action-value function

In general, an action-value function $q(s,a)$ describes how advantageous it is to take action $a$ at state $s$.