Chapter Six

General Discussion

Over 7,000 animals, including primates, dogs, cats and rodents, were subjected to maternal deprivation experiments (Table V). The deprivation experience was often coupled with stressful tests or surgery. The deprivation experience alone was enough to cause psychological damage, which ranged from acute distress to generalized psychological crippling. Many animals succumbed to the experimental procedures or were killed outright.


The price tag of these experiments was over $57 million (Table VI). Virtually all of this money came from agencies of the federal government, such as the National Institutes of Health and the National Institute of Mental Health. This indicates that taxpayers have unwittingly financed most of these experiments.

The search for direct clinical impacts of these animal experiments has nearly been in vain. Only one experiment had had a direct clinical impact. It involved habilitating monkeys reared in isolation. The habilitation procedure entailed pairing these disturbed monkeys with younger, more-normal individuals. This procedure has been applied successfully to shy children and children at risk for schizophrenia.

There is a double irony in this application of animal data. Not only did the habilitation procedure involve well-known psychiatric principles, it also could have been safely developed in human studies, without prior animal studies. Moreover, in the case of the shy children, the reason for the procedure’s success was the reverse of that in the animal study.

Other experiments have had little or no clinical impact. At best, the scores of deprivation and separation experiments simply added support to pre-existing human data. Only a handful of experiments went beyond the human data to make novel suggestions about clinical problems.

To date, several lines of physiological research on maternal deprivation or separation have failed to have any clinical impact. Except for McKinney’s research, much of this work appears to be basic research far removed from clinical concerns.


Hence, despite the intentions of the researchers, their experiments have made only a small contribution to clinical practice. This is a serious charge. The animal modelers could attempt to offset this by highlighting the theoretical implications of their research. Animal modelers could present at least two other defenses of their maternal deprivation experiments. The first is that the potential value of these experiments has been largely overlooked by clinicians, and, that, in time, this value will be appreciated. The second is that more experiments are needed before human benefits can be realized.

All three of these arguments have been made in a general defense of psychological experiments on animals (Gallup & Suarez 1980). However, skeptics would regard these defenses as excuses or vacuous enticements to lure further funding. A larger measure of skepticism is warranted. None of the reports cited in this survey provides examples of theoretical implications that have led to human research with practical application.

Three decades of maternal deprivation experiments on animals have produced over 350 research reports. Not all of these reports constitute separate studies; some discuss previously published experiments. However, there are at least 250 separate studies. This figure does not include unpublished experiments that are completed or underway. Some of these experiments will never be published.

Why were so many maternal deprivation experiments conducted? The seminal finding of separation experiments was that rhesus monkeys exhibit, under certain conditions, a protest-despair response similar to humans. Subsequent experiments were attempts to demonstrate the same response in other species or in rhesus monkeys reared in different conditions. These types of studies are still being conducted.

Other experiments were attempts to determine the influences of several variables on an infant’s behavioral responses to separation. These variables included the infant’s age, length of separation, degree of separation, previous history of separation, nature of mother-infant relationship, nature of separation environment, and other factors. The effect of these variables on a variety of physiological responses to separation was also studied. Different species, as well as different responses, were studied.

Mother-infant separation experiments led the Wisconsin researchers into their depression research. Suomi began his peer separation experiments to find a new way of inducing depression. The same sorts of experiments that involved mother-infant separation could be conducted again using infant-infant separations. Harlow introduced his vertical chamber as another means of inducing depression.

Mother-infant separation experiments also led the Wisconsin researchers into their recent anxiety research.

The clear message of this scenario is that these studies could drag on indefinitely, with yet more species, more diverse situations, and more procedures being tested. Indeed, recent reviews of separation experiments state that many more experiments need to be conducted. One is reminded of Levy’s (1952) adage: The “animal psychologist . . . is often wedded forever to his animal or his instrument” (p. 485).

The point is not that all maternal deprivation experiments were exact duplicates of others, or that none of these experiments had scientific merit. The point is that most seem trivial in light of their harsh exploitation of animals, their dismal record on clinical applications, their mediocre quality, and the value of alternative, human studies.

Consider the problem of research quality. Several lines of research used poor models of human phenomena. Total isolation was not a good model of institutionalization; separation of infant monkeys from their mothers when the latter were the infants’ only companion was not a good model of the more complex human situation; vertical chamber confinement was not a good model of depression, etc. Some poor models were quickly abandoned but others were used again and again. For example, infant monkeys were subjected to just about every conceivable schedule of total isolation.

The quality of maternal deprivation research was compromised a priori when poor models were used. Research quality was further compromised by problems with experimental design. Numerous experiments had defective designs. Moreover, the designs of separation experiments were nonstandardized, so that results differing from study to study could have resulted from any of a number of factors.

The problem of self-perpetuation of experiments, at least insofar as maternal deprivation experiments are concerned, is not one of inadequate dissemination of scientific information. Researchers propose their new, slightly different experiments, not out of ignorance of previous research, but in an effort to add to previous research (however trivial the additional information). In some ways, past research encourages new research, producing a bandwagon effect. Hence, this survey supports Gendin’s (1984) belief that passing legislation to better disseminate scientific information will not necessarily reduce duplicative research, but might even encourage it.

The topic of alternatives is almost moot with respect to models of maternal deprivation and maternal separation. After all, human studies predated these animal models. One wonders what conclusions the modelers would have drawn without the human data for direction and comparison. Recall that animal species differed in their basic behavioral response to deprivation and separation, not to mention responses to subtler variations in procedures.

The maternal deprivation modelers have overly criticized human studies. To be sure, the human research contained numerous examples of poor design, and it relied substantially on retrospective studies. Examples of faulty research can be found in any field. However, the animal modelers were attempting to throw the baby out with the bathwater. The human research contained numerous examples of well-designed and/or prospective studies (see Yarrow 1961). With all their purported drawbacks, human studies of maternal deprivation and maternal separation were the first to yield the conclusions that the animal studies later “supported.”

Research strategies for studying maternal deprivation and separation in human infants often involve “natural experiments.” Yarrow (1968) discusses two such situations:

[1] Child-care institutions that differ in basic child-rearing philosophies. There are great variations among these institutions in caretaker ratios and in various characteristics of the physical-sensory environment, such as . . . complexity of visual and auditory stimulation. Even though past studies of these settings have been theoretically unsophisticated and have had serious methodological shortcomings, they have produced some significant findings. With sharper conceptualizations and more refined observational techniques, more precise data on the effects of very early sensory and social deprivation in human infants can be obtained. Moreover, such settings also lend themselves to simple enrichment studies involving experimental intervention, e.g., . . . training caretakers to give measured degrees of language or social stimulation, or to respond contingently to specific aspects of the infant’s behavior.

[2] Several kinds of separation situations that occur with moderately great frequency in the life of young children allow exploration of important aspects of early experiences. Among the many kinds of situations in which variables such as loss of a significant relationship figure, broad environmental changes, or novelty, can be teased out and studied independently or in interaction are the following. A) Situations in which the mother leaves the infant temporarily in familiar surroundings, as distinguished from temporary separation associated with gross environment change. B) Situations in which the infant is temporarily moved to another home with an attentive caretaker, as distinguished from separation followed by gross deprivation of maternal care, as in institutionalization. C) Separation experiences associated with adoptive placement involving a change in mother figure (p. 107-108).

Yarrow adds:

The use of natural situations in research has inherent difficulties because of the many variables in complex and uncontrolled interactions, only some of which we are aware of and can specify. To use natural situations effectively, it is necessary that we conceptualize as clearly as possible the theoretical issues, and identify clearly the variables in the situation relevant to the theoretical issues , . . (p. 108). 

Observational techniques for such human studies have become increasingly sophisticated over the years. Ainsworth (1976) has incorporated ethological techniques into her naturalistic studies of mothers and infants. Unfortunately, “naturalistic studies of [human] mother-infant interactions . . . are not popular with grant-awarding panels” (p. 46).

Naturalistic studies examine a process in the context of its ecological setting. In contrast, experimental studies with animals are often attempts to remove influences of all variables except one.

The problem with this approach is that in order to keep all other factors constant, infants must be reared in relatively sterile environments that do not promote normal social development (Suomi 1977b, p. 212).

Yarrow (1968) highlighted the importance of naturalistic studies in unraveling human development.

The values and limitations of experimental studies with animals under closely controlled conditions should be weighed against the values and limitations of investigations with humans in naturalistic settings. Some kinds of problems clearly lend themselves to experimental manipulation of a limited number of variables; for other kinds of problems, such an approach can yield only suggestive hypotheses that must be tested in more complex settings. There are obvious limitations in the extent to which relationships established in a controlled “pure situation” can be generalized to the complex naturalistic setting. The action of a single variable in a hypothetical vacuum may be very different from the action of this same variable in a network of other variables. In environments in which there are many variables in complex interaction, as in most human environments, the significance of any given variable or limited set of variables may be radically altered by the larger context. . . As the behavioral sciences become more mature, I think we will have to give up the inordinate value placed on experimental manipulation of a few simple variables, and give increasing attention to the development of research designs for controlled analyses of complex environments (p. 112-13). 

Greater understanding of mother-infant interactions and child development can also be achieved through increased use of cross-cultural human studies. Although this kind of research has inherent weaknesses, these seem no greater than those handicapping animal studies. Cross-cultural studies can also reveal that conclusions from animal data are culture-bound. What animal data tell us about “the human case” depends on what we regard as the human case. With respect to the monkey separation studies, “what we have been pleased to call the human case” is Western society (Lehrmann 1974, p. 194).

Animal Models in Psychiatry

Most of the animal models discussed above can be considered to fall within the domain of psychiatry. These include models of mental illness, maternal deprivation, maternal separation, and depression. These models were part of a move toward using animal models of psychiatric disorders. This move was prominent during the 1960s and was an attempt to mimic animal model research in other fields of medicine (McKinney et al. 1975, Reite 1977).

Psychiatrists have traditionally been skeptical of animal models of mental disorders. This skepticism stems partly from the belief that human mental capacities are necessary to become neurotic (McKinney 1974a), and was fueled by careless analogies in early animal research (Reite 1977). This skepticism among psychiatrists still lingers. For example:

Human nature can not be reduced to experimental models of laboratory animals. . . . lf, however, any scientific conclusions are to be drawn from the animal model, the animal should be observed in the natural habitat . . . (Serban 1976, p. 280).

Again, “mechanical concepts [of experimental psychologists] are totally unsuited to man’s psychosocial existence” (Serban, in Serban et al. 1976, p. 2).

However, according to Suomi (1984a), “animal models of human psychopathology have been gaining increasing respectability and use” (p. 226). Several of the maternal deprivation modelers themselves have been psychiatrists, including McKinney, Reite, Kaufman, Levine, Jensen and Hofer.

Ironically, as the psychiatrists’ skepticism of animal models is diminishing, the animal modelers themselves are beginning to acknowledge that animal models at best can make only tentative suggestions about human parallels (e.g. Suomi 1982). This implies that poor animal models yield suggestions about humans that are indeed shaky. Consider a recent separation study of rhesus monkeys by Suomi et al. (1983). The researchers wrote:

What relevance, if any, do these findings have for . . . human mothers (and fathers) and infants? We clearly acknowledge that cross-species generalizations should be made with great caution and according to well-thought-out principles and that true one-to-one matches of phenomena between species are relatively rare. . . . Furthermore, the living conditions of our monkeys during separations are not analogous to the living conditions of most human infants during parental separations and it seems unlikely that exactly the same pattern of results would occur if our infant monkeys had been with peers or other adults during their separations, as human infants usually are. Nevertheless, the present monkey data raise some issues that might profitably be kept in mind when pondering possible consequences of frequent parental separations in young children (p. 784). 

Other statements in the maternal deprivation literature also do not inspire confidence in animal models. For example, Levine (in Harlow & Suomi 1976) once pointed out to Harlow that animals in laboratories are abnormal, even before experimental manipulations. Harlow’s response included the following statement: “You pray that [the research measures] have validity; . . . you just have to gamble!” (p. 74).

In spite of the shortcomings of animal models, the Wisconsin researchers have vigorously promoted their use in psychiatry.

The major areas of interest for psychiatrists in primate models probably include the development of kinds of attachment behavior, the effects of social isolation, separation studies, the possible experimental simulation of learned helplessness, and various biological approaches being developed. It is in these areas that work has been and is being conducted that has potential clinical usefulness. Investigators may be close to developing a viable animal model for depression that may facilitate a more comprehensive understanding of this particular syndrome and enable studies to be done that are currently impossible to perform utilizing human beings. The rest of medicine has long used nonhuman primates to advance knowledge about their fields and there is no logical reason why psychiatry should not do the same (McKinney et al. 1975, p. 330).

Yet, skepticism of animal models should remain firm. At least three reasons are apparent. First, experiments have had very little clinical impact. Second, they siphon money away from acceptable research on the human condition. Third, they subject animals to harsh treatment.

Maternal deprivation researchers have almost completely avoided discussing humane and ethical issues raised by their callous treatment of animals. There is virtually no indication that any procedures were designed or modified out of compassion for the animal subjects. The use of anesthesia and similar procedures was based on efficacy, not humanity.

Maternal deprivation researchers raise ethical issues only when “explaining” that ethical constraints dictate that animals be used instead of humans. The implicit premise is that the experiments can be performed on animals because animals are different from humans. The irony is that the scientific rationale for these experiments emphasizes the similarities between animals and humans. This similarity is demonstrated by the experiments themselves. A few researchers were struck by this similarity (e.g. Suomi 1977b, p. 222). Consider the conclusion from Bronfenbrenner’s comparison of the effects of deprivation-rearing in human and nonhuman primates:

. . . the most surprising and possibly sobering outcome of [our] . . . analysis [is] the similarity we humans bear to our fellow primates — at least when it comes to our need for stimulation and for attachment to others (1968, p. 756).

Unfortunately, mounting evidence of the similarities between human and nonhuman primates has not engendered a reevaluation of maternal deprivation experiments on animals. Instead, the similarities are considered an incentive for further research.

This situation puts the animal modelers in an odd position. On the one hand, their experiments raise disturbing questions about the propriety of this research. Moreover, public concern over these experiments is mounting. On the other hand, the modelers want to continue their research. The same dilemma applies to animal research in general. Researchers and funding agencies have responded by proposing guidelines for animal research. However, as in many cases of self-regulation, the strength of these guidelines has been exaggerated. Consider Suomi’s (1984b) statement:

. . . in most respects, current scientific and ethical guidelines for care and treatment of nonhuman primate subjects are at least as strict as those pertaining to research utilizing human subjects (p. 120).

Suomi does grant that the privilege to use primates in research does not mean that researchers have free reign over what they can do to their subjects.

What it does mean is that experimenters can selectively breed specific rhesus monkey pairs, and they can perform cross-fostering manipulations in order to separate possible genetic and rearing environ-mental effects. Moreover, they can monitor their monkey subjects’ behavioral and physiological activities to a degree and duration almost never feasible for any humans . . . (Suomi 1984b, p. 120).

Suomi neglected to mention harsher procedures that apparently are still allowed under the research guidelines. These include traditional deprivation experiments (e.g. Coelho & Bramblett 1984) and separation experiments (e.g. Nadler & Codner 1983).

The National Institute of Mental Health (NIMH) is worried about the public response to psychology experiments on animals. After NIMH recently hired Suomi, its director of intramural research praised him as “an animal protectionist of the first class” for his concern about the care of research animals (Cordes 1984). This concern for animals is not revealed in Suomi’s research reports. Consider the following passage, which was a discussion of his research plans:

We are also interested in the tunnel of terror in that it may be utilized in various ways in combination with the pits [vertical chambers], to shorten, intensify, and possibly stereotype the depressive syndrome. For example, individual monkeys or social pairs may be driven to the pits by the gradual or rapid approach of the marching monster. Depression in individual monkeys already incarcerated may be further enhanced by the appearance or advance of the monster from the top or ends of the pit chambers. Also, pairs of monkeys may be separated after tunnel terror or monster machinations in order to assess the possible complementary contribution of social separation to psychopathological production (Suomi & Harlow 1969, p. 249).

Characterizing people who conduct this type of research as animal protectionists does not instill confidence in the self-regulation of psychological research on animals.

Summary and Conclusions

Over 250 maternal deprivation experiments on animals have been conducted to date. These experiments cost over $57 million, provided primarily by the federal government. Over 7,000 animals were subjected to procedures that induced distress, despair, anxiety, general psychological devastation, or even death. The results of this research have had little impact on clinical practice, and the potential for future advances seems limited. Many experiments were trivial extensions of past research, or simply were attempts to reproduce in animals what was already known about humans.

Given research such as this, it is not surprising that psychology experiments on animals have been criticized as cruel, wasteful or misguided (e.g. Bowd 1980, Bannister 1981, Drewett & Kani 1981, McArdle 1984). Rebuttals of these charges (Gallup & Suarez 1980; Coile & Miller 1984, Miller 1985) illustrate the need for in-depth evaluations of psychological research. The present evaluation provides substantial support for the critics, at least with respect to maternal deprivation experiments. If this area of research is representative of others within psychology, then the use of animals in psychological research faces a formidable challenge.


Please cite as:

Stephens, Martin L. (1986) Maternal Deprivation Experiments in Psychology: A Critique of Animal Models.
Retrieved from