Brain Research~ 92 (1975) 103-121 © Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
103
T H E S U B F R O N T A L LOBE A N D T O U C H L E A R N I N G I N THE OCTOPUS
M. J. WELLS ANDJ. Z. YOUNG Department of Zoology, Downing St., Cambridge and Department of Anatomy, University College, London W.C.1 (Great Britain)
(Accepted January 31st, 1975)
SUMMARY Octopuses with the supraoesophageal lobes of the brain divided longitudinally can be taught to discriminate using the arms on either side. If there is no further lesion the two sides behave alike. Lesions limited to one side did not affect the performance of the contralateral, 'control' side. Lesions made in the vertical (n = 7) lobes led to a slight drop in the quality of performance in training to take a smooth sphere, in discrimination training (rough vs. smooth spheres) and in subsequent extinction and transfer tests. After removal of the median inferior frontal lobe (n = 10) there were somewhat greater effects in the same direction. Much larger effects followed interference with the subfrontal lobe (n = 20). Removal of parts from this always led to a marked loss of capacity for touch learning, broadly dependent on the amount of tissue removed. Removal of the whole of the subfrontal lobe (n = 6) produced animals that showed, at best, only very slight signs of learning. Such animals can adjust their overall level of response as a result of training but they seem incapable of adjusting response levels to two objects independently. These results are discussed in relation to the function of the subfrontal lobe as a memory store.
INTRODUCTION Octopuses can learn to make simple tactile discriminations between objects such as spheres or cylinders that differ in texture. There is evidence that this capacity is lost if all the small cells of the subfrontal lobe* are removed 3,9,1°. Any conclusion * We use the term 'subfrontal' to indicate the whole mass of tissue containing amacrine ceils on the median side of the inferior frontal system. The ventral part of this tissue is continuous with the larger cells of the posterior buccal lobe17.
104 about the memory record needs qualification, however, because octopuses with certain brain lesions and particularly those to the inferior frontal region tend to differ from their controls in two important respects. One is that they sometimes show a marked 'preference' for rough objects (grooved as opposed to smooth), which is not shown by normal (blinded but otherwise undamaged) octopuses 12. The other is that they are liable to accept objects presented at a higher (or more rarely at a lower) proportion of trials than normal animals. While every indication has been that these effects are due to interference with the tactile learning system, it has been impossible to be certain that they are not indirect consequences of interference with other aspects of the animal's physiology. The take level, or tactile sensitivity, might for example vary because the octopus was sick or hungry. In the experiments described below an attempt has been made to overcome these difficulties by using animals with the supraoesophageal part of the brain split by a vertical longitudinal cut. Experiments reported elsewhere have indicated that there is little or no lateral transfer under these conditions ~°,n. This means that the performance of one side of the split-brain animals can be used to control the effect of a lesion in the touch learning system of the other. Comparison of the two should reveal whether one is dealing with specific touch learning or more general physiological effects of the brain damage. Pretraining of the lesioned side to take smooth should eliminate the rough preference associated with lesions to the tactile system, if that side of the animal can learn at all. METHODS
Octopus vulgaris from the Bay of Naples was used. The animals were of both sexes and between 200 and 400 g weight. They were kept for about 4 days after bringing from the sea, until they were feeding readily upon crabs. They were then operated upon under urethane anaesthesia. Each animal was blinded by cutting the optic nerves. The cranium was opened and the supraoesophageal part of the brain split in the mid-line by means of a narrow blade entering from the front. Bilateral control animals were left in this condition. In the others lesions were made on the righthand side of the brain, one advantage of the split-brain technique being that it is then possible to approach structures like the subfrontal lobe, which are normally inaccessible without damage to overlying parts of the brain is. Details of the lesions made are given in the relevant sections of the report on 'Results' below. A test for untrainedpreferences After operation, the blinded animals were kept for several days, during which they were fed twice daily with 6 small pieces of fish, similar to those used later as rewards in the training experiments. When they were feeding regularly, each animal was tested to discover its untrained rough-smooth preference on the two sides. Two perspex spheres of 3 cm diameter were used. One was smooth (OR = no rings), the other was rough, with 13 latitudinal grooves, 1 mm wide and 1 mm deep cut into it (13R). Each sphere was suspended on nylon line. To test the preference these objects
105 were presented 64 times in all, one at a time, to arms on the left and right sides of the animal, the order of presentation being (OR shown as + , 13R as - - ) : L RL R L RL + +----+----+
RLR L RL +----+----+
RL
RL RL +----+
R LRL +--+ +
R L R L R L R L R ----++--++---(32tests, andrepeat, making64testsinall). This sequence was chosen to ensure as nearly as possible equality of the various possible successions. Test were at 3-min intervals. No rewards or punishments were given. The animals either took an object, grasping it and passing it under the interbrachial web towards the mouth (the 'positive' response after which the object was jerked away), or rejected it, dropping it after examination with the suckers and/or pushing it away (the 'negative' response).
Experiment 1: positive training and testing On the day after the initial test, positive training to take the smooth ball on the right side was begun. This ball was presented, as before, by touching it against one of the arms. If the octopus took the object it was given a piece of fish. If it rejected it, the object was presented again, together with the piece of fish. This 'correction procedure' normally ceased to be necessary after the first few trials. The objects were given only on the right and wherever possible to the second right arm. There were 8 such positive trials at intervals of 5 or 10 min in each of 2 sessions/day. This technique limits training to the right side of the body, and if care is taken to snatch the object away at the moment when the octopus bends its arm to move the sphere under the web, learning can be unilateral x0. In the present experiments it was not essential to avoid any effect on the left side, and so no special precautions were taken to keep it from contact with the smooth sphere. As a result, the left side as well as the right was generally found to have increased its proportion of takes of OR. After 4 days (8 positive training sessions) the animal was again tested with a series of 64 unrewarded tests, 32 to each side of the body, 16 with OR and 16 with 13R, carried out exactly as before. Four further days of positive training followed, and then a further test. The cycle was repeated 4 times, giving 5 sets of unrewarded tests (including the initial series) and a total of 256 positive training trials in which the right side was rewarded for taking OR.
Experiment 2: discrimination training The positive training phase of the experiment was followed by a period during which both sides of the animal were trained to discriminate between OR, continuing as 'positive' object, and 13R as 'negative'. If 13R was taken, the octopus was given a 10 V AC shock, through electrodes attached to a probe. Discrimination training sessions were of 16 trials, 4 with each object to each side of the body thus:
106 morningsession(readingffomleRtoright): L RL RL RL + +----+----+----+
RL
RL
RL RL R +--+ +--
eveningsession: R L R L ++----
R L + - - and so on.
The morning and evening patterns were interchanged next day, so that every other day began with L + , R + , L - - , R - - . Trials were at 5-min intervals, and sessions were not less than 6 h apart. Training continued for 10 days, 160 trials in all, 80 with each side of the split-brain animal. Experiment 3: extinction and transfer tests
On the day after the end of discrimination training the animals were given a series of extinction and transfer tests. In these they were presented not only with OR and 13R but also with similar spheres having 3, 6 or 9 incised grooves. Each side of each animal was presented with each object 15 times in a series of cycles, every cycle including one presentation of each object to the left hand side (LHS) and one to the right hand side (RHS). The order of presentation within each cycle was determined by reference to a table of random numbers and was not the same for LHS and RHS. Tests to LHS and RHS alternated, the interval between tests to any one side being 3-5 min. Statistical treatment
In considering the results of positive training (experiment 1) an index, 0, was calculated14; 0 is a measure of discrimination that is largely independent of the level of take. It is derived for each animal for each series of tests by scoring the number of occasions during the series when the octopus changed from taking ( + response) to rejecting ( - - response) or vice versa. A change of response from q- to - - is scored as 'correct' if the object rejected is 13R, as 'incorrect' if it is OR; - - to -+- is correct if the object taken is OR, incorrect if 13R. I f the animal continues to take everything or reject everything, there is no score. Thus, in any one series of 32 tests on one side with the sequence shown on p. 105 there may be a change of response from taking smooth to not taking rough, or vice versa on 27 occasions. I f this change is always made correctly the score is 100 ~ or P -- 1. This means of scoring reduces the contribution to the mean score of periods when the animal is either taking or rejecting both objects. The ratio of correct to all changes in sign of response gives a probability (P) of correct response for the animal concerned. The variance of this probability will, however, depend on its magnitude and on the number of changes on which it is based. The effect of the former can be largely eliminated by transforming P into 0 where 0 = sin-l~/ff (see ref. 1). The distribution of 0 (which runs from 0 to 90 °) approximates to the normal distribution rather better than does P and has a variance which depends only on the number of observations on which it is based. The average value of 0 for any group of animals in any session was calculated by multiplying each
107
A
C
D
RIG~t1" OPERAvTED SIDES 1
CONTROLS N=8
100
.
-
N=7
NMIF N=10
7
°z 80/ < 60
//"
4O
/
./
E 2o
i
8
0
LEFT GONTROL SLOES
~1oo 80 //
< 60
~ .o A,"/ ~
20
9 6,412,s,,922,~ 0
0i r64 1,28 1,92256
0641281922561 I I
GROUPS
D
OF UNREWARDED
E
TESTS
G
F
RIGHT OPERATED SLOES NSFD N=4
10C
NSFV N=6
/
O
Z 80 Z
uJ 4 0 >
O a.
lr.°-A'
20
SF
NSF N=6
dam d
N=4
//"
.K /
"A
/
w
0
LEFT GO TROL SIDES 100
80
¥'/
6O
•~
/' '\
d0
,
/
20
.A -I
#t---A...
/
,,/
i 0 64 128 192 256 I
i
i
0 64 128192256 ? 64 1781192256 I I I I t GROUPS OF U N R E W A R D E D TESTS
64 128192256j I I
Fig. 1. Per cent takes of the rough sphere (open circles) and smooth sphere (closed circles) in tests given before and after various amounts of positive training. Triangles show the change in 0 (see p. 106) during the tests. Controls had split brains but no other lesions. Lesions on the right sides of the rest were as follows: NV, vertical lobe removed; NMIF, median inferior frontal removed; NSFD, dorsal subfrontal removed; NSFV, ventral subfrontal removed; SF dam.a, more extensive lesions to subfrontal, and NSF, no subfrontal lobe. The left, control, sides of the same animals were intact. In each case, N = number of animals.
108
z
A
O -
B
I
RHSOR 1 0 0 J - ~
C
p
NV N=7
BILATERAL CONTROLS N:8
NMIF
oo: N.,O
-~ 4o
013~ 6,496 178150 0I TRAININO TRIALS
3I
)
13264~151~811600
9I 1~
EXTINCTION RINGS
RAINING T R I A LS
TRAINING T R I A LS
EXTINCTION RINGS
NSFD N=4
100
80-
0
oo,
D
3I 6I 9I ~3 EXTINCTION RINGS
x w
E
4O
o Z 20
"916128132 ~4~I~13R0, 6, ~~~ ~ TRAI NING EXTINCTION TRAI NING EXTINCTION TRIA L$ RINGS TRIALS RINGS
Z 0
z 100
"e...e.._.e_._~.-~ OR
S F dared N=4
t
NSF N=6
tJ
7,
F
80
Z
60
~
40
G 20
~.~13R
3t 6#961281160173161911332 64 96 128160 0 I
I
I
i
I
3
6
i
i
9 13 ]
i
Fig. 2. Per cent takes of the rough (13R) and smooth (OR) spheres during discrimination training and in extinction tests. Open circles show the performance of the lesioned right hand sides (RHS), closed circles show the performance of the control left hand sides (LHS). Labels for lesioned groups as in Fig. 1.
individual value by the number of observations on which it was based and dividing the total of these scores by the total number of observations. Because animals that happen to take all (or none) of the objects presented in the tests make no score further comparison of these results necessitated a nonorthogonal analysis of variance using left and right and the successive tests as classifications. The program to carry this out was devised for us by the A.R.C. statistics group at the Department of Applied Biology in Cambridge, and the results of this
109
..
,.~ ,.~ ,..~ ~ . . . .
.< 0
cq ,~
0 , - ~~
I ~1 0
z~
~9~0 ~
~
I
~ ,~
,,o~
~ ~
"I.
~
~
I
I
| ,~
~ o
~. o
"¢
.-,
.--,
I
I
I
I
I
I
°°v
o
~.v
~
~
©
~ ~ ~'~'2
©
,~.~
~ ~ ~
= ~
.
~
. ~
I rm
z
~
Io
[,~
~
Io -v
~
I z
~
~
d.,~
103
T H E S U B F R O N T A L LOBE A N D T O U C H L E A R N I N G I N THE OCTOPUS
M. J. WELLS ANDJ. Z. YOUNG Department of Zoology, Downing St., Cambridge and Department of Anatomy, University College, London W.C.1 (Great Britain)
(Accepted January 31st, 1975)
SUMMARY Octopuses with the supraoesophageal lobes of the brain divided longitudinally can be taught to discriminate using the arms on either side. If there is no further lesion the two sides behave alike. Lesions limited to one side did not affect the performance of the contralateral, 'control' side. Lesions made in the vertical (n = 7) lobes led to a slight drop in the quality of performance in training to take a smooth sphere, in discrimination training (rough vs. smooth spheres) and in subsequent extinction and transfer tests. After removal of the median inferior frontal lobe (n = 10) there were somewhat greater effects in the same direction. Much larger effects followed interference with the subfrontal lobe (n = 20). Removal of parts from this always led to a marked loss of capacity for touch learning, broadly dependent on the amount of tissue removed. Removal of the whole of the subfrontal lobe (n = 6) produced animals that showed, at best, only very slight signs of learning. Such animals can adjust their overall level of response as a result of training but they seem incapable of adjusting response levels to two objects independently. These results are discussed in relation to the function of the subfrontal lobe as a memory store.
INTRODUCTION Octopuses can learn to make simple tactile discriminations between objects such as spheres or cylinders that differ in texture. There is evidence that this capacity is lost if all the small cells of the subfrontal lobe* are removed 3,9,1°. Any conclusion * We use the term 'subfrontal' to indicate the whole mass of tissue containing amacrine ceils on the median side of the inferior frontal system. The ventral part of this tissue is continuous with the larger cells of the posterior buccal lobe17.
104 about the memory record needs qualification, however, because octopuses with certain brain lesions and particularly those to the inferior frontal region tend to differ from their controls in two important respects. One is that they sometimes show a marked 'preference' for rough objects (grooved as opposed to smooth), which is not shown by normal (blinded but otherwise undamaged) octopuses 12. The other is that they are liable to accept objects presented at a higher (or more rarely at a lower) proportion of trials than normal animals. While every indication has been that these effects are due to interference with the tactile learning system, it has been impossible to be certain that they are not indirect consequences of interference with other aspects of the animal's physiology. The take level, or tactile sensitivity, might for example vary because the octopus was sick or hungry. In the experiments described below an attempt has been made to overcome these difficulties by using animals with the supraoesophageal part of the brain split by a vertical longitudinal cut. Experiments reported elsewhere have indicated that there is little or no lateral transfer under these conditions ~°,n. This means that the performance of one side of the split-brain animals can be used to control the effect of a lesion in the touch learning system of the other. Comparison of the two should reveal whether one is dealing with specific touch learning or more general physiological effects of the brain damage. Pretraining of the lesioned side to take smooth should eliminate the rough preference associated with lesions to the tactile system, if that side of the animal can learn at all. METHODS
Octopus vulgaris from the Bay of Naples was used. The animals were of both sexes and between 200 and 400 g weight. They were kept for about 4 days after bringing from the sea, until they were feeding readily upon crabs. They were then operated upon under urethane anaesthesia. Each animal was blinded by cutting the optic nerves. The cranium was opened and the supraoesophageal part of the brain split in the mid-line by means of a narrow blade entering from the front. Bilateral control animals were left in this condition. In the others lesions were made on the righthand side of the brain, one advantage of the split-brain technique being that it is then possible to approach structures like the subfrontal lobe, which are normally inaccessible without damage to overlying parts of the brain is. Details of the lesions made are given in the relevant sections of the report on 'Results' below. A test for untrainedpreferences After operation, the blinded animals were kept for several days, during which they were fed twice daily with 6 small pieces of fish, similar to those used later as rewards in the training experiments. When they were feeding regularly, each animal was tested to discover its untrained rough-smooth preference on the two sides. Two perspex spheres of 3 cm diameter were used. One was smooth (OR = no rings), the other was rough, with 13 latitudinal grooves, 1 mm wide and 1 mm deep cut into it (13R). Each sphere was suspended on nylon line. To test the preference these objects
105 were presented 64 times in all, one at a time, to arms on the left and right sides of the animal, the order of presentation being (OR shown as + , 13R as - - ) : L RL R L RL + +----+----+
RLR L RL +----+----+
RL
RL RL +----+
R LRL +--+ +
R L R L R L R L R ----++--++---(32tests, andrepeat, making64testsinall). This sequence was chosen to ensure as nearly as possible equality of the various possible successions. Test were at 3-min intervals. No rewards or punishments were given. The animals either took an object, grasping it and passing it under the interbrachial web towards the mouth (the 'positive' response after which the object was jerked away), or rejected it, dropping it after examination with the suckers and/or pushing it away (the 'negative' response).
Experiment 1: positive training and testing On the day after the initial test, positive training to take the smooth ball on the right side was begun. This ball was presented, as before, by touching it against one of the arms. If the octopus took the object it was given a piece of fish. If it rejected it, the object was presented again, together with the piece of fish. This 'correction procedure' normally ceased to be necessary after the first few trials. The objects were given only on the right and wherever possible to the second right arm. There were 8 such positive trials at intervals of 5 or 10 min in each of 2 sessions/day. This technique limits training to the right side of the body, and if care is taken to snatch the object away at the moment when the octopus bends its arm to move the sphere under the web, learning can be unilateral x0. In the present experiments it was not essential to avoid any effect on the left side, and so no special precautions were taken to keep it from contact with the smooth sphere. As a result, the left side as well as the right was generally found to have increased its proportion of takes of OR. After 4 days (8 positive training sessions) the animal was again tested with a series of 64 unrewarded tests, 32 to each side of the body, 16 with OR and 16 with 13R, carried out exactly as before. Four further days of positive training followed, and then a further test. The cycle was repeated 4 times, giving 5 sets of unrewarded tests (including the initial series) and a total of 256 positive training trials in which the right side was rewarded for taking OR.
Experiment 2: discrimination training The positive training phase of the experiment was followed by a period during which both sides of the animal were trained to discriminate between OR, continuing as 'positive' object, and 13R as 'negative'. If 13R was taken, the octopus was given a 10 V AC shock, through electrodes attached to a probe. Discrimination training sessions were of 16 trials, 4 with each object to each side of the body thus:
106 morningsession(readingffomleRtoright): L RL RL RL + +----+----+----+
RL
RL
RL RL R +--+ +--
eveningsession: R L R L ++----
R L + - - and so on.
The morning and evening patterns were interchanged next day, so that every other day began with L + , R + , L - - , R - - . Trials were at 5-min intervals, and sessions were not less than 6 h apart. Training continued for 10 days, 160 trials in all, 80 with each side of the split-brain animal. Experiment 3: extinction and transfer tests
On the day after the end of discrimination training the animals were given a series of extinction and transfer tests. In these they were presented not only with OR and 13R but also with similar spheres having 3, 6 or 9 incised grooves. Each side of each animal was presented with each object 15 times in a series of cycles, every cycle including one presentation of each object to the left hand side (LHS) and one to the right hand side (RHS). The order of presentation within each cycle was determined by reference to a table of random numbers and was not the same for LHS and RHS. Tests to LHS and RHS alternated, the interval between tests to any one side being 3-5 min. Statistical treatment
In considering the results of positive training (experiment 1) an index, 0, was calculated14; 0 is a measure of discrimination that is largely independent of the level of take. It is derived for each animal for each series of tests by scoring the number of occasions during the series when the octopus changed from taking ( + response) to rejecting ( - - response) or vice versa. A change of response from q- to - - is scored as 'correct' if the object rejected is 13R, as 'incorrect' if it is OR; - - to -+- is correct if the object taken is OR, incorrect if 13R. I f the animal continues to take everything or reject everything, there is no score. Thus, in any one series of 32 tests on one side with the sequence shown on p. 105 there may be a change of response from taking smooth to not taking rough, or vice versa on 27 occasions. I f this change is always made correctly the score is 100 ~ or P -- 1. This means of scoring reduces the contribution to the mean score of periods when the animal is either taking or rejecting both objects. The ratio of correct to all changes in sign of response gives a probability (P) of correct response for the animal concerned. The variance of this probability will, however, depend on its magnitude and on the number of changes on which it is based. The effect of the former can be largely eliminated by transforming P into 0 where 0 = sin-l~/ff (see ref. 1). The distribution of 0 (which runs from 0 to 90 °) approximates to the normal distribution rather better than does P and has a variance which depends only on the number of observations on which it is based. The average value of 0 for any group of animals in any session was calculated by multiplying each
107
A
C
D
RIG~t1" OPERAvTED SIDES 1
CONTROLS N=8
100
.
-
N=7
NMIF N=10
7
°z 80/ < 60
//"
4O
/
./
E 2o
i
8
0
LEFT GONTROL SLOES
~1oo 80 //
< 60
~ .o A,"/ ~
20
9 6,412,s,,922,~ 0
0i r64 1,28 1,92256
0641281922561 I I
GROUPS
D
OF UNREWARDED
E
TESTS
G
F
RIGHT OPERATED SLOES NSFD N=4
10C
NSFV N=6
/
O
Z 80 Z
uJ 4 0 >
O a.
lr.°-A'
20
SF
NSF N=6
dam d
N=4
//"
.K /
"A
/
w
0
LEFT GO TROL SIDES 100
80
¥'/
6O
•~
/' '\
d0
,
/
20
.A -I
#t---A...
/
,,/
i 0 64 128 192 256 I
i
i
0 64 128192256 ? 64 1781192256 I I I I t GROUPS OF U N R E W A R D E D TESTS
64 128192256j I I
Fig. 1. Per cent takes of the rough sphere (open circles) and smooth sphere (closed circles) in tests given before and after various amounts of positive training. Triangles show the change in 0 (see p. 106) during the tests. Controls had split brains but no other lesions. Lesions on the right sides of the rest were as follows: NV, vertical lobe removed; NMIF, median inferior frontal removed; NSFD, dorsal subfrontal removed; NSFV, ventral subfrontal removed; SF dam.a, more extensive lesions to subfrontal, and NSF, no subfrontal lobe. The left, control, sides of the same animals were intact. In each case, N = number of animals.
108
z
A
O -
B
I
RHSOR 1 0 0 J - ~
C
p
NV N=7
BILATERAL CONTROLS N:8
NMIF
oo: N.,O
-~ 4o
013~ 6,496 178150 0I TRAININO TRIALS
3I
)
13264~151~811600
9I 1~
EXTINCTION RINGS
RAINING T R I A LS
TRAINING T R I A LS
EXTINCTION RINGS
NSFD N=4
100
80-
0
oo,
D
3I 6I 9I ~3 EXTINCTION RINGS
x w
E
4O
o Z 20
"916128132 ~4~I~13R0, 6, ~~~ ~ TRAI NING EXTINCTION TRAI NING EXTINCTION TRIA L$ RINGS TRIALS RINGS
Z 0
z 100
"e...e.._.e_._~.-~ OR
S F dared N=4
t
NSF N=6
tJ
7,
F
80
Z
60
~
40
G 20
~.~13R
3t 6#961281160173161911332 64 96 128160 0 I
I
I
i
I
3
6
i
i
9 13 ]
i
Fig. 2. Per cent takes of the rough (13R) and smooth (OR) spheres during discrimination training and in extinction tests. Open circles show the performance of the lesioned right hand sides (RHS), closed circles show the performance of the control left hand sides (LHS). Labels for lesioned groups as in Fig. 1.
individual value by the number of observations on which it was based and dividing the total of these scores by the total number of observations. Because animals that happen to take all (or none) of the objects presented in the tests make no score further comparison of these results necessitated a nonorthogonal analysis of variance using left and right and the successive tests as classifications. The program to carry this out was devised for us by the A.R.C. statistics group at the Department of Applied Biology in Cambridge, and the results of this
109
..
,.~ ,.~ ,..~ ~ . . . .
.< 0
cq ,~
0 , - ~~
I ~1 0
z~
~9~0 ~
~
I
~ ,~
,,o~
~ ~
"I.
~
~
I
I
| ,~
~ o
~. o
"¢
.-,
.--,
I
I
I
I
I
I
°°v
o
~.v
~
~
©
~ ~ ~'~'2
©
,~.~
~ ~ ~
= ~
.
~
. ~
I rm
z
~
Io
[,~
~
Io -v
~
I z
~
~
d.,~
