However, most people want to represent negative numbers. The question is how to do so.
To answer this question, let's go back to an older question. How many different bitstring patterns are there, given N bits. By now, you should know this answer very well: 2N.
That gives us 2N representations, which means up to 2N potential values (one per representation).
How many values should be negative, and how many values should be positive? One answer is to make half the representations stand for positive values, and half for negative values.
Why? Of course, one answer for why is, why not? If you don't have a better way to split the representations, then half for each seems very reasonable. If there's no compelling reason to have, say, 75% of the representations be positive, and the remaining be negative, then 50% for each seems quite sensible.
There is, however, a compelling reason to have half the representations be positive values and half be negative values, and that has to do with math. Very often, you will need to negate a variable. You have x. You want -x. If the numbers aren't split evenly, for example, if 75% of the representations are positive values, then you won't be able to negate 2/3 of the positive values (assuming 25% of the representations are negations of the positive values).
A negative sign would be yet another symbol, and we only have two.
The obvious solution is to make 0 or 1 stand for the negative sign, and that's what happens. Usually, 0 is positive, and 1 is negative. One reason to make 1 negative is so that the positive signed numbers and the unsigned numbers that have 0 in the most significant bit look the same. This isn't the greatest reason, but it's a reason.
This bit is often called the sign bit and is the most significant bit of an N bit representation.