Combined effects of moderate ethanol consumption and a low-protein diet during gestation on brain development in BALBc mice

EXPERIMENTAL

NEUROLOGY

9&422-433

(1985)

Combined Effects of Moderate Ethanol Consumption Low-Protein Diet during Gestation on Brain Development in BALB/c Mice PATRICIA WAINWRIGHT,* *Department

GLENN

R. WARD,?

AND KATHRYN

oiHealth Studies and tDepartment of Psychology, Waterloo, Waterloo, Ontario N2L 3Gl. Canada Received

April

2. 1985; revision

received

University

and a

BLOMt,’ of

June 24. 198.5

We investigated whether or not moderate ethanol consumption during gestation would interact with the effects of a low-protein diet in affecting brain development in BALB/c mice. The independent variables included fetal body and brain weights and cross-sectional area in midsagittal sections of the corpus callosum (CC) and anterior commissure (CA). Pregnant animals were fed either ethanol 12% v/v or an isocaloric sucrose solution from days 5 to 19 of gestation, when fetal development was assessed. In addition, the animals were fed semisynthetic isocaloric diets containing either 8 or 20% casein. All animals were pair-fed to those in the group receiving ethanol and 20% casein; an additional control group was fed lab chow ad libitum. There was clearly an interactive effect of diet and ethanol consumption on blood alcohol concentrations: those in the low-protein group were significantly higher than in the normal-protein group. Similarly. the effect on body weight in the group receiving low protein plus ethanol was greater than the additive effect of either treatment alone. although this may have been due partly to differences in litter size. Brain weight in this group was also significantly less than in the other three groups, which did not differ from each other. Covariance analysis, adjusting brain weight for body weight, suggested a brainsparing effect of low protein but not ethanol. Neither treatment affected the incidence of the CC being absent at midline. The low-protein treatment decreased the crosssectional area of both the CC and CA: the effect on the CC was independent of brain weight. There was no effect of ethanol on either of those measures. 0 1985 Academic PI-e,

Inc.

Abbreviations: CA-anterior commissure, CC-corpus callosum. ’ The authors thank Mr. W. Zagaja for animal maintenance, Miss M. Gagnon for technical assistance, and Mrs. S. Hurlburt for secretarial support. This research was a portion of a thesis completed by G. Ward in partial fulfillment of the requirements for an M.A. degree in psychology. The work was supported in part by a Natural Sciences and Engineering Research Council of Canada award A76 I7 to P. Wainwright, to whom reprint requests should be addressed. 422 0014-4886/85 $3.00 Copyright 0 1985 by Academic Press. Inc. All rights of reproduction in any form reserved.

ETHANOL,

PROTEIN,

AND

BRAIN

DEVELOPMENT

423

INTRODUCTION Prenatal ethanol exposure in humans has been associated with a characteristic pattern of fetal defects which have been termed collectively the “fetal alcohol syndrome” (2). It is possible for components of FAS to be apparent in the absence of the complete syndrome and these are referred to as “fetal alcohol effects.” Of particular concern are the effects on the development of the brain; animal studies have indicated that ethanol retards brain growth in a dose-related manner ( 1 l), as well as effecting neurochemical changes (8. lo), structural changes (3-5), and alterations in the organization of neuronal networks (20). The advantage of animal studies is that it is possible to study the effects of ethanol, while at the same time controlling for the effects of other variables” which may be associated with alcohol use in human populations, such as malnutrition or the concomitant use of other drugs. Nutritional variables are of particular interest in this regard, because it has been shown that the nutritional quality of the diet declines as alcohol intake increases (6). As well as being associated with malnutrition, the use of ethanol may in itself induce malnutrition, either by displacing other food in the diet due to its high caloric content (7.1 kcal/g), or by affecting nutrient absorption, distribution, or metabolism. The adverse effects of malnutrition on the brain have been well documented (9,22) and it is therefore possible that some effects of ethanol on fetal development may be attributable to malnutrition per se, or to the interaction between ethanol and malnutrition. A thorough discussion of these issues was provided in a recent review article ( 18). The use of a pair-fed control group in an experimental design addresses the question of whether or not there is an effect of ethanol independent of reduced food intake. However, pair-feeding does not speak to the problem of an interaction between ethanol and the nutritional status of the dam in affecting fetal development. This question was first addressed by Wiener et al. (2 1) who varied the concentration or protein in the diet; however, in that study the ethanol content of the diet decreased as the protein content increased which confounded interpretation of the results. In a recent study (19), Weinberg showed that 36% ethanolderived calories reduced fetal body and brain weights and resulted in high blood alcohol concentrations in both marginally nourished and well nourished dams. In that study no group could be considered excessively deprived of protein (18%. 25%, or 32% kilocalories as protein) and the importance of the results was in demonstrating that high alcohol consumption affected fetal growth, even under conditions of optimal nutrition. It was suggested that, because low-protein diets reduce hepatic aldehyde dehydrogenase (7) and microsomal enzyme activities ( 12) animals that are protein-malnourished may have a decreased rate of alcohol metabolism, which would result in correspondingly higher blood alcohol concentrations than those in well nourished animals consuming an equivalent amount of alcohol. Because alcohol has

424

WAINWRIGHT,

WARD,

AND

BLOM

been shown to affect fetal growth in a dose-related manner (1 l), these higher blood alcohol concentrations should be accompanied by larger fetal effects. Our purpose was therefore to investigate the effects of moderate alcohol intake on brain development under conditions of protein malnutrition. We compared fetal outcome in animals which consumed equivalent amounts of ethanol but differed widely in the protein intake (8% vs. 20%). The expectation was that there would be a synergistic effect between ethanol and diet with the ethanol effect potentiated in the low-protein condition, and correlated with higher blood alcohol content. The experimental model used was the development of the brain in BALB/ c mice. This strain is prone to deficiency of the corpus callosum (CC) which in some animals is either missing at the midline or is greatly reduced in size ( 13). Recent studies in our laboratory using these animals demonstrated that both protein malnutrition (15) and moderate ethanol consumption (16) resulted in smaller fetal brain weight and reduced growth of the CC and anterior commissure (CA), with some of the effects on fiber tract growth appearing to be independent of the differences in brain weights. It therefore seemed appropriate to use these measures as dependent variables in addressing the experimental hypothesis outlined above. METHOD Subjects. The subjects were virgin female mice, 100 to 160 days of age, from the BALB/cCF strain obtained from a breeding colony maintained at the University of Waterloo. They were housed in groups of five in 29 X 18 X 13-cm opaque plastic mouse cages with stainless-steel lids and “Betta Chip” hardwood bedding and several sheets of toilet tissue for nesting material. The temperature was 22 -t 1‘C and the 12-h light: 12-h dark schedule was reversed with the dark cycle beginning at 0800 h. Prior to the experimental treatment, the animals had free access to tap water and laboratory chow (Master rodent food cubes from Maple Leaf Mills, Toronto). Experimental Design. A 2 X 2 factorial design with two levels of ethanol (12% v/v in drinking water vs. an isocaloric sucrose solution) and two levels of protein (8% casein vs. 20% casein) resulted in four groups: 1, normal protein/ ethanol (NP/E); 2, normal protein/sucrose (NP/S); 3, low protein/ethanol (LP/E); and 4, low protein/sucrose (LP/S). As ethanol consumption has been shown to depress food intake in addition to supplying “empty” calories, it was important to control for a possible nutritional confounding ( 18). To this end each group consisted of animals matched by weight on day 5 postconception with an animal in the NP/E group and fed the same amount of food and volume of fluid ingested on the same day of pregnancy. Each group consisted of seven pregnant animals. In addition, three animals were assigned

ETHANOL,

PROTEIN,

AND

BRAIN

DEVELOPMENT

425

to a group fed lab chow ad libitum. The purpose of this group was to compare it with the NP/S group to determine if there were effects due to the lowered nutrient intake in the experimental groups. Materials. The experimental diets were obtained from ICN Biochemicals (Cleveland, Ohio) and consisted of either the American Institute of Nutrition AIN 76 semipurified mouse-pelleted diet (No. 905453) containing 20% casein, or a modified version which contained 8% casein with the deficit made up by the addition of an isocaloric amount of sucrose. The drinking fluid consisted of either a 12% (v/v) solution of ethanol (prepared from 95% ethanol obtained from Commercial Alcohols Ltd., Toronto) or 16.8% (w/v) solution of sucrose (J. T. Baker Chemical Co., Phillipsburg, N.J.). Both solutions were isocaloric, yielding 0.672 kcal/ml. Blood alcohol concentrations were determined using the alcohol dehydrogenase method provided as a diagnostic kit by Sigma Chemicals (Sigma Technical Bulletin No. 332~UV). Procedure. Animals were mated daily by caging the male with the female mice for a 6-h period, starting an hour after the beginning of the dark cycle. When pregnancy was ascertained by the presence of a vaginal plug, the pregnant animal was housed separately and maintained on the normal protein diet and tap water ad libitum. The day a plug was detected was designated day 0 and animals were weighed on days 0, 5, 12, and 19. The experimental dietary regimen commenced on day 5. Animals were assigned to experimental groups on the basis of their match in body weight on day 5 to a member of the NP/E group. From days 5 to 19 fluid and food were renewed between 0800 and 1000 h daily and intake recorded, except for those animals fed lab chow ad libitum. Between 1000 h and 1200 h on day 19 the females were killed by an overdose of sodium pentobarbital, and a blood sample was withdrawn by cardiac puncture and refrigerated 24 to 48 h until assayed for alcohol concentration. The fetuses and placentae were removed from the uterus and immersed in 10% buffered Formalin. Each fetus was weighed and the skull cap removed to allow better penetration of the fixative. Twenty-four hours later the brain was removed, trimmed by cutting perpendicularly through the brain stem at the base of the cerebellum, and weighed. The brains were then embedded in gelatin, frozen sections cut in the sagittal plane at 33 pm, and stained with hematoxylin and eosin. The cross-sectional area of the CC and CA were determined from tracings of midsagittal sections as described (15). All viewing was without knowledge of the experimental group from which the tissue had originated. Analysis. The data were analyzed using the GLM (general linear model) procedure provided by SAS (Statistical Analysis System) for analysis of variance and covariance. Because of the hierarchic nature of the design, i.e., litters nested within treatment groups, the between-litter variance was used as the error term in the analysis of variance ( 1). Individual group means were com-

426

WAINWRIGHT,

WARD,

AND BLOM

pared using preplanned t tests. Qualitative data were analyzed using the chisquare statistic. The significance level was set at (Y = 0.05. RESULTS Fetal Variables. These data are presented in Table 1 and Fig. 1. Body weight. Two-way ANOVA revealed a significant main effect for both protein and ethanol, but also a statistically significant interaction (one-tailed). Each of the normal-protein groups was heavier than each of the low-protein groups, but they were not different from each other, i.e., there was no effect of ethanol in the normal-protein condition. In contrast, the group receiving both low protein and ethanol weighed significantly less than any of the other three groups. The interaction indicated that the effect of the two treatments combined was synergistic, i.e., greater than the sum of the average effects of two treatments separately. Because the number of pups per litter in the LPfE group was significantly larger than that in the LP/S group (see section on maternal variables) the experimental effect on body weight was reanalyzed using analysis of covariance with litter size as the covatiate for litter mean body weight. Comparison of the adjusted means demonstrated that although the body weight of the LP/E group remained the least of all groups, it no longer differed significantly from the LP/S group. In contrast with body weight there were no significant differences in placental weight among the groups. Bruin weight. Again there were significant main effects, with both ethanol and protein causing a reduction in brain weight. Although the interaction was not statistically significant, the only group to differ significantly from any other group was that receiving both treatments. When body weight was used as a covariate, the adjusted means, which are an indication of whether the effect on brain weight was independent of that on body weight, showed that the LP/E group no longer differed from the others, but that animals in the NP/E group had significantly smaller brains for their body size than those in the LP/S group. This suggests that in the low-protein group there was a brain sparing effect, whereas this was not the case in the ethanol-treated animals. In the group receiving both treatments the effects would cancel each other, resulting in no overall effect. Corpus callosum. Chi-square analysis indicated no difference among treatment groups in terms of the number of animals missing the CC at midline. There was a significant main effect due to protein deprivation causing a reduction in the cross-sectional area of the CC. Although the animals in the LP/E group did have the smallest CC area, there was no significant effect due to the ethanol treatment, nor was there an interaction between the treatments. When brain weight was used as a covariate for CC area, only those groups which received different amounts of protein were significantly different from each other.

ETHANOL,

PROTEIN.

AND

BRAIN

DEVELOPMENT

6 8 6

F8 d 0 +I m ;r: 6

s 8 d

B 8 6 m tl

+I

$1

427

428

WAINWRIGHT,

--SUCRO

WARD, AND BLOM

-

ETHANOL

LOW PROTEIN

NORMAL PROTEIN DIETARY

CONDITION of DAM

FIG. 1. Body and brain growth in BALB/c fetusesat 19.0 days postconception as a consequence of maternal ethanol and protein intake during gestation.

Anterior commissure. Similarly, the only significant effect on the area of the CA was that it was significantly smaller in those groups receiving the lowprotein diet. The analysis using brain weight as a covariate could not be carried out for this variable because of heterogeneity of slope among the treatment groups. Maternal Variables. These data are presented in Table 2. There were no significant differences in body weights among the groups on either day 5 or day 12, but by day 19 there was a significant main effect of protein, with the dams in the LP/S group being significantly lighter than those of the NP/E group. This decrease in body weight during the last week resulted in the food intake (per gram body weight) being significantly higher in the low-protein than in the normal-protein groups. The two low-protein groups did not, however, differ from each other. Fluid intake did not differ among the groups. There was a significant main effect of ethanol on number of pups per litter, with ethanol-treated groups having larger litters. As mentioned above, litter

48.98 + 9.75’

0.9 1.0 1.0 0.6’ 0.5

15.13 f 6.78’

f f f f f

0. I38 f 0.006 0.144 + 0.010

25.2 27.1 36.3 8.9 2.0

0.144 f 0.004 0.146 + 0.007

0.7 0.5 o.96 0.4 0.6 0.140 + 0.003b 0.165 f 0.003’

f f f f +

1

Low protein

0.137 + 0.001 0.152 f 0.003b~d

24.6 27.8 40.3 7.9 2.6

Normal protein 7 0.8 0.7 1.5’ l.lC 0.6

0.154 f 0.007 0.166 + 0.014

f k + f +

0.143 f 0.004 0.147 rfr 0.005

25.0 26.4 35.3 5.7 1.7

Low protein 7

0. I36 f 0.004 0.158 + 0.004”’

+ 1.2 f 1.1 f 2.0 I? 0.8 f 0.3

No ethanol

0.130 + 0.003’ 0.146 + 0.003d

24.4 25.7 38.9 1.6 1.1

Normal protein 7

Note. p-Ethanol effect, q-protein effect. a Data are presented as means t SE. b~c,d Means with different superscripts in the same row are significantly different, P < 0.05.

Days 5- 12 Days 12-19 Blood alcohol (mg/lOO ml)

Body weight (g) Day 5 Day 12 Day 19 Litter size Resorptions Food intake/g body weight (day) Days 5-12 Days 12-19 Fluid intake ml/g body weight (day)

N

Ethanol

Dietary condition of dam

Effect of Consumption of Ethanol and Low-Protein Diet durina Gestation on Maternal Variables in BALB/c Mice“

TABLE 2

P q

9 P

effects (P < 0.05)

Main

R \o

9 s E; 2

g

z

8 2

z 5 5 s $

430

WAINWRIGHT,

WARD,

AND

BLOM

size in the LP/E group was significantly larger than in the LP/S group. There were no significant differences in resorptions among the groups. There was a significant effect of protein on blood alcohol concentrations, with the lowprotein group showing significantly higher values than the normal-protein group. The average daily intake of ethanol was 13.4 g/kg. Nutritional ej2ct.s.The group fed lab chow ad iibitum can be considered a normal reference group against which to compare specifically the NP/S group and thereby evaluate the nutritional effects of pair-feeding the synthetic diet. A one-way ANOVA was conducted on all five groups, followed by comparisons of group means. The preplanned t tests, comparing the ad libitum with the NP/S group, showed that for none of the variables reported above, both fetal and maternal, were the means of the two groups significantly different. The only exception to this was that the area of the CC was smaller in the ad libitum fed group (0.218 + 0.012 mm2 vs. 0.259 + 0.006 mm*). It should be noted that this effect is opposite in direction from that predicted had the NP/S group been malnourished relative to the ad libitum group. Simultaneous comparisons of all the groups, using Tukey’s t test to protect the overall (Ylevel, indicated that the ad Zibitum group differed with respect to body and brain weight only from the LP/E group; in each case the LP/E group was lighter. This is to be expected in light of the fact that the ad libitum fed group and the NP/S group did not differ from each other on these variables. DISCUSSION These results indicate that moderate ethanol consumption interacted with a low-protein diet to reduce fetal body weight, although part of this effect may have been mediated by the larger litter size in the ethanol-treated animals. Although the combined effect of the two treatments on brain weight was statistically additive, the only group that showed a significantly lower brain weight was that which received both treatments. In this regard, it should be noted that, using an ANOVA model, the test for the interaction is extremely conservative. This is because the interaction term is the variance remaining after the variance accounted for by the main effects has been removed. Thus, paradoxically, although the group receiving both treatments was the only one to differ, when the contribution made by this group to each of the main effects had been removed, the interaction was not significant. Covariance analysis suggested a brain-sparing effect in animals receiving low protein, but not in those receiving ethanol. This contrasts with research in the rat in which high ethanol consumption was associated with a brain-sparing effect (19). This specific effect of ethanol on brain growth is noteworthy, especially in view of the low blood ethanol concentrations. These data support other work in this laboratory using BALB/c mice with a similar method of alcohol administra-

ETHANOL,

PROTEIN,

AND

BRAIN

DEVELOPMENT

431

tion, in which smaller brain weights were found on both 19 (16) and 26 and 50 days ( 17) postconception. The low-protein diet resulted in a smaller cross-sectional area of both the CC and the CA, but there was no indication of an increase in the number of animals in which the CC was absent. Previous work had shown an increase in this measure as well as reduction in the size of the fiber tract (15). A possible explanation for the discrepancy between the two studies is that the previous study was conducted on 18.5-day fetuses, whereas those in the present study were 19.0 days old. Since, in the BALB/c strain, the first fibers of the CC cross midline between days 17 and 18 (14) a growth-retarding effect would be seen as an increase in the number of acallosal animals on day 18.5, but by day 19 these slower growing axons would have eventually crossed, and the retardation would then be reflected only by the smaller cross-sectional area of the CC. There was no apparent effect of ethanol on the growth of the CC and CA, in contrast with previous findings (16). The regimen of ethanol treatment differed between the two studies, 10% v/v having been administered in the first study and 12% v/v in the present study. Comparison of the dose of ethanol in g/kg indicated that the animals in the first study consumed more ethanol (17.7 vs. 13.4 g/kg). This finding suggests that the animals in this study found the higher concentration aversive and decreased their intake accordingly. The diets in the two studies also differed between lab chow and the present semisynthetic diet. However, comparison of the pair-fed control groups from each study showed that the measures on the dependent variables were very close in value. The one exception to this was CC area, which was larger in the present study. This result is interesting in light of the fact that the present study also showed that the CC of animals fed the synthetic diet was the only variable that was larger than in animals fed lab chow ad libitum. This suggests subtle dietary influences acting specifically on the growth of the CC. In light of these findings it perhaps would have been better to have used a control group fed synthetic diet ad libitum rather than lab chow. With respect to brain growth, it is interesting to note that the low-protein treatment resulted in a sparing effect on brain weight but nevertheless affected the growth of specific areas such as the CC and CA, whereas the effect of ethanol appeared to be more general. As predicted, there were significant differences in blood alcohol concentrations, with the low-protein groups showing much higher values of ethanol. As the intake of alcohol was equivalent in all groups, this supports the explanation of reduced metabolic capacity in the protein-malnourished animals. An alternative explanation is that there were differences in total body water between the groups which would affect the distribution of ethanol, but this is rather unlikely, in view of the small difference in maternal body weights on

432

WAINWRIGHT,

WARD, AND BLOM

day 19 relative to the large differences in alcohol concentrations. The slightly greater food consumption per gram body weight by the low-protein groups during the last week of gestation might, if anything, be expected to have contributed a conservative bias to the data, in that the effect of low protein may be greater than was seen here. The contribution made by these results is in demonstrating that moderate ethanol consumption interacts with the nutritional status of the dam to result in higher blood alcohol concentrations in protein-malnourished animals, and that these correlate with greater effects on fetal body and brain growth in the group receiving both treatments. These findings suggest that nutritional status may be an important consideration when predicting the effects of moderate maternal alcohol consumption on fetal outcome. REFERENCES 1. ABBEY, H., AND E. HOWARD. 1973. Statistical procedure in developmental studies on species with multiple offspring. Dev. Psychobiol. 6: 329-336. 2. ABEL, E. E. (Ed). 1984. Fetal Alcohol Syndrome and Fetal Alcohol Effects. Plenum, New York. 3. BARNES, D. E., AND D. W. WALKER. 198 1. Prenatal ethanol exposure permanently reduces the number of pyramidal neurons in rat hippocampus. Dev. Brain Res. 1: 333-340. 4. BAUER-MOFFETT, C., AND J. ALTMAN. 1977. The effect of ethanol chronically administered to preweanling rats on cerebellar development: a morphological study. Brain Res. 119: 249-268. 5. BORGES,S., AND P. D. LEWIS. 1982. A study of alcohol effectson the brain during gestation and lactation. Teratology 25, 283-289. 6. HILLERS, V. N., AND L. K. MASSEY. 1985. Interrelationships of moderate and high alcohol consumption with diet and health status. Am. J. Clin. Nutr. 41: 356-362. 7. HORN, R. S., AND R. W. MATHEI. 1965. Ethanol metabolism in chronic protein deficiency. J. Pharmacol.

Exp.

Ther. 147: 385-390.

8. RAWAT, A. K. 1980. Biochemical aspects of the neuroteratogenic effects of ethanol. Neurobehav.

Toxicol.

2: 259-265.

9. RESNICK, O., M. MILLER, W. FORBES,R. HALL, T. KEMPER, J. BRONZINO,AND P. J. MORGANE. 1979. Developmental protein malnutrition: influences on the central nervous system of the rat. Neurosci. Biobehav. Rev. 3: 233-246. 10. REYES, E., J. M. RIVERA, L. C. SALAND, AND H. M. MURRAY. 1983. Effects of maternal administration of alcohol on fetal brain development. Neurobehav. Toxicol. Teratol. 5: 263-267. 11. SAMSON, H. H., AND K. A. GRANT. 1983. Ethanol induced microcephaly in the neonatal rat: relation to dose. Alcohol: C/in. Exp. Res. 8: 201-203. 12. VARMA, D. R. 1981. Protein deficiency and drug interactions: a review. Drug Dev. Res. 1: 183-198. 13. WAHLSTEN, D. 1974. Heritable aspectsof anomalous myelinated fibre tracts in the forebrain of the laboratory mouse. Brain Res. 68: 1- 18. 14. WAHLSTEN, D. 1984. Growth of the mouse corpus callosum. Dev. Brain Res. 15: 59-67. 15. WAINWRIGHT, P., AND R. STEFANESCU.1983. Prenatal protein deprivation increases defects of the corpus callosum in BALB/c laboratory mice. Exp. Neural. 81: 694-702. 16. WAINWRIGHT, P., AND M. GAGNON. 1985. Moderate prenatal ethanol exposure interacts

ETHANOL,

PROTEIN,

AND BRAIN DEVELOPMENT

433

with strain in affecting brain development in BALB/c and C57BL/6 mice. Exp. Nercrol. 88: 84-94. 17. WAINWRIGHT, P., AND G. FRITZ. 1985. Effect of moderate prenatal ethanol exposure on postnatal brain and behavioral development in BALB/c mice. Exp. Neural. 89: 237-249. 18. WEINBERG, J. 1984. Nutritional issues in perinatal alcohol exposure. Neurobehav. To.~icol. Teratol.

6: 26 I-269.

19. WEINBERG,J. 1984. Effectsof ethanol and maternal nutritional status on fetal development. Alcohol:

Clin. Exp. Rex

9: 49-55.

20. WEST, J. R., AND C. A. HODGES-SAVOLA. 1983. Permanent hippocampal mossy fiber hyperdevelopment following prenatal ethanol exposure. Neurobehav. To.uicol. Terutol. 5: 139150.

WIENER, S. G., W. J. SHOEMAKER, L. Y. KODA, AND F. E. BLOOM. 198 I. Interaction of ethanol and nutrition during gestation: influence on maternal and offspring development in the rat: J. Pharmacol. Exp. Ther. 216: 572-579. 22. ZAMENHOF. S., AND E. VAN MARTHENS. 1978. Nutritional influences on prenatal brain development. Pages 149-l 86 in G. GOTTLIEB, Ed. Studies on the Development ofBehavior and the Nervous System. Vol. 4. Early Influences. Academic Press, New York.

21.