This study examines whether honeybees can learn to fly through complex
mazes, in the presence or the absence of specific visual cues. The re
sults are summarized as follows: 1. Bees can learn to fly through a co
mplex maze by following a trail of colored marks. 2. Bees, initially t
rained to follow color marks through an initial part of the maze, are
immediately able to use the same sign-tracking cue to find their way t
hrough the rest of the maze, which is unfamiliar to them. 3. Bees trai
ned to follow color marks through a particular maze can use the same c
ue to negotiate a novel maze. 4. Bees trained to use a particular colo
r to negotiate a maze can immediately use a novel color to negotiate t
he same maze or even a novel maze. 5. After learning to negotiate a ma
ze by following colored marks, bees can find their way through the maz
e even when the marks are removed, albeit at reduced levels of accurac
y. Thus, the trained bees do not rely solely on sign-tracking to find
their way through the maze: they also acquire a spatial memory of the
maze or at least a sequence of motor commands describing the correct p
ath through it. 6. Bees can learn to use color as a signal even when i
t indicates the path through the maze in a symbolic way, for example,
blue indicating a turn to the light and green a turn to the left. 7. B
ees can learn an unmarked maze. Performance under these conditions is
poorer than when marks are provided, but is still significantly better
than chance level. 8. Control experiments rule out the use of externa
l landmarks in all of these situations. (C) 1996 Academic Press, Inc.