During the 1950s peanut butter came in notoriously hard-to-close pry-top jars, and an enterprising rat in my family’s home took advantage. At night, after my parents’ bedroom door clicked shut, they would hear a clatter as the rat removed the metal lid and dropped it to the floor. Eventually mom and dad poisoned the ingenious beast. My brother skinned it, tanned its hide, and nailed it to a bulletin board—a stark warning to future marauders.

Lately I’ve been thinking about that rat. How did it figure out how to open the jar? How did it learn that the coast was clear when my parents retired for the night? Did it have a lid-opening gene? A peanut butter gene? Was it predisposed to explore, or did its family and rodent companions teach it to investigate? Why did that individual, and not, for example, its twin or its sister, exploit this food source?

It is a quirky case, the condiment-thieving rat, but the question of how individual differences arise is important to us all. In his State of the Union address, President Obama made a big push for personalized medicine—based on the idea that if we know an individual’s genome, we can predict medical outcomes and tailor individual treatments—and the success of that enterprise relies on the right answer.

This is a complicated challenge because, as biologists first proposed in the early twentieth century, a genotype—all the genes in the cells of an organism—does not guarantee a phenotype—what the organism looks like and how it behaves. A vast territory links genes in their cellular starting environment to individual phenotypes. Recently behavioral biologist Julia Freund and her colleagues published a fascinating study that directly addresses this problem.

Individuality can emerge even when genes and environment are constant.

Freund’s group took forty genetically identical mice and raised them in an elaborate environment—an almost thirty-six-square-foot enclosure, more than six feet high, with five levels connected by clear tubes through which the mice moved freely. Researchers placed cardboard tubes, flowerpots, wooden scaffolds, and other toys on each level and used tiny radio devices implanted on the mice to record individuals’ exploratory behavior. Control mice were kept in small, unadorned cages.

At the start of the three-month experiment, all the mice behaved in pretty much the same manner, but by the end each had evolved a distinct behavioral pattern. Some maintained a small territory and moved around a lot within it. Others roamed the entire enclosure. The mice that moved around more also grew more new nerve cells in the hippocampus, a brain region especially responsive to environmental complexity. Freund’s study, then, links brain plasticity in genetically identical mice to individual experiential differences.

Freund’s team offers several possible explanations for their results. Different experiences might modify gene expression by changing the control systems in a cell’s DNA. Such changes are known to happen in human identical twins as they age. Other factors that might generate individual differences in genetically identical animals include intrauterine position during fetal development, nutrition, maternal stress, and early postnatal events, such as researcher handling of individual mice. Finally, there is mere chance. If the mice entered the cage in the same general spot, some might have been on the outside of the crowd, others on the inside. Tiny effects of initial placement could have been magnified over time via some version of the butterfly effect, whereby starting conditions have unpredictable but deterministic influence on outcomes in complex systems.

Studying individual variation in genetically identical organisms has a long history. The oldest tradition is to measure a phenotype under varying environmental conditions. For example, certain genes affecting eye development in fruit flies produce a typical eye at one growth temperature but an aberrantly small eye at another. For any particular genotype, scientists can produce a norm of reaction—a map of the phenotype under different growth conditions. Freund’s work carries this a step further, showing that individuality can emerge even when both the genes and the environment are held constant.

Enough about rodents. What about humans? Unless one is an identical twin, each of us has a unique genotype, and each of us experiences different environments. Consider genes for estrogen receptors (ER) in breast cancer tumors. Estrogen receptors are medically important because cancers that have them—ER+ tumors—can be treated with anti-estrogenic drugs, considerably improving odds of survival. But even those who have the estrogen receptor genes may be stricken with an ER– tumor. Public health specialist Nancy Krieger has noted a number of ways in which the relationship between ER and breast cancer defies simple causal connections between genotype and phenotype.

First, historically the incidence and individual distribution of ER+ tumors is not fixed. For example, many studies have suggested that because white women have a higher incidence of ER+ cancers than do black women, there must be intrinsic racial genomic differences at work. Krieger and her colleagues, however, show that incidence of ER+ tumors in white women rose during the period when hormone replacement therapy, which was more available to affluent populations, was in vogue. It dropped again when hormone relacement therapy lost medical favor. Thus differences in ER+ tumors are time-sensitive rather than invariant.

Second, individual life histories affect the appearance of ER+ tumors. A person who develops an ER+ tumor at one point in her life may later form a second tumor that is ER–. Again ER status is not a fixed genomic trait.

Third, the tumor has its own natural history. Cells in environmentally different locations within a tumor may express different genes. In lab experiments, ER– tumors can be transformed into ER+ cells.

Ultimately Krieger suggests that understanding the evolutionary origins of traits such as estrogen receptors can offer insight into why ER expression is so flexible.

So the relationship between genotype, gene expression, and phenotype is complex, without clearly understood causal linkages between the presence of genes and phenotypic outcomes. Yet public discussion often pretends otherwise. Good things may come of personalized medicine, but backers and the press deeply confound the concepts of genotype and phenotype.

In a recent article on the president’s initiative, the New York Times reports that personalized medicine involves a doctor prescribing treatment that targets a particular gene. National Institutes of Health Director Francis Collins talks excitedly about how inexpensive it has become to sequence individual genomes, creating new opportunities for such therapies. But whether someone has an ER+ tumor does not depend just on the presence of the ER gene in the genome. It hinges instead on the expression of that gene in cancerous cells. And as Krieger shows, this depends, in ways that are not yet fully understood, on individual life histories—prior chemical and drug exposures, stress, nutrition, and more.

This is not just a picky point. Understanding of genotype and phenotype affects where we invest our research dollars. Krieger did a quick word count on a 2012 NIH report devoted to minority health. Words such as “genetic,” “gene,” and “genome” appeared eighty-seven times, while “social determinants” appeared only once, and there was no mention of poverty or racism, which cause stress that affects basic biology.

By focusing on genotypes rather than the development of phenotypes, we turn our dollars and our best research minds away from the process of disease formation and thus fail to learn how a more or a less dangerous phenotype becomes embodied. The less we know, the less likely it is that we can interfere with disease formation before it gains the upper hand in an individual body.