Without reading the context, I expect that 'computational modeling' refers to a largely mathematical model that is potentially agnostic about underlying causes.

Without reading the context, I expect that 'computational modeling' refers to a largely mathematical model that is potentially agnostic about underlying causes. You can run simulations that accurately reflect the real world using one set of variables, then run another model (or somehow test another model) that confirms the original simulation's results from a different perspective; ie, you have two theories, and you validate the second by proving its equivalent to the first, even though it uses a different set of variables, or looks at it using a different theoretical framework. of course, you're right that this can be troublesome. just because you can validate a theory as accurately modeling reality doesn't mean that the components of the model are true or useful. and necessarily, the first 'computational model' would have needed to be already validated before something else might be validated against it...in case that wasn't clear.