Chapter Four

Experiments in Deprivation

The severe psychological problems that developed in Harlow’s monkeys led him to consider the effects of maternal deprivation. Hence, Harlow’s experiments in the affection category led to his experiments in the deprivation category. These experiments concern the effects of social deprivation on infant development. In virtually all of these experiments, infants were not only deprived of their mothers, but also denied contact with any other individuals of their species. They were kept alone in bare, wire cages or in isolation chambers.

Several rationales for deprivation experiments have been offered by those conducting this line of research. These experiments have been considered to be animal models of social isolation, mental illness, or even normal development. Some have been considered to be relevant to child abuse. In addition, some were attempts to assess the importance of early life experiences to behavioral development. Rearing an animal in a depriving environment was a means of nearly obliterating its life experiences. Hence, any behavioral abnormalities that developed were attributed to deficient experience, not deficient genetic endowment.

The following section is a summary of these deprivation experiments. It should be emphasized that our concern here is not with deprivation experiments in general, but with those that deprive infants of their mothers. Deprivation experiments on juveniles and adults are beyond the scope of this report.

Overview of Deprivation Experiments

Deprivation experiments have been conducted primarily on nonhuman primates and dogs, and only these studies are considered here. Some deprivation experiments have been conducted on farm animals and rodents but these studies are all but ignored by researchers studying dogs and primates. The studies of farm animals are outdated and those on rodents mostly involved animals older than infants. For reviews of the experiments on farm animals and rodents, see Moore (1968), Porter (1974) and Thoman and Arnold (1968).

Experiments on Dogs

Several deprivation experiments were conducted at McGill University during the 1950s. These experiments involved rearing terriers in isolation chambers for several months (Photo 3). Isolation began when the puppies were separated from their mothers at four weeks of age. The emphasis of these experiments was on the consequences of early experience for behavioral development.

The results indicated that isolation-reared dogs had bizarre “whirling fits” (Thompson et a/. 1956) and aberrant, highly excited responses to innocuous objects (Melzack 1954). Hebb, a McGill researcher, wryly described these animals as “permanent screwballs” (Hebb 1958, p. 109). The isolates were poor at problem-solving, “a result in agreement with previous animal experimentation” (Thompson & Heron 1954, p. 31). They were also inferior to nonisolated dogs in social behavior, even when reared with a companion in their isolation chambers (Melzack & Thompson 1956).


Photo 3
Isolation chambers for dogs at McGill University. From R. Melzack 1965, “Effects of early experience on behavior,” in Psycho-pathology of Perception (P H. Hoch & J. Zubin, eds.), Grune & Stratton, Inc. Reproduced by permission of author and publisher.

Melzack continued these experiments when he moved from McGill to the Massachusetts Institute of Technology (Melzack 1962, 1965, Melzack & Burns 1963). Isolation-reared beagles, like the McGill terriers, were hyperactive when removed from their chambers. They displayed abnormal responses to visual, auditory and olfactory stimulation, and seemed unperturbed after inadvertently crashing into things. Melzack attributed their difficulties to an inability to filter out irrelevant sensory information. This led to hyperexcitement, which in turn, interfered with testing performance. Preliminary findings from physiological experiments tended to support this view.

Several deprivation experiments involving dogs were conducted at the Jackson Laboratories, a supplier of animals to research institutions. Puppies were reared in total isolation for three to five months, beginning at three or four weeks of age. As in the McGill and MIT experiments, the puppies were hyperexcited on emerging from their chambers. Fuller and Clark (1966a) characterized this behavior as a “panic response.”

…its behavioral manifestations can vary from cowering withdrawal, rigid immobility in unusual postures, through stereotyped approach-avoidance, and violent tantrums with loss of [bladder] control (p. 256).

Fuller and Clark’s experiment suggested that the effects of isolation resulted from the trauma of emerging from isolation, rather than from deprivation during the isolation period. They called their explanation the “emergence trauma hypothesis,” and contrasted it with the “critical period hypothesis.” The latter, when applied to isolation experiments, states that the ill effects of isolation result from a lack of normal experience during a “critical period” in development.

It is important to note that two breeds were tested and found to differ in the severity of their response to isolation. Hence results from one breed do not necessarily generalize to another breed, let alone to another species. The same conclusion was supported by another experiment (Fuller & Clark 1968).

Fuller and Clark (1966b) found that various forms of stimulation during isolation-rearing did not prevent abnormal social and exploratory behavior from developing. The authors interpreted these results as further support for the emergence trauma hypothesis.

Later experiments, reviewed in Fuller (1967), showed that less severe deprivations can have less severe effects. Fuller also reviewed experiments on the effects of isolation on problem solving. The results admittedly were inconsistent.

Several isolation experiments on dogs were also conducted at Galesburg State Research Hospital by Fox and coworkers (Fox & Stelzner 1966a). Isolates were caged individually and kept in a darkened room for one week, beginning at four weeks of age. Thus the pups experienced sensory as well as social isolation.

One week of isolation produced less drastic and persistent abnormalities than the long-term isolation in the McGill, MIT and Jackson experiments (Fox & Stelzner 1966b). For example, the isolates did not exhibit the “mass-fear response” (extreme avoidance and fearfulness) seen in the long-term isolates. The researchers attributed the differences between the long- and short-term isolates to the timing and the duration of isolation.

One week of isolation was sufficient to produce intense behavioral arousal on emergence. This arousal was associated with changes in brain physiology (Fox 1967). One group of dogs in this study had electrodes chronically implanted in their brains. The results lent support to the interpretation of the McGill, MIT and Jackson studies; namely, that dogs reared in isolation are unable to readily adapt to the sensory stimulation of normal environments.

Another experiment attempted to correlate behavioral and neurochemical effects of isolation (Agrawal et al. 1967). After isolation, the puppies were decapitated and their brain chemistry was analyzed. Several changes had occurred in the concentrations of free amino acids. The interpretation of some of these neurochemical changes was unclear. As the researchers noted, some may have resulted from the darkened isolation room, rather than the social isolation itself.

Experiments on Primates

University of Wisconsin and Wisconsin Regional Primate Research Center

Deprivation experiments on primates began in the 1930s (Foley 1934, 1935) but did not become popular until the 1960s. Most deprivation experiments on primates were conducted at the University of Wisconsin, under the direction of Harry Harlow. All of the Wisconsin deprivation experiments involved rhesus monkeys. These experiments were viewed as animal models of human development— both normal and abnormal. The model of normal development was viewed as a means of assessing the importance of early social experiences. Most of the attention, however, was on abnormal development, with the deprivation experiments being considered as models of maternal deprivation, social isolation and mental illness.

The Wisconsin researchers distinguished two types of isolation conditions. ‘Partial isolation’ involves rearing infants alone in bare wire cages. The infants can see and hear other monkeys, but not touch them (Photo 4). ‘Total isolation’ involves rearing infants in isolation chambers that preclude all forms of interaction between monkeys (Photo 5). The partial isolation experiments are summarized first, followed by the total isolation experiments, and then various other kinds of deprivation experiments are considered.


Photo 4
Rhesus monkey infants being reared in ‘partial isolation.’ Each infant is housed alone in a wire cage (the front panel of the cage is plexiglass). From Harlow and Harlow (1966) with permission.

Partial Isolation

Mason, who at the time was Harlow’s student, conducted the first Wisconsin experiments on the effects of partial isolation (Mason 1960, 1961a, 1961b, 1963a, 1963b, Mason & Green 1962). Rhesus monkey infants were subjected to partial isolation for their first 28-29 months. They were then tested in a variety of situations. A control group consisted of monkeys captured in India, shipped to a zoo, and later taken into the laboratory.


Photo 5
Rhesus monkey infant in an isolation chamber. The photograph was taken when the chamber door was raised for the first time after half a year of ‘total isolation.’ Reproduced from Harlow (1964) with permission; copyright by the Williams & Williams Co., Baltimore.

The isolates were socially abnormal. They responded poorly to social cues, and their grooming, sexual, and aggressive behavior were impaired. They were also more disturbed than wild-born monkeys when placed in novel situations. Mason concluded on the basis of a variety of tests that “orderly and harmonious social relations in rhesus monkeys are dependent upon previous socialization experience” (Mason 1961a, p. 290).

Harlow and Harlow (1962b) described the abnormalities of a much larger group of monkeys reared in partial isolation. Fifty-six rhesus infants were available for study during the early 1960s because husbandry conditions that were routine at the time called for partial isolation. The unanticipated abnormalities of these isolates included blank staring and stereotyped, repetitive circling of cages. These infants often had compulsive habits, including self-mutilation. Harlow and Harlow (1971) described the self-mutilation of limbs:

Ordinarily [self-biting does] not break the skin, but under conditions of stress or threat they may literally tear their limbs to pieces. In a considerable number of instances, injury has been so severe that the animals had to be sacrificed (p. 210).

Surprisingly, partial isolates were nearly as proficient on learning tasks as were control monkeys (Gluck et al. 1973).

Isolates that were left in their cages did not outgrow their affliction (Cross & Harlow 1965; Suomi et al. 1971). They remained withdrawn and self-abusive although some developmental changes occurred (e.g. Arling et al. 1969). The control group in the Suomi et al. study had been born in India, captured and brought into the laboratory before their second birthday, and subjected to solitary confinement for 9 to 15 years.

Total Isolation

In the earliest series of Wisconsin experiments on the consequences of total isolation, newborn rhesus monkeys were confined for 3, 6 or 24 months (Harlow & Harlow 1962b), as well as for 18 months (Mason & Sponholz 1963). Three months of confinement had no noticeable effect on seven of eight infants. However, longer periods of confinement had increasingly severe effects. The 24-month old infants “were totally destroyed . . . ” (Harlow 1964, p. 159).

These initial experiments demonstrated that total isolation left monkeys extremely fearful (Photo 6). Indeed, “A stimulus such as a tennis ball . . . is apt to elicit intense terror from an isolate monkey,” even one reared in partial isolation (Suomi & Harlow 1976, p. 20).


Photo 6
Rhesus monkey infant in isolation chamber. The photograph was taken after the chamber door was raised. Note the self-clutching, a characteristic of rhesus monkeys reared in isolation. From Harlow and Harlow (1965) with permission.

According to Harlow et al. (1965), newborns were also “sentenced” to 3, 6, or 12 months of total isolation. A fourth group was subjected to 6 months of isolation beginning at 6 months of age (“6-month late isolates”). Following the clinical literature, Harlow referred to the condition of the 6-month late isolates as ‘deprivation’ and that of the other isolates as ‘privation.’ No mention was made of the earlier Wisconsin experiments that sentenced newborns to 3, 6, 18, and 24 months of total isolation. Apparently, these experiments were regarded as preliminary.

The goal of the newer research was a better understanding of the effects of “human social isolation,” which was said to occur in orphanages, foster homes, and abnormal private homes.

The results of these experiments, which were reported by Harlow et al. (1964), Harlow et al. (1965), and Griffin and Harlow (1966) were as follows. When released from their isolation chambers, the infants initially suffered “emotional shock.” The three-month isolates were especially traumatized at release, showing “extreme withdrawal” and “severe depression.” This reaction was so severe that one [monkey] died of starvation because it refused to eat .. . (Griffin & Harlow 1966, p. 534). Another three-month isolate would have died of this “emotional anorexia” had it not been force-fed.

Total isolation had dramatic, debilitating effects on the survivors (Photo 7). The severity and persistence of these effects were related to the duration and timing of isolation. The effects were eventually reversed in the three-month isolates. The six months of early isolation “severely impaired” the potential for socialization, whereas “12 months of isolation almost obliterated the animals socially . . . ” (Harlow et al. 1965, p. 94). These asocial animals were severely disturbed. “Testing . . . had to be discontinued . . . because the control animals were mauling and abusing the helpless isolates to the point that they could not have survived continuing interaction” (Harlow & Harlow 1969, p. 18).


Photo 7
A rhesus monkey infant reared in an isolation chamber. Reproduced from Harlow (1974) with permission.

The 6-month late isolates showed neither the severe short-term reaction of the 3-month isolates nor the severe long-term reaction of the 6-month early isolates and the 12-month isolates.

Learning tests revealed little evidence of intellectual impairment in any group; extensive follow-up testing supported this conclusion (Harlow et al. 1971a).

These monkeys were 12 to 20 months old when tested initially. The same isolates were tested again at 28 to 44 months and showed few signs of improvement (Mitchell et al. 1966). They were fearful and aggressive and showed little or no interest in sex or play. Their acts of aggression were often suicidal attacks against huge adult males or brutal beatings of juveniles . ” (p. 578). Throughout testing, the 12-month isolate showed “practically no positive social behavior” (p. 568). The isolates were also characterized by “infantile disturbance” and bizarre ritualistic movements.

Mitchell (1968a) tested these isolates again when they were 42 to 54 months. There were some signs of improvement, though the isolates remained socially inept. Fear was increasingly replaced by hostility.

Mitchell and Clark subjected rhesus infants to six months of total isolation beginning at three months of age, instead of at birth or at six months of age as in the earlier Wisconsin studies. This study also differed in other ways from most earlier Wisconsin studies. One difference was that the isolates were reared with cloth surrogate mothers. Another difference was that the isolates were preadapted to the post-isolation test apparatus. The latter procedure was an attempt to reduce the trauma associated with emerging from isolation chambers. This part of the experiment was a test of the “emergence trauma hypothesis” developed from isolation studies on dogs (see above).

The isolates were “inferior” to controls in sexual, emotional, and dominance behavior. However, one wonders to what extent these differences resulted from factors other than total isolation. The isolates differed from one or both control groups in several ways other than being in total isolation for six months. These differences involved access to real mothers, access to peers, and separation from peers.

Sackett (1966) kept newborn rhesus infants in isolation chambers for nine months. The aim was not to induce abnormal behavior, but to see how “socially naive” infants would respond to pictures of monkeys. The pictures depicted monkeys of various ages and sexes engaged in different activities. The infants responded in a species-typical manner to certain pictures, and this was taken as evidence for unlearned or inborn responses to social cues. The abnormalities of these ‘picture isolates’ were described in later reports (Sackett 1967a, 1968a, b).

Follow-up Studies

Numerous follow-up studies were conducted on monkeys that had been reared in partial or total isolation at the University of Wisconsin. Several of these studies described the maternal behavior of females who had been deprived of their own mothers throughout infancy (e.g. Harlow & Seay 1966, Suomi 1977a). Harlow termed these females “motherless mothers.” These females had been reared in wire cages either alone, with cloth surrogates, or with peers. They were so disturbed by their deprived upbringing that few were impregnated without experimental intervention. Most of them were artificially inseminated or impregnated by males while restrained in a ‘rape rack’ — a device designed and named by Harlow.

All four motherless mothers in the first study were “totally inadequate” in caring for their first offspring (Seay, Alexander & Harlow 1964). Two infants died from neglect or abuse, one after becoming blind and partially paralyzed and the other after being refused mothers’ milk. All of the infants probably would have been killed if the experimenters had not intervened.

Most of the motherless mothers ignored their infants, but other motherless mothers abused their babies by crushing the infant’s face to the floor, chewing off the infant’s feet and fingers, and in one case by putting the infant’s head in her mouth and crushing it like an eggshell (Harlow et al. 1971b, p. 545).

(See Photo 8). Surprisingly, the motherless mothers in the Seay et al. study behaved more or less normally with their second infants.


Photo 8
Typical behavior of an isolation-reared female towards her infant. The mother is crushing her baby against the cage floor. From H. F. Harlow, 1962, “The Heterosexual Affectational System in Monkeys,” Am. Psychol. 17:8. Copyright (1962) by the American Psychological Association. Reprinted by permission of the publisher.

Despite neglect and abuse, infants of motherless mothers persisted in attempts to cling to, and suckle from, their mothers. In a rare display of sympathy, Harlow wrote:

One of the things that would unnerve experimenters was to watch the desperate efforts of these babies to make contact with the abnormal mother. She would beat them and knock them down; they would come back and make contact; the mothers would rub their faces into the floor; they would wriggle free and again attempt contact. The power . . . and demandingness of the infant to make contact and the punishment the infant would accept would make strong men reach the point that they could hardly bear to observe this unmaternal behavior (Harlow 1963, p. 25).

The social development of these infants was adversely affected by their traumatic upbringing (Arling & Harlow 1967). They “were deficient in social play and sexual behavior and hyperaggressive in peer interaction” (p. 371).

Several factors influenced the adequacy (or inadequacy) of maternal care by motherless mothers (Ruppenthal et al. 1976, Suomi 1978). These factors included (1) mother’s age at first social contact, (2) mother’s previous experience with peers or their own infants, and (3) sex of infant. For example,

… females with minimal . . . contact [with their previous infants] rarely showed maternal care improvement even after as many as five births. In most of these cases these females were hyper-abusive, either killing their baby soon after birth or mutilating it . . . (Ruppenthal et al. 1976, p. 345).

Gross ineptitude, not hyperaggression, was behind much of this abuse. For example, a female was seen “calmly chewing on its [infant’s] digits” and trying to “rip the remainder of the umbilical cord out of the infant’s midsection” (p. 346). However, more than ineptitude was involved. Abusive mothers “sometimes sought out the baby to strike or bite it without provocation” (Harlow et al. 1966, p. 59).

Other follow-up studies at Wisconsin examined different aspects of deprivation rearing. Sackett (1967a) found that the adverse effects of various deprivation experiences were persistent. He tested four-year-olds that had been reared in a variety of conditions, including solitary rearing in cages or isolation chambers and rearing with motherless mothers. The greater the deprivation experience, the greater the damage evident years later. The most deprived individuals were the most fearful and withdrawn and the least sociable and active.

Sackett also found that partial isolates and, especially, total isolates were less active and exploratory in novel environments than control monkeys (Sackett 1972a). However, female isolates were more normal than males on these measures, suggesting that in some respects, males are more vulnerable than females to isolation.

Sackett and coworkers conducted several studies of the social preferences of monkeys that had been reared in partial or total isolation (Sackett et al. 1967; Pratt & Sackett 1967; Suomi et al. 1970a; Sackett & Ruppenthal 1974). The aim was to relate these preferences to the isolates’ limited social experience or to their “innate” tendencies. Preferences were assessed in the “self-selection circus,” which had a central compartment that was separated by windows from several outer compartments. The chooser was placed in the center and the animals to be chosen were placed in the outer compartments. Preferences for particular “stimulus animals” were measured by the relative amounts of time the chooser spent near each candidate. Preferences were considered to be uninfluenced by the behavior of the stimulus animals, but this assumption was not necessarily true.

In one of these preference studies (Sackett et al. 1965], monkeys that had been reared alone in wire cages, but handled by humans during their first month, preferred humans to monkeys. The conclusion was “that the very early experience of a monkey with a specific type of stimulus can have lasting effects upon later preferential choice concerning that stimulus” (p. 306). Other results indicated that this very early experience can be overcome by later experience.

Sackett (1970a) concluded that the results of his preference studies do not support theories that emphasize the role of learning in the formation of social bonds. Sackett argued that monkeys are born with social tendencies, or they inevitably acquire these tendencies as they mature. These “response biases” lead to the acquisition of social attachments in the absence of learning opportunities. A second conclusion was that early rearing experiences are crucial in determining (1) “whether or not social attachments to species members will be formed” (p. 112) and (2) the nature of the resulting preferences (e.g. infants over adults). Although Sackett downplayed the importance of learning in the formation of social bonds, he believed that learning is important in the maintenance of social bonds.

The follow-up experiments that have been considered so far examined the maternal behavior and social preferences of maternally deprived monkeys. Follow-up experiments on these monkeys also investigated responses to stress. In one study, partial or total isolates did not differ from control monkeys in any hormonal measure associated with stress (Meyer & Bowman 1972). Monkeys were stressed by being placed in “primate restraint chairs” for two hours. Somewhat contradictory results were found in a second study (Sackett et al. 1973).

Habilitation Experiments

Wisconsin researchers made several attempts at habilitating monkeys reared in isolation (e.g. Chamove 1978). None of these attempts was motivated by sympathy for the disturbed monkeys. Rather, the rationale was to consider the “post-isolation syndrome” of rhesus monkeys to be a model of human mental disturbance, and to develop therapeutic techniques that may apply to humans.

The first attempted therapy simply involved giving isolation-reared monkeys opportunities to interact with one another. Partial isolates were moved to a zoo and placed on a monkey island that was surrounded by alligators (Harlow & Harlow 1962b, Harlow & Griffin 1965). This social experience had mixed results. “Fighting was severe at first” (Harlow & Harlow 1962b, p. 141). Two females were so injured that they were returned to the lab. One male drowned. Some friendships developed but sexual behavior remained impaired. The monkeys were returned to the lab, where friendly interactions ceased and fighting increased. “After two years we had to conclude that the island experience was of no lasting value” (Ibid., p. 141).

Sackett (1968b) attempted to habilitate socially withdrawn isolates by training them to approach one another. The isolates were placed in an apparatus with an electrified floor. The only way an isolate could escape shock was to jump onto a stool that already contained a monkey. The aim was to induce the monkey being shocked to join the monkey on the stool. This rather limited aim was met. However, the induced social contact failed to habilitate the monkeys.

The Wisconsin researchers eventually were successful in their efforts at habilitation (Harlow & Suomi 1971a; Suomi & Harlow 1972a; Suomi et al. 1972). Their subjects had spent their first six months in isolation chambers. Therapy involved pairing these isolates with ‘therapist’ monkeys (Photo 9). The therapists were younger and more normal than the isolates, but they too had been deprived of their mothers from birth. The isolates showed significant recovery in social and personal behavior. Success was attributed to the choice of therapists. These younger individuals sought contact without being threatening. They also could tutor isolates as they themselves developed social behaviors.

“Social therapy” was successfully extended to monkeys that had spent their first twelve months in isolation chambers (Novak & Harlow 1975, Novak 1979).

The success of the social therapy cast doubt on the Critical Period Hypothesis as an explanation for the devastating effects of total isolation on rhesus monkeys. This hypothesis states that certain social experiences facilitate behavioral development, but in order to do so, must occur during a critical period. If isolation prevents this, then the individual’s development should be permanently impaired. This does not appear to hold for rhesus monkeys.

Another possible explanation for the effects of isolation is the Emergence Trauma Hypothesis, derived from experiments on dogs (see above). Although emergence trauma may play a role in rhesus monkeys (Novak 1979), the Wisconsin researchers favor a learning deficit explanation (Suomi et al. 1974; Novak & Harlow 1975). According to this explanation, isolation precludes opportunities for social learning and leaves individuals severely disturbed. Under certain facilitating conditions, these disturbed individuals can learn species-typical behavior from more normal monkeys.


Photo 9
Rhesus monkeys in a therapy experiment. On the left, a monkey reared in an isolation chamber. On the right, a ‘therapist’ monkey. Reproduced from Harlow (1974) with permission.

Rhesus monkeys in a therapy experiment. On the left, a monkey reared in an isolation chamber. On the right, a ‘therapist’ monkey. Reproduced from Harlow (1974) with permission.

The process of social learning is extremely difficult because isolates are so withdrawn. Their self-directed behavior interferes with social interaction. Sackett (1970b) argued that isolates have difficulty inhibiting their self-directed behavior, as well as most existing behavior patterns. He proposed that this inability was responsible for the persistence of the isolates’ disturbance. Testing revealed that isolates do have difficulty inhibiting response patterns (Gluck & Sackett 1976, p. 174).

The success of the social therapy technique overshadowed the limited successes achieved by other techniques. These include exposure to cloth surrogate mothers (Suomi 1973a) and administration of drugs (McKinney et al. 1973; Nobel et al. 1976). In the latter study, one infant was so severely disturbed by the isolation procedure that “his disturbance behaviors included headbanging, convulsive jerks and spasms . . . and he seemed more susceptible to injury and illness” (p. 1169).

The long-term effects of both surrogate therapy and social therapy (involving ‘therapist’ monkeys) were assessed by Cummins and Suomi (1976). The habilitated monkeys showed no signs of regression, although the surrogate-habilitated group still exhibited only limited recovery.

Social Factors in Normal Development

Harlow and Harlow (1962b) discussed the development of rhesus monkey infants in various rearing environments. Their aim was to assess the ability of mothers and infants to facilitate normal development in infants. For example, Harlow and Harlow contrasted two social environments: mothers and infants versus cloth surrogate mothers and infants. Infant development in the two environments suggested that an infant’s relationship with its mother facilitates its social and sexual interactions with peers. Hansen (1966) reached a similar conclusion.

Various rearing experiments suggested that interactions with peers (other infants) were important for infant development. For example, Harlow and Harlow reared infants with one another, but not with mothers. They appeared to develop normally in their play and sexual behavior, despite the absence of real mothers. Harlow and Harlow suggested that interactions among infants “may compensate for lack of mothering” (p. 146). A similar conclusion was reached by other Wisconsin researchers (e.g. Hansen 1966, Sackett 1968a, Suomi & Harlow 1971).

However, later experiments provided less cause for optimism about the fate of peer-reared monkeys (Chamove et al. 1973). Peer-reared monkeys developed abnormally, perhaps because they clung excessively to one another. Moreover, their abnormality seemed to be persistent; for example, males often displayed incomplete mounting patterns and females were as abusive to their first-born infants as were isolation-reared females (Suomi & Harlow 1977a).

Miscellaneous Deprivation Experiments

Several deprivation experiments conducted at Wisconsin do not readily fit into the categories discussed above. These miscellaneous experiments are discussed in this section.

Sackett (1965a) presented a theory to account for the effects of early social deprivation on rhesus monkey behavior. According to this theory, normal behavioral development proceeds by gradual exposure to increasingly complex environmental stimulation. Animals are said to prefer stimulation that is a little more complex, but not much more, than what they are used to. Inappropriately paced exposure to environmental complexity is thought to result in some of the behavioral abnormalities of deprived monkeys.

Sackett conducted indirect tests of his theory (Sackett 1965b), using monkeys that had been subjected to varying degrees of isolation in previous studies. These tests offered admittedly limited support for the theory. The strongest conclusion consistent with the results is that animals reared under stimulating conditions prefer to interact with complex stimulation, whereas animals reared under impoverished conditions prefer simple stimulation. This conclusion received further support in follow-up testing (Sackett 1972a).

Kerr et al. (1969) conducted a statistical study of the physical growth of monkeys subjected to partial and total isolation in previous experiments. The isolates grew normally during and after isolation. The experimenters related their findings to a human syndrome known as “deprivation dwarfism” (see Discussion).

Lichstein and Sackett (1971) examined the reaction of isolation-reared monkeys to “noxious stimulation.” Isolation-reared monkeys and socially-reared monkeys were subjected to electric shock from the drinking spouts in their cages. The isolation-reared monkeys were found to tolerate higher levels of shock than the controls. The same monkeys were later given a choice between an electrified and non-electrified drinking tubes. Neither group preferred the electrified tube. This suggested that electric shock was a noxious stimulus for both groups.

Lichstein and Sackett concluded that rhesus monkeys reared in isolation react anomalously to noxious stimulation. The same phenomenon had already been shown in dogs and rats, but according to Lichstein and Sackett, the special feature of their study was that one explanation for the anomalous reaction, namely, that the stimulation was not perceived as noxious, was ruled out.

University of Pittsburgh School of Medicine

Researchers at the University of Pittsburgh conducted follow-up experiments on rhesus monkeys obtained from Wisconsin, where the subject had been reared in total isolation. The isolates were poor at sending social cues via facial expressions, and poor at utilizing cues sent by others (Miller et al. 1967). This was taken as support for Mason’s hypothesis that the social inadequacy of isolation-reared monkeys stems from lack of opportunities to acquire communication skills. This experiment subjected monkeys to electric shock and restraint in a primate chair for up to 72 hours.

Later experiments documented the eating and drinking habits of the isolates. The isolates consumed more food and fluids than socially-reared monkeys (Miller et al. 1969). This was attributed to a greater consumption of food and fluid during each meal, rather than, for example, more frequent meals (Miller et al. 1971). No reason was apparent for this phenomenon.

US. Public Health Service

Several studies of rhesus monkeys reared in partial isolation were conducted by the U.S. Public Health Service (PHS). Two results contrasted with the Wisconsin data. First, partial isolation had little or no effect on adult reproductive performance (Meier 1965a). Second, all ‘motherless mothers’ exhibited normal maternal behavior after normal delivery of their offspring (Meier 1965b). However, maternal behavior was inadequate following surgical delivery.

The contrast between the Wisconsin and PHS data was decreased when a later study (Missakian 1969) found that isolation did impair reproductive performance. Sexual behavior remained disoriented following social therapy, although some animals showed improvement in other aspects of social behavior (Missakian 1972). Therapy involved pairing isolates with younger individuals, as in the Wisconsin studies.

California’s Regional Primate Research Center at Davis

Mitchell and others at Davis conducted several deprivation studies on rhesus monkeys. Mitchell had conducted similar experiment as a student of Harlow’s.

Monkeys reared in isolation at Davis did not drink excessively, unlike those at Pittsburgh (Fittinghoff et al. 1971). One possible explanation, not mentioned by the researchers, is that polydipsia occurs in total isolates (Pittsburgh), but not partial isolates (Davis).

The Davis researchers made follow-up observations on a group of isolate monkeys obtained from Wisconsin (Fittinghoff et al. 1974). The 12 to 13 year-old partial isolates retained the abnormal personal and social behavior that characterized their younger years. The isolates’ abnormalities were viewed as normal responses to an aberrant environment “. . . the study provides few surprises by comparison with previous work . . .” (p. 136).

Baysinger et al. (1972) compared the behavioral development of mother-reared and isolation-reared infants during the rearing period. Both groups were housed in wire cages, but the isolates’ cages were covered. Unfortunately, the researchers removed the coverings during observations and made the dubious assumption that the isolates’ behavior was not affected.

The Davis researchers tried their hand at “social therapy” after the Wisconsin group successfully habilitated isolate-reared rhesus monkeys. Therapy involved pairing isolates with older males and females. “Although some of the isolate-reared animals appeared to have benefited from pairings with preadolescents [Brandt & Mitchell 1973], the isolates clearly gained little of lasting value” (Erwin et al. 1974, p. 814).

Capitanio (1984) recently conducted a deprivation experiment at Davis under the direction of Mason who — like Mitchell— had conducted deprivation experiments as a student of Harlow’s. Capitanio compared the social behavior of rhesus monkeys that had been reared by either a dog or a plastic hobbyhorse. The aim was to demonstrate that the animate, interactive characteristics of mothers enhances an infant’s social abilities. Dogs, but not hobbyhorses, have these characteristics. “Although members of both groups were clearly abnormal in the extent of their social interactions,” the dog-reared individuals were less aberrant than the hobbyhorse-reared individuals (p. 35).

Other deprivation experiments conducted at Davis examined the “looking behavior” and social organization of isolates (Mitchell 1972, Anderson & Mason 1974), and the affiliative behavior of an adult male isolate toward an infant (Gomber & Mitchell 1974).

Washington Regional Primate Research Center

Sackett, yet another of Harlow’s former students, continued his deprivation experiments at the Washington Primate Center after leaving Wisconsin. He began using pigtailed monkeys, a close relative of rhesus monkeys, expecting that pigtails and rhesus would react similarly to total isolation. Much to his surprise, Sackett found that “rhesus monkeys, as a species, appear to be more susceptible to isolation rearing effects than pigtails” (Sackett et al. 1976a, p. 128). Moreover, exposure to age-mates appeared to be an effective therapy for pigtails but not rhesus (Sackett et al. 1976b).

The results of a second study also emphasized species differences (Sackett et al. 1981). Rhesus and pigtailed monkeys, as well as the closely related crab-eating monkeys, were reared in total isolation. Effects on personal behavior, social behavior and exploration varied depending on species. The results cast doubt on the notion of an isolation syndrome and the generality of this syndrome.

Yerkes Regional Primate Center

Several deprivation experiments were conducted at the Yerkes Laboratories in Florida, and their successor, the Yerkes Regional Primate Center in Georgia. Only one Yerkes study involved rhesus monkeys. in that experiment, isolation, apparently total, “did not grossly affect basic learning abilities” (Mason & Fitz-Gerald 1962). Other studies all involved chimpanzees.

Davenport, Menzel, and Rogers conducted a series of deprivation experiments designed to address human clinical problems. For example, Davenport et al. (1 961) considered isolation-reared chimpanzees to be models of inadequate maternal care in humans. In humans, such care had been associated with poor growth and survival. The chimps were reared in bare wire cages or small isolation cubicles, and their growth and survival were compared to those of mother-reared chimpanzees. Rearing in wire cages was said to mimic rearing in impersonal human institutions. No explicit analogy was made for rearing in isolation chambers. Unfortunately, food intake—an important determinant of growth—was not controlled from group to group. The chimp data were contrary to the human data. Deprived chimps grew faster and survived better than nondeprived chimps. Harlow’s group reached the same conclusion in a later study (Kerr et al. 1969) and did not even mention the Yerkes study. These studies are discussed further in the Discussion section.

The Yerkes group also reared chimpanzees in different degrees of restriction. The most restricted infants were reared in bare “cribs” whose walls precluded visual access to the outside world. Other restricted groups were given more opportunities for visual, social, or manipulatory experience. Infants from these restricted groups were compared to infants that had been born in the wild.

The restricted infants were abnormal in several respects. First, they developed unusual stereotyped behavior and postures that persisted for years (Davenport & Menzel 1963). Second, they were unresponsive or fearful of objects (Menzel et al. 1963a) or unfamiliar environments (Menzel et al. 1963b). Third, they were slow learners, but improved with training (Davenport & Rogers 1968; Davenport et al. 1969; Rogers & Davenport 1971; Davenport et al. 1973). Their poor performance stemmed from inflexibility and “distractibility.” These results are consistent with human studies but inconsistent with rhesus monkey studies (see above). Fourth, sexual performance at adulthood was impaired initially, but recovered when isolates were aired with normal age-mates. Again, the chimp data do not agree with the rhesus data.

Some of this research was repeated using older chimpanzees. One to two-year-old infants were imported from Africa and subjected to six months of deprivation in isolation cribs (Davenport et al. 1966). Unlike chimps subjected to such procedures at birth, they did not develop stereotyped behavior. The researchers concluded that some as-yet-unknown forms of maternal stimulation, if provided early, can prevent stereotypy in chimpanzees later subjected to isolation.

Another deprivation experiment by the Yerkes researchers was a failed attempt to habilitate the isolation-reared chimpanzees (Turner et al. 1969). The subjects were adolescent chimps who avoided social contact and displayed abnormal behavior. These aberrations persisted despite various manipulations, including social therapy (pairing isolate males with same-aged females or older females), and drug therapy (including LSD).

Social therapy apparently was successful in another experiment (Rogers & Davenport 1969). Isolates were grouped with normal, same-aged males and females and showed marked improvement in sexual performance. The researchers claimed that recovery extended to social behavior in general, but unfortunately this was not documented.

The same chimpanzees reared in isolation by Davenport et al. were also studied by Berkson and Mason, who focused on the isolates’ abnormal stereotyped behavior. Stereotyped behavior was prominent in situations that induced excitement or “arousal”, as expected, and such behavior was reduced in situations that elicited alternative behavior (Berkson et al. 1963; Berkson & Mason 1964). One of these studies involved food deprivation and administration of amphetamines.

Another group of Yerkes chimpanzees was reared in partial isolation for two to four years and then tested in mildly stressful situations. Isolates gave fewer “distress vocalizations” (whimpers and screams) than wild-born chimps (Randolph & Mason 1969).

Delta Regional Primate Research Center, Louisiana

Some of the deprivation experiments conducted at Delta concerned self-recognition.The capacity for self-recognition was judged from the subjects’ responses to their reflection in a mirror. Unlike chimpanzees reared in the wild, those reared in partial isolation showed no signs of self-recognition, except when given remedial social experience (Gallup et al. 1971). The researchers argued that the self concept is developed through personal interactions, which are precluded by isolation. See Gallup and McClure (1971) for a related study on rhesus monkeys.

Another set of experiments compared the development of rhesus infants reared with surrogate mothers that were either mobile or stationary. Only the “stationary infants” developed stereo-typed body rocking (Mason & Berkson 1975). This was regarded as experimental support for human studies that suggested an association between maternal stimulation and body rocking. The “mobile infants” were less aroused by novel situations (Mason & Berkson 1975, Eastman & Mason 1975) and were more socially responsive (Anderson et al. 1977) than the “stationary infants.”

Miscellaneous Institutions

Researchers at numerous institutions conducted one or two deprivation experiments on nonhuman primates. For convenience, these experiments are grouped here, rather than under separate institutional headings.

Green examined some effects of partial isolation on rhesus monkeys. Compared to controls, the isolates were less exploratory and more fearful and hyper-reactive to environmental changes (Green & Gordon 1964). This interfered with their responses on learning tasks (Green 1965). Similar results were found in the Wisconsin and Yerkes studies.

Angermeier et al. (1967) reared rhesus monkeys in various degrees of deprivation and paired them with socially-reared monkeys in “dominance fights.” Isolates were not subordinate to the controls, unlike the situation in rats. The researchers concluded that studies of social behavior may yield different answers depending on the species being investigated.

Several experiments conducted at miscellaneous institutions involved species other than the popular rhesus monkey. Studies of partial isolation were extended to pigtailed monkeys (Evans 1967), cebus monkeys (Elias & Samonds 1973), and bonnet monkeys (Rosenblum & Smiley 1984). All three species resembled rhesus monkeys in their response to isolation. Some of the isolation-reared bonnet monkeys recovered substantially when they were required to gather their own food from a foraging device. The success of this therapy depended on the monkeys’ status in the dominance hierarchy. Kawabe (1969) reared Japanese monkeys in isolation, but he did not specify whether isolation was partial or total. In any event, his analysis was restricted to vocalizations.

Coelho and Bramblett (1984) reared yellow baboons in partial isolation except that each was allowed to interact with peers for a few hours per day. This arrangement was said to be subtly different from rearing with mothers and peers. This amazing assertion may have some validity, however, given that few differences in social behavior were found between the two groups.

Berkson continued his experiments on abnormal stereotyped behavior when he moved from Yerkes to the Illinois State Pediatric Institution (see Berkson 1967). Crab-eating macaques, but not marmosets, developed abnormal stereotyped acts when reared in partial isolation (Berkson et al. 1966; Berkson 1968). The macaques isolated earliest in infancy developed the most stereotypies.

Kaplan deprived squirrel monkey infants of their mothers and reared them with inanimate surrogate mothers. The infants formed an attachment bond to the surrogates and used them as a security base, much like surrogate- reared rhesus monkeys. However, unlike rhesus, the squirrel monkey infants did not exhibit stereotyped body rocking (Kaplan 1974).

Huebner and colleagues also reared squirrel monkeys with inanimate surrogate mothers. Surrogate-reared infants developed behavioral abnormalities, but recovery was substantial when infants were allowed to interact with a kitten (Huebner & King 1984). The kittens were thought to be important in providing a source of stimulation contingent upon the monkey’s behavior. “The importance of stimulation contingent upon an infant’s own behavior has been emphasized in human child development” (p. 241). The researcher noted that, for squirrel monkeys, this stimulation did not have to come from another member of the same species.

Several physiological experiments have been conducted on primates reared in isolation. In one experiment, Reite and Short (1977) examined the sleep physiology of pigtailed monkeys reared in total isolation. No differences from controls were found. Two other physiological experiments were based on the clinical observation that a biochemical abnormality in human schizophrenics was related to a monotonous rearing environment. Beckett considered the Wisconsin rhesus monkeys and the Yerkes chimpanzees to be animal models of humans reared in monotonous environments. Biochemical data from both rhesus (Beckett et al. 1963a) and chimpanzees (Beckett et al. 1963b) provided support for the link between the biochemical abnormality and a monotonous upbringing.

Another physiological experiment (Beasley & Seal 1968) was also related to schizophrenia. Partial isolation in rhesus monkeys was associated with alterations in the levels of several blood constituents. However, these alterations did not resemble those found in human schizophrenics. Indeed, many of the alterations were the opposite of those found in humans.


The Nature and Extent of Suffering

Clearly, deprivation researchers have inflicted a great deal of damage on their animal subjects. Some animals were killed by the deprivation experience. Virtually all of the survivors suffered long-term psychological disturbance. For example, monkeys reared in isolation were social misfits, fearful and withdrawn, encapsulated in a world of abnormal stereotyped behavior and self-abuse.

The deprivation experience itself was not the only source of damage to the animals. A few of them were killed after their isolation. Many isolates were subjected to stressful procedures, such as electric shock and the “rape rack.” Because the isolates were so disturbed, virtually any procedure was distressing. Hence, the isolates were extremely fearful when exposed to novel environments, or were helpless when paired with aggressive individuals.

The damage inflicted by some experiments extended to the offspring of the participants. Because the maternal behavior of female isolates was grossly abnormal, many infants were neglected or abused; some died as a result.

These consequences of deprivation experiments were endured by over 1,300 animals. This figure includes 1,110 primates and 202 dogs. These figures are minimum estimates; many more animals actually were used. Unfortunately, numerous reports failed to mention the number of animals used in particular experiments.

Financial Cost

The estimated total cost of deprivation experiments was $20.7 million (Table 1). Most of this money was provided by the National Institutes of Health and the National Institute of Mental Health. Both organizations are federal institutions; hence, U.S. taxpayers ultimately paid for much of this research.


Benefits to Humans

Only one deprivation experiment has had a tangible impact on clinical practice. This was Suomi and Harlow’s (1972a) habilitation of rhesus monkey infants that had been reared in isolation chambers. Recall that these infants were habilitated during interactions with younger, more normal ‘therapists.’ This technique “has been applied successfully to treatment of socially …withdrawn children as well as those at high risk for schizophrenia” (Suomi 1982, p. 268).

The irony in this application of animal data is that the therapeutic technique could have been safely tested on humans, without using animals first. No potentially harmful drugs or surgical procedures were involved. Furthermore, the principles behind the success of the monkey habilitation were already familiar to psychiatrists. In discussing the habilitation, Suomi et al. (1972) observed.

The role of gentle . . . contact and model-serving exhibited by the therapists was likely of …importance for . . . recovery. Many of these principles are not unfamiliar to human psychotherapists (p. 931).

Clinicians had simply overlooked the potential therapeutic value of pairing certain kinds of disturbed children with younger, normal children. Suomi and Harlow’s contribution was to suggest this possibility.

Let us consider one of the two human studies that Suomi cited as having used his habilitation technique. This was a study of socially withdrawn preschoolers at a day care center (Furman et al. 1979). These children were labeled “social isolates” by Furman et al. and “socially retarded” by Suomi. Both labels are suggestive of a serious behavioral disorder; however, according to Furman et al., these children

were simply in the low end of peer interactions, compared to their classmates . . . none was described by the teachers as mentally retarded or emotionally disturbed (p. 917).

Hence, shyness in preschoolers was compared to the profound disturbance in rhesus monkeys following isolation-rearing.

These shy preschoolers became more sociable during interactions with younger individuals. The reason for this change was that the “isolates” became leaders in interactions with the younger individuals. The reverse was true in the monkey case. Even Furman et al. conceded that . . . the analogue between this study and the animal model . . . can easily be overstated . .” (p. 922). Accordingly, it can be argued that Suomi’s social therapy technique did not generalize to this human situation, and that the success of human study was serendipitous. Moreover, numerous techniques for overcoming shyness in human children already exist.

Suomi claimed that his habilitation study has had general theoretical impact in addition to alleged practical impact (Suomi & Harlow 1978, p. 268; Suomi 1984a). Specifically, the study was said to have had

considerable impact on theoretical considerations of developmental plasticity and the potential reversibility of early cognitive and social deficits at the human level . . . (Suomi 1984a, p. 232).

This assertion was unsupported. No mention of such an impact was made in numerous recent books on human development (see Reference Note, p. 40).

Aside from Suomi’s habilitation experiment, other clinically-oriented experiments reviewed have had little or no clinical impact. At best, some have added support to pre-existing human data or have suggested that more attention be paid to certain social factors, such as peer interactions.

Mental Illness in Humans

The limited impact of isolation experiments on the diagnosis, treatment, prevention, and understanding of human mental illness stems partly from the vagueness of the isolation model. The researchers usually failed to specify what human syndrome they were modeling. They talked vaguely of “mental disturbance” (Harlow & Harlow 1962a), “mental illness” (Mitchell 1974), “abnormal behavior” (Sackett 1968a), and “psychosis” (McKinney 1974a). Each of these terms refers to a heterogenous assortment of human clinical syndromes. Accordingly, no single animal model for any of these assortments is likely to go far. With respect to mental illness, the post-isolation syndrome remains an animal model in search of a human counterpart.

Even Mason, himself a Wisconsin researcher, believed that the use of animal models of human behavior works best for well-circumscribed processes, not for example, “the sources of neurotic behavior” (Mason 1968, p. 71). Sometimes the isolation researchers did draw parallels to specific human problems, such as autism (e.g. Fox 1966), depression (e.g. Griffin & Harlow 1966), and schizophrenia (Harlow et al. 1964; Sackett 1972b). Some analogies were more convincing than others, but no researcher has yet demonstrated that the post-isolation syndrome is a good animal model (both valid and practical) of any human mental illness. Such a demonstration is unlikely given that the response to isolation varies both within and across animal species (e.g. Fuller & Clark 1968; Meier 1970; Sackett et al. 1976).

Maternal Deprivation in Humans

Recall that isolation experiments were viewed as animal models of maternal deprivation as well as mental illness. Both the Wisconsin and Yerkes researchers explicitly related their isolation experiments to the clinical literature on maternal deprivation.

The term ‘maternal deprivation’ was used rather loosely in the human literature (see Ainsworth 1962). It referred primarily to an insufficient level of maternal care. Such insufficient care prevailed in non-nurturing orphanages, foster homes, and families. Numerous clinicians reported that children reared in impersonal institutions and other depriving environments are prone to develop abnormally (Bowlby 1952). Some of these children, but by no means all, develop emotional, intellectual, social, or physical problems.

Some of the early human studies were criticized for methodological or conceptual inadequacies (e.g. Yarrow 1961, Andry 1962). Some studies were criticized for ignoring non-maternal variables. For example, deprivation of sensory stimulation as distinct from maternal care —appeared to be a contributing factor in the deprivation syndrome (Casler 1961). This was especially so for infants less than six months old, which is the period before an infant becomes emotionally bonded to its mother. However, the significance of maternal deprivation, especially for infants over six months of age, was apparent from the human studies (Ainsworth 1962, Lebovici 1962, Thompson 1962, Bronfenbrenner 1968).

The isolation experiments that came closest to mimicking conditions in impersonal institutions for children were (1) the Wisconsin experiments with rhesus monkey infants housed alone in wire cages (‘partial isolation’) and (2) the Yerkes experiments with chimpanzee infants similarly housed. The Wisconsin cage-reared monkeys, insofar as they resembled children being reared in impersonalized institutions, provided support for the negative effects of maternal deprivation in humans (Bowlby 1976, Mussen et al. 1984). The same can be said for the Yerkes chimpanzees.

Harlow attempted to relate his total isolation experiments to human maternal deprivation (Harlow & Harlow 1962a; Harlow et al. 1965), but his attempt was far from convincing. A panel of the National Academy of Sciences concluded that total isolation comes closest to modeling cases of drastic neglect in human infants that probably would lead to death (Hamburg 1970). Monkeys reared in isolation chambers are deprived of all social experience, maternal or otherwise. Moreover, total isolates are deprived of normal sensory experiences. Hence, isolation effects could stem from lack of sensory experience, as well as maternal deprivation per se. Yarrow (1961, p. 480) noted that even “in the most extreme institutional environments the degree of sensory deprivation is less severe than in animal studies.”

It is ironic that Harlow did not control for sensory experience in his total isolation experiments. Diminished sensory experience had been a criticism of some of the early human studies (Caster 1961). Harlow was fond of criticizing clinical studies for their lack of experimental rigor, and he touted his own ostensibly well-controlled experiments. Harlow et al. (1964) dismissed sensory experience as a confounding factor by claiming that “. . the monkeys [inside the chambers] had constant stimulation, . . . including contact stimulation from manipulation of their [cage’s] iron bars . .” (p. 120). This meager level of stimulation clearly did not approach the level that monkeys usually experience in the wild or even in the laboratory.

These were not the only problems with the total isolation model of maternal deprivation. Sackett’s group found that the effects of total isolation vary across monkey species:

The results . . . suggest a word of caution concerning generalization of behavioral effects of isolation rearing across species . . The generality of rearing condition effects, even across very closely related non-human primate species [cannot be] simply assumed (Sackett et al. 1976a, p. 129). 

These results cast considerable doubt on Harlow’s belief

…that the same basic laws operate for [human and rhesus infants] and that social isolations which produce abnormality in one species will have comparable effects on the other (Harlow et al. 1965, p.90).

A decade after Harlow’s initial experiments on total isolation, he reassessed the contribution of these studies to the understanding of human maternal deprivation (Harlow & Novak 1973). His extremely weak assessment was couched in terms of his trumpeted experimental control.

The information we have obtained . . . differ[s] from human data only in the fact that the precise controls which we could exercise over our monkey subjects make the differential concepts [of privation and deprivation] plausible and precise (p. 467).


. . . our data give clarity and classification to the earlier clinical efforts (p. 477).

A decade before Harlow’s isolation experiments, Ainsworth and Bowlby (1954, cited by Lebovici 1962) distinguished between deprivation of the mother from birth (privation’) and deprivation after an infant developed an emotional bond to its mother (‘deprivation’). The differential effects of privation and deprivation were so striking in the human studies that no animal studies were necessary to add plausibility or precision to the human data. Moreover, Harlow’s experiments actually confounded privation and deprivation. The monkeys subjected to privation were all incarcerated in isolation chambers at birth, whereas the monkeys purportedly subjected to deprivation were not incarcerated until six months of age. However, even the latter monkeys were subjected to privation because they spent their first six months without their mothers. The inescapable conclusion is that Harlow’s total isolation experiments added nothing to our under-standing of maternal deprivation in humans.

It is puzzling that Wisconsin researchers came to view their isolation experiments (partial or total) as models of human maternal deprivation. The earliest isolation studies at Wisconsin were said to be investigations of what was already known for humans: “deprivation of normal social experiences in human children results in a wide range of personal and social deficiencies and aberrations . .” (Mason 1960, p. 582). The rationale for Mason’s experiments was that no comparable data existed for non-human primates. Sackett (1967b) also acknowledged prior human studies. After summarizing the effects of various deprivation experiences on rhesus monkeys, he wrote:

a wealth of clinical data and data on institutionalized children suggests that many of the conditions discussed in this paper have similar, if not identical, effects in human children (p. 75). 

One of few isolation experiments to go beyond existing human data suggested that isolation-rearing does not impair physical growth or survival, at least in chimpanzees (Davenport et al. 1961). Later studies on humans (Whitten et al. I 968) and rhesus monkeys (Kerr et al. 1969) yielded similar results. These studies suggested that “deprivation dwarfism” in institutionalized children results from malnutrition, not social isolation. However, the conclusions of both animal studies were tempered by methodological flaws.

The biggest breakthrough in maternal deprivation research apparently was the realization that impersonal institutions and other depriving environments can impair child development. Bowlby (1952) called attention to this problem back in 1952, when he reviewed the existing human data. His monograph was largely responsible for a marked improvement in institutional care for children (World Health Organization 1962, p. 7-8) The Wisconsin and Yerkes data came after this.

To summarize: Deprivation experiments made little contribution as animal models of maternal deprivation. Many of these experiments were little more than sensational illustrations of what had already been established in humans.

Development in Human Infants

The Wisconsin researchers sought to justify their deprivation experiments not only as models of mental illness and maternal deprivation, but also as models of infant development. Harlow and Harlow (1962a), for example, aimed to help “resolve the baffling complex roles of various kinds of early experience in the development of human personality” (p. 146).

The Wisconsin researchers apparently were not successful. In 1976, Suomi wrote:

. . . whether actual data obtained from nonhuman primates have added measurably to our understand-ing of human development is another matter The record to date has not been particularly impressive. To be sure, there are a few findings derived from nonhuman primates that have clearly advanced knowledge of human development [Cites affection experiments]. Overall, however, such cases are relatively rare. Most monkey data that readily generalize to humans have not uncovered new facts about human behavior; rather, they have only verified principles that have already been formulated from previous human data.

A case in point is in the area of the development of mother-infant relationships. To date the monkey data have added little to knowledge of human mother-infant interactions (Suomi 1976, p. 203). 

A year later, Suomi (1977b) implied that Wisconsin research has influenced research in child development.

The importance of monkey peer relationships for social development has been stressed for years (e.g. Harlow 1969); only after many years of disinterest have human development researches [sic] directed more than passing attention toward the development of infant and childhood friendships . . . (p. 222).

The importance of peer relationships among monkeys was emphasized in a viewpoint that Suomi (1979) offered to students of human infant development. Suomi provided several examples of how this viewpoint — derived from his rhesus monkey experiments — might be used to plan social policy:

[This] viewpoint would emphasize greater focus on the study of peer interaction patterns in children and it would encourage a policy of providing children with ample opportunities to develop mutual relationships in stable and non-threatening environments. It would argue that peer relationships in younger children are unlikely to thrive if problems with parents (and perhaps teachers and other adults) remain substantial, leading to prevention programs that focus on individuals’ entire social environment, not simply their relationships with peers or with parents. Finally, such a viewpoint would argue for careful monitoring of children’s relationships with peers as a barometer of their current level of social competence and as a possible predictor of subsequent psychopathology (p. 146).

These policies may well be beneficial. However, to base such policies on Wisconsin’s monkeys seems unwise, if not absurd. All of these monkeys are admittedly abnormal (Harlow, in Cullen 1974, p. 422; Suomi 1977b, p. 212) and living in highly artificial conditions.

There is a more specific reason to be skeptical. One of the most important features of any model of human development is the nature of mother-infant interactions. Here the rhesus monkey model falls short. According to the Wisconsin researchers, play is one of an infant’s most important activities; yet these researchers acknowledge that “rhesus mothers engage in little interactive play with their infants, . . . unlike human mothers” (Harlow & Harlow 1969, p. 21, emphasis added). Given this model, perhaps it is not surprising that the Wisconsin researchers have emphasized the role of infant-infant interactions over that of mother-infant interactions. Consider Suomi and Harlow’s (1975a) conclusion:

. . . the primary vehicle for shaping genetically acquired potential into competent and adaptive social activity lies in the development and maintenance of peer friendships [among infants] . . . we suspect it is also . . . true for . . . man” (p. 176).

While this suspicion may be true, differences between the model and humans do not inspire confidence.

The Wisconsin researchers have tried to get students of normal social development in humans to consider the implications of their rhesus monkey data (Mason 1968, Suomi 1979). For example, Suomi and Harlow (1978) offered two implications: (1) a child’s environment can have significant effects on behavioral development and (2) these environmental influences are particularly important at the beginning of various maturational stages. Suomi and Harlow proclaimed these implications as if they were novel, but of course they were not. Indeed, the first (the importance of a child’s environment) is regarded as a truism. Beach and Jaynes (1954) reviewed numerous experiments documenting the importance of early experience in animals. All of these were conducted well before Harlow’s research. Back in 1952, Levy wrote that the importance of early experience in humans was recognized by “psychiatrists generally, child psychologists, educators, and a variety of investigators in other disciplines” (p. 489). Levy added that the importance of early experience is “easily verified by the study of case records in any number of schools and clinics” (p. 489). In his view, experiments to “document” this are attempts “to verify the obvious” (p. 489). Hence, the Wisconsin researchers weren’t the first to document the importance of early experience; their contribution was to develop a new model for research on this topic (Hamburg 1970).

Suomi and Harlow’s second implication is that environmental influences become particularly important at the beginning of certain maturational stages. This had already been shown, for example, by early human studies of maternal deprivation (reviewed in Ainsworth 1962). One such study (Ainsworth & Bowlby 1954, cited in Lebovici 1962) distinguished between deprivations occurring before and after the infant forms an emotional attachment to the mother. This occurs at around 6 months of age, and at this maturational stage the consequences of deprivation seem most severe.

While primate deprivation models have yielded numerous implications about human development, these implications are largely ignored by clinicians, who rely primarily on human studies. A few implications from deprivation experiments have filtered into clinical thinking, but these do not appear to have influenced clinical practice. Only one recent book out of several that were consulted even mentioned an assessment of the clinical impact of the Wisconsin models of normal infant development (Reference Note, p. 40). Referring to experiments by Harlow and others, Kolb and Brodie (1982) wrote:

From animal studies it has been inferred that the establishment of the . . . relationship [of an infant to its mother] comes about through the presence of the mother, whether her behavior may be judged as gratifying or thwarting, rewarding or punishing (p. 69).

This statement provides little evidence that Harlow’s models of infant development made significant contributions to the understanding of human development.

Child Abuse in Humans

Recall that in the Wisconsin experiments, female rhesus monkeys who were reared without mothers often became inadequate mothers themselves. The poor maternal behavior of motherless mothers has been mentioned in the psychiatric literature by Kolb and Brodie (1982, p. 69) and White (1974). White went so far as to say that the “demonstration of the intergenerational effects of deprivation [has] had a significant influence on psychiatric thinking . .” (p. 18). It is difficult to imagine the basis for this glowing assessment. Would White have been surprised if human females who had been reared in isolation became poor mothers?

Suomi has related the Wisconsin deprivation experiments to human child abuse (e.g. Suomi 1977a). His assessment has been cautious. For example, he evaluated the primate model according to McKinney and Bunney’s criteria for a valid animal model (see Ch. 2) and judged the model to be weak:

The existing data on monkey neglect and abuse of offspring scarcely begin to satisfy all of these criteria (Suomi 1978, p. 32).

Hence, there is little basis for expecting that preventative or therapeutic measures that work for rhesus monkeys will work for humans.

Suomi recently summarized Wisconsin data on infant abuse and drew several conclusions (Suomi & Ripp 1983). First, several factors contribute to the problem of infant abuse. These include mothers’ social history and infants’ demandingness. Second, mothers who were the most adversely affected by particular social experiences are likely to be the most abusive. Third, therapy/prevention can be helpful by reducing exposure to risk factors and giving special attention to high-risk females. None of these conclusions is likely to come as a surprise to students of human child abuse.

Horenstein (1977) was more sanguine than Suomi regarding the potential of Wisconsin experiments to contribute to clinical aspects of child abuse. Horenstein faulted fellow clinicians for ignoring the Wisconsin data. He and others distinguished “motherliness” (physical contact, warmth, emotional involvement, etc.) and practical mothering (feeding, cleaning, etc.). He claimed that monkey infants who were denied motherliness, but given practical mothering, developed deficits in their own motherliness. This is incorrect; these monkeys were denied practical mothering for all but the first few weeks of their lives. Throwing monkey chow into a cage hardly counts as practical mothering (breastfeeding, grooming, etc.). Moreover, a link between deprivation of motherliness as a child and deficits in motherliness as a mother has been shown by human studies, as Horenstein notes.

Horenstein thought that early Wisconsin experiments suggested two treatment programs: supervised mothering for abusive mothers, and peer interactions for abused children. He conceded that both types of treatment were developed independently of Harlow’s work, but he pointed out that some clinicians do not treat both infants and mothers in their programs.

Horenstein also claims that theoreticians, unlike clinicians, regarded Harlow’s work as important to understanding child abuse. He cited Steele and Pollack (1968) and Fontana (1973) to support this claim. However, Horenstein exaggerated these theoreticians’ assessments of Harlow’s work. Consider Steele and Pollack’s assessment. After quoting Harlow, these authors write:

These observations do suggest that there may be some connection between the lack of early “being mothered” experience and deficits in later parental functions (p. 117).

That is the only statement regarding the positive contribution of Harlow’s work. The authors had previously noted that human studies (not to mention common sense) revealed the same connection between early experience and later maternal behavior. The authors then go on to point out differences between Harlow’s monkeys and abusive parents. They conclude by suggesting that Harlow’s monkeys are more like infants reared in depriving institutions, rather than like mothers prone to abuse. Hence, Steel and Pollack’s paper provides little support for the claim that Harlow’s work made important contributions to the understanding of child abuse.

The same conclusion holds for Fontana’s book. Fontana referred to the monkeys reared on cloth surrogate mothers and those reared on wire surrogate mothers (see Ch. 3 above). The former were said to develop more normally than the latter, who, denied any semblance of maternal affection, sometimes developed into abusive mothers. Fontana then added “Similarly, the indifference of a human parent can be shattering to a child . .” (p. 57, emphasis added). In other words, what we already know about abusive human mothers also appears to be the case in rhesus monkeys.


What is the extent of unnecessary duplication of experiments in deprivation studies? A reasonable assessment of unnecessary duplication is possible by examining the goals, results, and number of studies in this field.

Consider the premises upon which the following assessment is based. First, most of the deprivation experiments that we have considered were directed at human problems. These experiments were not designed, or funded, to amass knowledge for its own sake. For example, Harlow designed his research to address human problems (Harlow, in Tavris 1973). Most of the money for deprivation research was supplied by the National Institutes of Health, whose “mission . . . is to protect and improve the health of the American people” (National Institutes of Health 1984, p. iii).

A second premise is that deprivation experiments that were not justified on the basis of human benefits should not have been conducted, given their damaging effects on animals. A third point concerns the argument that no animal experiment, however poorly conceived, is unnecessary because it can serendipitously yield important results. Research on animal subjects should follow the same ethical principle that applies to research on human subjects; namely, an experiment is either justified or unjustified at the outset, regardless of outcome (Levine 1985). Self-serving appeals to the vagaries of luck are not excuses for bad research.

Deprivation studies had several goals; most of these studies were attempts to document the effects of isolation, the persistence of these effects, and their experimental reversal. In many cases, all three of these aspects of isolation could have been documented using the same subjects.

Over 175 reports of deprivation experiments have been published through 1984. To be sure, each of these reports does not represent a separate study; 36 contain reviews of experiments published in other reports. This leaves 146 studies, nearly all of which involved primates (Table II).

The obvious question is why so many studies were conducted. The 146 studies do not comprise the same type of study conducted on species after species; most studies involved rhesus monkeys. Rather, they comprise study after study of different aspects of the deprivation experience. Various kinds, durations and consequences of deprivation were examined. For example, Harlow and colleagues subjected monkeys to total isolation 14 times and devoted part or all of 21 follow-up studies to total isolates. The differences among several of the original 14 studies involved the timing and duration of isolation. The relevance of any of these studies to humans is suspect, given that human infants probably would perish if subjected to total isolation (Ham-burg 1970). The total isolation studies appear to provide an example of what Rowan (1984) labels “parametric tinkering.”

Research Quality

Several aspects of deprivation research are open to question on scientific grounds. As we have seen, the experimental design of many of these studies was seriously flawed. Harlow himself criticized numerous deprivation experiments for faulty design. His target was research on the association between deprivation and learning ability. He all but dismissed the studies that did not involve primates. For example, he wrote:

The subprimate literature possesses a wealth of massive . . . and often meaningless researches .. . The primary contribution of these earlier researches is the illustration of errors in experimental design and procedure (Harlow et al. 1971a, p. 126).

Harlow et al. thought these errors led researchers to the wrong conclusion (p. 147).

A second scientific aspect of deprivation experiments that is open to criticism is the choice of research questions. At the time of the primate experiments, the adverse consequences of rearing human infants in relative isolation, without mother love, were well established. The more controversial issue of the day was what aspects of maternal care facilitated infant development at various stages. Specifically, was the general sensory stimulation provided by mothers more important than emotional aspects of mother love? Rather than concentrate on this question (or similar issues), the animal modelers played it safe by “experimentally demonstrating” the ill effects of maternal deprivation.

A third criticism of deprivation research is that dogs and other nonprimates are not good choices for animal models. The dissimilarity to humans is too great. No evidence has been found that any of the experiments on nonprimates has influenced clinical thinking or practice. Although nonhuman primates were better models, the excessive reliance on one species (rhesus monkeys) was poor research strategy. The relatively few studies of monkeys other than rhesus monkeys demonstrated substantial variability in response to deprivation from species to species. This variability cast doubt on the generality of the results from these studies.


The issue of alternatives is almost moot with respect to deprivation experiments. It does not apply to the bulk of deprivation experiments because of the following ironies. First, animal data on the effects of maternal deprivation came after the human data (see above). Second, insofar as the animal data related to mental illness or child abuse, it was of dubious value (see section on clinical impact). Even some of the modelers themselves conceded that the models violated criteria of validity. Third, some of the animal data, though possibly interesting, made little difference to human affairs.

Only one deprivation experiment had a direct clinical impact, and as already mentioned, the procedure derived from that experiment could easily have been derived from human studies.

One of the themes of the Wisconsin deprivation experiments is the importance of peer interaction for infant development. Here again, animals did not have to be deprived of their mothers to determine this. The important comparison was not monkey infants reared only with peers versus only with mothers. (No one would suggest that human infants be reared by their peers instead of their mothers, regardless of the outcome of the monkey experiment.) The relevant comparison was mother-peer rearing versus mother-only rearing. In this perspective, alternative studies with humans are readily apparent and undoubtedly have already been conducted. Researchers can compare the development of mother-reared infants who have various degrees of contact with peers. Infants could be given enhanced levels of peer contact, or researchers could take advantage of naturally occurring variability in peer contact, such as differential enrollment in day care programs.

Animal studies not only were unnecessary, in light of the accomplishments and potential of human studies, they also were excessively harsh. This translated into more damage to animals and less relevance to humans. Mason recognized this when he observed that deprivation studies can be “likened to the blow of a sledge hammer, when a few gentle taps might have been more instructive” (Mason 1968, p. 79). For example, total isolation studies were more damaging and, as has been argued above, less relevant to humans than were partial isolation studies. Similarly, animals did not have to spend their entire lives in isolation in order to demonstrate the importance of social experience in early life.


Experiments in the deprivation category involved rearing infant animals in various degrees of social isolation. The aim of the experiments was to determine the effects of this isolation on development.

Over 1,300 animals, mostly primates, were subjected to these experiments. Most suffered long-term psychological damage, and some were physically harmed or even killed. These experiments cost over $20 million. Most of this funding was supplied by federal institutions.

Although nearly 150 deprivation experiments were conducted, only one had a direct impact on clinical practice. The procedure involved in this study could have been easily derived from human studies, without prior testing in animals.

This field of research contains many experiments that were either poorly designed or unnecessarily duplicative of prior research, including human studies. Moreover, alternatives to many of these animal experiments were either already in use or could have been devised.

Reference Note

These books included Handbook for the Practice of Pediatric Psychology (Tuma 1982), Child Development and Personality (Mussen et al. 1984), The Child: Development from Birth Through Adolescence (Harris 1984), Social and Personality Development: Infancy Through Adolescence (Damon 1983), Child Development (Abron 1984), Modern Clinical Psychiatry (Kolb & Brodie 1982), and Handbook of Research in Early Childhood Education (Spodek 1982).


Please cite as:

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