[ACT-R-users] Is learning possible without self-explanation?

[MacLaren]

I am trying to differentiate scientific reasoning from self-explanation and 
other meta-cognitive behavior, and it got me thinking.

Chi and company claim people learn BETTER if they self-explain.  They learn 
better if they study "carefully" and ask "better" questions when 
elaborating examples.

It seems to me that at the very least if ANY elaboration is needed you MUST 
be self-explaining.  Perhaps you could learn simple declarative facts, but 
it seems almost impossible to learn anything without doing something
that should be called self-explanation.

One definition I found for self-explanation was "generating explanations to 
oneself to clarify a worked out solution" but this seems way too specific 
for the process I imagine being employed more generally.

Am I defining self-explanation too broadly?  Perhaps Chi et al are talking 
only about verbalized self-explanations?  Only the type that can be 
taught???  These seem a bit arbitrary.

So is the type of self-explanation being taught merely those subsets of 
questions that students don't often ask themselves?   If I am right then if 
they couldn't generate these questions they couldn't solve problems at all.

Thoughts?