Brain Research, 178 (1979) 89-98 © Elsevier/North-Holland Biomedical Press
89
ENRICHMENT-ISOLATION, CORTEX LENGTH AND THE RANK ORDER EFFECT
ROBERT ASHLEY CUMMINS and PETER JAMES LIVESEY Institute of Special Education, Burwood State College, Burwood, Victoria 3125, and Department of Psychology, University of Western Australia, Nedlands, Western Australia 6009 (Australia)
(Accepted March 22nd, 1979)
Key words:
enrichment - - isolation - - cortex length - - rank order effect - - brain development - - theoretical model
SUMMARY In a previous paper we have proposed a developmental theory to account for the neurological changes which result from differentially rearing animals in either enrichment or isolation. On the basis of brain weight measurements we suggested that the primary cause of the differential development could be traced to retarded neurological growth in the isolated animals. The present 9 studies test the generality of this theory by applying it to the cortex length changes induced in rats by differential rearing periods of between 18 and 160 days. In the light of this new data the theory has been revised to the extent that the developmental ceiling for the dependent variable is now considered to change with age instead of being fixed. Two major consequences of this revision are as follows. Firstly that the environmentally induced changes in cortical development are seen as persistent. Secondly that the Rank Order Effect is shown to be a transitory phenomenon which exists only when some, but not all of the isolate values have been restricted in their development.
INTRODUCTION It is now generally accepted that the anteroposterior length of the rat cortex can be altered by differential rearing, although for some time the reliability of this effect was in doubt. In 1968, Altman et al. 1 were the first to report that environmental enrichment caused the cortex to grow longer, but in the following year Rosenzweig and Bennett 15 were unable to detect a cortex length change in either rats or gerbils. However, the two studies were not strictly comparable since Rosenzweig and Bennett had used 30 days of differential rearing, a duration previously considered optimal for the demonstration of enrichment-isolation change, whereas Altman et al. had used 90 days.
90 This issue was resolved when it was found that, unlike the parameters of forebrain weight 6, cortex weight13,16 and cortex depth 7, the enrichment-isolation differences in cortex length continued to develop beyond 30 days19, 2°. We have recently proposed a developmental theory to account for the neurological changes which result from differentially rearing animals in either enrichment or isolation 6. This theory was based on studies of forebrain weight where we proposed that the primary cause of differential development could be traced to retarded development in the isolated animals. This conclusion was based on an observation derived from rank-order comparisons of the enriched and isolated groups. We reported that the differences between the enriched and isolated values were not evenly distributed throughout all ranks but tended to be largest between the lowest ranking values from each group. We labelled this phenomenon the Rank Order Effect. The data base upon which the developmental theory was formulated comprised forebrain weight measurements on rats differentially reared for 18-120 days after weaning. A critical feature of the forebrain weight data was that optimal effects were evident with enrichment-isolation periods of 18-40 days. Beyond this time the environmentally induced differences became less evident. On the other hand, cortex length findings strongly suggest that the optimal effect becomes evident only after much longer periods of differential rearing. The present studies were designed to test both the generality of the Rank Order Effect in the parameter of cortex length and to describe the temporal characteristics of environmentally induced changes in this variable. MATERIALS AND METHODS Of the following 9 studies, those differentially reared from between 18 and 120 days comprise animals used previously in the temporal description of forebrain weight changes 6. All animals were male random-bred albino rats. For each study, the rats were separated at weaning into one of two groups. In the enriched condition, the rats lived in social groups of between 6 and 12 animals in large open-mesh cages supplied daily with a variety of 'toys'. In the isolated condition, the rats were confined to small, individual, solid-walled cages with mesh floors and ceilings. The duration of exposure to these differential rearing conditions is given by the group number, from which it can be seen that the range was from 18 to 160 days. Three separate studies were run for a period of 30 days (30a, b, c). Procedure At the end of each rearing period the animals in that group were killed and weighed, and their brains were removed. After fixation in formol-saline, the olfactory bulbs were removed and the brains sectioned coronally immediately behind the posterior pole of the cerebrum. Each brain was then placed on the stage of a binocular dissecting microscope fitted with a cross-wire in one eye-piece. Both lateral and vertical movements of the microscope stage could be recorded on a Vernier scale accurate to 0.1 ram. The length
91 of each cerebral hemisphere, from anterior to posterior pole, was determined at a magnification of 12.5. The two results for each b r a i n were then averaged to give a single estimate of cortex length. As a test of m e a s u r e m e n t reliability, 42 brains were remeasured yielding a reliability coefficient of 0.92. All m e a s u r e m e n t s were m a d e w i t h o u t knowledge o f b r a i n identity. As variable times of f o r m a l i n fixation were used for the separate studies, n o comparisons o f absolute cortex length could be made between studies. However, within any one study the enriched a n d isolated brains were treated identically. Results
The e n r i c h m e n t - i s o l a t i o n difference in cortex length was significant in 6 of the 9 studies. There was a tendency for the youngest animals to show the least a m o u n t of difference. While Studies 80 to 160 all showed a significant difference, at least at the 1 ~ level, three of the shorter studies (18, 30a a n d 60) were non-significant, while Studies 30b a n d 30c were significant only at the 5 ~ level. As in the previous investigation of the forebrain weight changes 6 the influence of b o d y weight as a d e t e r m i n i n g factor for cortex length was estimated by the productm o m e n t correlation between these two variables. I n general, the two measures were largely i n d e p e n d e n t of one a n o t h e r (r 0.04-0.32), except in Studies 30a a n d 160, where TABLE 1 Cortex length differences between the environmental groups
The test for the Rank Order Effect is a comparison of regression line slope between the rank ordered enriched and isolated values (see text). L, cortex length; E, enriched; I, isolated; N, number of animals. Study
Parameter
N
18
30a
30b
30e
60
80
120
154
160
L 36
L/Body L 36 40
L 36
L 36
L 36
L 38
L 40
L/Body 36
2.0 1.310 n.s.
15.0 1.9 2.1 1.3 2.5 4.9 1.804 2.175 1.989 0.659 2.954 4.051 0.05 0.05 0.05 n.s. 0.01 0.001
1.9 15.0 2 . 5 2 0 4.185 0.01 0.001
1.1 1.322 n.s.
8.0 1.163 n.s.
1.7 2.138 0.05
2.8 2.321 0,05
1.2 3.2 0.957 3.530 n.s. 0.01
5.6 9.375 0.001
2.2 3.819 0.01
15.2 4.634 0.001
2.9 3.192 0.01
26.1 2.0 7.650 2.996 0.001 0.01
1.4 2.846 0.01
1.3 2,031 0.05
1.7 3,171 0.01
4.2 3.068 0.01
1.6 2.230 0.05
14.9 3.685 0.01
13.57 0.001
0,291 n.s.
16.14 0,001
8.262 0.01
1,016 n.s.
5.933 0.05
E~I
E~I
Total
E > I (~) t P I (~) t P < Low ranks
E ~ I (~) t P
E
E~I
E~I
92 significant correlation was found within both of the environmental treatments (r 0.52-0.87). The influence of this confounding variable on the cortex lengths of these two studies was removed by forming a cortex length/body weight ratio. Using this ratio the enriched animals exceeded the isolates in both Studies 30a and 160. For Study 160, the result was similar to that found for cortex length alone, the two environmental treatments differing significantly from one another. For Study 30a, the result was markedly changed from that derived from cortex length alone, in which no overall difference between the environmental groups had been found. The net result of these manipulations was that, of the 9 studies, only Studies 18 and 60 failed to display an overall enrichment-isolation difference in cortex length (Table I). In order to examine the 9 studies for evidence of the Rank Order Effect, each of the environmental groups were separated into high and low ranks with the median rank being chosen as the point of demarcation. Two separate enrichment isolation comparisons were then performed within each study, one between the high ranking values and one between the low ranking values. From Table I it can be seen that only the high ranking values of Studies 18, 30a and 60 failed to support a significant a
LENGTH
30a LENGTH/BODY
,..,
30b
LENGTH
/
=w,
150
./
26%
8%
13
I00
;,'.. 5
(0
15
/
J
,.,.,. 50
''5
50c LENGTH
;o. . . .
~ ....
80
LE L.
90
....
LENGTH '--"
8C
;;%
16
"2,,,
15
1.26%
5
I0
15
5
120
I0 154 LENGTH
5
15
IO
15
160 LENGiH / B O ~ _ 500
14-5
400.~
13-5
2-2% 12"5 'fJ. . . . . 5
I ......... 0 15
5
I0
15
•
.~ 3(2( 2O
15'2%
5
I0
15
Fig. 1. The rank order comparisons between the enriched and isolated cortex lengths for each study. Ordinate = length; abscissa = rank order. Each rank order comparison has been divided by a vertical line at the median of each enriched (solid line) and isolated (broken line) group. The extent of the enrichment-isolation difference (EC > IC ~) is indicated separately for those values lying below the median (low ranks) and above the median (high ranks).
,
a0
93 T A B L E II
Variance differences between enriched and isolated subjects in both high ranking and low ranking animals
Study
Parameter N
18
30a
Length
Length/ Length Length Body 9 10 9
Length Length
14.825 0.05
15.330 0.05
21.261 0.01
43.415 0.001
12.477 n.s.
I > E
I > E
E > I
E > I
9
High ranks Z~ P < Direction o f difference Low ranks Z~
P< Direction o f difference
30b
30c
60
9
80
9
120
154
Length
Length Length
9
25.097 0.01
20.946 0.05
E > I
E > I
12.514
10.978
10.963
10.240
12.008
15.600
n.s.
n.s.
n.s.
n.s.
n.s.
0.05
9.381
n.s.
10 10.970 n.s.
160
9 32.818 0.001
E > I 13.391
14.518
n.s.
n.s.
E ~ I
environmental effect. For each of these three studies, the proportion of the overall enrichment-isolation difference (EC~IC~o) contributed by the low ranks exceeded that of the high ranks. In 5 out of the 6 remaining studies, the proportion of the overall difference contributed by the high ranking values predominated. Tests for the significant occurrence of the Rank Order Effect were performed by comparing the slopes of the enriched and isolated regression lines of cortex length to rank order within each studylL In Studies 18 and 30a the slope of the isolates was found to exceed that of the enriched. For Studies 30c, 80, 120 and 160, this effect was significantly reversed, the slope of the enriched lengths exceeding that of the isolates. The rank order comparisons between the enriched and isolated cortex lengths are represented in Fig. 1. In Table II the within-group variances for enriched and isolated subjects are compared separately for high and low ranking values using the formula (n-l) variancex variance y ' whose values conform to the chi-square distribution (Sokal and Rohlf, ref. 17, p. 171). Only in Studies 18 and 30a did the variance of the high ranking isolates exceed that of the high ranking enriched. In the remaining studies, there was a tendency for the high rank comparisons to show a greater variance for the enriched animals. This was significant for Studies 30b, 30c, 80, 120 and 160. Only one significant variance asymmetry was found within the low ranking subjects. This was in Study 80, where the variance of the enriched exceeded that of the isolates. DISCUSSION
In order for the results to be compatible with the model of brain development previously described for forebrain weight6, the environmentally induced differences in
94
cortex length should be most pronounced between the low ranking values from each group and the within group variance of the isolates should exceed the enriched. In fact only two of the briefest studies, 18 and 30a, provided such data. The results for the remaining 7 studies tend to be either non-significant or incompatible with the model. In the light of this new data the model was revised. This revision is presented in Fig. 2 and differs from the original in that the developmental ceiling is now represented as a variable which changes with age rather than being fixed. The extent of the environment dependent neural development is severely limited in young animals by temporal constraints upon growth. These constraints are imposed upon all neural development by metabolic factors. However, as the animal ages there is a greater potential for such environmentally induced development to accumulate with the result that the developmental ceiling is raised. For the isolates the level of environment support is insufficient for this potential to be realized, and as the ceiling for development is raised there will be little corresponding environment dependent neural development. The enriched animals, on the other hand, will continue to accrue such development, albeit at a diminishing rate as the animals progressively adapt to their environment. There are two major consequences of these temporal considerations. (a) The development consequences of limited environment will become more pronounced with IOO%
AGE A i ~ .
(AGE A + 2 X )
...'''"
100% (AGE A+X)
AGE A+X
,.....,-
I i
~ VARIANCE A4 2X 'IARIANCE A~.X
AGE A I00 ~te (AGE A)
ii
..... ;---- ~
~ - - - - ~
DIAN A
I
:l
i.. q.
-ENRICHMENT
IANCE A
k~ e~ c~ q: k. ILl ~SOLATION
k
OIL
MINIMAL
OPTIMAL LEVEL OF
SENSO RY
STIMULATION
Fig. 2. The revised theoretical model of environment dependent neural development. Three stages of development are represented (Ages A, A + X, A + 2X) and the relative extent of enriched or isolated development at each stage is depicted in terms of the median and variance for each environmental group.
95 increasing durations of exposure. (b) The previously described Rank Order Effect6 will appear as a transitory phenomenon, which exists only during the intermediate stage of environmental influence when some, but not all of the isolate values have been restricted in their development. At the stage when all of the isolate values are reduced below the enriched the Rank Order Effect will disappear. The revised model in Fig. 2 leads to three characteristics of the within-group and between-group variance distributions which are explained as follows. (1) As the animals age in their respective environments there should be a tendency for the variance pattern to reverse, i.e. for the variance of the enriched lengths to exceed that of the isolates. The reason for this is that the continued development of the isolates is limited by the environment to a greater extent than the enriched (Fig. 2). In the early stages of differential rearing, it has been argued that a few of the isolates would be able to achieve the relatively small amounts of environment dependent development necessary to approximate the early developmental ceiling6. As the ceiling is raised, however, even the most responsive isolates are unable to achieve the necessary development to realise this potential. Thus, the variance of the isolates will tend to remain stable once this initial environment dependent development is achieved. By analogous reasoning, it is postulated that for the enriched animals the initial period of enrichment will provide sufficient stimulation for most animals to achieve their developmental potential at that stage. However, as the animals progressively adapt to this enriched environment with prolonged exposure periods, the number of animals able to extract sufficient further stimulation for the maintenance of full development will diminish. The increasing influence of these individual differences in the neurological response to enrichment will be reflected by an increasing variance of the dependent parameter as the period of differential rearing is extended. The cortex length data are consistent with this description. In Table II it can be seen that, while for two of the shortest exposure times, the variance of the isolate lengths exceeds that of the enriched, for all longer exposures either the variance of the enriched lengths exceeds that of the isolates, or the inter-group comparison was nonsignificant. (2) As the animals age, there is a tendency for the variance of the high ranking enriched cortex lengths to exceed that of the high ranking isolates, while the variance between the low ranking lengths will remain largely unchanged. The development of the elevated variance with prolonged rearing should be confined to the high ranking enriched values, where some animals are likely to approximate the developmental ceiling. The low ranking enriched lengths, on the other hand, show less change with prolonged rearing. It was argued previously6 that the low ranking values in the isolated group were most likely to be contributed by animals retarded in their environment dependent neural development. By analogous reasoning, the proportion of the low ranking enriched values representing sub-optimal development will increase as the animals adapt to their environment and the level of sensory stimulation becomes progressively the limiting factor to their continued development. One effect of this would be that the median for the enriched group would rise at a slower rate than the developmental ceiling. The variance below the median
96 will therefore show progressively less temporal change as the animals in this region increasingly represent those which are no longer extracting any benefit from their continued exposure to enrichment. The variance of the isolates, both above and below the median, should be unchanged with prolonged rearing. Their development has already been limited during the early period of exposure. The data are consistent with a progressive development of an increasing variance restricted to the high ranking enriched values. Table II compares the variance differences between the enriched and isolated subjects for the high ranking and low ranking cortex lengths separately. Only in two early studies, 18 and 30a, did the variance of the high ranking isolates exceed the high ranking enriched. There was a marked tendency for the high rank comparisons in the remaining studies to indicate a greater variance for the enriched animals. This was significant for Studies 30b, 30c, 80, 120 and 160. On the other hand, only one significant variance asymmetry was found within the low rank comparisons. (3) As the animals age, there is a tendency for the Rank Order Effect to indicate a larger environmental influence between the high ranking lengths (Positive Rank Order Effect) than between the low ranking lengths (Negative Rank Order Effect). This follows from the previous argument that the predominant environmental influence with prolonged rearing will be to increase the magnitude of the high ranking enriched lengths. The slope of the enriched regression line will gradually increase as the high ranking enriched follow the upward drift of the developmental ceiling, whilst the lengths of the other ranks, enriched and isolated, remain relatively stable over the same period. However there is a sine qua non associated with the occurrence of a Positive Rank Order Effect. This is that it must always be associated with a significant enrichment-isolation difference within both high and low ranking values. In order for the Positive Rank Order Effect to develop the isolate lengths must remain retarded while some of the enriched continue to develop. Thus the intergroup differences will be enhanced to include a difference between the high ranking values, while the differences between the low ranking values will be retained. The data is consistent with this description. In Table I it can be seen that a Positive Rank Order Effect was found for Studies 30c, 80, 120, and 160 and that, for each of these studies, a significant difference was present between the low and high ranking values.
The revised model and forebrain weight The preceding three requirements of the revised model of environment dependent neural development appear to be consistent with the data from the cortex length measurements. The question now arises as to why the two parameters of forebrain weight and cortex length behave so differently. The environmental influence on forebrain weight was shown to be dissipated with prolonged rearing 6 but the effect on cortex length would appear to become more marked with increasing maturity (Table I, Fig. 1). One explanation may lie in the degree to which each parameter is a true
97 reflection of environment dependent development, or Js contaminated by concomitant growth, unrelated to the differential rearing per se. Whilst the most rapid phase of the brain weight development has occurred prior to weaning tl, there is a continuous, negatively accelerating increase in brain weight for a considerable portion of the animal's life3,9,10. However the cortical and sub-cortical components do not make an equal contribution to this age-related brain weight increase. It has been reported by Bennett et al. 3, that while both the cortex and sub-cortex (whole brain minus cortex) continue to grow after weaning, the growth of the cortex occurs at a much slower rate than the rest of the brain. As a result, maturation leads to a marked decrease in the cortex/sub-cortex weight ratio4,12,13. It has also been reported that the dimensions of the sub-cortex are generally not susceptible to change as a result of enrichment-isolation experience. Neither the weight of the sub-cortexZ,9,14, nor the dimensions of the diencephalon s have been found to change, even though the same studies reported marked cortical responses. In 1969 we reported that the dimensions of the hippocampus changed with differential rearing is. In an extensive series of studies Diamond et al.7 were unable to repeat this result, so it would seem unlikely that the hippocampal dimensions change reliably with differential rearing. It would therefore seem that the parameter of forebrain weight is subject to two influences acting to determine its postweaning development. The slow, differential accumulation of environment dependent development is superimposed upon a continuous increase in overall brain size which may be related to general somatic growth. It has previously been reported that there is a consistent positive correlation between brain weight and body weight, which is very marked in young animals, becoming less so in adultsS, 11. In the present series of studies there was a consistent positive correlation between body and forebrain weight which, on average, amounted to about 20 ~ shared variance. The forebrain weight changes with different durations of rearing can then be explained. The initial environmental influence is to retard cortical development in some of the isolates. This induces the Negative Rank Order Effect. As the period of differential rearing is extended, the capacity of the environments to induce further change is diminished and the changes in environment dependent development are gradually overshadowed by the continuing process of general brain growth. The majority of this continued growth occurs in the sub-cortex which is minimally responsive to the environmentally induced changes. The extent of this development may be highest in the isolated animals, since in the present series of studies the isolated animals consistently weighed more than the enriched (unreported data). The transitory nature of the forebrain weight changes may therefore not represent a diminution of differential environment dependent development as we previously suggested but, rather, the difference in development may be maintained but be obscured by brain growth related to general somatic development. ACKNOWLEDGEMENTS This research was supported by a grant from the Australian Research Grants Committee.
98 REFERENCES 1 Altman, J., Wallace, R. B., Anderson, W. J. and Das, V. G., Behaviorally induced changes in the length of the cerebrum in rats, Develop. PsychobioL, 1 (1968) 112-I 17. 2 Bennett, E. L., Krech, D. and Rosenzweig, M. R., Reliability and regional specificity of cerebral effects of environmental complexity and training, J. comp. PhysioL Psychol., 57 (1964) 440-441. 3 Bennett, E. L., Rosenzweig, M. R. and Diamond, M. C., Time courses of effects of differential experience on brain measures and behavior of rats. In W. L. Byrne (Ed.), Molecular Approaches to Learning and Memory, Academic Press, New York, 1970, pp. 55 89. 4 Bennett, E. L. and Rosenzweig, M. R., Chemical alterations produced in brain by environment and training. In A. Lajtha (Ed.), Handbook ofNeurochemistry, Vol. 6, Plenum Press, New York, 1971, pp. 173-201. 5 Clark, G. M. and Zamenhof, S., Correlations between cerebral and cortical parameters in the developing and mature rat brain, Int. J. Neurosci., 5 (1973) 223-229. 6 Cummins, R. A., Livesey, P. J., Evans, J. G. M. and Walsh, R. N., A developmental theory of environmental enrichment, Science, 197 (1977) 692-694. 7 Diamond, M. C., Ingham, C. A., Johnson, R. E., Bennett, E. L. and Rosenzweig, M. R., Effects of environment on morphology of rat cerebral cortex and hippocampus, J. NeurobioL, 7 (1976) 75-85. 8 Diamond, M. C., Law, F., Rhodes, H., Lindner, B., Rosenzweig, M. R., Krech, D. and Bennett, E. L., Increases in cortical depth and glial numbers in rats subjected to enriched environments, J. comp. Neurol., 128 (1966) 117-125. 9 Ferchmin, P. A., Eterovic, V. A. and Caputto, R., Studies of brain weight and R N A content after short periods of exposure to environmental complexity, Brain Research, 20 (1970) 49-57. 10 Fuller, J. L. and Geils, H. D., Brain growth in mice selected for high and low brain weight, Develop. Psychobiol., 5 (1972) 307-318. 11 Kobayashi, T., Brain-to-body ratios and time of maturation of the mouse brain, Amer. J. Physiol., 204 (1963) 343-346. 12 Riege, W. H., Environmental influences on brain and behavior of year-old rats, Develop. Psychobiol., 4 (1971) 157-167. 13 Rosenzweig, M. R., Mechanisms by which environmental complexity may alter brain chemistry and anatomy, In Int. Congr. PsychoL, London, July 29, 1969. 14 Rosenzweig, M. R. and Bennett, E. L., Drugs modulate effects of environment on brain growth,
Proc. 76th Ann. Convention Amer. Psychol. Ass., 1968. 15 Rosenzweig, M. R. and Bennett, E. L., Effects of differential environments on brain weights and enzyme activities in gerbils, rats and mice, Develop. Psychobiol., 2 (1969) 87-95. 16 Rosenzweig, M. R., Bennett, E. L. and Diamond, M. C., Effects of differential environments on brain anatomy and brain chemistry. In J. Zubin and G. Jervis (Eds.), Psychopathology of Mental Development, Grune and Stratton, New York, 1967, pp. 45-56. 17 Sokal, R. R. and Rohlf, F. J., Biometry, Freedman, San Francisco, 1969. 18 Walsh, R. N., Budtz-Olsen, O. E., Penny, J. E. and Cummins, R. A., The effects of environmental complexity on the histology of the rat hippocampus, J. comp. NeuroL, 137 (1969) 361-366. 19 wa•sh• R. N.• Budtz-O•sen• •. E.• T•r•k•A. and Cummins• R. A.• Envir•nmenta••y induced changes in the dimensions of the rat cerebrum, Develop. PsychobioL, 4 (1971) 115-122. 20 Walsh, R. N., Cummins, R. A. and Budtz-Olsen, O. E., Environmentally induced changes in the dimensions of the rat cerebrum: a replication and extension, Develop. Psychobiol., 6 (1973) 3-8.
89
ENRICHMENT-ISOLATION, CORTEX LENGTH AND THE RANK ORDER EFFECT
ROBERT ASHLEY CUMMINS and PETER JAMES LIVESEY Institute of Special Education, Burwood State College, Burwood, Victoria 3125, and Department of Psychology, University of Western Australia, Nedlands, Western Australia 6009 (Australia)
(Accepted March 22nd, 1979)
Key words:
enrichment - - isolation - - cortex length - - rank order effect - - brain development - - theoretical model
SUMMARY In a previous paper we have proposed a developmental theory to account for the neurological changes which result from differentially rearing animals in either enrichment or isolation. On the basis of brain weight measurements we suggested that the primary cause of the differential development could be traced to retarded neurological growth in the isolated animals. The present 9 studies test the generality of this theory by applying it to the cortex length changes induced in rats by differential rearing periods of between 18 and 160 days. In the light of this new data the theory has been revised to the extent that the developmental ceiling for the dependent variable is now considered to change with age instead of being fixed. Two major consequences of this revision are as follows. Firstly that the environmentally induced changes in cortical development are seen as persistent. Secondly that the Rank Order Effect is shown to be a transitory phenomenon which exists only when some, but not all of the isolate values have been restricted in their development.
INTRODUCTION It is now generally accepted that the anteroposterior length of the rat cortex can be altered by differential rearing, although for some time the reliability of this effect was in doubt. In 1968, Altman et al. 1 were the first to report that environmental enrichment caused the cortex to grow longer, but in the following year Rosenzweig and Bennett 15 were unable to detect a cortex length change in either rats or gerbils. However, the two studies were not strictly comparable since Rosenzweig and Bennett had used 30 days of differential rearing, a duration previously considered optimal for the demonstration of enrichment-isolation change, whereas Altman et al. had used 90 days.
90 This issue was resolved when it was found that, unlike the parameters of forebrain weight 6, cortex weight13,16 and cortex depth 7, the enrichment-isolation differences in cortex length continued to develop beyond 30 days19, 2°. We have recently proposed a developmental theory to account for the neurological changes which result from differentially rearing animals in either enrichment or isolation 6. This theory was based on studies of forebrain weight where we proposed that the primary cause of differential development could be traced to retarded development in the isolated animals. This conclusion was based on an observation derived from rank-order comparisons of the enriched and isolated groups. We reported that the differences between the enriched and isolated values were not evenly distributed throughout all ranks but tended to be largest between the lowest ranking values from each group. We labelled this phenomenon the Rank Order Effect. The data base upon which the developmental theory was formulated comprised forebrain weight measurements on rats differentially reared for 18-120 days after weaning. A critical feature of the forebrain weight data was that optimal effects were evident with enrichment-isolation periods of 18-40 days. Beyond this time the environmentally induced differences became less evident. On the other hand, cortex length findings strongly suggest that the optimal effect becomes evident only after much longer periods of differential rearing. The present studies were designed to test both the generality of the Rank Order Effect in the parameter of cortex length and to describe the temporal characteristics of environmentally induced changes in this variable. MATERIALS AND METHODS Of the following 9 studies, those differentially reared from between 18 and 120 days comprise animals used previously in the temporal description of forebrain weight changes 6. All animals were male random-bred albino rats. For each study, the rats were separated at weaning into one of two groups. In the enriched condition, the rats lived in social groups of between 6 and 12 animals in large open-mesh cages supplied daily with a variety of 'toys'. In the isolated condition, the rats were confined to small, individual, solid-walled cages with mesh floors and ceilings. The duration of exposure to these differential rearing conditions is given by the group number, from which it can be seen that the range was from 18 to 160 days. Three separate studies were run for a period of 30 days (30a, b, c). Procedure At the end of each rearing period the animals in that group were killed and weighed, and their brains were removed. After fixation in formol-saline, the olfactory bulbs were removed and the brains sectioned coronally immediately behind the posterior pole of the cerebrum. Each brain was then placed on the stage of a binocular dissecting microscope fitted with a cross-wire in one eye-piece. Both lateral and vertical movements of the microscope stage could be recorded on a Vernier scale accurate to 0.1 ram. The length
91 of each cerebral hemisphere, from anterior to posterior pole, was determined at a magnification of 12.5. The two results for each b r a i n were then averaged to give a single estimate of cortex length. As a test of m e a s u r e m e n t reliability, 42 brains were remeasured yielding a reliability coefficient of 0.92. All m e a s u r e m e n t s were m a d e w i t h o u t knowledge o f b r a i n identity. As variable times of f o r m a l i n fixation were used for the separate studies, n o comparisons o f absolute cortex length could be made between studies. However, within any one study the enriched a n d isolated brains were treated identically. Results
The e n r i c h m e n t - i s o l a t i o n difference in cortex length was significant in 6 of the 9 studies. There was a tendency for the youngest animals to show the least a m o u n t of difference. While Studies 80 to 160 all showed a significant difference, at least at the 1 ~ level, three of the shorter studies (18, 30a a n d 60) were non-significant, while Studies 30b a n d 30c were significant only at the 5 ~ level. As in the previous investigation of the forebrain weight changes 6 the influence of b o d y weight as a d e t e r m i n i n g factor for cortex length was estimated by the productm o m e n t correlation between these two variables. I n general, the two measures were largely i n d e p e n d e n t of one a n o t h e r (r 0.04-0.32), except in Studies 30a a n d 160, where TABLE 1 Cortex length differences between the environmental groups
The test for the Rank Order Effect is a comparison of regression line slope between the rank ordered enriched and isolated values (see text). L, cortex length; E, enriched; I, isolated; N, number of animals. Study
Parameter
N
18
30a
30b
30e
60
80
120
154
160
L 36
L/Body L 36 40
L 36
L 36
L 36
L 38
L 40
L/Body 36
2.0 1.310 n.s.
15.0 1.9 2.1 1.3 2.5 4.9 1.804 2.175 1.989 0.659 2.954 4.051 0.05 0.05 0.05 n.s. 0.01 0.001
1.9 15.0 2 . 5 2 0 4.185 0.01 0.001
1.1 1.322 n.s.
8.0 1.163 n.s.
1.7 2.138 0.05
2.8 2.321 0,05
1.2 3.2 0.957 3.530 n.s. 0.01
5.6 9.375 0.001
2.2 3.819 0.01
15.2 4.634 0.001
2.9 3.192 0.01
26.1 2.0 7.650 2.996 0.001 0.01
1.4 2.846 0.01
1.3 2,031 0.05
1.7 3,171 0.01
4.2 3.068 0.01
1.6 2.230 0.05
14.9 3.685 0.01
13.57 0.001
0,291 n.s.
16.14 0,001
8.262 0.01
1,016 n.s.
5.933 0.05
E~I
E~I
Total
E > I (~) t P I (~) t P < Low ranks
E ~ I (~) t P
E
E~I
E~I
92 significant correlation was found within both of the environmental treatments (r 0.52-0.87). The influence of this confounding variable on the cortex lengths of these two studies was removed by forming a cortex length/body weight ratio. Using this ratio the enriched animals exceeded the isolates in both Studies 30a and 160. For Study 160, the result was similar to that found for cortex length alone, the two environmental treatments differing significantly from one another. For Study 30a, the result was markedly changed from that derived from cortex length alone, in which no overall difference between the environmental groups had been found. The net result of these manipulations was that, of the 9 studies, only Studies 18 and 60 failed to display an overall enrichment-isolation difference in cortex length (Table I). In order to examine the 9 studies for evidence of the Rank Order Effect, each of the environmental groups were separated into high and low ranks with the median rank being chosen as the point of demarcation. Two separate enrichment isolation comparisons were then performed within each study, one between the high ranking values and one between the low ranking values. From Table I it can be seen that only the high ranking values of Studies 18, 30a and 60 failed to support a significant a
LENGTH
30a LENGTH/BODY
,..,
30b
LENGTH
/
=w,
150
./
26%
8%
13
I00
;,'.. 5
(0
15
/
J
,.,.,. 50
''5
50c LENGTH
;o. . . .
~ ....
80
LE L.
90
....
LENGTH '--"
8C
;;%
16
"2,,,
15
1.26%
5
I0
15
5
120
I0 154 LENGTH
5
15
IO
15
160 LENGiH / B O ~ _ 500
14-5
400.~
13-5
2-2% 12"5 'fJ. . . . . 5
I ......... 0 15
5
I0
15
•
.~ 3(2( 2O
15'2%
5
I0
15
Fig. 1. The rank order comparisons between the enriched and isolated cortex lengths for each study. Ordinate = length; abscissa = rank order. Each rank order comparison has been divided by a vertical line at the median of each enriched (solid line) and isolated (broken line) group. The extent of the enrichment-isolation difference (EC > IC ~) is indicated separately for those values lying below the median (low ranks) and above the median (high ranks).
,
a0
93 T A B L E II
Variance differences between enriched and isolated subjects in both high ranking and low ranking animals
Study
Parameter N
18
30a
Length
Length/ Length Length Body 9 10 9
Length Length
14.825 0.05
15.330 0.05
21.261 0.01
43.415 0.001
12.477 n.s.
I > E
I > E
E > I
E > I
9
High ranks Z~ P < Direction o f difference Low ranks Z~
P< Direction o f difference
30b
30c
60
9
80
9
120
154
Length
Length Length
9
25.097 0.01
20.946 0.05
E > I
E > I
12.514
10.978
10.963
10.240
12.008
15.600
n.s.
n.s.
n.s.
n.s.
n.s.
0.05
9.381
n.s.
10 10.970 n.s.
160
9 32.818 0.001
E > I 13.391
14.518
n.s.
n.s.
E ~ I
environmental effect. For each of these three studies, the proportion of the overall enrichment-isolation difference (EC~IC~o) contributed by the low ranks exceeded that of the high ranks. In 5 out of the 6 remaining studies, the proportion of the overall difference contributed by the high ranking values predominated. Tests for the significant occurrence of the Rank Order Effect were performed by comparing the slopes of the enriched and isolated regression lines of cortex length to rank order within each studylL In Studies 18 and 30a the slope of the isolates was found to exceed that of the enriched. For Studies 30c, 80, 120 and 160, this effect was significantly reversed, the slope of the enriched lengths exceeding that of the isolates. The rank order comparisons between the enriched and isolated cortex lengths are represented in Fig. 1. In Table II the within-group variances for enriched and isolated subjects are compared separately for high and low ranking values using the formula (n-l) variancex variance y ' whose values conform to the chi-square distribution (Sokal and Rohlf, ref. 17, p. 171). Only in Studies 18 and 30a did the variance of the high ranking isolates exceed that of the high ranking enriched. In the remaining studies, there was a tendency for the high rank comparisons to show a greater variance for the enriched animals. This was significant for Studies 30b, 30c, 80, 120 and 160. Only one significant variance asymmetry was found within the low ranking subjects. This was in Study 80, where the variance of the enriched exceeded that of the isolates. DISCUSSION
In order for the results to be compatible with the model of brain development previously described for forebrain weight6, the environmentally induced differences in
94
cortex length should be most pronounced between the low ranking values from each group and the within group variance of the isolates should exceed the enriched. In fact only two of the briefest studies, 18 and 30a, provided such data. The results for the remaining 7 studies tend to be either non-significant or incompatible with the model. In the light of this new data the model was revised. This revision is presented in Fig. 2 and differs from the original in that the developmental ceiling is now represented as a variable which changes with age rather than being fixed. The extent of the environment dependent neural development is severely limited in young animals by temporal constraints upon growth. These constraints are imposed upon all neural development by metabolic factors. However, as the animal ages there is a greater potential for such environmentally induced development to accumulate with the result that the developmental ceiling is raised. For the isolates the level of environment support is insufficient for this potential to be realized, and as the ceiling for development is raised there will be little corresponding environment dependent neural development. The enriched animals, on the other hand, will continue to accrue such development, albeit at a diminishing rate as the animals progressively adapt to their environment. There are two major consequences of these temporal considerations. (a) The development consequences of limited environment will become more pronounced with IOO%
AGE A i ~ .
(AGE A + 2 X )
...'''"
100% (AGE A+X)
AGE A+X
,.....,-
I i
~ VARIANCE A4 2X 'IARIANCE A~.X
AGE A I00 ~te (AGE A)
ii
..... ;---- ~
~ - - - - ~
DIAN A
I
:l
i.. q.
-ENRICHMENT
IANCE A
k~ e~ c~ q: k. ILl ~SOLATION
k
OIL
MINIMAL
OPTIMAL LEVEL OF
SENSO RY
STIMULATION
Fig. 2. The revised theoretical model of environment dependent neural development. Three stages of development are represented (Ages A, A + X, A + 2X) and the relative extent of enriched or isolated development at each stage is depicted in terms of the median and variance for each environmental group.
95 increasing durations of exposure. (b) The previously described Rank Order Effect6 will appear as a transitory phenomenon, which exists only during the intermediate stage of environmental influence when some, but not all of the isolate values have been restricted in their development. At the stage when all of the isolate values are reduced below the enriched the Rank Order Effect will disappear. The revised model in Fig. 2 leads to three characteristics of the within-group and between-group variance distributions which are explained as follows. (1) As the animals age in their respective environments there should be a tendency for the variance pattern to reverse, i.e. for the variance of the enriched lengths to exceed that of the isolates. The reason for this is that the continued development of the isolates is limited by the environment to a greater extent than the enriched (Fig. 2). In the early stages of differential rearing, it has been argued that a few of the isolates would be able to achieve the relatively small amounts of environment dependent development necessary to approximate the early developmental ceiling6. As the ceiling is raised, however, even the most responsive isolates are unable to achieve the necessary development to realise this potential. Thus, the variance of the isolates will tend to remain stable once this initial environment dependent development is achieved. By analogous reasoning, it is postulated that for the enriched animals the initial period of enrichment will provide sufficient stimulation for most animals to achieve their developmental potential at that stage. However, as the animals progressively adapt to this enriched environment with prolonged exposure periods, the number of animals able to extract sufficient further stimulation for the maintenance of full development will diminish. The increasing influence of these individual differences in the neurological response to enrichment will be reflected by an increasing variance of the dependent parameter as the period of differential rearing is extended. The cortex length data are consistent with this description. In Table II it can be seen that, while for two of the shortest exposure times, the variance of the isolate lengths exceeds that of the enriched, for all longer exposures either the variance of the enriched lengths exceeds that of the isolates, or the inter-group comparison was nonsignificant. (2) As the animals age, there is a tendency for the variance of the high ranking enriched cortex lengths to exceed that of the high ranking isolates, while the variance between the low ranking lengths will remain largely unchanged. The development of the elevated variance with prolonged rearing should be confined to the high ranking enriched values, where some animals are likely to approximate the developmental ceiling. The low ranking enriched lengths, on the other hand, show less change with prolonged rearing. It was argued previously6 that the low ranking values in the isolated group were most likely to be contributed by animals retarded in their environment dependent neural development. By analogous reasoning, the proportion of the low ranking enriched values representing sub-optimal development will increase as the animals adapt to their environment and the level of sensory stimulation becomes progressively the limiting factor to their continued development. One effect of this would be that the median for the enriched group would rise at a slower rate than the developmental ceiling. The variance below the median
96 will therefore show progressively less temporal change as the animals in this region increasingly represent those which are no longer extracting any benefit from their continued exposure to enrichment. The variance of the isolates, both above and below the median, should be unchanged with prolonged rearing. Their development has already been limited during the early period of exposure. The data are consistent with a progressive development of an increasing variance restricted to the high ranking enriched values. Table II compares the variance differences between the enriched and isolated subjects for the high ranking and low ranking cortex lengths separately. Only in two early studies, 18 and 30a, did the variance of the high ranking isolates exceed the high ranking enriched. There was a marked tendency for the high rank comparisons in the remaining studies to indicate a greater variance for the enriched animals. This was significant for Studies 30b, 30c, 80, 120 and 160. On the other hand, only one significant variance asymmetry was found within the low rank comparisons. (3) As the animals age, there is a tendency for the Rank Order Effect to indicate a larger environmental influence between the high ranking lengths (Positive Rank Order Effect) than between the low ranking lengths (Negative Rank Order Effect). This follows from the previous argument that the predominant environmental influence with prolonged rearing will be to increase the magnitude of the high ranking enriched lengths. The slope of the enriched regression line will gradually increase as the high ranking enriched follow the upward drift of the developmental ceiling, whilst the lengths of the other ranks, enriched and isolated, remain relatively stable over the same period. However there is a sine qua non associated with the occurrence of a Positive Rank Order Effect. This is that it must always be associated with a significant enrichment-isolation difference within both high and low ranking values. In order for the Positive Rank Order Effect to develop the isolate lengths must remain retarded while some of the enriched continue to develop. Thus the intergroup differences will be enhanced to include a difference between the high ranking values, while the differences between the low ranking values will be retained. The data is consistent with this description. In Table I it can be seen that a Positive Rank Order Effect was found for Studies 30c, 80, 120, and 160 and that, for each of these studies, a significant difference was present between the low and high ranking values.
The revised model and forebrain weight The preceding three requirements of the revised model of environment dependent neural development appear to be consistent with the data from the cortex length measurements. The question now arises as to why the two parameters of forebrain weight and cortex length behave so differently. The environmental influence on forebrain weight was shown to be dissipated with prolonged rearing 6 but the effect on cortex length would appear to become more marked with increasing maturity (Table I, Fig. 1). One explanation may lie in the degree to which each parameter is a true
97 reflection of environment dependent development, or Js contaminated by concomitant growth, unrelated to the differential rearing per se. Whilst the most rapid phase of the brain weight development has occurred prior to weaning tl, there is a continuous, negatively accelerating increase in brain weight for a considerable portion of the animal's life3,9,10. However the cortical and sub-cortical components do not make an equal contribution to this age-related brain weight increase. It has been reported by Bennett et al. 3, that while both the cortex and sub-cortex (whole brain minus cortex) continue to grow after weaning, the growth of the cortex occurs at a much slower rate than the rest of the brain. As a result, maturation leads to a marked decrease in the cortex/sub-cortex weight ratio4,12,13. It has also been reported that the dimensions of the sub-cortex are generally not susceptible to change as a result of enrichment-isolation experience. Neither the weight of the sub-cortexZ,9,14, nor the dimensions of the diencephalon s have been found to change, even though the same studies reported marked cortical responses. In 1969 we reported that the dimensions of the hippocampus changed with differential rearing is. In an extensive series of studies Diamond et al.7 were unable to repeat this result, so it would seem unlikely that the hippocampal dimensions change reliably with differential rearing. It would therefore seem that the parameter of forebrain weight is subject to two influences acting to determine its postweaning development. The slow, differential accumulation of environment dependent development is superimposed upon a continuous increase in overall brain size which may be related to general somatic growth. It has previously been reported that there is a consistent positive correlation between brain weight and body weight, which is very marked in young animals, becoming less so in adultsS, 11. In the present series of studies there was a consistent positive correlation between body and forebrain weight which, on average, amounted to about 20 ~ shared variance. The forebrain weight changes with different durations of rearing can then be explained. The initial environmental influence is to retard cortical development in some of the isolates. This induces the Negative Rank Order Effect. As the period of differential rearing is extended, the capacity of the environments to induce further change is diminished and the changes in environment dependent development are gradually overshadowed by the continuing process of general brain growth. The majority of this continued growth occurs in the sub-cortex which is minimally responsive to the environmentally induced changes. The extent of this development may be highest in the isolated animals, since in the present series of studies the isolated animals consistently weighed more than the enriched (unreported data). The transitory nature of the forebrain weight changes may therefore not represent a diminution of differential environment dependent development as we previously suggested but, rather, the difference in development may be maintained but be obscured by brain growth related to general somatic development. ACKNOWLEDGEMENTS This research was supported by a grant from the Australian Research Grants Committee.
98 REFERENCES 1 Altman, J., Wallace, R. B., Anderson, W. J. and Das, V. G., Behaviorally induced changes in the length of the cerebrum in rats, Develop. PsychobioL, 1 (1968) 112-I 17. 2 Bennett, E. L., Krech, D. and Rosenzweig, M. R., Reliability and regional specificity of cerebral effects of environmental complexity and training, J. comp. PhysioL Psychol., 57 (1964) 440-441. 3 Bennett, E. L., Rosenzweig, M. R. and Diamond, M. C., Time courses of effects of differential experience on brain measures and behavior of rats. In W. L. Byrne (Ed.), Molecular Approaches to Learning and Memory, Academic Press, New York, 1970, pp. 55 89. 4 Bennett, E. L. and Rosenzweig, M. R., Chemical alterations produced in brain by environment and training. In A. Lajtha (Ed.), Handbook ofNeurochemistry, Vol. 6, Plenum Press, New York, 1971, pp. 173-201. 5 Clark, G. M. and Zamenhof, S., Correlations between cerebral and cortical parameters in the developing and mature rat brain, Int. J. Neurosci., 5 (1973) 223-229. 6 Cummins, R. A., Livesey, P. J., Evans, J. G. M. and Walsh, R. N., A developmental theory of environmental enrichment, Science, 197 (1977) 692-694. 7 Diamond, M. C., Ingham, C. A., Johnson, R. E., Bennett, E. L. and Rosenzweig, M. R., Effects of environment on morphology of rat cerebral cortex and hippocampus, J. NeurobioL, 7 (1976) 75-85. 8 Diamond, M. C., Law, F., Rhodes, H., Lindner, B., Rosenzweig, M. R., Krech, D. and Bennett, E. L., Increases in cortical depth and glial numbers in rats subjected to enriched environments, J. comp. Neurol., 128 (1966) 117-125. 9 Ferchmin, P. A., Eterovic, V. A. and Caputto, R., Studies of brain weight and R N A content after short periods of exposure to environmental complexity, Brain Research, 20 (1970) 49-57. 10 Fuller, J. L. and Geils, H. D., Brain growth in mice selected for high and low brain weight, Develop. Psychobiol., 5 (1972) 307-318. 11 Kobayashi, T., Brain-to-body ratios and time of maturation of the mouse brain, Amer. J. Physiol., 204 (1963) 343-346. 12 Riege, W. H., Environmental influences on brain and behavior of year-old rats, Develop. Psychobiol., 4 (1971) 157-167. 13 Rosenzweig, M. R., Mechanisms by which environmental complexity may alter brain chemistry and anatomy, In Int. Congr. PsychoL, London, July 29, 1969. 14 Rosenzweig, M. R. and Bennett, E. L., Drugs modulate effects of environment on brain growth,
Proc. 76th Ann. Convention Amer. Psychol. Ass., 1968. 15 Rosenzweig, M. R. and Bennett, E. L., Effects of differential environments on brain weights and enzyme activities in gerbils, rats and mice, Develop. Psychobiol., 2 (1969) 87-95. 16 Rosenzweig, M. R., Bennett, E. L. and Diamond, M. C., Effects of differential environments on brain anatomy and brain chemistry. In J. Zubin and G. Jervis (Eds.), Psychopathology of Mental Development, Grune and Stratton, New York, 1967, pp. 45-56. 17 Sokal, R. R. and Rohlf, F. J., Biometry, Freedman, San Francisco, 1969. 18 Walsh, R. N., Budtz-Olsen, O. E., Penny, J. E. and Cummins, R. A., The effects of environmental complexity on the histology of the rat hippocampus, J. comp. NeuroL, 137 (1969) 361-366. 19 wa•sh• R. N.• Budtz-O•sen• •. E.• T•r•k•A. and Cummins• R. A.• Envir•nmenta••y induced changes in the dimensions of the rat cerebrum, Develop. PsychobioL, 4 (1971) 115-122. 20 Walsh, R. N., Cummins, R. A. and Budtz-Olsen, O. E., Environmentally induced changes in the dimensions of the rat cerebrum: a replication and extension, Develop. Psychobiol., 6 (1973) 3-8.
