Thursday, May 27, 2010

Logical Inference in Science

A certain protein is absent in the brain, but then becomes present at a certain point of brain development and then disappears when it is done. A scientist would then ask, what is happening at this stage of development?

Perhaps a group of cells diverges into two different kinds of cells at this stage. The scientist might then hypothesize: this protein is probably the case of this divergence. Then he could devised an experiment to test the hypothesis.

When the brain reaches this stage of development, the scientist could remove this protein and see what happens. If the divergence does not take place, then probably the protein is the cause of the divergence.

But what if the absence of the protein affects something else, which is the real cause of the divergence? The scientist can devise other experiments to isolate the protein as the cause.

For example, instead of removing the protein, the scientist could increase it and see what happens. Perhaps instead of diverging all of the cells turn into other kinds of cells. The scientist could also focus on the part of the cell that receives the protein. If that part of the cell is de-activated, and then the divergence does not take place, then that would be additional evidence that this protein is the cause of the divergence.

The reliability of the results would increase by publishing the results, permitting other scientists to re-duplicate the results, and also to suggest other possible causes to be tested.

So scientists can determine with greater and greater certainty that the protein is the cause of the divergence, by increasing the number of experiments that point to this protein as the cause in a variety of different ways. The conclusion would have a very high degree of certainty--beyond a reasonable doubt.

So how can the thought process be represented in logical form?

1. If x is followed regularly and predictably by y, then x is probably the cause of y
2. Divergence into two kinds of cells follows regularly and predictably the presence of this protein.
3. Therefore, this protein is probably the cause of the divergence.

1. If y does not happen when x is absent, then x is probably the cause of y.
2. Divergence into two kinds of cells does not happen when this protein is absent.
3. Therefore, this protein is probably the cause of the divergence.

etc.

So what seems to be happening is that the conclusion is made more certain by multiplying the number of proofs, the amount of evidence.

This is not so much the combining of evidence, but it takes the definition of what it means to be a cause, and then tests to see to what extent a certain substance fits this definition.

No comments: