Learning involves taking past experience (or knowledge) and applying it to a
new situation.  Instinct does not require "learning", although practice
(experience) can improve results and "learning" can improve the results
further (in some circumstances).  As an example, for man to fly, not only is
learning required (how to operate the controls of whatever craft is chosen),
but a certain base of experience (not to mention an aircraft) is required to
start.  For a bird to fly requires a simple, perhaps panicked, muscle
response to falling out of the nest.  Of course, one hopes that the practice
of each results in better flights after a period of time.

Bees do many things on instinct, some become better through practice.  But,
memorizing the correct route through a maze is more of a measure of
patterning, rather than of intelligence (the experience of a single maze
does not really help in the next one, unless pattern clues are used).  The
bee's general behavior of foraging ensures that eventually the solution to
the maze is found ... with some visual or other sensory clue, the bee is
capable of memorizing the "solution" -- much as the location of good
foraging grounds and how to get there and back are "learned".  Most measures
of non-humans are along the lines of this type of "learned behavior", as are
many tests for very young human subjects (also known as students).

The problem is determining the difference between learning (education) and
intelligence: think of intelligence as the capability to learn something.
We can test if you have learned "it", assigning a _minimum_ level of
intelligence to the subject.  It is harder to test the capability itself, so
we generally don't try, instead coming up with "standardized" tests that
"anyone" should be able to pass.  Of course, that's where the cultural bias
problems kick in -- not just because of non-exposure of certain groups to
the information being tested (easily overcome by standardized curriculums --
you think that middle class kids have much more in common with the culture
of the greek classics or really speak the language of shakespeare, than
other income groups?), but because of the peer-enforced intentional lack of
learning in certain cultures and age groups.  Too bad that our ability to
learn quickly/easily decreases as we age, forever limiting those that do not
apply themselves or that do not receive sufficient challenges when young.

As to the brains of drones, the increased sensors from antenna and
especially from the eyes, would require a much larger volume of processing
power, just to stay at the same "speed" as the other bees' brains (for
examples of this, look at how the increased resolution of your new digital
camera eats up memory on those high res pictures and how much slower your
computer is when applying filters to higher res images -- then multiply
immensely when talking about image processing in real time).  The processing
power required for locating a small, bee sized object against either the sky
or the ground, at a distance, then calculating the best trajectory to get
there before the competition would be much greater than that required for
flying up in the sky and avoiding the (much) larger bird objects there.  If
this excess brain power were being used for much of anything else, one would
expect the drones to have figured out how to quit getting tossed out of the
hive in the fall, by now.