For the $u$-spinor solutions $u_{a}(s,p)$, which you mention in the comments, the spinor index $a$ runs over $a = \{1,2,3,4\}$, whilst independently $s$ refers to the two different spin states. And we have the $v_{a}(s,p)$ for the negative energy solutions.
Take $\mathbf{p} = (0,0,p_z)$, then the positive energy spinor solutions for the different values of $s=1,2$ are
\begin{align}
u_{a}(1,p) = \begin{bmatrix}
\sqrt{E+m} \\
0 \\
\sqrt{E-m} \\
0
\end{bmatrix}
\quad \quad
u_{a}(2,p) = \begin{bmatrix}
0 \\
\sqrt{E+m} \\
0 \\
-\sqrt{E+m}
\end{bmatrix} \ .
\end{align}
The spinor index $a$ still runs over $\{1,2,3,4\}$, so for example $u_{1}(1,p)=\sqrt{E+m}$. As you probably know, when you write the fields $\Psi_a (\mathbf{x},t)$ explicitly we include the $u$- and $v$-spinor solutions, whilst summing over $s$, with the spinor index $a$ being free.
If you're asking more generally about the space of the Dirac/spinor indices, I think there's already questions on that here, e.g. spinor vs vector indices of Dirac gamma matrices. There's also quite a lot of resources online about spinors generally, e.g. https://arxiv.org/abs/1312.3824