This question is at the heart of an intriguing case study just published in the journal Infant and Child Development. The author, the developmental psychologist and therapist Aletha Solter, had been working with the family of a little boy, Michael, who at the age of five months had had a short stay in hospital while he underwent cranial surgery. The therapy he was receiving was aimed at relieving the behavioural symptoms of traumatization. Michael had had no further experience of hospitals since his stay as a baby. Solter took advantage of this fact in planning a study of Michael's memory for his time in hospital. Crucially, she also asked Michael's family not to discuss his hospital stay with him.
Solter then conducted two follow-up interviews in which she asked Michael about his memory for the event. The first interview, which took place when Michael was 29 months old, showed him to have some strikingly detailed memories of his stay in hospital two years before. He recalled that his eyes had been closed for a time (as a result of the surgery), that the nurse had been wearing a red blouse and scarf, and that his grandfather had sung the carol 'Silent Night'.
At the second interview, conducted when Michael was 40 months, the picture was very different. This time, the little boy appeared to have no memory of his time as an in-patient. When Solter prompted him about the details he had recalled a year before, he had entirely forgotten them. As a two-year-old, Michael had limited but detailed and accurate memories of this event from his infancy. At three, those memories had vanished.
These findings are intriguing for a number of reasons, but not least because they demonstrate that 29-month-old Michael was able to use language to describe events for which he would not have had the relevant language at the time. Although he is noted to be a verbally precocious child, at the age of five months he would presumably have had no language at all. Solter notes some other studies that have shown that children can later apply verbal labels to preverbal experiences. For example, the researchers Gwynn Morris and Lynne Baker-Ward showed two-year-old children an event (the activation of an interesting bubble-making machine) that was critically related to colour (only a particular colour of bubble soap activated the machine). Children who did not have colour words at the time of the event were then, over a period of two months, given instruction in using colour words. When they were tested for recall of the original event, a significant percentage of children who did not have colour words at the time of the event nevertheless used their new colour labels in recalling what had happened.
The weight of the evidence, though, points to very limited verbal access to preverbal memories. A major force behind childhood amnesia is undoubtedly that children are trying to relate in language events that happened before they had language. What Michael's case shows is that there might be a sensitive period for access to such memories. If your language is good enough in that period between about 2 and 3 years of age, you might be able to gain verbal access to traumatic memories for events in early infancy. Most children, though, are not as verbally precocious as Michael is reported to be. For them, the horrors (and joys) of infancy are lost for ever.