This happens to me fairly often while public speaking and (much less often) low-stress social settings. Not a complete mind blank, I do not lose awareness of my surroundings and so on, just completely lose my train of thought. It was enough of a concern that I started keeping a diary of the circumstances. It nearly always boiled down to mind-wandering and its relatives: either focusing on a point I wanted to make down the line and not paying close enough attention to what I was saying at the moment (especially in public speaking), or thinking about something else during a social conversation. Sort of, operating on conversational auto-pilot, which seems to break down as soon as anything goes amiss (I make a spontaneous speech error & need to correct it but then suddenly am unsure of what I was saying; or someone makes an unexpected point and I wasn't paying close enough attention). In public speaking I reduced this by switching from a highly rehearsed literal approach (writing up a script and then close-to-memorizing it), to something more based on a narrative arc, as in this post on public speaking ... I still get tongue-tied occasionally but not nearly as often, since I am only trying to get to each important point in the narrative but not trying to keep to a specific script. In social conversation, my diary notes suggested this happened mostly during conversations that I wasn't so engaged with - not really such a concern except for social reasons it's not a great thing to zone out when someone else is talking to you. So my feeling from an n=1 diary study (confounded with practice effects in public speaking), is that this is not a working memory problem for me, but more about distraction and focus.

This usually indicates a badly-initialised pointer - either it's zero and shouldn't be, or it is fooling a test-for-zero check through being filled with random garbage, and thus preventing program flow from correctly returning to the calling function. Suggest running the debugger to see if the problem goes away - this indicates the second issue, as the debugger will initialise all pointers not given an explicit value to zero; if it persists, you should at least be able to get a stack trace.