Do postdocs still take courses?
Adding to the other answers to reply to the comment: sure, a postdoc would have the skills to learn the subject on their own. But do they have the time? It's (usually) faster to learn when you get a world expert to teach you, than trying to do it on your own in your office. You can ask questions to the expert, and depending on the format of the course and your questions, the expert can even adapt the course to the audience (a book can't do that). You also have a few other motivated people learning the same thing beside you, and you can ask questions to them. It also provides motivation: if you have an appointment every Monday at 10 to learn something, it's harder to blow off than "I'll read this book some day during the week... maybe...".
Finally, well, learning is fun. Maybe the course isn't strictly necessary for what the postdoc is working on right now. But it's always nice to take a couple of hours every week to learn about something new, open up new perspectives. And it's quite refreshing when you know you are doing this for yourself and not for some exam at the end of the semester. (Almost?) nobody has learned all there is to learn during their PhD. And who knows, maybe you will have a revolutionary idea by attending the course and thinking "wait a minute, I've seen this sort of phenomenon before...!"
Of course, as the others said, I don't think any postdoc would "take" the course to get credits or something like that. Postdocs aren't students.
I run short courses and have post-docs (and professors and PhD students and...) as participants regularly. For example, you may need a methodology you haven't used before to deal with some aspect of your research question. You don't stop learning just because you've finished your PhD. Conference workshops, short courses and other training opportunities are full of postdocs. It's unlikely a postdoc would attend a course 'for credit', but auditing a course where you work is an easy way to find out new ways of doing research.
I'll just drop it here in a response to Maarten's fine answer (see the Dunning-Kruger effect):
