The learning theory of attachment suggests that attachment is a set of learned behaviors instead of innate biological behavior. The basis for the learning of attachments is the provision of food.
This account is often called the cupboard love theory. Because feeding is the engine of the bond, the caregiver is thought to gain her emotional value the way a cupboard does: by being where the food is.
This theory encompasses two types of learning: classical conditioning, where an infant learns to associate the caregiver with comfort and eventually forms an attachment.
Operant conditioning, on the other hand, assumes that infants are in a drive state of internal tension or discomfort, and their actions focus on removing this discomfort.
Key Takeaways
- Cupboard Love: Learning theory holds that attachment is learned through feeding, not innate; the caregiver becomes valued because she is the source of food.
- Classical Conditioning: The caregiver, initially a neutral stimulus, becomes a conditioned stimulus for pleasure and security through repeated pairing with feeding.
- Operant Conditioning: Feeding reduces the infant’s hunger drive, and the caregiver becomes a secondary reinforcer the infant learns to seek out.
- Dollard and Miller: Their secondary-drive hypothesis (1950) gave the theory its fullest form, fusing Freudian and Hullian ideas.
- Evidence Against Feeding: Harlow’s monkeys preferred a comforting cloth surrogate over a feeding wire one, and Schaffer and Emerson found many infants attached to non-feeding caregivers.
- Modern Evidence: A 2024 meta-analysis of over 22,000 child-caregiver pairs confirms caregiver sensitivity, not feeding, predicts secure attachment, though learning may still shape how secure a bond becomes.
Classical Conditioning
Classical conditioning involves the formation of associations between different events and stimuli.
This process explains how an emotional bond is formed through associations with comfort and security, contributing to the infant’s attachment to their mother.
Classical conditioning, as explained in the context of attachment theory, posits that infants learn to associate their caregivers, usually the mother, with satisfying their needs. This association produces pleasure.
Over time, the infant forms an attachment to this person due to their association with pleasure. The mother becomes the conditioned stimulus, and happiness becomes the conditioned response…attachment has formed.
Here is a basic breakdown of how it works:
- Unconditioned Stimulus (UCS): The infant’s biological needs, such as hunger, creates discomfort. When these needs are satisfied, such as feeding, it leads to pleasure.
- Unconditioned Response (UCR): The infant feels pleasure when their needs are satisfied.
- Neutral Stimulus (NS): Initially, the mother is a neutral stimulus, as she is not innately associated with satisfying the infant’s needs.
- Association: Over time, the infant begins associating the mother (NS) with satisfying their needs (UCS). The mother is present when feeding results in the infant associating the comfort and pleasure of feeding with her.
- Conditioned Stimulus (CS) and Conditioned Response (CR): The mother becomes a conditioned stimulus. This means that her presence alone is enough to trigger a sense of security and pleasure in the infant (now a conditioned response), even without the original unconditioned stimulus (feeding). This formed association is what we understand as the attachment between the infant and the mother.
Dollard and Miller
John Dollard and Neal Miller gave the learning theory of attachment its most complete form.
Their 1950 model, the secondary-drive hypothesis, fuses Freud’s and Hull’s drive theories into a single account of how feeding creates a bond.
Who Were Dollard and Miller?
John Dollard and Neal Miller created the learning theory of attachment, combining elements from Freud’s and Hull’s drive theories and unifying psychoanalysis and behaviorism in one model.
Their theory holds that attachment is a learned behavior, built through classical and operant conditioning around drive reduction.
The bond itself is real, they argued.
But on Dollard and Miller’s account it is downstream of feeding, not an innate pull toward the caregiver herself. The caregiver becomes emotionally important because she reliably satisfies the infant’s hunger, again and again, feed after feed.
Remove the feeding history entirely.
On this theory, the attachment would then never have formed at all. It remains the most systematic statement of the “cupboard love” idea in developmental psychology, and the version Bowlby spent much of his own work arguing against.
Drive Reduction and Operant Conditioning
Operant conditioning explains the infant’s behavior, not just its feelings. Dollard and Miller proposed that infants live in a drive state: a form of internal tension or discomfort that motivates them to act so as to reduce it.
When an infant is hungry, this discomfort creates a drive state. Feeding removes it. The drive is reduced, hunger is satisfied, and the infant feels comfortable again.
Over time, the infant learns to view food as a primary reinforcer, rewarding in its own right because it directly reduces the drive. The theory runs on four linked processes: drive, cue, response, and reinforcement. That is the core mechanism.
Through repetition, infants learn to associate the relief of discomfort with their caregiver. That learned association is, on this theory, the attachment itself.
The Secondary-Drive Hypothesis
The secondary drive hypothesis is the core of Dollard and Miller’s (1950) theory. It explains how complex secondary drives, like emotional closeness, develop out of primary ones, like hunger.
Primary drives are innate.
Secondary drives are learned. They are acquired only because they have repeatedly been paired with a primary drive’s relief. As infants grow and meet different situations, this pairing builds into a complex secondary drive for the caregiver’s company specifically, not for food itself any longer.
She becomes something more than a feeder.
The person providing the food becomes associated with the comfort that follows, and so becomes a secondary reinforcer in her own right. The infant then seeks her out. Not because of who she is, but because of what her presence has come to predict.
Reciprocal Reinforcement
The hypothesis treats attachment as a two-way learning process between child and caregiver. Ending the infant’s distress is negatively reinforcing for the caregiver too, since it satisfies her own wish to soothe him.
The infant’s own behavior reinforces the caregiver in turn.
Her smiles and settling reward the caregiver directly, giving her real pleasure in the exchange. The reinforcement process runs both ways, and it is this two-way loop, not the feeding alone, that strengthens the bond between them over time.
This mutual exchange does not create the bond from nothing, learning theory argues.
It strengthens a connection that feeding has already established. Reinforcement and association, on this view, are what let a bond deepen once it already exists, feed after feed, day after day. That is the theory’s real insight: not that love begins in reinforcement, but that reinforcement helps it grow.
Critical Evaluation
Before looking at each criticism in detail, here are the main problems researchers have raised about the learning theory of attachment:
- Not Built on Feeding: Schaffer and Emerson (1964) found that in 39% of cases, an infant’s main attachment figure was not the person who fed and cared for them most.
- Contact Comfort Beats Food: Harlow’s study of monkeys clung to a soft cloth surrogate mother even when a separate wire surrogate provided all their food.
- No Explanation for the Critical Period: the theory does not explain why attachment must form within a set developmental window, or why infants reach the same stages at similar ages.
- Imprinting Without Feeding: Lorenz’s research involving goslings demonstrated imprinting on the first moving object they perceived, with no feeding involved at all.
- Over-Reliance on Animal Research: much of the key evidence comes from monkeys and geese, raising questions about how far it generalises to humans.
- An Oversimplified Mechanism: drive reduction alone cannot explain behaviors, such as secondary reinforcement, that are not really about reducing discomfort.
- Contemporary Research: a large 2024 meta-analysis confirms that caregiver sensitivity, not feeding, predicts secure attachment, though modern theory still gives learning a supporting role.
Not Built on Feeding
In a prospective study of 60 Glasgow infants, Schaffer and Emerson (1964) tracked attachment formation across the first eighteen months of life. Specific attachments typically appeared at seven to eight months.
Multiple attachments were the rule, not the exception.
The result that mattered most for learning theory concerned who each infant’s main attachment figure actually was. In 39% of cases, it was not the person who fed, bathed, and changed the infant most. Many were strongly attached to a non-caregiving figure, often the father.
Feeding alone did not decide it.
Responsiveness, not feeding, predicted the bond. This is a direct contradiction of the cupboard-love prediction, which says the feeder should always be the attachment figure.
The design has real strengths: a genuine, real-world, longitudinal study, not a lab experiment. But the data rest on mothers’ self-reports, open to bias. The sample was small and working-class. It came from just one 1960s Glasgow population, which limits how far it generalises.
Contact Comfort Beats Food
Harlow’s study of monkeys (1958) reared infant rhesus monkeys with two surrogate mothers: a bare wire cylinder and a soft, terry-cloth-covered one.
Crucially, the feeding bottle was fixed to the wire mother for some infants and to the cloth mother for others.
The result was the same either way. Monkeys spent most of their time clinging to the cloth mother, regardless of which surrogate actually fed them. When frightened, they ran to the cloth mother and used her as a secure base.
Even monkeys fed exclusively by the wire mother went to her only to feed. They returned to the cloth mother immediately after. Food did not buy attachment.
This is a controlled experiment, and a clean one: it separates food from comfort directly. But it is also open to real objections. Harlow’s monkeys are not children, and the study inflicted lasting, deliberate harm on the animals involved.
No Explanation for the Critical Period
The learning theory of attachment does not explain why a critical period exists in most animals and humans.
After this window closes, infants find it far harder to form an attachment, or cannot form one at all.
Nor does it explain why infants seem to pass through the same stages at about the same age. Drive reduction says nothing about developmental timing. It is silent on the question entirely.
Bowlby borrowed the idea of a critical period directly from ethology.
Later work softened it to a less absolute “sensitive” period, but the basic timing pattern still needs an explanation learning theory does not supply.
This omission matters because Bowlby’s rival account treats the critical period as central evidence for an innate, biologically-timed system, precisely the kind of explanation drive reduction cannot offer.
Imprinting Without Feeding
Lorenz’s research involving goslings (1935) studied imprinting in greylag geese.
He observed goslings in the hours immediately after hatching, manipulating which moving object they encountered first. Each gosling formed a strong bond with, and then followed, the first large moving object it saw.
For one group, this was Lorenz himself, not the mother goose. No feeding was involved at all.
The bond formed only within a brief critical period. It was largely irreversible once formed.
In at least one species, a strong and selective bond can form entirely without feeding, through perception alone.
That makes a pure cupboard-love account look incomplete.
Generalising from geese to human infants remains a real limitation, and imprinting alone cannot explain the emotional richness of human attachment.
Over-Reliance on Animal Research
Two of the three classic pillars of evidence against cupboard love are animal studies: Harlow’s rhesus monkeys and Lorenz’s goslings. That raises a standing question about generalising from animals to humans.
Behaviorists argue that human and animal learning processes are similar enough to justify the comparison. Critics see this as overly simplistic. It overlooks much of what makes human attachment distinctly human.
Harlow’s work also raises serious ethical concerns. The monkeys suffered real and lasting harm for the sake of the findings.
The human evidence, chiefly Schaffer and Emerson’s study, does converge with the animal findings on the central point: feeding is not what drives attachment. That convergence is what makes the cross-species generalisation defensible at all, despite the ethical and generalisability concerns raised above.
An Oversimplified Mechanism
Drive reduction, the mechanism at the heart of the learning theory, is no longer widely accepted as a complete account. It is seen as too simple on its own.
It struggles to explain behaviors driven by secondary reinforcers rather than primary ones. It also cannot explain actions that cause discomfort rather than reduce it, which a pure drive-reduction model should rule out entirely. That is a real gap.
Reducing a rich, reciprocal bond to a set of conditioned responses leaves out most of what attachment actually is. The secure-base behaviour, the separation distress, the felt specialness of the bond: none of it is really captured.
A theory built to explain lever-pressing struggles to capture love. This is the theory’s deepest weakness.
Contemporary Research
The classic studies date from the 1930s to 1960s.
A modern meta-analysis brings the question up to date: is it responsiveness, or something else, that builds a secure attachment?
- Aim: to quantify how strongly caregiver sensitivity predicts child-caregiver attachment security across the published literature, and to test whether the effect differs for mothers versus fathers.
- Method: a large meta-analysis pooling 230 effect sizes from 174 samples (N = 22,914 child-caregiver pairs), with separate estimates for maternal and paternal sensitivity (Madigan et al., 2024).
- Results: sensitivity was significantly linked to attachment security overall (r = .25). Paternal sensitivity predicted security about as strongly as maternal sensitivity did (r = .21 versus r = .26), even though fathers typically do far less routine feeding.
- Conclusion: across more than twenty thousand pairs, sensitive responsiveness, not feeding, reliably predicts a secure attachment, and it works this way for fathers as well as mothers.
Learning theory is not without a modern defence, though.
Bosmans et al. (2020) argue that individual differences between secure and insecure attachment partly follow the principles of classical and operant conditioning. These are linked to the body’s oxytocin, cortisol, and dopamine systems.
They still treat attachment as innate.
Conditioning, on this modern account, shapes how secure a bond becomes, not whether it forms at all.
Learning Check
Question 1
Andrea provides most of the care for her son, Oliver, feeding, comforting, and playing with him. She has noticed that, whilst he is happy to spend time with his father, Oliver seems most content when he is with her.
Use your knowledge of the learning theory of attachment to explain Oliver’s behavior. (2 marks)
| Answer As Andrea provides most of the care for Oliver, he associates her with the satisfaction of primary drives, such as hunger and thirst. She has also become associated with other pleasurable things, such as comfort and security, through classical conditioning. In contrast, his father provides little care, so Oliver has not associated him with pleasure and comfort. Therefore, Oliver prefers to be with Andrea than with his father. |
Question 2
Miguel’s mother gave up work when he was born and stayed home to look after him. Miguel’s father works far away, so he is rarely home before his bedtime. Miguel is nine months old and has a very close bond with his mother.
Use learning theory to explain how Max became attached to his mother rather than to his father. (6 marks)
| Answer According to the learning approach, attachment is a learned behavior. It can be learned through classical conditioning, learning through association, and operant conditioning, learning through consequences. According to classical conditioning food (UCS) produces pleasure (UCR). Miguel’s mother started as a neutral stimulus but was associated through repeated pairings with the food and became a conditioned stimulus. So Miguel felt pleasure when he saw his mother (CR). According to operant conditioning, food satisfied Miguel’s hunger and made him feel comfortable again (drive reduction). Food was, therefore, a primary reinforcer. His mother was associated with food and became a secondary reinforcer. Miguel became attached to his mother because she was a source of reward. However, his father being absent often was not associated with pleasure as he did not feed Miguel or make him feel comfortable, so the child shows no attachment to him. |
FAQs
Who developed the learning theory of attachment?
The learning theory of attachment is most fully developed in Dollard and Miller’s (1950) secondary-drive hypothesis. It applies the general laws of conditioning that Pavlov and Skinner established to the infant-caregiver bond.
This differs from John Bowlby‘s attachment theory, which incorporates cognitive and evolutionary elements.
What is the learning theory of Dollard and Miller?
Dollard and Miller’s learning theory suggests that attachment behavior is learned through conditioning principles. Infants learn to associate caregivers with the comfort of meeting their needs, forming an attachment. Positive interactions reinforce this attachment, while negative experiences may lead to avoidance.
How are Dollard and Miller different from Skinner?
Skinner’s theory focuses on reinforcement: a caregiver’s response can encourage or discourage attachment behavior. Dollard and Miller’s theory integrates both behavioral and psychoanalytic concepts, holding that attachment forms from learned habits driven by the need to reduce the discomfort of hunger.
References
Bosmans, G., Bakermans-Kranenburg, M. J., Vervliet, B., Verhees, M. W. F. T., & van IJzendoorn, M. H. (2020). A learning theory of attachment: Unraveling the black box of attachment development. Neuroscience & Biobehavioral Reviews, 113, 287–298. https://doi.org/10.1016/j.neubiorev.2020.03.014
Dollard, J., & Miller, N. E. (1950). Personality and psychotherapy; an analysis in terms of learning, thinking, and culture. New York, NY, US: McGraw-Hill.
Harlow, H. F. (1958). The nature of love. American Psychologist, 13(12), 673–685. https://doi.org/10.1037/h0047884
Harlow, H. F., Dodsworth, R. O., & Harlow, M. K. (1965). Total social isolation in monkeys. Proceedings of the National Academy of Sciences of the United States of America, 54 (1), 90.
Harlow, H. F. & Zimmermann, R. R. (1958). The development of affective responsiveness in infant monkeys. Proceedings of the American Philosophical Society, 102,501 -509.
Lorenz, K. (1935). Der Kumpan in der Umwelt des Vogels. Der Artgenosse als auslösendes Moment
sozialer Verhaltensweisen. Journal für Ornithologie, 83, 137–215, 289–413.
Madigan, S., Deneault, A.-A., Duschinsky, R., Bakermans-Kranenburg, M. J., Schuengel, C., van IJzendoorn, M. H., Ly, A., Fearon, R. M. P., Eirich, R., & Verhage, M. L. (2024). Maternal and paternal sensitivity: Key determinants of child attachment security examined through meta-analysis. Psychological Bulletin, 150(7), 839–872. https://doi.org/10.1037/bul0000433
Schaffer, H. R., & Emerson, P. E. (1964). The development of social attachments in infancy. Monographs of the Society for Research in Child Development, 29(3), 1–77. https://doi.org/10.2307/1165727
