英语论文网

could be forgiven at this point for thinking that, unless you are doing advanced work in logic or mathematics, you will only ever be concerned with inductive reasoning, or inductive arguments. However in truth you will probably be involved in a mix.

Consider the following argument:

(Argument 6)

This thin copper wire conducts electricity.
This thick copper wire conducts electricity.
This copper rod conducts electricity.
(Hypothesis) Copper conducts electricity.
So:
(Prediction) A copper surface will conduct electricity.

This is actually a deductive argument. The hypothesis (added as the fourth premise from the top) has been reached inductively. The predictive conclusion follows logically from the premise set.

This is widely believed to be typical of the way much science (including the social sciences) actually works. The prediction is a way of testing the hypothesis. If it turns out false we must do one of two things: either reject the hypothesis (which can be by rejecting it outright or by incorporating a qualification, thereby modifying it) or reject the conclusion (e.g. this is not ‘coppery’ enough. This suggests that, after further testing with larger samples of copper, we might once again modify the hypothesis to something like ‘Things with a constituency of Q percent copper conduct electricity.’

How many correct predictions do we have to make to justify saying we have proven our hypothesis? In practice all we can really say is ‘A lot’. And, in practice, we need to make correct predictions in relation to the sorts of circumstances which, based on general experience, we think might affect predictive accuracy (e.g. the temperature of the copper sample, the height above sea level, the geographical and geological source of the copper, the extent and the way in which it has been prepared for the conductivity test. When a hypothesis is universal in scope, and applies to a class which is open to the future (e.g. copper that does not exist yet) as well as the past (all the copper that has ever been, most of which nobody has ever seen) we can never even say we have examined a particular percentage of the class (things made of copper). For it is certainly less than 50 percent, certainly less than 10 percent and, for all we know (because we do not know how much there is, not to mention how much, over the next few centuries, will come into being) a lot less than one percent.

How many incorrect predictions do we have to make to establish that our hypothesis is false? For many years (and some still say this) the standard answer, following Sir Karl Popper, was one. The mantra was No number of correct predictions will prove an inductively derived hypothesis to be true, but it only takes one incorrect prediction to prove a hypothesis false.

This looks reasonable. The first part is uncontroversial. That is why people prefer to speak of successful predictions as confirming a hypothesis rather than verifying the hypothesis. An accurate prediction confirms a hypothesis if, and only if, it is exactly what you would expect to happen if the hypothesis is true. That is a long way short of conclusive verification which, if the hypothesis is open-ended to the future, is an impossible dream.

So what is wrong with the (once) standard mantra? People soon realised this was not the way in which natural and social scientists behaved. At first this was dismissed, somewhat arroga本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。