When I was little, I liked to play with toy trains. But when I got tired of those, I wiled away my days with cowboy and Indian figures, an erector set, a chemistry set, and wooden puzzles. There was one in particular, of the forty-eight states (yes, I’m that old). I liked to shut my eyes and try to figure out which state I was holding by feeling its distinctive shape. I didn’t have much interest in dolls; nor did my feminist mother, who wanted me and my older brother to become scientists. Later, at summer camp, a group of friends jokingly wrote each other’s epitaphs. Mine: “In memory of Anne, who liked bugs better than boys.” I was eleven and had spent the summer creating a nature museum filled with snakes, frogs, and bird nests.

And then there were memories that became mine because my mother regularly repeated them to me. According to her, whenever she asked me what I did at nursery school, I would reply, “Just pegs.” The moral was that she had hoped I would make friends and socialize, but I was always a loner. In many ways I still am.

But was I really? What relationship do these memories—mine, hers, hers that have become mine—bear to the “objective” events of my childhood? Can I read them as evidence that I was born to become who I am today?

The question of memory is of far-reaching importance. Can we use adult recollections of childhood as a source of scientific data or as a basis for charging someone with child abuse (sexual or otherwise) years after the alleged offense? Can we trust eyewitness identifications in the courtroom? Can we put an understanding of how memory works to practical use by, say, devising treatments for people with disabling posttraumatic stress disorder? Recent work on the biology of memory retrieval offers some disturbing answers and some encouraging ones.

Long before cognitive science was a glint in Donald Hebb’s eye, Freud understood that memories do not precisely mirror the past. More recently the idea that memory is dynamic and constructive rather than passive and fixed has gained currency among neurobiologists. Among animals as different as crabs, rats, slugs, and humans, new memories are shaky. If something disruptive, for example systemic drug treatment or a major electric shock, happens shortly after training a rat or a mouse on some task, it can’t remember the training. But let matters settle for a few hours or a day and then give the shock or drug treatment, and the rodent remembers its training. Such experiments—and there have been many of them—define the concept of consolidation: specific memories undergo a period of solidification that renders them stable.

But there’s a rub. Retrieve that memory, and it becomes temporarily labile. Consider a rat that has been given a mild electric shock while a bell rings. After consolidation the rat freezes in fear whenever it hears the bell, even if there is no further shock. Upon hearing the bell it retrieves the memory of the shock. If, however, it hears the bell and then receives an antibiotic injection that inhibits protein synthesis, the rat forgets the association between the bell and the shock. If it just receives the injection, without retrieving the consolidated memory, the memory remains intact. In other words when someone remembers something, it is possible that, during recall and before the memory gets tucked back into bed (reconsolidated), something can interfere with its storage. This knowledge has potential applications in the treatment of posttraumatic stress disorder. By inducing selective amnesia of traumatic events, suffering may be reduced.

But wait! There’s more. When a retrieved memory reconsolidates, it may incorporate new information and a contemporary context. That is, the new “old” memory is not identical to the original old memory. Scientists have just begun to study this phenomenon in humans. In one experiment students learned a set of numbers on day one and a different set on day two. If they were reminded of the events of day one before taking on the second task, some day-two numbers found their way onto the recalled day-one list. But if the events of day one were not retrieved, the students remembered a “purer” set of day-one numbers. In a different experiment, scientists found that stress actually stabilized memories during reconsolidation.

What might these findings mean for researchers who use adult memories of childhood events to study the development of specific behaviors? Let’s return to childhood play patterns. Many studies link gender non-conformant play behaviors to adult homosexuality. Among boys, a love of dolls or solitary sports or female companions correlates about 50 percent of the time with adult male homosexuality. Among girls, engaging in socially defined masculine play correlates with adult homosexuality about 40 percent of the time. An oft-drawn conclusion is that future homosexuality can be detected via the play patterns of young children, with the suggestion that play preferences themselves may be progenitors of adult sexual preference. This line of study and the interpretation of the findings are contested, but I am not arguing here about interpreting results. Instead I suggest that, given what we are learning about memory reconsolidation, adult memories of early childhood play preferences cannot provide a veridical data set.

I would definitely score high on a play-based nonconformity test. I did what the boys did: climbed trees, played cowboys and Indians and hide and seek. In my day we said I was a tomboy. I did not play with dolls or jump rope, although I did play jacks with one of my girlfriends. My gang were all boys and all older than me. I did what I could to keep up. And the funny thing is, I knew I was different. We all were—not because of how we played, but because we were the children of leftists during the height of the McCarthy period. The point being that context contributes to the shape of childhood activities, and the same behaviors and feelings can have more than one origin. Layer onto that the fact that memories change and re-associate over time, and reading cause from recall becomes a very tricky business.

Let me climb out on a limb here: adult memories cannot provide causal evidence of the childhood origins of adult behavior. The fact that many gay men and women score high on childhood gender nonconformity tests reflects how they have integrated memories—via reconsolidation—into narratives of adult identity.

Until recently we did not speak of children as being gay because what emerges as sexual desire develops slowly during childhood and youth, and comes intertwined with—both driving and being driven by—an evolving sexual identity. It would therefore be interesting to study the processes by which gendered patterns of play become linked in some children to adult forms of desire. This requires that we understand both adult and childhood behaviors not as essences but as softly assembled patterns that may be stable but are not necessarily fixed. What are the contexts in which play memories are evoked and discussed? How do the memories change and connect to other experiences when a child, youth, or adult revisits them? At what ages do children or youth link their play interests to specific desires? How does the dynamic differ for boys and girls?

If memories adjust to current context when evoked, then they become a part of the developmental process rather than a record of the past. This makes studying childhood more difficult, but it refocuses us on the importance and fascination of development itself. Rather than try to relate events recalled during a specific moment of adult retrospection to an adult’s sexual desire as assessed during that moment of recall, we need to take multiple dips into a set of children’s lives to examine a wider variety of boyhoods, girlhoods, and developing identities as they emerge and evolve in real time.