Infant Learning and Psychoanalysis

New studies show that infants learn by watching adults in complex ways that may have important implications for psychoanalytic theory and treatment. Much of the history of how we understand infant learning is dominated by the influence of the Swiss philosopher and developmental theorist Jean Piaget. He explained child development through stages progressing from understanding the world predominantly via movement and the senses (the sensorimotor stage, birth to age 2) followed by the gradual evolution of logical thinking and manipulation of abstract ideas (pre-operational, ages 2-7; concrete operational, ages 7-11; and formal operational, ages 11 and above). Piaget's theories, though powerful in describing a normative pathway, conspicuously minimized the role of social interaction and attachment figures in cognitive development. Thus, they have been difficult to integrate with theories of pathological social and emotional development and have limited relevance to more sophisticated models of parent-infant interactions.

Several contemporary developmental theorists have gone to an opposite extreme by explaining infant learning through the concept of “primary intersubjectivity.” As represented in the work of Daniel Stern, Colin Trevarthen, Andrew Meltzoff, and proponents of “mirror neuron” theories, led by Vittorio Gallese, this approach suggests that infants learn by first sharing the subjective physiologic and mental states of the other, and then abstracting rules from that experience. Though bearing some resemblance to Freud's notion of “primary identification,” these theories take this notion further by emphasizing the role of primary shared experience as preceding and even in adults often overriding more nuanced and psychologically processed ideas about self versus other. As such, they have direct implications for our understanding of psychopathology and therapeutic action, and have recently been cited as supporting an intersubjective or self-psychological view of psychoanalysis.

Motherese

However, recent experimental work suggests that the most accurate model of infant learning adheres neither to the description of Piaget nor to that of primary intersubjectivity. In a classic experiment, Piaget showed that if a one-year-old infant is repeatedly shown that an object is hidden in one location he continues to look for it there even if he is subsequently shown the object being hidden in a different location. (For the sake of simplicity, we will use the male pronoun to refer to infant subjects and the female pronoun to refer to adult demonstrators, even though in practice both male and female infants and demonstrators were used). Piaget explained this “perseverative search error” by postulating that the infant's development of “object permanence” is not yet complete by one year of age and that the infant conceives of the location of the object as a consequence of the search process, not the hiding of the object. In other words, the infant believes that the act of searching “produces” the object in the location where he has repeatedly seen it appear, so it makes sense for him to continue to look there.

Gyorgy Gergely (pronounced “Joory Gairgay”), a psychoanalyst and developmental psychologist working at the Central European University in Budapest, reported a surprising twist on this experiment in a 2008 Science paper. He and colleagues compared the results of the typical task, in which the infant and demonstrator interact normally during the hiding of the object, to two variants: (1) the demonstrator does not look at or talk to the infant while she is hiding the object, and (2) the demonstrator's body and face are hidden behind a screen while she is hiding the object. In Piaget's version of the experiment, when there is normal social interaction, the infant searches for the object in the wrong place 81 percent of the time. However, in both variants of the task, where there is no social communication, the infant's error rate drops to 41 percent. Gergely and colleagues interpret this result, which has no ready explanation in Piaget's framework, in the context of a theory they call “natural pedagogy.” They hypothesize that when infants are directly addressed with social cues (such as through eye contact or through the typical way of speaking that care-givers spontaneously use with infants, often called “motherese”), they automatically expect to be taught about generalizable properties of an object type (i.e., “where is this kind of an

object usually found?”). In contrast, when an infant does not receive social cues, he automatically expects to be taught specific facts about the current environment (i.e., where is the object found in this particular situation?). In the typical version of the experiment, social communication with the demonstrator fools the infant into thinking that he is being taught (incorrectly) to look for the object in the same place each time. In Gergely's variants, the absence of social cues frees the infant to look for the object (correctly) in the place that it has just been hidden. Thus, the social context of how an infant learns is highly significant to what he learns.

…is transference change a kind of generalizable learning that sometimes is helped by and is sometimes inhibited by social cues?

Just Like Me?

Gergely is equally skeptical of theories that suggest that learning is entirely based on social context, as is sometimes predicted by primary intersubjectivity. In a 2002 Nature paper, he described a clever twist on Andrew Meltzoff's classic demonstration of imitation in 14-month-old infants. In the original Meltzoff experiment, a demonstrator showed an infant that he could turn on a light by touching his forehead to it. A week later, two-thirds of these infants spontaneously tried to turn the light on by touching their own foreheads to it, whereas infants who had not seen the previous demonstration never did so. This result was interpreted as meaning that an infant learns through imitation that is induced by identification of the other who is perceived as “just-like-me.” Gergely first repeated this experiment in two different contexts: (1) while the demonstrator used her head to turn on the light, her hands were occupied by holding a shawl around her shoulders (suggesting that she used her head to turn on the light because it was the best available body part in this context, and not because it was the best general way to accomplish this goal) and (2) while the demonstrator used her head to turn on the light, her hands were free. When the demonstrator's hands were free, the results were the same as Meltzoff's, i.e., 69 percent of infants tried to turn the light on with their heads. But when the demonstrator's hands were occupied, only 21 percent of infants tried using their heads and the rest used their hands. Thus, Gergely argues, when an infant observes a person performing an action, he does not simply identify the demonstrator as “like me” and directly imitate her (Meltzoff's account). Rather, the social context of the interaction between demonstrator and infant engages the system of natural pedagogy and induces the expectation that he is going to learn new and relevant information. When the demonstrator's hands are free, and therefore could have been used to turn on the light (which the infant knows is the typical thing to do), the infant learns “this light ought to be turned on by touching it with the head!” However, when the demonstrator's hands are otherwise occupied, the infant concludes that this is the reason she is using her head, and that if her hands were free (as the infant's are) she would have used her hands. Gergely demonstrated that social context is necessary for inducing this pedagogical stance by repeating both versions of the experiment (i.e., with and without the demonstrator's hands occupied), this time without any social cues between demonstrator and infant. In both these situations, infants used their hands. Thus, without the proper social cues, the pedagogical stance is not activated.

In both experiments, Gergely's deceptively simple modifications of classic and well accepted studies have subtle but important implications for our understanding of the mind. Children as young as one year of age are already applying rational learning principles to observations they make of people and objects. They do this neither by simple learning principles nor by direct imitation. Hence, there is also no reason to believe (as would seem highly unlikely given the sophistication that such a talent would require) that infants have an inborn ability to identify their own subjective internal states and compare these to those of other individuals, as is sometimes hypothesized by intersubjectivists. Rather, healthy infants appear to be good at choosing the correct level of generalization when they are observing and learning from another individual's actions. It seems likely that this skill is highly adaptive and leads later on in development to the ability to identify mental states in the self and others.

Of course, many questions remain. What kind of variation exists in this learning flexibility and how does it affect emotional and social development in both healthy and pathological ranges? How are the process and content of unconscious fantasy and defense mechanisms related to the role that social context plays in learning in infants, children, and ultimately in adults? What does this teach us about the social context of learning in psychoanalysis and psychotherapy? For example, is transference change a kind of generalizable learning that sometimes is helped by and is sometimes inhibited by social cues? Finally, how might we use our understanding of these systems to choose the best therapeutic parameters for individual patients? The implications for psychoanalytic theory and treatment are exciting to consider.

Readers are invited to contact the author at for suggestions on further readings.