We want move from the [[Conditional Expectation]] to the unconditional [[Expectation]]. Therefore we write the conditional expectation in its functional form $g(Y)=\mathbb E[X\vert Y]$.
$
\begin{align}
\mathbb{E} \big[ \mathbb E[X \vert Y] \big]
&= \mathbb E[g(Y)] \tag{1}\\[8pt]
&= \sum_Y g(y)*p_Y(y) \tag{2}\\
&= \sum_Y \mathbb{E}[X \vert Y=y]* p_Y(y) \tag{3}\\
&= \mathbb{E}[X] \tag{4}
\end{align}
$
where:
- (1) Substituting the function $g(Y)$ with the conditional expectation.
- (2) Writing out the [[Expectation#^32925b|expectation]] as a sum.
- (3) Substituting back the conditional expectation for the function $g(Y)$.
- (4) Recognizing this as the unconditional expectation by [[Total Expectation Theorem#^38d58d|total expectation theorem]].
Thus:
$\mathbb{E} \big [ \mathbb{E}[X \vert Y] \big ] = \mathbb{E}[X] $