With the Myers-Briggs there are just 4 axis to measure: extroversion vs introversion, thinking vs feeling, sensing vs intuiting, judging vs perceiving. That means you can have a simple struct or class that has 4 booleans, or 4 integers, and be effectively able to record the personality type. Yes, you could keep those 4 booleans or integers in an array, but that is the old fashioned FORTRAN or COBOL way doing things. The modern C# object oriented approach is to actually have 4 separate well named fields.
Your code then can take an instance of these classes or structs, and examine each of its fields and decide what to do with what it finds in them. Yes, you could create a bitmap of the 4 booleans to come up with the 16 permutations and use that as jump table (e.g. an array with pointers to code to be executed next), but again, that is the old fashioned C way of doing things. In modern C#, the first preference would be to use a chain of if-else statements, or a C# 8.0 pattern matching switch statement. (The C# 8.0 compiler will just convert the switch statement into if-else's anyway.) Some people may setup a dictionary with the permutation values as the key and delegates to be executed to use a more data structure approach, but again it's not going to be an array. Only a really old school C programmer transitioning into C# would setup an array of delegates where the index into the array is the equivalent bitmap value -- but legibility of the code will really suffer, as well as, completely puzzle modern C# programmers as to why the author would be bending over backwards to not write idiomatic C# code.