These are heady days for psychiatry. The neuroscience and social science research underlying our specialty has flowered spectacularly. Although the goal of integrating brain and behavior continues to be elusive, it is now possible to measure brain events that are synchronous with affective and cognitive processes. We are in an era in which it is possible to practice "evidence-based psychiatry." The concept had not yet been invented when I began my career; more to the point, there was precious little evidence to support the effectiveness of what we did 50 years ago (1).

I have tried to capture the intellectual excitement of contemporary psychiatry in the title of this paper: "Nature, Niche, and Nurture: The Role of Social Experience in Transforming Genotype Into Phenotype." My aim is to highlight the ways "nature, niche, and nurture" interact to give rise to an unique human being.

My medical education, insofar as it acknowledged genetics at all, stressed the dichotomy between heredity and environment, commonly phrased as "nature vs. nurture." Its premises are that the source of a given characteristic is to be found in one or the other, that the effects of heredity and environment are additive rather than multiplicative, and that effects are absolute rather than relative. All the premises are misleading. The effects of "nature" are dependent on the quality of "nurture" during development of the organism, and the outcome of "nurture" is dependent on the "nature" of the persons exposed to it. Yet the "heritability" of a trait is defined as the proportion of variance in that trait attributable to genetic effects. But heritability is not an invariant and absolute value. The heritability calculated in a given study applies only to the population examined at that time. If external circumstances change, so will heritabilities (2). Gene effects depend on the environment in which development occurs. Environmental effects depend upon the constitution of the organism on which they act.

The concept of nature and nurture as antonyms derives from Sir Francis Galton (3), the founder of biometrics. In his monograph English Men of Science: Their Nature and Nurture, written in 1874, he said

There is an important caveat in that statement: the phrase "on equal terms." As Galton went on to note,

The men Galton studied in his English Men of Science inherited England along with their genes; without England (that is, without English culture), "English nature" would not have produced eminent men of science.

So trapped have we been by the either/or point of view that psychiatry has oscillated aperiodically from a total disregard for heredity to a preoccupation with it. When the distinguished Danish psychiatrist, the late Eric Stromgren (4), visited the United States for the first time in 1948, most American psychiatrists would not discuss "the possibility of genetic contributions to the etiology of mental disorders. Psychoanalysis had become a religion." But there was more to it than the sway of analysis. After World War II, "genetics had become a dirty word" because of Nazi racial policy. The pendulum has swung back with a vengeance. Today, just about every issue of major psychiatric research journals is replete with studies of the genetic basis of psychopathology (5), studies that take the environment for a given rather than as an independent variable.

The knowledge that disease runs in families has pragmatic value for the clinician and the public health officer when mechanisms have been identified. Dissecting out the key pathways through which nature and nurture interact makes all the difference. Control of phenylketonuria (PKU) followed on the discovery that classic PKU results from an inability to produce phenylalanine hydroxylase because of a genetic defect. Any one of some 250 mutations in the gene can cause the abnormality, but what matters is the final common pathway: an enzymatic block. Affected individuals are unable to convert phenylalanine to tyrosine; as a result, precursors accumulate and damage the brain. Once screening was made feasible by the discovery of a low-cost method for measuring blood phenylalanine, establishing public health programs for early detection, followed by the provision of a low-phenylalanine diet to affected children, permitted clinical disease to be prevented. Although PKU is an inherited disease and although the underlying metabolic defect remains uncorrected, a change in nurture—a special diet—permits normal, or near normal, development.

Nature and nurture stand in reciprocity, not opposition. Offspring inherit, along with their parents' genes, their parents, their peers, and the places they inhabit. Meredith West and Andrew King (6), of the University of Indiana, coined the term "ontogenetic niche" to emphasize that the organism develops within an ecological and social setting that, like its genes, it shares with its parents. The word "niche" stems from the Latin for nest. It helps us recognize that neighborhood and neighbors matter along with parents and siblings. The ontogenetic niche is a legacy that guides development, a crucial link between parents and offspring, an envelope of life chances. Replacing Galton's jingle "nature and nurture" with "nature, niche, and nurture" emphasizes the conjunctions rather than the oppositions between nature and nurture in shaping the developmental process.

Nature and nurture jointly mold brain structure. The ground plan of the central nervous system (CNS) is specified in the human genome, but the precise details of connectivity result from activity-dependent competition between presynaptic axons for common postsynaptic target neurons. Because the environment that is typical for the species—the niche, if you will—reliably supplies the necessary input for connectivity, major CNS structures are as uniform as if all the wiring plans preexisted in the genome. That, however, is not the case; it would be "wasteful" if were so, as François Jacob has pointed out: nature, the tinkerer, reuses what is already at hand (7). The predictable regularities of the system result from interactions among the organ systems of the embryo, between the embryo and its uterine environment, and between the organism and its postnatal niche (8).

As to the first, interactions between simultaneously developing organ systems, consider the formation of the ocular alternation layers in the geniculate nuclei. In the early stages of geniculate formation, processes from both retinas enter the nucleus and intermingle with one another. The separation of layers for each eye, essential for normal function, depends upon asynchronous waves of electrical activity from within retinal ganglion cells, transmitted over their axons. These spontaneous electrical storms arise from the instability of retinal cell membranes. But if this random retinal activity is abolished by the experimenter, separation of inputs does not occur (9). It is neither the genetic instructions governing the anatomy of the retina nor those governing the anatomy of the geniculate that specify the ocular alternation layers—it is the interaction between retina and geniculate during embryogenesis that bring it about.

As to the second, interactions between the embryo and its uterine environment, Gandelman et al. (10) have shown that female fetuses nidated next to male rather than female fetuses in utero exhibit subtle differences in morphology and behavior. Relatively mild stress during pregnancy affects birth weight and neuromotor maturation in rhesus monkey infants (11). Human auditory learning begins before birth. The infant in utero hears its mother's voice repeatedly; on testing after birth, it is able to discriminate that voice from other female voices (12). The infant has learned more than voice recognition. It has begun to learn the prosody of its parents' tongue. Four-day-old French infants will suck harder in order to hear a recording of spoken French instead of Russian because of in utero auditory experience (13).

As to the third, interactions between organism and environment, the ocular dominance columns in the occipital cortex come into being as a result of postnatal stimulation (14). Both eyes of the newborn must receive precisely focused stimulation from the visual environment during the early months of postnatal life to fine-tune the final structure. If one eye is occluded by an opaque cover, or if its acuity is blurred by a translucent cover, or if it is made strabismic by severing the extraocular muscles, the unimpaired eye "captures" most columns in the absence of competition from the deprived eye. The change is irreversible if occlusion is maintained throughout a sensitive period. Deprivation need not be total; unrecognized and uncorrected human astigmatism leads to permanent deficits in visual acuity in the abnormal meridional orientations. Human amblyopia, in which there are incongruent visual images from the two eyes, results in permanent loss of effective vision from the unused eye if not corrected within the first 5 years of life.

The substrate that makes brain structure responsive to the environmental inputs is the overproduction of brain cells and synapses by a factor of two or more. Which cells live and which cells die is determined by the amount and consistency of stimulation they receive. Interaction between organism and environment results from patterned neuronal activity, activity that determines which synapses will be preserved (15).

Greenough and Black (16) have proposed two mechanisms by which new synapses result from experience. In the first, which they term experience-expectant synaptogenesis, synapses form after minimal experience; the prototype is the development of stereoscopy. It requires normal visual input but, because that input is an expectable part of the niche, synapses are preset to require minimal reinforcement. The second mechanism, experience-dependent synaptogenesis, optimizes adaptation to relatively unique features of the environment, features that vary from one environment to another and must be learned. The substrate of experience-expectant learning is the pruning back of synapses without appropriate excitation. Experience-dependent neural activity selects a functionally appropriate subset out of the abundant connections present in the infant brain.

The evidence for this structure—function relationship is by now extensive. Rodents reared in "enriched environments" (that is, large cages with many objects to manipulate) show thicker patterns of interconnection in the cerebral cortex than do those confined in small bare cages (17). Rats forced to perform skilled motor acts have more synapses in the cerebellum than do animals performing the same amount of motor activity on an exercise wheel (16).

Structural differences based on differences in experience are found in humans. Schlaug and colleagues (18) mapped the cortex of normal adults. They compared magnetoencephalographic recordings from experienced violinists with those from nonmusicians and found a substantially larger cortical representation of the fingers of the left hand (the one used to play the strings) than of the fingers of the right (or bowing) arm and more brain area dedicated to representation of fingers in the musicians than in the corresponding recordings from the nonmusicians. A second example of the relationship among brain structure, musical ability, and experience is that the planum temporale is larger on the left than on the right in the musicians; the asymmetry is most marked in those with perfect pitch (19). The cortex has a remarkable capacity for remodeling after environmental change. This is strikingly evident in a study by Sterr et al. (20). By using magnetic source imaging, Sterr et al. contrasted the cortical somatosensory representation of the fingers in blind braille readers who read with three fingers on both hands simultaneously with that for one-finger braille readers and for sighted nonbraille reading subjects. The investigators found a substantial enlargement of hand representation in the three-finger Braille readers, accompanied by changes in the topographical arrangements of finger representation on the postcentral gyrus. This, in turn, was associated with errors in identifying the finger being touched during tactile sensory-threshold determinations. Loss of sensory input resulting from surgery is followed by dynamic changes; intact areas enlarge their representation at the expense of the cortex from which innervation has been removed (21—25). Does this phenomenon, analogous to disuse atrophy, have any clinical significance? Population studies of Alzheimer's disease (AD) suggest that it may.

AD is a multifactorial disease. On the one hand, defects in at least 3 genes, the amyloid precursor protein gene on chromosome 21, and presenilin genes 1 on chromosome 14 and 2 on chromosome 1, are associated with early-onset AD. Most cases, however, are not familial. In late-onset AD, a far more common disorder, one of the four alleles of the gene for apolipoprotein E (APOE) on chromosome 19, APOE 4, is associated with a fourfold increased risk; this association holds for multiple ethnic populations worldwide (26), but not for the African Americans and Hispanics studied in New York City (27). Thus, the allele is neither necessary nor sufficient to cause the disease; what it does is increase risk. Half of AD patients do not carry this allele, and only one in six have two copies (28).

Whatever genetic epidemiology may ultimately reveal, population studies in the United States, France, Italy, Sweden, Finland, Israel, and China reveal a robust inverse correlation between the amount of schooling received in youth and the prevalence of dementia in old age (29). There are as yet no data on interactions between APOE 4 and educational level; that is, does APOE 4 differentially increase risk for AD among the less educated persons who carry the allele in contrast to those who do not?

In the Manhattan Aging Project (sample size 1,500), prevalence rates for dementia among those 85 and older were 60% among Hispanics, 55% among African Americans, and 36% among whites. A similar ratio was found among 75- to 84-year-olds: 30% to 22% to 10% for each ethnic group, respectively (30). However, the apparent relationship between race/ethnicity and dementia was accounted for by differences in educational level and age stratification among the three demographic groups. Multivariant regression analysis demonstrates that, once the effects of age (odds ratio 6.1) and educational level (odds ratio 4.3) were accounted for, neither ethnicity nor gender was significantly associated with rates for dementia. However, in a second report by the Columbia investigators (27), African Americans and Hispanics were found to have an increased frequency of AD with age, compared with whites, regardless of APOE phenotype.

The investigators set forth three hypotheses for effect of education. First, is selection at work? That is, do persons who will develop dementia in late life have reduced cognitive reserve early in life that will lead them to poor performance and school dropout? There is one piece of evidence compatible with this view. Snowden and his co-workers (31) examined the autobiographical essays written at a mean age of 22 by women whose cognitive performance was assessed almost 60 years later. Those whose essays had been characterized by "low idea density" had a tenfold greater risk of poor cognitive function in old age. This sample (nuns from the School Sisters of Notre Dame religious congregation) is unique; the study calls out for replication because its implications are so important, but can comparable data be found anywhere?

Second, is poor education a proxy for the repeated episodes of malnutrition, trauma, alcohol abuse, poor health care, and so on associated with poverty that damage the brain progressively? Third, does the intellectual stimulation provided by school lead to increased synaptic density during development, as is the case with animals reared in "stimulus-rich" environments? Could additional "brain reserve" delay the appearance of clinical symptoms even after amyloid protein is deposited in the brain? Hypothesis Three has a presumptive public health implication—namely, the importance of the quality and duration of education. Persons with graduate degrees are at lower risk for AD. Why? Is it because of education that risk is lower? To assert that the relationship is causal is to extrapolate well beyond the data. But what is to be lost by promoting enhanced public education as sound policy for senility prevention? At the least, we Americans will have become more competent before we become senile!

The impact of neighbors and neighborhood as niche is stunningly illustrated in the findings of The Project on Human Development in Chicago Neighborhoods (32,33). Tony Earls and his colleagues knew—and confirmed—that certain characteristics of neighborhood structure—the concentration of poverty, the extent of ghettoization, residential instability—account for a significant amount of the variance in adolescent antisocial behavior. However, what the researchers were able to show by the use of sophisticated statistical methods is that, after adjusting for prior levels of neighborhood crime, informal social control emerged as a significant deterrent to adolescent delinquency. "Informal social control" refers to the likelihood that adults in the community will monitor spontaneous children's play groups, intervene to prevent truancy and street-corner hanging out by teenagers, and confront persons misusing or disturbing public space. Further, informal social control reflects the ability of cohesive communities to demand needed resources from city authorities for police patrols, fire stations, garbage collection, and housing-code enforcement. The importance of this power is apparent from the correlation between abandoned housing, burned-out buildings, graffiti and litter in an area, and more serious crime.

"Collective efficacy" is the term proposed for the social cohesion among neighbors willing to act on behalf of the common good. Unstable and poverty-stricken neighborhoods with high concentrations of recent immigrants display low collective efficacy. In turn, low efficacy itself mediates a substantial part of the association between disadvantage and violence. The ecology of neighbors and neighborhood interacts with family characteristics to determine behavioral outcomes.

Turn now from cognitive to affective development. The dependence of human infants on parenting for sheer survival creates the social context in which we become human. Not only does the growth of the child's social intelligence result from such dyad relationships, but so also does the organization of its neuroendocrine axis. The multiple sensory inputs that accompany parenting—warmth, odor, touch, sound, proprioception, vestibular stimulation—all influence the hypothalamic-pituitary adrenal system. Infant homeostasis is the outcome of a collaborative process. Infant body temperatures are regulated by caregivers when they respond to signals such as crying or changes in color by holding the infant more closely. Maternal touch and warmth modify infant growth-hormone production.

Animal and human studies demonstrate the enduring impact of maternal—infant interaction. Early experience modulates the aging of the brain (34). The amount of tactile stimulation a rodent pup receives from its mother by licking before it is weaned modulates the cortisol response to stress. Pups that are handled by the experimenter or licked more often by their dams have more glucocorticoid receptors in the hippocampus and hence better regulation of hormone levels. Less-stimulated pups produce more cortisol in response to stress, exhibit early degeneration in the hippocampus, and display premature memory loss as they age (35,36). Monkeys reared in social isolation not only show the severe emotional dysregulation Harry Harlow demonstrated many years ago, but also display corresponding neuroanatomical abnormalities in the hippocampal formation (37). In human subjects, persistent high cortisol levels over a 5-year period correlate with a significant 14% decrease in the volume of the hippocampus compared with low-cortisol control subjects (38).

How far do psychiatric disorders embody the brain structure—function relationships emphasized by this developmental analysis? Answers are only beginning to come in, but they are fascinating. The data on familial aggregation in depression argue compellingly for the genetic transmission of susceptibility. At the same time, the evidence for the importance of life events in precipitating depressive episodes is equally compelling (39—41). Perhaps most remarkable of all to a generation raised on psychopharmacology have been the multiple trials comparing cognitive therapy, interpersonal therapy, and antidepressant medication, trials that demonstrate similar outcomes from drugs and from the two psychotherapies (42—44). Is combination treatment superior to either alone? The jury is still out.

Neither depression nor its subtypes are pure entities. Clinicians undoubtedly confront "phenocopies" of several "genotypes." In a United Kingdom—United States collaborative study, significantly more patients classified as "nonendogenous" than those diagnosed as "endogenous" had experienced a severe life stress in the 6 months before the onset of clinical depression (45,46). Epidemiologic studies demonstrate secular trends toward increasing rates of major depression, findings that defy genetic explanation (47).

Anatomical data are beginning to weigh in. By using positron emission tomography to measure cerebral blood flow and glucose metabolism and using magnetic resonance imagings (MRIs) for anatomical localization, Drevets and his collaborators (48) identified an area of abnormally decreased activity in the prefrontal cortex ventral to the genu of the corpus callosum in familial unipolar and bipolar depressed patients. MRIs in such patients demonstrate reductions in gray-matter volume in the area corresponding to the metabolic changes. What makes this finding exciting is that somewhat the same region has been implicated in the mediation of emotional and autonomic responses to provocative stimuli; patients with subgenual lesions demonstrate abnormal autonomic responses to emotional experiences, inability to experience the appropriate emotion, and impaired understanding of anticipated consequences (49).

The relationship between brain and behavioral abnormalities is even more provocative in obsessive-compulsive disorder (OCD). The symptoms of OCD are associated with changes in cerebral metabolism in the basal ganglia, the limbic system, and cortical projections from both (50). Yet symptoms are reduced by behavior therapy about as often as they are by serotonin reuptake inhibitors. When the patient improves, what happens to the abnormal brain metabolism? The improvement, whether by drug or psychotherapy, is associated with a relative "normalization" (51). Schwartz et al. (52) obtained similar (but not identical) changes in brain metabolism in a replication after successful behavior therapy. Starting from the opposite direction (symptom provocation), Rauch et al. (53) have demonstrated that psychological stimuli that arouse obsessional thoughts in particular patients lead to increases in regional blood flow.

There is no mind without brain; structure and function are inextricably intertwined.

Nine years ago, Sam Guze (54), in an address to the Royal College of Psychiatry, asked "Biological Psychiatry: Is There Any Other Kind?" The question was rhetorical; Sam's point was that there is not any other kind. He was, and is, right; all psychiatry is biological. Pursuing the same rhetoric, if I ask "Social Psychiatry: Is There Any Other Kind?" or "Psychological Psychiatry: Is There Any Other Kind?", the same answer applies. Psychiatry is at one and the same time social and psychological and biological. The distress and dysfunction that our patients experience occur at all levels of organization (55), though one or another level may be predominant in a given patient with a particular disorder, and a particular level may provide the most effective point for therapeutic leverage. This is precisely the raison d'étre for psychiatry as a medical specialty.

If we allow ourselves to become mere pill pushers—or psychotherapists who fit all patients into one Procrustean bed—or social advocates claiming to have discovered the exclusive road to mental health, we abandon the essence of physicianhood: an awareness of the evidence for existing therapeutic alternatives, a commitment to help the patient choose among them, and the flexibility to fashion solutions in accord with patient preference. To the extent we put technique ahead of patients, we are indeed technicians. Only if we defend our patients—and the integrity of our profession—by insisting on the provision of comprehensive and integrated care, care that is at once biological, social, and psychological will we honor the great traditions of our calling.

This paper was presented as the Harvey Schein Lecture at the annual meeting of the American Association of Directors of Psychiatric Residency Training, Orlando, Florida, January 16, 1998.