The definability of $\mathbb{E}$ in self-iterable mice

Let $M$ be a fine structural mouse and let $F\in M$ be such that $M\models$``$F$ is a total extender'' and $(M||\mathrm{lh}(F),F)$ is a premouse. We show that it follows that $F\in\mathbb{E}^M$, where $\mathbb{E}^M$ is the extender sequence of $M$. We also prove generalizations of this fact. Let $M$ be a premouse with no largest cardinal and let $\Sigma$ be a sufficient iteration strategy for $M$. We prove that if $M$ knows enough of $\Sigma\upharpoonright M$ then $\mathbb{E}^M$ is definable over the universe $\lfloor M\rfloor$ of $M$, so if also $\lfloor M\rfloor\models\mathrm{ZFC}$ then $\lfloor M\rfloor\models$``$V=\mathrm{HOD}$''. We show that this result applies in particular to $M=M_{\mathrm{nt}}|\lambda$, where $M_{\mathrm{nt}}$ is the least non-tame mouse and $\lambda$ is any limit cardinal of $M_{\mathrm{nt}}$. We also show that there is no iterable bicephalus $(N,E,F)$ for which $E$ is type $2$ and $F$ is type $1$ or $3$. As a corollary, we deduce a uniqueness property for maximal $L[\mathbb{E}]$ constructions computed in iterable background universes.