The term secondary drive hypothesis is used to describe the process of learning an attachment through operant and classical conditioning. It explains how primary drives which are essential for survival (eating when hungry), get associated with secondary drives (comfort when the mother is present).
The learning theory provides a plausible logical explanation of how attachments are formed through essential survival needs and attaching to the one person who satisfies those needs, however there is little research that supports it, and lots of evidence that infants are able to form multiple attachments, and attachments with those who don’t feed them.
For example, although Bowlby’s evolutionary theory of attachment suggests that infants only initially for one primary attachment (monotropy) most often with the mother, he also suggests that millions of years of evolution have produced an innate behaviour for infants to form attachments, called social releasers (behaviours that are intended to initiate interaction and unlock adults’ tendency to care for infants), as if they didn’t form attachments with caregivers, they would not be fed and comforted and therefore would not survive to reproductive age. Attachments would also have to be formed within a critical period (up to 2 and a half years of age) or no attachment would be formed. This undermines the learning theory of attachment that infants have to learn to form attachments; they are not innate and biologically pre-programmed.
For example, Schaffer and Emerson (1964) did a longitudinal study of the attachments formed by 60 infants in the first 18 months of life, at monthly intervals. They found that a significant number of the infants formed multiple attachments with a person other that the primary feeder. They found that it was the quality of the interaction that determined who they attached to the most, the ones who responded quickly to the infants behaviours and interacted with them most were the most attached.
Lorenz did a study with geese (1935). He took a group of goose eggs and kept them until they were about to hatch. Half of the eggs were placed with the mother and the other half was placed with Lorenz. He found that when the geese hatched they instinctively followed the first figure they saw, and for half of them this was Lorenz. This process happened before any feeding had taken place, and is known as imprinting. Lorenz also ensured imprinting had taken place by putting all the geese together enclosed, and when the enclosure was removed, Lorenz’s half of the geese returned to him. This study goes against the learning theory of attachment, and suggests that attachment is in fact innate and biologically pre-programmed. It also supports the idea of the critical period; as if no attachment had developed within a certain number of hours then it was unlikely that any attachment would ever develop.
Harlow (1958) also experimented with attachments formed between rhesus monkeys and surrogate mothers, made from wire. In each case there were two surrogate wire mothers, one provided food, and the other provided comfort, made from soft cloth. He found that the monkeys would cuddle up to the cloth monkeys when distressed, and spend the majority of their time with them, only leaving to go to the wire surrogate when hungry. This undermines learning theory, as the monkeys did not spend much time with the mother that provided them with food, they instinctively went to the cloth mother for comfort and when distressed, suggesting that attachment in rhesus monkeys is innate, and they don't link food with pleasure.