A M . ZOOLOGIST, 9:145-152 (19G9).
Learning in Spiders
Louis LEGUELTE
Division of Research, North Carolina Department of Mental Health,
Raleigh, North Carolina
SYNOPSIS. After the spider, Zygiella x-notata Cl. has been enticed to leave the retreat
and catch a fly in a web turned upside-down, it takes a relatively long time to return
to the retreat. The return-time is divided into periods of active searching and
motionlessness. Several factors have been found to shorten significantly the searching
and motionless time; among them are youth of the animals, small number of webs
previously built, the position of the web before and between experiments, and
repetition of trials at certain time intervals. The observed phenomenon and variations
in time are explained in terms of current concepts of memory function and information processing.
It is well known that many invertebrates, like vertebrates, change certain
parameters of their behavior with experience. Very little data are available, however, on such changes or "learning" in spiders.
Peckham and Peckham (1887) observed
that spiders dropped from their webs
when a tuning-fork was brought close, but
modified their reaction after several
presentations of the stimulus. The authors
suggested, from their observation of one
animal, that spiders probably remember
their former experience but that this memory is very short (less than a day). Bays
(1962) "trained" spiders to respond differentially to two different frequencies by
pairing presentation of tuning-forks with
edible and inedible objects thrown into
the web. Our earlier observations (Le
Guelte, 1966s, b) indicated that spiders of
the species, Zygiella x-notata, did not
immediately find their way back to the
retreat if the web was turned around before they were enticed out of their hiding
place. On a second trial, following immediately, the return was much faster. All of
Author's present address: Laboratoire de Psychophysiologie, Universite de Nancy, 42, Avenue de
la Liberation, Nancy, France.
This work was supported in part by Grant No.
B7-0865-R from the National Science Foundation to
Peter N. Witt.
Discussions with Dr. P. N. Witt and Dr. C. F.
Reed in the course of the experiments and the
preparation of this manuscript are gratefully acknowledged.
the preceding observations dealt with qualitative data.
It is the purpose of this report to show
that the number of repetitions of such an
experience and the time between trials
have some influence on the time a spider
takes to return to its retreat.
Peters (1932) has shown that after an
Araneus diadematus has run on a thread
from hub A to B, it returns to A (using
direction as a cue). If the frame is turned
around in the meantime, the spider will go
in the direction of A's previous rather than
its present location. Many spiders which
characteristically await prey in the hub of
their webs, are unable to return to the hub
immediately if the web is turned around
while they are fetching prey. However, Zygiella x-notata never stays on the hub in
daylight but hides in a retreat which is
usually built above and to one side of the
web. Zygiella seems to have difficulties relocating its hiding-place on a reversed
web, even if the turning occurs before the
spider leaves its retreat.
During the day, this spider remains in
its retreat (A) (Fig. la) in the upper
portion of the frame with the front legs on
the signal-thread (A-B). When a fly is
thrown into the web at C the spider runs
along the signal-thread from A to B and
proceeds to C. After enwrapping the prey,
it returns rapidly to B and A (1 or 2
seconds).
In experimental situations, the frame is
145
146
FIG. la.
Zygiella
every 24
phy and
Louis LEGUELTE
Building position. The "normal" web of
x-notala Cl. as it is built generally once
hours. The web was sprayed for photogralighted up from the sides. Pictures were
taken with high contrast photographic film (Witt,
1963). A indicates the retreat in which the spider
sits, B the hub, and C the position in which the fly
is entangled in the web.
turned 180° so that the retreat is now in
the lower portion of the web (Fig. lb);
the spider is in the retreat (A) as before.
A fly is thrown into the web at C: the
spider proceeds from A over B to C at the
same speed as in la. After enwrapping the
prey, return is now toward the upper end
of the frame, and a period of alternate
"searching" and "motionlessness" follows,
until the animal finally arrives at A.
The time during which the spider moves
around in the web is considered to be the
LEARNING IN SPIDERS
147
FIG. 1b. Experimental position. This is the same
web as in Figure la but the frame has been turned
180". The retreat of the spider is now at the
bottom of the frame in A, B is again the hub, and
C the place where a fly is entangled. After the
spider has wrapped the prey at C, it returns to B,
and finally to A, but only after a long "searching
time" in the upper part of the web.
"active searching-time". The searchingsequences are divided into two to five periods with intervals during which the spider
does not move (motionless-time).
"EXPERIENCED" VS. "NON-EXPERIENCED"
ANIMAL
"Experience" in catching prey on webs
in the normal position increased the time
148
Louis
LEGUELTE
of searching on the first trial in the reversed position: spiders 10, 17, 30, and 45
days old took 19, 40, 40, and 45 seconds,
respectively, on a reversal test.
The time required to find the way back
to the retreat in the reversed position
seems to increase with age. In order to
learn whether maturation or experience
determines this increase, the following experiment was performed: a group of spiders from two cocoons was kept in small
individual containers in which they were
able to build retreats but not webs (n=6).
Another group of spiders from the same
cocoons (n=9) were kept for the same
length of time on frames on which they
daily built webs and retreats (45 webs);
they caught their prey and otherwise
moved on the web. Sufficient food was
provided to maintain equivalent growth in
the "container-group" and in the "framegroup".
At 1.5 months of age, the "container
spiders" were put on frames. After they
had built webs the following day, both
groups were tested in the reversed position. The spiders of the "frame-group"—
experienced in moving about the web and
returning to the upper corner—were significantly slower (median 45 sec) on the
first trial in the reversed position than
spiders of the "container-group" (median
3 sec).
For this aspect of behavior, the halfmature spiders without experience can be
compared to newly-hatched spiders. Old
spiders appear to have a firmer "recollection" of the location of the retreat than
the newly-hatched spiders. When they
were 45 days old without experience of
catching prey on webs, having built only
one web, the spiders on the reversed webs
did not behave as spiders do on webs in
the usual position: they searched for a few
seconds.
One could interpret these experiments
as showing that the early experience of the
spider influences its later behavior by increasing the difficulty of acquisition of a
new behavior which competes with a habit
(going up) developed in response to a
particular situation (retreat in one of the
upper corners). But even after the same
amount of experience in the usual situation, spiders show a wide range of variation in speed of return to hub on the first
reversal. This variation is probably an aspect of the strength of the association
"going-up-returning" and of the behavioral
plasticity of each individual.
INTER-TRIAL INTERVAL AND RETURN-TIME
A group of 78 "experienced" spiders (i.e.,
having an age of 1.5-2 months and a mean
of 45 retrievals of prey) were timed for
the first reversal of the web. An average of
43 sec was spent "searching" and 53 sec
motionless before they were able to return
to the retreat. The time of the first trial is
indicated in Table 1, columns 1 and 2, for
several sub-groupings of these spiders.
There are no significant differences between the sub-groujas in searching or motionless time as far as the first trial is concerned.
The second trial followed the first at
different intervals for the several groups,
and the times are listed in columns 3 and 4
of Table 1.
There is no significant difference in times
of return for spiders tested 24 hr after
the first trial. These trials occurred on new
webs, one trial to a web. No change occurred in the third through sixth trials on
webs built in subsequent days.
Similarly, there is no significant difference in return-time for animals given a
second trial 12 hr after the first (Table 1,
line 2). By experimenting at different times
of day, it could be analyzed that ten of
these spiders had built a new web between trials, and nine were tested on the
same web; neither condition shows a significant time-difference.
However, the decrease on the second
trial is significant if the interval between
trials is 4 hr or less (Table 1, line 3-6)
down to a two-minute inter-trial interval.
The 30-min interval shows the greatest
number of spiders going immediately to
the retreat. Perhaps the spiders need a
149
LEARNING IN SPIDERS
TABLE 1. Variations between the first and the second trials for different intervals (spiders in
the original position between trials).
Interval between
first and second
trials in hours
24
12
4
1
0.5
0.03
Trial 1 (time in sec)
motion
motionless
54
52
19
34
46
25
89
50
108
117
167
23
Trial 2 (time in sec)
motion
motionless
53
31
18
58
15+
3+
2+
9+
15+
0+
0+
Number of
spiders tested
11
19
10
11
10
17
26
The times varied over a wide range, but testing the medians (as listed) with non-parametric
tests resulted in significant differences below the 0.01 level ( + ).
minimum interval to process the information gathered on a first trial (Fig. 2).
The shortest inter-trial interval does not
show the most rapid return to the retreat.
This result is important in evaluating
possible cues to improve performance. The
spiders receiving both trials in the same
web have the potential cue provided by
the drag-line left by the spider in trial 1.
The 12-hr spiders tested on the same web
failed to improve in time, as seen above.
The 2-min spiders do improve, but their
performance is not the most rapid of all
groups. The presence of the dragline of
the first trial does not seem to be the single
factor mediating improved performance,
although it may have some role. In keeping with this argument, it is to be noted
that "motionless" time for the 2-min animals is not significantly reduced. The "motionless" time is an index of disorientation
as much as is active "searching" time.
Performance improves on repeated trials,
using speed of return as an index of
improvement. Another criterion for improvement is the number of trials required
for all animals in a group to return
immediately to the retreat.
Acquisition and retention of the improved performance appear to depend
upon repetition of trials and upon the interval between trials. Immediate repetition
does not necessarily result in more rapid
acquisition than that following a longer
interval. When the tests were repeated every 30 minutes, six trials were necessary to
bring all the spiders in one group (n=10)
to the criterion of immediate return to
the retreat (Fig. 2a). When another group
100
CO
«
CO
75
50
25
30
60
90
120
150
Intervals in minutes
FIG. 2a: Increase of correct responses by repetition
of the test in the reversed position every 30
min, original position between tests.
(n=:17) was tested by immediate repetition, eight trials were necessary to bring all
animals to criterion (Fig. 2b). Both
groups improved performance; the difference in rate of improvement is not significant; larger samples might show otherwise.
For spiders which have reached the
150
Louis LEGUELTE
criterion of immediate return, repetition
24 hr later does not yield immediate return
on the first trial. However, significantly
fewer trials are needed to recover the criterion level. Moreover, repetition of these
daily multi-trial training sessions does
seem to show results by the fourth day,
even on the first trial. That is, the first
trial on day 4 produces shorter returntimes than the first trial on day 1.
The increased experience in the reversed position adds up to counteract the
past experience in the ordinary position:
the experience in the reversed position becomes a long-term memory.
It is possible that most of the improvement of the first day for these animals was
lost because the spider built a new web in
the ordinary position during the night.
The animal stayed on this web after building and then climbed up to the retreat.
This new experience, which recalls the
past experience, is perhaps enough to
make it forget the experience in the reversed position.
INTER-TRIAL FUNCTION OF THE RETREAT
The influence of the position of the spiders between tests can be shown in the
following experiments:
(a) A group of spiders was tested and
the webs then were kept for 4 hr in the
reversed position before retesting. The improvement on the second trial was significant: a median of 2 sec as against a firsttrial median of 25 sec. Six of the spiders
returned to the retreat without delay. The
improvement shown by these spiders is also
greater than that shown by spiders which
stayed in the web for 4 hr after the first
trial, but in the normal position.
(b) A group of 10 spiders was kept in
the reversed position for 4 hr before the
first trial was given. Their performance on
the first trial (median 15 sec) was faster
than that of spiders which were tested
immediately after turning, but not as fast
as that of the spiders which had been in
the reversed position for 4 hr after the first
trial. A Kruskal-Wallis one-way analysis
of variance shows the three groups to differ
100
S
75
50
25
I
2
3
4
5
Trials
6
7
8
FIG. 2b: Increase of correct responses by immediate repetition of the test (no in-between reversal).
significantly. Thus, the altered position of
the web does lead to improved performance even without active exploration.
These results lead us to conclude that
three forms of experience may lead to improved performance, the specific course
taken to leave and return to the retreat,
active general exploration of the web, and
immobile positioning in the reversed web
for some period of time.
Agranoff (1967) has reported similar environmental effects on long-term memory
consolidation which may be comparable to
this effect.
CHANGE IN RETURN ROUTE
In the preceding experiments of turning
the web around, the courses the spiders
move along to return to the retreat and to
go on the web are the same and the spiders
use the information gathered when going
on the web to find the way back.
To 12 spiders trained to go back immediately to their retreat when the web is
turned 180°, a more difficult problem was
presented:
151
LEARNING IN SPIDERS
Building
position
Starting
position
Returning
position
Time in Situation
seconds
42
II
35
13
FIG. 3: Various combinations of web-structure,
and starting and returning position of the web, and
their influence on the time it takes the spider
to return from the web to the retreat. See analysis
of data in text.
152
Louis LEGUELTE
The spider starts in one position and the
web is turned around when the animal is
on the web (Fig. 3).
All of the spiders tested are slower in
situation 1 than in situation 2. Independent of the past experience, if the retreat
is switched while the spider is on the web
and is now below, the return takes longer
(Wilcoxon test T = 0, N = 12, P.<.005).
This hypothesis can be tested with experiments on the same spiders on webs
with the retreat below, of a different structure (LeGuelte, 1966b) (Fig. 3).
Comparing the situations 1, 2, 5, and 6
(in which web-position is switched while
the spicier is catching the fly so that the
information gathered in the web when
going would lead the spider in a wrong
direction on return) gives the following
results. When the retreat is above the web
on return (situations 2 and 6), the searching times are short. LeGuelte (1966£>)
has shown that the web-structure depends
on the building position. The different
structures in 2 and 6 do not seem to affect
the speed of return. When the retreat is
below the web, contrary to usual, the searching-times are long (situations 1 and 5).
We conclude that details in webstructure are not of great importance for
orientation. Two interrelated factors influence the speed of return. A preference for
return in the upper direction (where the
retreat usually happens to be), and a competing tendency to return in the direction
from which it came. The most rapid return is made when there is no switch in
web-position while prey is caught. The
next most rapid times occur if a switch is
made so that the return to the retreat is in
the upper direction. The slowest reaction
times occur when the return to the retreat
is swi tched to downward.
The usual behavior (returning to the
upper corner) can be reinforced, even after complete training on the reversed web,
by means of a recall of the past experience: going down the web.
The problem of the orientation of the
spider in its web seems of special interest
because, in contrast to other training experiments, it does not involve any reward
or punishment (the fly which is given to
the spider is only a means to entice the
spider onto the web, and it is caught before the test begins).
The phenomenon which has been
studied is a negative transfer from a task A
(going back into the upper corner) to task
B (going back into the lower corner). The
improvement is a short-term memory
effect. It is hoped that we have established
first points of a "learning" as well as a
"forgetting" process for the spider. The
influence of drugs and other factors which
may affect speed of learning and forgetting
in this animal can be tested eventually
with our procedure.
REFERENCES
Agranoir, K. W. 1967. Memory and protein synthesis. Scientific American 216:115-122.
Hays, S. 1962. A study o£ the training possibilities
of Araneus diadematus Cl. Experientia 18:423.
LeGuelte, L. 1966a. Note preliminaire sur un apprentissage chez Zygiella x-notata Cl. a l'etat
adulte. (Araignee, Argiopidae). Coinpt. Rend.
Acad. Sci., D, 262:689-691.
LeGuelte, L., 1966b. Structure de la toile de Zygiella x-notata Cl. (Araign£es, Argiopidae) et facteurs qui regissent le comportement de l'Araignee pendant la construction de la toile. These.
Publ. Univ. Nancy: 1-77.
I'eckham, G. W., and E. G. Peckham. 1887. Some
observations on the mental powers of spiders. J.
Morphol. 1:383-419.
Peters, H. M. 1932. Experimente iiber die Orientierung der Kreuzspinne Epeira diademata CL im
Net?.. Zool. Jahrb. 51:239-288.
Witt, P. N. 1963. Environment in relation to behavior of spiders. Arch. Envir. Health 7:4-12.