Chapter Three


Maternal deprivation experiments in the affection category were attempts to determine which maternal characteristics elicit an infant’s emotional attachment to its mother. All but two of these experiments were conducted on rhesus monkeys at the University of Wisconsin and Wisconsin Primate Center. Harlow directed this research, labeling the topic “the nature of love.” His experiments involved depriving rhesus monkey infants of their mothers at birth and rearing them with inanimate surrogate mothers.

Overview of Affection Experiments

Harlow began his affection research after depriving newborn monkeys of their mothers and noting that they became psychologically attached to cloth diapers placed in their cages. The infants’ attachment to cloth diapers gave Harlow the idea of using cloth surrogate mothers to demonstrate which maternal attributes elicited filial affection.

Harlow’s initial study (Harlow 1958, 1959a, 1962a, 1962b, Harlow & Zimmerman 1959) attempted to determine which of two factors was more important in eliciting filial affection. One factor was the physical contact associated with clinging to the mother. Harlow referred to this as “contact comfort.” The other factor was the breast and activities associated with nursing.

Harlow’s observations of infants becoming attached to cloth objects led him to suspect that contact comfort was an important factor in filial bonding. He pitted contact comfort against nursing as a challenge to prevailing views on the importance of nursing in filial attachment. For example, the prevailing view of psychologists was that human infants learned to love their mothers by associating the mother with the satisfaction gained from suckling. In the terminology of learning theory, mother’s milk was the primary reinforcer that satisfied the infant’s hunger, and mothers themselves were the secondary reinforcer. Accordingly, this view of filial bonding was known as the Reinforcement Theory. It was also known as the Drive Reduction Theory, a reference to the infant’s hunger as a biological drive.

In Harlow’s initial study he deprived rhesus monkey infants of their mothers and reared them with both a cloth surrogate mother and a wire surrogate mother (Photo 1). A nursing bottle was attached to one surrogate or the other, but not both. The comparison of greatest interest was the cloth surrogate without the bottle versus the wire surrogate with the bottle. This pitted contact comfort against nursing. A control group, referred to as “orphans” (Harlow 1959a), was denied physical contract with any surrogate mothers or other monkeys for about eight months from birth.

Several lines of evidence suggested that contact comfort was more important than nursing in eliciting filial attachment. For example, infants spent more time with the cloth surrogates (CS) than with the wire surrogates (WS). Other measures of the infants’ greater attachment to CSs clearly involved distressing procedures. In the “home cage fear test,” an opaque screen in the infants’ cage was suddenly lifted, exhibiting a “mechanical monster” (Harlow & Harlow 1965). The monster moved towards the infant while making loud noises. This produced “abject terror” in infants, who frantically ran to their CSs (Harlow & Zimmerman 1959).


Photo 1
Rhesus monkey infant housed with surrogate mothers made of wire and cloth. From Harlow (1966) with permission of Harper & Row.

Another distressing test involved placing infants in a large, unfamiliar room containing several objects. In the presence of their CSs, infants would rush to the surrogates in fear. If the CSs were absent, sometimes the infants would run to where the CSs customarily were and then run

from object to object, screaming and crying all the while. Continuous, frantic clutching of their bodies was very common . . . (Harlow 1958 p. 680).

Also, the infants would

rush across the room and throw themselves facedown on the floor, clutching their heads and bodies and screaming their distress (Harlow 1959a, p. 72).


Photo 2
Rhesus monkey infant after being placed in a strange room without its cloth surrogate mother. From H. F. Harlow, 1958, “The Nature of Love,” Am. Psychol. 13:680. Copyright (1958) by the American Psychological Association. Reprinted by permission of the publisher.

These infants eventually were separated from their CSs and later tested to see if they had retained their affection for the surrogates. The tests administered during early infancy were repeated and yielded similar results. One variation of the test in the unfamiliar room involved shielding the CSs, when present, by a plexiglass box. Hence, the youngsters could not reach them. The infants crashed into the plexiglass in frantic efforts to reach their surrogates. A more frightening variation of the fear test was also devised: a plexiglass panel forced infants to approach the mechanical monster in order to reach their cloth mothers.

The “orphans” initially deprived of real and surrogate mothers were later housed with CSs and WSs, neither of which supplied milk. The orphans’ initial reaction to the strange surrogates was one of “extreme disturbance.” “All infants screamed violently and made repeated attempts to escape the cage whenever the door was opened” (Harlow 1958, p. 683). However, the orphans eventually became attached to the CSs.

According to Harlow (1959a), the results of his initial experiments indicated the “surprising importance of contact comfort as a prime requisite in the formation of an infant’s love for its mother” and the “unimportant role of the breast and act of nursing” (p. 74). These results were considered a fatal blow to the Drive Reduction Theory, which stressed the importance of nursing.

Harlow’s follow-up experiments were less consequential. These involved rearing infants with (1) a CS or a nursing WS, but not both (Harlow & Zimmerman 1959), (2) a CS and a nursing CS (Harlow 1960), (3) a rocking and a non-rocking surrogate, (4) a “clingable” and a non-clingable surrogate (Harlow 1959a, Harlow 1962), and (5) surrogates covered by cotton, rayon, vinyl or sandpaper (Furchner & Harlow 1969).

Harlow and Suomi (1970) conducted the most recent affection experiments at Wisconsin. Much was made of the new, simplified design for surrogate mothers a terry cloth torso with an optional head. The researchers again assessed the importance of several maternal variables in the development of filial affection.

One such variable was nursing. Harlow and Suomi admitted that one of Harlow’s major conclusions that activities associated with the breast were of no importance—was “undoubtedly incorrect.” The new experiments showed that infants preferred a nursing surrogate over a non-nursing one, though the preference waned over time. Harlow and Suomi did not note that Harlow reported the same finding in 1965 (Harlow 1965).

Another experiment reported by Harlow and Suomi showed that body warmth is an important affectional variable. One of these experiments involved surrogate mothers whose bodies could be warmed and chilled experimentally. An infant huddled in a corner and cried “piteously” when its surrogate was chilled. Harlow and Suomi, “feeling somewhat guilty,” restored the warmth, but later resumed alternating warm and cold temperatures. A second infant was subjected to a modified procedure and was so devastated that it did not even cling to the surrogate when the latter was warm.

One affection study was conducted on dogs (Igel & Calvin 1960). “Mongrel” puppies were separated from their mothers at birth and reared with a CS, a WS, or both. The amount of time spent on each surrogate was the sole criterion for the strength of attachment.

As in the Wisconsin studies, ‘contact comfort’ was found to be a highly important variable, but in the present study, contrary to the Wisconsin finding, lactation was found to have a very marked effect for the subjects with cloth mothers (Igel & Calvin 1960, p. 305).

Hence, the researchers believed that their results revived the notion that nursing is a factor in filial bonding. Harlow reached a similar conclusion after this dog experiment was published (Harlow & Suomi 1970, see above).


The Nature and Extent of Suffering

Over 62 animals were involved in the affection experiments (Reference Note 1, p. 11). Most of these animals were the monkeys in Harlow’s experiments. Defenders of Harlow’s experiments might claim that these monkeys were not harmed. However, Harlow’s experiments were quite stressful and psychologically damaging. Virtually all of the infants were, in Harlow’s own words, “terrorized” in various tests.

Much of the trauma in these experiments could have been avoided. For example, one can argue that the tests used to demonstrate the superiority of cloth surrogates over wire surrogates were unnecessarily stressful and excessive. Simply the amount of time the infants spent on each surrogate would have been a sufficient measure of the cloth surrogate’s superiority (as in the dog study). Even if the supplementary tests were considered essential, they could have been modified to be less stressful but equally convincing.

The full extent of the damage inflicted in these experiments was not revealed in Harlow’s original reports. The extensive disturbance of surrogate-reared infants was realized only when these infants grew up. Rhesus infants reared with inanimate surrogate mothers became as psychologically disturbed as infants reared alone (see Ch. 4). One manifestation of their abnormality was “chewing or tearing at [their own limbs] with the teeth to the point of injury” (Harlow & Harlow 1982a, p. 217). Ironically, Harlow initially boasted that his surrogate mothers were “mother machines” that were superior to real mothers.

Some of Harlow’s infants not only were deprived of their real mothers, but also prematurely separated from their surrogate mothers. Such separations can induce anxiety and grief.

Finally, the suffering of monkey mothers should also be considered. These mothers no sooner bore their offspring than the latter were taken from them. The removal procedure involved a struggle between the mothers and the researchers. In the aftermath of this struggle, the mothers undoubtedly grieved over the loss of their infants.

Unnecessary Duplication

Do affection studies contain unnecessary duplication? This issue is separate from the question of whether or not this line of research should have been conducted at all. Several reports of affection research published by Harlow were largely different versions of the same manuscript, making the research appear more duplicative than it actually was. However, some of Harlow’s latest experiments (Harlow & Suomi 1970) duplicated earlier research.

Financial Cost

The Wisconsin affection experiments cost approximately $2.4 million to conduct (Reference Note 2, p. 12). (No funding information is available on two studies not conducted at Wisconsin). Ninety six percent of this amount was supplied by the National Institutes of Health. This means that American taxpayers paid for virtually all of Harlow’s affection research. Lesser amounts were supplied by the Ford Foundation ($100,000) and the University of Wisconsin (exact amount unknown).

Benefits to Humans

Harlow’s affection experiments brought him scientific acclaim. The most widely cited finding was that rhesus monkey infants preferred a cloth mother to a wire mother, even when the latter provided milk. This finding discredited the Drive Reduction Theory of how humans develop an attachment bond to their mother.

The idea that infants’ emotional ties to the mother . . . [was] based on the reduction of biological drives dominated American theories of infancy from World War I until the early 1960’s. Because the feeding situation was considered so important, child development experts and parents devoted a great deal of attention to whether a child was breast-fed or bottle-fed, whether fed on schedule or demand, [etc…]. Although these important questions were investigated extensively by scientists, no consistent relations between feeding patterns and the child’s subsequent social and emotional development were discovered. . . . But a more fatal blow to theories emphasizing biological drive reduction was dealt by Harry Harlow and his colleagues (Mussen et al. 1984, p. 170).

Hence, Harlow’s affection experiments on rhesus monkeys were considered to have advanced the understanding of human infancy. This theoretical shift undoubtedly influenced later research on infant development in humans. However, there is no apparent evidence that Harlow’s affection experiments have had any impact on clinical practice, either directly or indirectly.

Harlow and Contact Comfort

The concept of contact comfort is most closely associated with Harlow, and rightly so. However, Harlow’s contribution in this area should not be exaggerated.

Harlow was not the first to recognize the phenomenon of contact comfort, nor the first to suggest it was important in bonding human infants to their mothers. Psychoanalysts did not use the term ‘contact comfort,’ but they clearly referred to the same phenomenon. Indeed, some psychoanalysts postulated a clinging drive that was capable of bonding infants to mothers (Bowlby 1958). This view, also shared by ethologists, was the major alternative to the Drive Reduction Theory in the period before Harlow began his affection experiments.

Contact comfort in infant monkeys and apes was recognized decades before Harlow’s work (Wallace, cited in Mason 1968; Foley 1934, 1935). Foley related this “contact clinging” and its emotional consequences to the rhesus monkey’s semi-arboreal existence. Harlow seemed reluctant to acknowledge the obvious connection between infant clinging and arboreality. Perhaps Harlow thought that this connection revealed that rhesus monkeys may not be a good model of filial attachment in humans. Harlow’s contribution to these earlier studies was to provide evidence that the importance of contact comfort in infant primates was counter to prevailing views of filial attachment.

This evidence was recently criticized, on methodological grounds, by L. Ainsworth, who conducted his own affection study (Ainsworth & Baker 1982). His myth-shattering conclusion was that:

Contrary to general belief, published reports of the original infant/surrogate work contain few data to support the idea of a need for contact comfort . . . (Ainsworth 1984, p. 943).

Harlow’s Conclusions and Their Implications for Humans

Perhaps Harlow’s most surprising and influential conclusion was the “unimportant role of the breast and act of nursing” in the formation of an infant’s love for its mother (Harlow 1959a, p. 74). Yet Harlow himself later admitted he was “undoubtedly wrong” in dismissing the importance of nursing (Harlow & Suomi 1970). Harlow came to recognize nursing as well as contact comfort as important variables binding rhesus monkey infants to their mothers. Bowlby had reached a similar conclusion regarding human infants in 1958, although he admittedly lacked experimental data.

Harlow was initially led astray on the importance of nursing probably because his nursing surrogate mother was made of wire. The wire may have so frustrated the pervasive clinging responses of newborn primates that the importance of nursing could not be manifested (Gerwitz, in Harlow 1965).

Even if clinging is more important that suckling in filial attachment in rhesus monkeys, the same is not necessarily true for humans. Mason, a former student of Harlow, cast doubt on the general relevance of monkey affection data for humans.

No human infant, of course, is able to cling, find the nipple, and to nurse without extensive maternal assistance. The same infantile responses that make the difference between life and death for the rhesus monkey [e.g. clinging] are often viewed in the human infant as mere behavioral curios, without obvious adaptive value (Mason 1968, p. 98).

Mason (1968) concluded that animal models of human behavior work best for well-circumscribed processes and not, for example, “the origins of love.” This candid statement is a clear reference to Harlow’s affection research, which Harlow did consider to be an animal model of love (Harlow, in Tavris 1973). Mason presumably would agree with psychoanalyst Dallas Pratt (1980), who wryly noted the following:

It has been pointed out that contact comfort is more important in monkeys than in man, and anyone who has watched an infant monkey clinging tightly to its mother as the latter swings through the trees can easily understand why (p. 61).

To summarize, Harlow’s animal model of filial affection overestimated the importance of contact comfort and underestimated the importance of nursing. Hence, any direct applications of Harlow’s conclusions to humans may have led human research astray.


It has already been suggested that Harlow could have modified his procedures to reduce the stress to which he subjected his monkeys. Could he have avoided using animals altogether? While no one would suggest depriving human infants of their mothers to conduct affection research, certain types of human studies can be conducted. For example, mothers could provide variable levels of contact comfort or nursing while keeping the other factor constant. Thus, all infants would thereby receive contact comfort and nursing, but some would receive enhanced levels. The strength of the infants’ developing attachment to their mother could then be monitored. Psychologist Mary Ainsworth has conducted studies along these lines. For example, on the basis of one of Ainsworth’s recent studies, Mussen et al (1984, p. 170) concluded that “the strength of a child’s attachment to either parent is not related in any simple way to the frequency with which that parent feeds [the child].”

In the 1950s, when Harlow’s first affection experiments were conducted, infants in many orphanages lived in severely deprived environments (Bowlby 1952). This unfortunate situation could have provided opportunities for positive, albeit experimental, intervention by caretakers and researchers into the nature of filial attachment.


Affection experiments involved rearing infant monkeys and dogs with artificial mothers having different characteristics. The aim of these experiments was to determine which maternal characteristics underlie the infant’s attachment to its mother.

Most of the 100 or so animals involved in these experiments were infant monkeys. They were distressed in various tests and psychologically damaged by maternal deprivation. The nine studies conducted by Harlow cost a total of about $2.5 million, supplied primarily by the federal government. The studies had an impact on child development theory, but no direct impact on human welfare. Some alternatives to using animals in this research are cited.

Reference Notes

1. Throughout this report, numbers of animals refer only to those animals subjected to deprivation experiences. Animals not subjected to these experiences, such as animals in some control groups, were not included in the totals. Furthermore, an effort was made to avoid recounting the same animals mentioned in two or more reports.

2. Determining the exact cost of particular studies is difficult. Individual studies are parts of larger research projects, and these projects are the recipients of research grants. Funding data cited throughout this report are therefore approximate. The cost of some studies undoubtedly has been overestimated, but this should be offset by studies whose cost has been underestimated, or by studies for which funding data are lacking.


Please cite as:

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