John,
Math is unlike religious belief. But it is also unlike science.
Aristotle is often viewed as the father of formal logic. Math employs this logic as follows:
State some "starting points", such as axioms or postulates.
Use "deduction" to derive some logical consequences of those axioms.
The axioms, as such, are neither true nor false. They are merely "interesting" or "uninteresting", depending upon whether or not they lead to interesting deductions.
Science, for Aristotle, was much the same:
State some assumptions, like "the cosmos is perfect", "the most perfect form is a perfect sphere", then deduce that the moon etc. must be a perfect sphere.
Unfortunately for Aristotle, and modern physicists that make the same mistake, in science, unlike math, the "truth" or "false" of the "starting points", are of interest. That is the most important distinction between math and science.
The truth/false question about the starting points cannot be resolved, ever, via deduction. So attempts to demonstrate their truth via induction, were made. But all such attempts failed.
Consequently, the "scientific method" rejects all such assumptions, axioms etc., as valid "starting points". In their place, it simply uses observations as the "starting points".
Observations are what they are, and like the "starting points" in math, they are neither true nor false.
But unlike math "starting points", the "starting points" of science, have other problems; they might be "repeatable" or not. They might be subject to misinterpretation or not. And attempting to use them, in science, is impossible, without dealing with the "Problem of Induction", as described by David Hume, a couple of centuries ago; "repeatable" observations necessarily assumes that the future will resemble the past.
From these scientific "starting points", theory merely "fits" mathematical equations to the observations, to create a concise, quantitative description of those observations. Then, as Karl Popper noted, good science, as opposed to pseudo science, makes risky predictions about yet-to-be made-additional-observations, collected under different circumstances (in the distant future, at very different temperatures, pressures, energies etc.), that can be falsified, or not, by future observations.
A major problem in modern physics, is that many physicists have reverted to using "starting points", other than observations (as in mathematics), and/or fail to make risky predictions (as in pseudo science).
As George Santayana said "Progress, far from consisting in change, depends on retentiveness... Those who cannot remember the past are condemned to fulfill it."
Rob McEachern