[Railstar]

Having a fondness for playing necromancers means it comes up for me. Cannot Learn is the difference between the mindless zombie who will always need micromanagement, and the trainable Improved Zombie who can develop sophisticated routines based around complex or extended decision trees, with “if X then Y” processes determined by their mental skills.

As I’ve used it, ‘standard’ skeletons or golems with Cannot Learn will never be more than they are when you create them. The failings they have will always be there. The problems of Automatons cannot be circumvented by the minion gradually learning “when I am given instruction A I should go about it through B, C & D, or E under conditions F.” Instead they stay literal-minded forever. Any increased efficiency comes from the master learning to give better orders, not the minion gaining a better understanding of what those orders mean.

One research project for my necromancer is creating more sophisticated undead servitors, not by giving them a more powerful template, but by using the absence of Cannot Learn on Improved Zombies to have them slowly start buying off certain limitations of the Automaton meta-trait (I assume normally it can’t be bought off, but that is why this is the subject of a necromancer’s research). For instance, they may replace Low Empathy with Callous & Oblivious, or buying down Incurious, becoming more able to interpret orders.

This might eventually backfire, if they start buying down Slave Mentality and become more difficult to control, but it is an acceptable risk in the name of science. The motivation isn’t entirely about efficiency anyway.