Yes.
The reason "a simple null test" won't resolve these questions is that we're talking about whether something is audible or not.
In the specific case of DAC filters, we already know the results of an electrical null test.... the outputs of the filters are quite obviously electrically different, which is easily seen on an oscilloscope trace, so those differences will show up on a null test. I don't think it takes a huge leap of faith to assume that, if you subtract two different electrical signals that look different on an oscilloscope trace, you will end up with some sort of difference (they won't null to zero). The only way that could fail to be true would be if all the chip vendors who make DACs with multiple filter choices (like Texas Instruments and Wolfson), and all the vendors who sell audio technology that offers multiple filters (like Dolby Labs), were deliberately publishing falsified graphs in their data sheets - which seems rather doubtful. (Find yourself a copy of the Texas Instruments PCM5102a DAC chip data sheet, and compare the Interpolation Filter Impulse Response graphs of the various filters; it's pretty obvious that they are quite different. Of course, since we're talking about impulse response, which only affects signals containing transient (non-steady-state) content, a pure sine wave should be identical when played through the various filters, but non-steady-state signals will not be.)