While there were already many questions on superposition in quantum mechanics, with great answers, these somehow leaves me uncertain about its physical meaning once I think about it a bit further.
Let me explain why the explanation with the double-slit experiment does not quite satisfy me. The system with the first (respectively, the second) slit open can be described by a Schroedinger equation of the form $$ ih\frac{\partial \Psi}{\partial t} = (-\Delta + V_{1,2})\Psi, $$ where the potential $V_{1}$ (resp. $V_2$) is, say, large negative at the first slit (resp., the second one) and zero elsewhere at the wall. If both slits are open, the equation can be written as
$$ ih\frac{\partial \Psi}{\partial t} = (-\Delta + V_{1}+V_2)\Psi. $$ If $\Psi_{1,2}$ are solutions to the former equation, then $\Psi_1+\Psi_2$ is approximately a solution to latter one. Note that we are actually adding states corresponding to two different physical systems, to obtain a state of a third system.
The same applies to the reflection example. We are using two different solutions to the free particle Schroedinger equation to cook up a solution to the equation with a barrier.
So far it all looks like merely a mathematical trick, exploiting nice features of the Schroedinger equation. We are not adding two states of the same physical system. So, the question is:
- What is a physical meaning of adding two states of the same system?
One thing that concerns me here is the following. As far as I understand, the state space of a quantum-mechanical system is not, actually, a Hilbert space, but a complex projective space over a Hilbert space: the wave function should be normalized, and multiplication by $e^{i\alpha}$ leads to a physically indistinguishable state. Now, if quantum superposition had a physical meaning, then adding two states should be well-defined over this projective space, which it is not. In other words, $$ \frac{1}{\sqrt{2}} (|\psi_1\rangle+e^{i\alpha}|\psi_2\rangle) $$ corresponds to physically different states for different choice of $\alpha$, while $e^{i\alpha}|\psi_2\rangle$ describes the same state.