There's not really a concept of notation in MIDI. It's a more abstract concept where the note length is determined by a number of "ticks per quarter note" aka "parts per quarter" which is the resolution that the beat can be divided into. So there's a PPQ resolution, a tempo, a time signature, and the number of ticks that occurred between the "note on" and "note off" messages. Given that info, any note length can be defined without even having to consider how you might notate it with sheet music.
MIDI kind of has to work this way because keep in mind that notation can't properly describe the way music is actually played. There's more nuance in a real performance like playing slightly ahead or behind for feel, etc. It would be pretty boring if people played exact quantized note lengths like a machine. So the MIDI format is optimized to be able to capture those nuances in performances in a way that storing the same data directly as sheet music couldn't.
But that's just how the underlying MIDI data works. Many user facing apps might use MIDI as an input mechanism and/or storage of the data but also add a traditional notation UI on top that underlying data for your convenience (kind of like syntactic sugar to use a programmer concept). So an app might let you choose whether you see ties vs dotted notes, but underneath the data is probably still MIDI.