This is a question on the Philosophy of Science examination paper for the BPhil at Oxford, as set last year.

"Does the success of a scientific theory justify belief in the entities that the theory posits?"

Indeed the paper has a related question:

"Must a good scientific explanation of the occurrence of an event refer to some of its causes?"

The student manual includes the helpful suggestion that all questions are, of course, liable to interpretation.

The Ptolemaic model of our planetary system may have looked a mess but was a fairly accurate predictor of celestial motion given the rudimentary observational techniques. It is a moot point whether anybody believed in any invisible machinery that resembled the cycles and epicycles Ptolemy described. The question came to the fore when Kepler discovered elliptical orbits and a sun-centred planetary system. However, the Church argued that just as Ptolemy's system was a purely predictive device then so too was the new-fangled heliocentric system. Both merely "saved the phenomenon" and had nothing to say about how angels went about their heavenly business.

Quantum mechanics has been one of the great successes of theoretical physics and yet its mathematical machinery is dependent upon wave functions that cannot be directly perceived. String theory seems to have tangled itself up into competing theories that are, perhaps, merely different formulations of one genuine overall theory. Is the mathematics itself leading physicists into metaphysical realms or are these just modern ways of saving the phenomena?

The theory looks beautiful, the predictions have been verified, but what does it mean?