0
Visual learning in the common cuttlefish,
Sepia officinalis
Keri Norton
Gilly Lab
Hopkins Marine Station
Spring 1997
Visual learning in the common cuttlefish, page 1
Abstract
Although many learning experiments on cephalopods have been carried out with
octopus, little is known about the learning ability of a close relative, the cuttlefish. The
capability of two common cuttlefish, Sepia officinalis, to associate either a flashing light or
a constant light with a reward of food was explored in this experiment. One animal was
presented with both a flashing and a constant light and had to choose the constant light in
order to receive the reward. In a simpler experiment, the other experimental animal was
presented with a flashing light and simply needed to react to the light to receive the reward.
Over the course of the trials, the animal in both experiments showed significant increases in
responses focused on the light associated with reward (pe0.001 for the choice experiment;
p20.05 for the simple experiment).
The animal trained initially to choose the constant light was then exposed to a
situation in which the flashing light meant a reward. Although a full reversal was not
achieved due to time constraints, the difference between response score before reversal and
response score after reversal was also significant (pe0.05). Similar experiments would
need to be done with a higher number of animals (n»1) to get statistically significant
results.
Introduction
Cuttlefish are visual predators. Therefore, it seems natural to assume that they are
able to pick up on visual clues to aid in the location and capture of prey. Once prey has
been located, the animal goes through a stereotypical, three part response. First, the animal
goes into what is called the attention position- two tentacles held in a vertical V formation
and their color darkened. Once in the attention position, the animal orients itself towards the
prey and approaches it. As soon as the animal has approached to within a certain range, it
shoots out two longer tentacles and grabs the food, pulling it in towards the mouth.
(personal observations)
Previous experiments with cuttlefish have shown that they have the ability to
associate sound with a negative reward (Packard et al., 1990). Two years ago, an
experiment was done by another Hopkins Spring Student that demonstrated the ability of
the cuttlefish to associate sound not only with a positive reward of food, but with the
complex task of moving around a barrier to receive the reward (Rawizza, 1995).
In order to test the ability of the cuttlefish to associate a visual stimulus with a positive
reward of food, two different experiments were set up. In the first (Tank One), the animal
was presented with two stimuli simultaneously, a constant light and a flashing one. Except
Visual learning in the common cuttlefish, page 2
for some later trials, there was always a food reward to be found below the constant light.
In order to receive the reward, the cuttlefish was required to approach the constant light. At
that point, the food would be exposed to view. In the second experiment, the animal was
presented with only one stimulus- a flashing light- and the reward was given when it had
reacted to that stimulus.
Once the animal in a learning experiment has demonstrated an apparent ability to
master the task, a typical next step is to try what is called a reversal. For the animal in Tank
One, this would entail a change in the stimulus that the cuttlefish needed to approach in
order to receive a reward. Since the initial learning task was approaching the constant light,
a reversal would mean the flashing light would now be the stimulus that had the food
below it. If an animal has truly learned, it should be able to accomplish a reversal in less
time than the original learning.
Materials and Methods
Rearing (Boletzky and Hanlon, 1983) of experimental animals
Cuttlefish were obtained from the National Cephalopod Resource Center in
Galveston, Texas. They were kept in individual circular tanks with a semi-closed
circulation system (half of the seawater was recirculated after passing through a heating
apparatus). Temperature was usually around 17 degrees Celsius. Lighting was artificial
with a 10:14 LD cycle. The animals were approximately 15 cm in mantle length.
Experimental animals were fed a combination of frozen commercial shrimp and live
crabs, primarily Petrolithes cinctipes, collected from the shore around Hopkins Marine
Station. The amount of food each animal received on a given day depended on performance
in experimental trials, but was generally two shrimp, four crabs, or some combination of
the two. An animal that did not perform in the trials for that day was not fed outside of the
trials unless such behavior had gone on for more than one day. In such a situation, a
shrimp was provided to avoid extreme starvation.
Experimental set-up
For a sketch of the experimental set-up, see Figure 1. Each tank was surrounded by a
black curtain to exclude visual stimuli other than that desired for the experiment.
Recordings were made using a Canon A-Digital camcorder on a tripod above the tank, and
a Panasonic AG-6730 TL (time lapse) VHS video cassette recorder.
Visual stimuli were red LEDs (peak emission wavelength of 660 nm) run off of two
AA batteries. The flashing light was created by interrupting the current with a relay switch
run by a function generator set to a square wave with a frequency of 1 Hz. The stimulus
Visual learning in the common cuttlefish, page 3
were located approximately 8 cm off of the bottom of the tank, inside of standard aquarium
air tubing. In the choice (complex) experimental tank (Tank One), food was place in a
white PVC pipe just below the level of the LED. In the simple experimental tank (Tank
Two), the food was dropped into the water through a long black PVC pipe that extended
from just below the surface of the water to outside the curtain.
Trial protocol
Ten to fifteen minutes before a trial in Tank One, the food reward was placed into one
of the two tubes at the base of the stimuli apparatus. Every attempt was made to shield the
food from the animal's field of view. The same actions used to place the food into the tube
were mimicked around the other food tube.
At the time of the trial, the stimuli were turned on for one minute before the food
tubes were raised to expose the food to the cuttlefish in Tank One, or before food was
dropped through the food chute into Tank Two. Both of the food tubes in Tank One were
raised simultaneously. The location of the rewarded stimulus- the constant light for Tank
One, and the flashing light for Tank Two- was randomized by tossing a coin before each
trial. Because only three to five trials were run each day, it was possible that all trials for
that day could be run with the rewarded stimulus in the same location, which could lead to
association of that location with food. To avoid this, if all trials for a given day had been
run with the rewarded stimulus in the same location, an additional trial would be run with
the location changed.
If, after five minutes, the animal had not reacted to the stimulus or consumed the
food, the lights would be turned off, the food tubes lowered in Tank One, and the food
removed from the tank.
Behavioral observation
To get a better idea of the regular pattern of activity for cuttlefish in this environment,
the activity of the experimental animals was recorded over a period of twelve hours using
time lapse video.
Data collection
The video of each trial was analyzed using NIH Image version 1.60 to measure
angles and distances during the trial. These factors were measured before the onset of the
stimuli and before the reward was presented (before the tubes were raised in Tank One, and
before food was dropped into Tank Two).
Visual learning in the common cuttlefish, page 4
Results
During a twelve-hour recording of activity, the position and orientation of the animal
was recorded every five minutes, and neither cuttlefish was found to approach the light in a
manner similar to that during trials. Figure 2 shows the data for Tank One, and Figure
Three shows the data for Tank Two. The scattered points along the perimeter of the tank
were mainly when the animal was swimming near the surface or in mid-water, whereas the
high density areas of points correspond to a position near the bottom of the tank.
Plots were also made of the orientation and position of each cuttlefish at the beginning
of the trials (see Figures 4 and 5), and no preferential orientation is apparent.
One component of a response to the stimulus, the normalized change in distance to
the stimulus from the beginning of the trial to shortly before the presentation of the reward
was calculated (see Figure 8). These results were placed into groups of ten trials each, and
analyzed using one-way ANOVA. Figures 6 and 7 show the change in normalized distance
over the course of the experiment in Tanks One and Two, respectively. For Tank One,
results were significant with p 0.05, but results for Tank Two were not.
A more complete analysis of the trials was also carried out. Three factors were
considered in the score for each trial: the onset of an attention position, the normalized
distance the cuttlefish moved toward the stimulus, and the final angle between the
cuttlefish's body orientation and the stimulus (see Figure 8). Each parameter was given a
score from 0-1, and the total response was a sum of the three parameters. Each trial,
therefore, was assigned a final score from 0-3. The scores were placed into groups of ten
trials, and a one-way ANOVA was performed. Both results, with pe0.01 for Tank One,
and pe0.05 for Tank Two. The results are shown in Figures 9 and 10, respectively.
Discussion
There were several potential problems with each of the experimental designs. First,
the animal could, perhaps, have come to associate a certain location with the food reward.
This was avoided by randomizing the location before each trial. Second, the animal in Tank
One could have heard the live crabs moving around in the tube before the trial, and
therefore known which of the two stimuli would have food below it. The trials run with
shrimp served as a control for this. There was also some concern that the cuttlefish could
be receiving some type of chemical signal from the food in the tube, be is crab or shrimp
However, trials run following the training protocol, but with no food in the tube also
resulted in positive responses as pronounced as those with a food reward present.
Visual learning in the common cuttlefish, page 5
The raising of the tubes in Tank One was also considered a potential problem in the
experiment. Perhaps the animal could be associating the reward with the tubes being raised
instead of the presence of the correct stimulus. In an attempt to control for this, there were
tubes placed at the base of each stimulus, and the tubes were raised simultaneously during
the course of the experiment. Observations of trials also led to the conclusion that the
animal was reacting to the stimulus before any motion of the tube. However, perhaps
because of an ability to associate a chain of events with the reward, or because of the
apparent wariness due to the raising of the tubes (discussed below), the animal was
observed several times waiting to approach until the tubes had actually been raised for a
few seconds.
The distance that the animal would approach the stimulus before the tubes were raised
was affected by this choice of experimental set-up. In many trials, the animal had
approached the stimulus, but was disturbed by the raising of the tubes. In such cases, the
animal would jet away, flash the pseudo-eye spots on the mantle characteristic of a
defensive position, or simply undergo a rapid color change.
This problem was avoided in the second experimental tank. However, the second set-
up had the problem of non-linkage between the stimulus and the food reward. In every
trial, the feeding chute was located in the same place, but the location of the stimulus was
changed. Because the opening of the food chute was near the surface of the water, the crab
was often able to swim quite a distance from the initial point of entry. Once, a crab was
observed to swim halfway across the tank from the chute before touching the bottom of the
tank. This could, perhaps, have contributed to the length of time it took the animal in this
experiment to learn. It was also not reflected in the scoring paradigm, as that included
simply the orientation and distance of the animal with respect to the light stimulus, and not
toward the food chute, as was observed in several of the later trials.
A more ideal experimental set-up would have a self-rewarding system that requires
the animal to approach within a certain distance of the stimulus, or to actually attack the
stimulus, in order to receive the reward. Perhaps, too, a negative factor could be
incorporated, as this has been shown to increase the speed at which animals learn (Boal,
1996). Within the time available for this experiment, however, it was not possible to
incorporate these features into the experimental design.
Despite these problems, both experimental animals seemed to have learned the task
set before them. for the complex task, the first set of scores (trials 1-10) is significantly
different (pe0.05) from the scores for all other sets of trials, with the exception of trials 11-
30. Based on this, then, the experimental animal presented with the complex learning task
Visual learning in the common cuttlefish, page 6
seems to have learned the task after 30 trials. The increase in scores, and the overall trend
of increase, is significant evidence that this particular animal did lear to choose the
constant light over the blinking light. There was also a significant difference between the
scores for the trials before the reversal and those after the reversal process had begun.
The second animal also learned the simple association task, but there was not
significant scores until after trial 40. This increased length of time could be due to variation
between the animals, or to the problems associated with the experimental design, as
discussed above.
Conclusion
This paper has provided evidence that the cuttlefish has the capability to associate
either a flashing light or a constant light with a reward of food. The experimental set-up
was also seen to have an important effect on the success of the learning process, and
further research would need to be done to correct some of the problems associated with this
experimental design. It would also be useful to continue the reversal experiment in order to
determine the speed at which a change in choice can occur.
Further research of interest would include studying the patterns of activity in the brain
at the onset of a visual stimulus that the animal has learned to associate with a reward. In
this way, perhaps more could be understood about the functioning of the cuttlefish brain.
Since the animals are able to perceive the difference between flashing and constant
stimuli, further research of interest could also focus on flashing lights of different
frequencies, to determine at which frequency the animal is no longer able to distinguish
between the two stimuli.
Visual learning in the common cuttlefish, page 7
Acknowledgments
I would like to thank: Dr. William Gilly for the use of his lab and equipment, as well
as his patience and help with my project; Dr. Thomas Preuss for the ideas and large
amounts of time he gave me; and Rebekah Harrison for her unending support and
friendship.
Visual learning in the common cuttlefish, page 8
Literature Cited
Boal, J. G. (1995). A review of simultaneous visual discrimination as a method of training
octopuses. Biological Reviews. 71, 157-190.
Boletzky, S. v., and Hanlon, R. T. (1983). A review of the laboratory maintenance,
rearing and culture of cephalopod molluscs. Memoirs of the National Museum
Victoria. 44, 147-187.
Packard, A., Karlsen, H. E., and Sand, O. (1990). Low frequency hearing in
cephalopods. Journal of Comparative Physiology A. 166: 51-55.
Rawizza, H. (1995). Hearing and associative learning in cuttlefish, Sepia officinalis.
Hopkins Marine Station Spring Papers.
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Visual learning in the common cuttlefish, page 9
Figure Headings
The experimental set-up.
The activity of the cuttlefish in Tank One over a twelve-hour period. Each »
represents five minutes, and the arrow points in the direction the cuttlefish
was oriented towards.
The activity of the cuttlefish in Tank Two over a twelve-hour period. Each »
represents five minutes, and the arrow points in the direction the cuttlefish
was oriented towards.
The position and orientation of the animal in Tank One at the beginning of
each trial. Each » represents the location at the onset of the stimulus, and
points in the direction of the animal’s orientation.
The position and orientation of the animal in Tank Two at the beginning of
each trial. Each » represents the location at the onset of the stimulus, and
points in the direction of the animal's orientation.
Mean proportional distance moved over the course of the experiment for the
animal presented with the complex (choice) learning task. Numbers were
calculated by subtracting the animal's final distance from the initial distance,
and dividing it by the initial distance.
Mean proportional distance moved over the course of the experiment for the
animal presented with the simple learning task. Numbers were calculated by
subtracting the animal's final distance from the initial distance, and dividing
it by the initial distance.
How a score was assigned to each trial. The final angle was scored as a
fraction of 180 degrees, and the distance factor was scored as the fraction of
the original distance from the stimulus that the animal approached. Each
factor had a range from O to 1 for positive responses. A trial was given a
score of 1 if the animal went into the attention position at the onset of the
stimulus, and a score of O otherwise.
Mean scores over the course of the experiment for the animal presented with
the complex (choice) learning task.
Mean scores over the course of the experiment for the animal presented with
the simple learning task.
Experimental Set-Up

- blinking or
constant light

food chura
for simple
experiment (Tank Two)
Stinulus tube
food fube
for complex
experiment
(Tank One)
Tank One: Position and Orientation
Over Twelve Hours




8


Each » represents five minutes
2
K
L
VS

L
V
V

A


5
A
A



J

2
Figure 6
O.30
0.25-
0.20
0.15.
0.10-
0.05
0.00
1-10 11-20 21-30
51-60 61-70 71-80
31-40 41-50
Trial Number
Figure 7.
O.30-
0.20-
0.10-
00
1-10
11-20
sersun

31-40 41-50 51-60
21-30
Trial Number
Scoring Trials
6
1. Finitial final
initial
2. 180-Ofinal
3. Attention position?
Figure 9

0-10 11-20 21-30
51-60
61-70
71-80
31-40 41-50
Trial Number
0-10
11-20
21-30
31-40
imber
41-50
51-60