>We can build them and priding how they react, but our explanations are all based on other things we don't actually understand.

You can say this about anything. This is science; making models and evaluating their predictive power. The models don't have to capture the underlying "truth", if it even exists.