Prospective Investigation of Stress Inoculation in Young Monkeys FREE

Emotional and neurobiological responses to psychosocial adversity showstriking individual variation.^1- 4 Somepeople who experience a stressful event develop mental health problems suchas depressive or anxiety disorders, whereas others do not. Researchers havesought to identify attributes associated with resilient or stress-resistantindividuals,^5- 6 with the expectationthat understanding the etiology of resilience could increase prevention ofstress-induced mental health problems by enhancing resistance to stress andadversity.^7- 8

Garmezy et al⁹ formulated 3 conceptualframeworks (compensatory, protective factor, and challenge) to investigatethe impact of stress. To date, research efforts have focused almost exclusivelyon the characteristic-based compensatory and protective factor approaches.On the basis of these 2 frameworks, empirical studies have identified thefollowing 3 categories of characteristics that ameliorate risk status: positivepersonality dispositions, a supportive family, and an extrafamilial supportsystem that reinforces active and successful coping efforts.^6,10

Far less researched, but of equal importance, is the challenge approach.In this process-oriented framework, stress is viewed as a potential enhancerof future competence, provided the type and degree of stress are not excessive.⁹ Severe stress often leads to dysfunction,^11- 14 whereasmoderate stress provides a challenge that, when overcome, produces competencein the management of and increased resistance to future stressful circumstances.^5,15 Variously described as inoculating,^2,16 immunizing,^15,17- 18 steeling,^7,19 toughening,^20- 21 andthriving,²² the notion that prior stressfulexperiences may strengthen an organism's resistance to subsequent stressorshas long been recognized.

Although frequently noted in the resilience literature, empirical investigationof stress inoculation has lagged behind theoretical conjecture. Nevertheless,several studies have documented, often unexpectedly, evidence supporting stressinoculation in humans. In adults, experienced survivors of floods and earthquakesexhibit lower anxiety²³ and less depressedaffect²⁴ after encounters with the same disastercompared with inexperienced survivors. Similarly, survivors of torture withprior psychological preparedness training and previous experience with similartraumatic stressors exhibit fewer symptoms of anxiety, depression, and posttraumaticstress disorder compared with unprepared, inexperienced torture survivors,despite the fact that individuals in the latter group experienced less severetrauma.²⁵ Although some sampling bias may occur(ie, individuals severely affected by the initial stressor may not be includedin studies of repeated exposure), this documented effect can be describedas specific inoculation because stress inoculation and manifest resistanceare associated with the same, or similar, stressor.

Of greater interest are findings that indicate that stressful experiencesmay confer general inoculation or cross-immunization.²⁶ Thatis, exposure to one stressor may strengthen resistance to different stressorsencountered later in life. In children and adolescents, prior stressful eventsare associated with diminished emotional distress associated with hospitaladmission,²⁷ attenuated fearfulness in a day-caresetting,²⁸ and decreased cardiovascular responsesto psychologically stressful laboratory tests (eg, mental arithmetic, videogame performance, and hand submersion in ice water).¹⁶ Inaddition, in adults, women are found to better cope with stressful events(eg, spousal loss, major accident, and illness) if they previously experiencedand successfully coped with stressful circumstances in childhood.²⁹ In contrast, unsuccessful coping efforts often leadto deleterious outcomes.¹

Despite their mostly retrospective and correlational nature, these reportsprovide important socioemotional evidence that documents the existence ofstress inoculation in humans. Although the neurobiological underpinnings ofthis effect are largely unknown, data from developmental studies of primatesby our group^30- 31 suggest thatthe hypothalamic-pituitary-adrenal (HPA) axis may provide a neural basis forprogramming stress resistance in the developing child. In human and nonhumanprimates, psychosocial stressors are perceived by the brain, which activateshypothalamic release of corticotropin releasing factor, stimulates pituitaryrelease of corticotropin (ACTH), and induces glucocorticoid (eg, cortisol)secretion from the adrenal cortex. During acute or manageable stress, glucocorticoidsexert negative feedback on the brain and pituitary to turn off the HPA stressresponse.³² However, during chronic or unmanageablestress, the HPA axis frequently becomes dysregulated, and this process hasbeen implicated in the pathophysiology of depression and anxiety disorders.^33- 34

It is now well established that chronic stress experienced during childhoodoften leads to impaired acquisition of appropriate coping skills, heightenedHPA responsivity to stress, and increased risk for the development of adultmental health problems.^35- 39 However,moderate exposure (ie, inoculation) to stressors early in life may protectagainst these deleterious effects. In particular, stress resistance may occurthrough manageable exposure to moderately stressful events that temporarilyactivate the HPA axis but permanently alter neuroendocrine sensitivity tosubsequent stressors by fostering the acquisition of coping strategies thatsafeguard against the development of stress-related disorders.^7,40 Becauseopportunities for prospective, controlled studies are limited in children,we prospectively tested the general inoculation theory of stress resistancein a primate model in the following experiments.

Twenty squirrel monkeys (Saimiri sciureus)of Guyanese origin, born at Stanford University, Stanford, Calif, served assubjects. Monkeys received dye marks and number tags worn on necklaces tofacilitate easy identification. Natal group composition was primarily determinedby birth dates to minimize developmental differences between infants and geneticrelatedness (ie, paternal half siblings were not assigned to the same natalgroup). Whenever possible, sex assignment was balanced across natal groups.

Subjects were housed in 1.8 × 1.2 × 1.8-m wire-mesh cagesin groups of 3 or 4 mother-infant pairs. Animals were housed and tested inclimate-controlled rooms on a light-dark cycle (12:12-hour ratio) with anambient temperature of 26°C. Monkeys had ad libitum access to water, commercialmonkey chow, fresh fruits, and vegetables. Monkey cages were cleaned daily.A sliding door in each home cage provided access to a small, portable capturecage. Monkeys were pretrained to enter the capture cage on voice commandsto facilitate the experimental manipulations. All procedures were approvedby the Stanford University Administrative Panel on Laboratory Animal Care.

In nature, squirrel monkeys locomote independently by postnatal week5, forage successfully at postnatal week 7, and are weaned by postnatal week16.⁴¹ By postnatal week 17, free-living monkeysare biologically independent,⁴¹ although theyremain emotionally attached to their mothers as assessed by behavioral andpituitary-adrenal responses to maternal separation.^42- 43 Inthis experiment, subjects remained undisturbed in natal groups until postnatalweek 17, at which time natal groups were randomly assigned to 1 of 2 experimentalconditions. In one condition, 11 offspring (7 female and 4 male) from 4 natalgroups were exposed to an intermittent stress inoculation (IS) protocol. Onceevery week from postnatal weeks 17 through 27, each young monkey was removedfrom the natal group for 1 hour, placed in a cage (46 × 46 × 46cm) adjacent to unfamiliar adult monkeys in a different room, and temporarilydeprived of all forms of contact with the natal group. Intermittent separationsinduce isolation calls, locomotor agitation, and acute increases in cortisollevels, which return to baseline soon after reunion.^44- 46 Nomore than 1 monkey from each natal group was separated on a given day. Inthe other condition, 9 offspring (8 female and 1 male) from 3 natal groupsremained undisturbed as nonstressed (NS) controls. Although sample sizes inthis experiment are inadequate to detect the small effects typically foundin clinical trials, samples of 9 to 11 monkeys per treatment condition arecommonly used in preclinical studies to detect large effects on behavioraland neuroendocrine measures.^45,47- 48

At postnatal week 35, each mother-offspring dyad was removed from thenatal group and transported to a novel test cage (60 × 60 × 90cm) in an unfamiliar room that did not contain other monkeys. The cage androom used for testing were different from those used for the IS protocol.The novel environment contained polyvinyl chloride perches and a variety offamiliar and unfamiliar objects. Monkeys also had access to biscuits, cantaloupe,marshmallows, and water throughout testing. All food was replenished and thecages were cleaned between tests. Tests lasted 30 minutes, and each mother-offspringdyad underwent testing once a day for 5 consecutive days. All tests occurredbetween 3 and 6 PM, and testing occurred at the same time eachday for a given dyad. Each week, 3 to 4 dyads underwent testing (no two fromthe same natal group), and the order of testing for IS and NS offspring wasevenly distributed across daily and weekly schedules.

Using a computer-aided recording program, one of us (K.J.P.) collectedbehavior. Offspring anxiety was inferentially assessed by dorsal contact (ie,the species-typical offspring riding posture on the mother's shoulders andupper back), exploration of unfamiliar objects, and food consumption. Thefollowing measures were analyzed: (1) latency to terminate dorsal contact,(2) latency to explore the first object, (3) total familiar and unfamiliarobject exploration counts (scored each time a monkey mouthed or touched anobject), and (4) total food consumption counts (scored each time a monkeyplaced food in its mouth). The number and quality of social transactions betweenmothers and offspring were also recorded as previously described.^49- 50 Briefly, transactions were initiatedby attempts to change the immediate state of association between the motherand offspring by means of breaking or making dorsal contact initiatives, ormaking or breaking affiliative contact (ie, side-by-side huddling) initiatives.Successful attempts were scored whenever initiatives were accommodated bythe target. Failed attempts were scored whenever initiatives were overtlyresisted by the target (eg, an infant tried to move off the mother's backand the mother prevented this initiative by means of species-typical "shoulderscooping").

Blood samples were collected from young monkeys 10 days before and 10days after experimental manipulations to establish baseline levels of ACTHand cortisol. Blood samples were also collected immediately after testingat the first and last sessions to examine stress hormone concentrations afterexperimental manipulation. Blood sampling occurred between 3:30 and 6 PM to control for circadian variation.⁵¹ Sampleswere collected from manually restrained monkeys while blood (0.8 mL) was drawnby means of femoral venipuncture with single-use polypropylene syringes containing20 µL of EDTA. Sixty-eight (85%) of the 80 plasma samples were collectedwithin 180 seconds from cage entry (median, 122 seconds; range, 63-403 seconds),and all but 4 samples (5%) were collected within 4 minutes. Each sample wasimmediately centrifuged at 4°C, and the plasma fraction was stored at−80°C. Hormone levels were measured in duplicate using commerciallyprepared ACTH (DiaSorin, Inc, Stillwater, Minn) and cortisol (Diagnostic ProductsCorporation, Los Angeles, Calif) radioimmunoassays as previously described.⁵² The intra-assay and interassay coefficients of variationwere 9.8% and 11.5%, respectively, for ACTH, and 6.0% and 5.5%, respectively,for cortisol. Assay sensitivity was 7 pg/mL (1.5 pmol/L) for ACTH and 3 µg/dL(82.8 nmol/L) for cortisol.

To assess the generality of rearing effects in a different test setting,a stainless steel wire-mesh box (46 × 46 × 46 cm) was attachedat floor level to each group's home cage during postnatal week 50. Free accessto the box was provided through an interconnecting opening (30 × 30cm). On initial presentation, the box was anxiety provoking to all monkeys,as it instigated robust alarm calling. Before testing, monkeys were acclimatedto the box for 1 h/d for 7 consecutive days, by which time alarm calling hadceased. Each natal group then underwent testing for 5 min/d for 4 consecutivedays. Test sessions occurred between 2:45 and 4:30 PM, andthe order of testing for the IS and NS natal groups was evenly distributedacross daily schedules.

Behavior was collected by one of us (K.J.P.) using the computer-aidedrecording program. Offspring anxiety was inferentially assessed by means ofvoluntary exploration and play behavior in the wire-mesh box. The followingbehavior measures were analyzed for each monkey: (1) entry latency (scoredwhen the monkey's entire body, excluding the tail, first entered the box),(2) total number of entries, and (3) duration of time spent in the box (thelatter two scored when the monkey's entire body, excluding the tail, was inthe box). The presence of play behavior was scored when monkeys wrestled,exhibited exaggerated leaping and chasing, and/or carried toys into the box.⁵³

We assessed the effect of postnatal rearing experience (IS vs NS) onbehavior and hormone measurements with repeated-measures analysis of varianceusing least squares estimates from general linear models in the MGLH moduleof Systat (Systat Software, Inc, Point Richmond, Calif). Postnatal rearingcondition was considered a between-subjects factor, and test session was consideredthe repeated, within-subjects factor. We used the Geisser-Greenhouse correctionto adjust for multiple comparisons across the repeated test-block factor.⁵⁴ For latency to terminate dorsal contact and totaldorsal contact duration, the offspring's body weight, expressed as a percentageof its mother's, was used as a covariate to examine whether heavier offspringwere carried for less time independent of rearing conditions. Hormone valueswere logarithmically transformed to stabilize the variance across groups andto satisfy the equal variance assumptions of parametric statistical tests.We also analyzed mother-offspring transactions (summed across sessions) usingindependent, 2-tailed t tests. The presence or absenceof play behavior for each test session was analyzed using Pearson χ² tests, and α error was adjusted to protect against multiplecomparisons (ie, differences were considered significant when P<.0125). Descriptive statistics are presented as mean ±SEM.

Consistent with the stress inoculation effect, the novel environmentwas less anxiety provoking for IS offspring as rearing-related differencesin dorsal contact termination latency, object exploration latency, unfamiliarbut not familiar object exploration, and food consumption were found. Duringthe first test session, 18 mother-offspring pairs (90%) entered the novelenvironment with the offspring clinging to the mother's back. At 35 weeksof age, young monkeys rarely cling to their mothers other than during timesof significant emotional distress. However, across consecutive days, the novelenvironment was less anxiety provoking for IS mother-offspring pairs, as ISpairs more rapidly terminated dorsal contact (F_4,68 = 4.56 [P = .02]) (Figure 1).These observed differences in dorsal contact termination were not due to differencesin offspring weight (IS and NS offspring weights did not differ), and heavieroffspring did not hasten dorsal contact termination.

The IS offspring were also faster to explore a first object than NSoffspring (F_1,18 = 7.21 [P = .02]), andthis effect was time dependent (F_4,72 = 3.21 [P = .03]) (Figure 1). Thisrearing condition by test session interaction occurred because NS offspringdemonstrated consistently high exploration latencies across test sessions,whereas IS offspring exhibited decreased first-object exploration latenciesacross consecutive test days (F_1,10 = 20.02 [P = .001]).

The IS and NS offspring exhibited similar amounts of familiar objectexploration and did not differ on familiar object exploration counts (Figure 2). However, IS and NS offspring demonstratedan overall difference in exploration of unfamiliar objects (F_1,18 =5.59 [P = .03]), and this effect was time dependent(F_4,72 = 4.92 [P = .005]) (Figure 2). Analysis within rearing condition showed that NS offspringremained reluctant to explore unfamiliar objects over time, whereas IS offspringdemonstrated an increase in unfamiliar object exploration across consecutivetest sessions (F_1,10 = 14.47 [P = .003]).

The IS and NS offspring also differed in food consumption counts overall(F_1,18 = 4.59 [P = .046]) and by testsession (F_4,72 = 3.77 [P = .02]) (Figure 2). As observed previously with unfamiliarobject exploration, NS offspring demonstrated consistently low food consumptionacross test sessions, whereas IS offspring showed an increase in food consumptionover consecutive test sessions (F_1,10 = 16.39 [P = .002]).

Mothers and offspring engaged in a total of 6195 social transactionsacross all test sessions. Transactions consisted of the following initiatives:breaking dorsal contact (10%), making dorsal contact (4%), making affiliativecontact (49%), and breaking affiliative contact (37%).

Mothers and offspring did not differ on attempts to terminate dorsalcontact, and there were no effects of rearing condition on offspring- or mother-initiatedattempts. Although IS and NS offspring did not differ on the total numberof times they attempted to leave their mothers, attempts by the IS offspringto break dorsal contact were more frequently accommodated by their mothers(93% ± 5%) than were those of NS offspring (45% ± 17%) (t₁₂ = 3.10 [P = .009]).

A total of 232 attempts to make dorsal contact were recorded after initialdorsal contact was broken. Most of these attempts (82%) were initiated byoffspring, and there were no effects of rearing condition on offspring- ormother-initiated dorsal contact attempts. Of the offspring that attemptedto regain the species-typical riding posture during testing, 37% ±8% of their initiatives to make dorsal contact were accommodated by theirmothers. The IS and NS offspring did not differ in the proportion of initiativesthat were accommodated.

As with dorsal contact attempts, offspring also made more affiliativeovertures than did mothers, and offspring initiatives represented 82% of totalaffiliative overtures. Although IS and NS offspring did not differ on totalaffiliative overtures, they differed on the percentage of time their overtureswere accepted (78% ± 5% for IS vs 61% ± 6% for NS; t₁₅ = 2.16 [P = .047]). Offspringand mothers did not differ on who terminated side-by-side affiliative contact,and there were no rearing condition effects on offspring- or mother-initiatedovertures. All targets accommodated termination of affiliative contact.

Main effects of rearing condition for ACTH (F_2,36 = 5.76[P = .01]) and cortisol (F_2,36 = 5.77[P = .009]) concentrations were found (Figure 3). At baseline, IS offspring had significantly lower ACTH(F_1,18 = 15.68 [P = .001]) and cortisol(F_1,18 = 16.47 [P = .001]) levels thandid NS offspring. Although the novel environment increased stress hormonelevels in all monkeys, IS compared with NS offspring demonstrated lower stresshormone levels on the first day of testing (for ACTH, F_1,18 = 4.94[P = .04]; for cortisol, F_1,18 = 5.47[P = .03]), consistent with a stress inoculationeffect. By the fifth day of testing, however, rearing-related differencesin stress hormone levels were no longer discerned.

Like the novel environment, the home-cage test was less anxiety-provokingfor IS compared with NS offspring (Figure4). Specifically, IS offspring exhibited decreased exploration latenciescompared with NS offspring (F_1,18 = 107.27 [P<.001]). The IS monkeys also more frequently entered (F_1,18 = 76.27 [P<.001]) and spent more time in(F_1,18 = 36.71 [P<.001]) the wire-meshbox compared with NS offspring. The IS natal groups also exhibited more playbehavior in the box than did NS offspring (χ²₁ =7.00 [P = .008] for each test session), such thatall IS but no NS natal groups played in the box during testing. Mothers didnot enter the test box, and no mother was observed to physically or vocallyrestrict offspring entry into the box.

Results from these experiments provide the first prospective evidenceof stress inoculation in primates. Specifically, squirrel monkeys previouslyinoculated with intermittent periods of moderate stress (IS condition) duringearly development demonstrated diminished anxiety in 2 different experimentalparadigms compared with NS control monkeys. These findings in monkeys parallelretrospective data in humans^{16,23- 25,27- 29} andsuggest that controlled exposure to moderately stressful events early in lifemay produce stress resistance in developing humans and nonhuman primates.

By postnatal week 35, young squirrel monkeys in their home cage rarelycling to their mothers and increasingly initiate social and environmentalexploration.⁵⁵ These developmental transitionsfrom filial, mother-directed behavior (eg, clinging and nursing) to exploitative,other-directed behavior (eg, visual exploration, object and social play, investigatoryactivities) promote species-typical learning opportunities and support theyoung primate's transition to independence.^56- 57 Nevertheless,throughout development, mother-directed contact seeking occurs in responseto highly stimulating events and typically leads to reduced arousal.⁵⁸ In experiment 1, IS and NS offspring remained inclose dorsal contact with their mothers during the first 2 days of testing(Figure 1), which indicates thatthe novel test environment was a source of significant emotional distressfor offspring from both rearing conditions.

During consecutive test days, however, the novel environment was lessanxiety provoking for IS offspring, as they spent less time clinging to theirmothers, initiated object exploration more rapidly, investigated more unfamiliar(but not familiar) objects, and consumed more food than did NS offspring (Figure 1 and Figure 2). Similarly, in experiment 2, IS offspring at 1 year ofage more rapidly acclimated to the anxiety-provoking test apparatus becausethey were faster to enter the wire-mesh box and spent more time exploringit (Figure 4). Moreover, IS butnot NS monkeys played in the box throughout testing. Insofar as exploitativebehavior typically occurs under conditions of low to moderate arousal,⁵⁶ these findings suggest that prior experience withmoderately stressful events early in life results in more effective arousalregulation when emotionally challenging circumstances are again encountered.

In bonnet macaque⁴⁸ and human⁵⁹ infants, behavioral responses to emotionally challengingcircumstances are also influenced by the quality of the mother-infant relationship.In experiment 1, mothers of IS and NS offspring differed on several measuresof maternal accommodation of offspring-initiated overtures. First, IS motherswere more permissive of offspring-initiated termination of dorsal contactthan were NS mothers. In Japanese macaques⁴⁷ andvervet monkeys,⁶⁰ a less permissive (ie, restrictive)maternal style is associated with decreased offspring exploration. Consequently,by more frequently accommodating exploitative tendencies in their offspring,IS mothers may have better facilitated offspring exploration of the novelenvironment compared with more restrictive and less accommodating NS mothers.Second, similar to human mothers of securely attached infants,^61- 62 ISmothers more frequently accommodated offspring-initiated bouts of brief side-by-sidehuddling that may have buffered offspring from overarousal by providing asecure base from which to explore. Why IS and NS mothers differ in measuresof maternal responsivity is unclear, but one possible explanation is thatrepeated exposure to highly aroused offspring after each separation periodfacilitated the development of increasingly effective maternal means by whichto assess and reduce offspring arousal. Over time, repeated exposure to manageablestressors combined with the presence of a reassuring and accommodating motherat reunion may have strengthened emotional self-regulation and promoted exploitativetendencies in developing IS monkeys. Because exploitative actions broadenthe range and complexity of situations that offspring tolerate and approachfreely,^56- 57 such cumulative experiencemay provide a socioemotional foundation for the development of stress resistance.

Maternal differences also present a potential confound, as it is unclearwhether rearing-related differences in offspring anxiety are the product ofstress inoculation or reflect maternally driven changes in offspring behavior.In experiment 2, when there was little direct maternal influence on offspringbehavior because of offspring maturity, IS offspring nevertheless continuedto demonstrate diminished anxiety compared with NS offspring. Moreover, additionaldata from these monkeys support the notion that rearing-related differencesare sustained well after removal of the mothers, as juvenile IS monkeys demonstrateenhanced cognitive and emotional control compared with NS monkeys on preliminarytests of prefrontal cortical function.⁶³

Another issue that warrants comment is whether these findings reflecta specific or a general stress inoculation effect. Experiment 1 and the inoculationprocedure shared several similarities, including separation from the natalgroup and confinement in an unfamiliar environment. Our experimental resultsmay thus reflect increased familiarization by IS monkeys with these commonprocedures. However, there were considerable differences between experiment2 and the inoculation procedure, including voluntary vs involuntary participation,present vs absent bnatal group, and home vs novel-cage environment. Despitethese differences, the home-cage test was less anxiety provoking for IS thanNS offspring, which suggests a general, rather than specific, stress inoculationeffect.

In addition to rearing-related differences in socioemotional behavior,IS monkeys exhibited lower basal and stress-induced increases in plasma concentrationsof ACTH and cortisol (Figure 3).These results parallel findings of studies of other monkey cohorts that examinedthe effects of early maternal availability on offspring development.^30- 31 In these previous experiments fromour group, we unexpectedly found that intermittently separated offspring respondedto the removal of mothers at weaning with smaller elevations in plasma cortisolconcentrations, fewer distress calls, and more time spent near peers relativeto offspring raised in high- or low-demand foraging conditions.³⁰ Moreover,in early adulthood, intermittently separated monkeys demonstrated enhancedglucocorticoid negative-feedback sensitivity compared with monkeys from otherrearing conditions.³¹

Data from these previous studies^30- 31 andthose presented herein support a role for the HPA axis as a potential neurobiologicalmediator of stress resistance in monkeys. However, the process by which earlyenvironmental events program stress resistance remains unknown. In rodents,increased maternal care received during infancy is associated with bluntedpituitary-adrenal hormone responses to stressful circumstances in adulthood.This is mediated, in part, by enhanced glucocorticoid negative-feedback regulationresulting from increased glucocorticoid receptor expression in the hippocampus.⁶⁴ As ACTH and cortisol levels are diminished in ISmonkeys, it is likely that, as in rodents, brain mechanisms above the pituitaryenhance the inoculated monkey's ability to more efficiently regulate stresshormone responses to emotionally arousing events.

Similar to securely attached children,⁶⁵ ISoffspring demonstrate low levels of circulating basal cortisol. In contrast,hypercortisolism is often observed in chronically stressed individuals (eg,former Romanian orphans,⁶⁶ children of emotionallyunavailable mothers⁶⁷). Although cortisol playsa critical role in an organism's ability to cope with stressors, the deleteriousconsequences of sustained cortisol overproduction are well documented.⁶⁸ Insofar as high basal cortisol levels associatedwith adverse early experiences permanently alter stress biology and confervulnerability to develop adult-onset depressive and anxiety disorders,^37,39,69 broadly speaking,comparatively low basal cortisol levels may serve a protective function duringdevelopment.

Because encounters with stress and adversity are unavoidable, theoristshave argued that stress resistance cannot reasonably reside in the avoidanceof risk experiences, but rather, in successful engagement with and masteryof them.^4,7 However, it is importantto remember that stressful events, even comparatively mild ones, may stillincrease vulnerability to the effects of subsequent stressors if they supersedethe developing organism's ability to cope with them. For example, in youngmarmoset monkeys,⁷⁰ initiation of a brief,repeated parental separation protocol during postnatal week 1 has stress-sensitizingeffects, whereas a similar protocol initiated during postnatal week 17 insquirrel monkeys, as described herein, has stress-inoculating properties.Thus, as with any developmental event, the type, timing, duration, and severityof a given stressor within a given species are likely to be important factorsin determining whether early experiences ultimately produce a protective ordeleterious outcome.

As with all studies, potential limitations should be considered. First,it was impossible for the observer who recorded behavioral data to be blindto the treatment conditions. However, hormone levels for experiment 1 weredetermined blindly, and robust group differences in the same direction predicteda priori were found. Thus, it seems unlikely that experimenter bias producedthe behavioral differences observed in these experiments. Second, 11 primaryoutcome measures were analyzed in experiment 1, and although almost all treatment-relateddifferences were highly statistically significant, the possibility of false-positiveresults due to compounded α error from multiple analyses should be noted.Last, these studies should be replicated by other researchers studying thisand other species to assess the generalizability of our findings.

Although no known studies have examined pituitary-adrenal responsivityin stress-inoculated humans, prospective evidence from our experiments inmonkeys supports the notion that the HPA axis may provide a neural basis forprogramming stress resistance in the developing child. Moreover, these findingsraise important clinical questions concerning the role of stress managementin therapeutic interventions. However, as with primates, the effects of stressinoculation in developing children are likely to be complex and context dependent.With continued investigation of stress inoculation in primate models, a comprehensiveunderstanding of the neurobiology of stress resistance may ultimately providea foundation for new approaches to the successful treatment and preventionof stress-induced depressive and anxiety disorders.

Correspondence: Karen J. Parker, PhD, Department of Psychiatry andBehavioral Sciences, Stanford University Medical School, 1201 Welch Rd, MSLSRoom P104, Mail Code 5485, Stanford, CA 94305-5485 (kjparker@stanford.edu).

Submitted for publication August 1, 2003; final revision received January30, 2004; accepted February 16, 2004.

Dr Parker was supported by a Stanford University School of MedicineDean's Fellowship and individual postdoctoral National Research Service AwardF32 MH066537 from the National Institute of Mental Health, Bethesda, Md. Thisstudy was supported by grants MH47573 and MH50604 from the National Instituteof Mental Health; grant DA016902 from the National Institute on Drug Abuse,Bethesda; and the Nancy Pritzker Network, New York, NY.

This study was presented at the annual meeting of the Society for BehavioralNeuroendocrinology; June 27, 2003; Cincinnati, Ohio; and the annual meetingof the Society for Neuroscience; November 11, 2003; New Orleans, La.

We thank Helena Kraemer, PhD, for her statistical guidance; David Spiegel,MD, Christopher Hayward, MD, MPH, Paresh Patel, MD, PhD, and 3 anonymous reviewersfor their thoughtful comments on this manuscript; Karan Sundlass for creatingour computer-aided scoring program; and Stuart Anhorn for excellent care ofour animals.