistency
8. Language Change
9 Next Chapters
10 Responsibility
11 Accidental Patterns
12 Knowledge Islands
13 Euclidean Logic
14 Deductive & Empirical 
     Views of Mathematics
15 Objectivity
16 Origin of Rules
and Patterns
17 Objective Ways
18. Waking up
19. Symbols  & Logic
20. Pronouns or Symbols
21. Truth Tables I.
22. Truth Tables II
22. Biconditional
22. Contrapositive
23. IF-THEN table
24. Indirect Reason Again

To reason often means to persuade someone of the need for an idea or action. That someone could be yourself. So be careful.

1A Logic Postscripts
- online only

+Proof by Absurdity alias proof by contradiction
+How the demand for consistency supports the law of the excluded middle
+Reality versus or with the aid of Imagination
+Links for reason, logic and crtical thinking
+Three Remarks
+History Lost or Missing

There is a difference between
knowing how to spend money,
and having money to spend.

There is likewise a difference
between mastering a skill
and having meeting a situation in which it applies.

 .

[ Back ] [ Area Intro ] [ Up ] [ Next ] [Top of this Page]  
Road Safety Message  Do not walk on a road with your back to the traffic. 
  Favourite Sites:  BBC News  and mathematics portion of  English National Curriculum  
[Site Author Contact Form]  If you find an error, please cut and paste its 
web page location (URL) into its report.

All trademarks and copyrights on this page are owned by their respective owners.
Copyright to comments & contributions are owned by the Poster. 
The Rest © 1995 onward by site author,   Alan Selby (email) All Rights Reserved. 

www.whyslopes.com