First published online March 28, 2008
Journal of Experimental Biology 211, 1180-1186 (2008)
Published by The Company of Biologists 2008
doi: 10.1242/jeb.016683
Honeybees can recognise images of complex natural scenes for use as potential landmarks
Adrian G. Dyer1,2,*,
Marcello G. P. Rosa1 and
David H. Reser1
1 Centre for Brain and Behaviour, Department of Physiology, Monash University,
Clayton 3800, VI, Australia
2 Institut fur Zoologie III (Neurobologie) Johannes Gutenburg Universität,
Mainz 55099, Germany
View larger version (81K):
[in this window]
[in a new window]
|
Fig. 1. Bees provided with differential conditioning can learn to discriminate
between and recognise perceptually very similar complex natural stimuli. (A) A
target stimulus with which bees received differential conditioning. (B) A
distractor stimulus with which bees received differential conditioning. (C) A
novel distractor stimulus. (D–F) Fast Fourier transforms (FFT) of images
in A–C where vertical and horizontal axes show relative distribution of
low (towards the centre) and high spatial (towards the edges) information in
the respective images. The FFTs are almost identical (compared with images
M–O below) showing that there is an approximately equivalent
distribution of spatial information in the stimuli. (G–I) A
representation of the images in A–C, respectively, considering the
visual acuity of bee spatial vision which is approximately limited to viewing
frequencies less than about 0.3 cycles deg.–1. (J–L)
Angular high contrast geometric images including a diamond, square wave
grating and a figure `Y' that bees generalise to if only provided with
absolute conditioning (see text for references). (M–O) FFT of angular
high contrast geometric images (J–L) show relatively large differences
in the distribution of low and high spatial frequency information (compare
with D–F), but bees do not make use of this information to make
discriminations if provided with absolute conditioning.
|
|
View larger version (111K):
[in this window]
[in a new window]
|
Fig. 2. Video composite at 33 ms intervals of a honeybee flying from right to left
to approach a distractor stimulus (upper image), correctly rejecting it, and
then viewing and correctly choosing a target stimulus (lower image). The
sequence illustrates how bees make decisions to reject or select stimuli
following a visual inspection.
|
|
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 3. Frequency of stimulus selection plotted against the total number of
response decisions (landings and rejections). (A) Honeybee acquisition
(±1 s.d.) whilst being trained with differential conditioning to target
and distracter stimuli representing similar complex scenes that might be
encountered in a forest. Acquisition in experiment 1 is indicated by circles
and in experiment 2 by squares. Experiment 2 used a different target and
distractor combination; acquisition was very similar, and triangles and bold
line show pooled data. (B) Pooled data for the frequency with which bees made
abort flights (bees approached a stimulus and then turned away and left
without making any contact); the solid line shows a significant negative
correlation of aborts to target stimuli with increasing experience (see text
for statistics), and the broken line shows no significant correlation of
aborts to the distractor stimuli.
|
|
View larger version (6K):
[in this window]
[in a new window]
|
Fig. 4. Bee acquisition of target from distractor. The bold line shows bee
(N=10) acquisition with only 40 responses to stimuli, and the broken
line bee (N=10) acquisition for 120 responses with quinine
hemisulphate used as a punishment for landings on distractor stimuli (same
data as in Fig. 3). Insert
shows frequency of correct choices in subsequent non-rewarded tests where bees
trained for only 40 responses were significantly poorer at recognising the
target, but were still able to perform significantly better than chance (see
text for statistics).
|
|
© The Company of Biologists Ltd 2008
