Um, you said interpolation, but you never said what kind of interpolation was used. The crystal ball is very foggy today, so I am having trouble reading your mind.
Anyway, this is a common question, with almost always the answer being no, you cannot. Well, you could, but it would just not be worth the effort. The interpolation function will just be a mess of coefficients, none of which would make any sense to you, or be of any value. And as I said, there are LOTS of different ways to interpolate, even if you restrict that to interp1, or to tools like interp2, griddata or scatteredInterpolant.
For example, reading the help for interp1, I see all this:
'linear' - (default) linear interpolation
'nearest' - nearest neighbor interpolation
'next' - next neighbor interpolation
'previous' - previous neighbor interpolation
'spline' - piecewise cubic spline interpolation (SPLINE)
'pchip' - shape-preserving piecewise cubic interpolation
'cubic' - same as 'pchip'
'v5cubic' - the cubic interpolation from MATLAB 5, which does not
extrapolate and uses 'spline' if X is not equally
'makima' - modified Akima cubic interpolation
Every single one of those methods would result in a completely different "function". And every one of them would be just a mess of coefficients that would be of no real value to you. (Ok, 'nearest', 'next', and 'previous' would be relatively easy to write down. But even then, the "function" would be mainly just a list of numbers, just the data you passed in.)
If you truly, desperately wanted to see the actual cubic polynomials generated in a spline interpolant, for example, I have provided tools that can output those functions, written in symbolic form. But why would you bother?