Meaningful differences in ghrelin response that were indepen-dent of changes in actual nutrients have important implicationsgiven recent interest in exploring the physiological and pharma-cological use of gut peptides as a means to manage weight (Bayneset al., 2006; Zigman & Elmquist, 2006). We argue that, much likeplacebo effects, alterations in mindset—what one believes andexpects to be eating—have the potential to elicit a seeminglyinappropriate sense of satiation. This is an intriguing addition tothe growing literature supporting the mindset’s pervasive, yet oftenunexpected, influence on physiological states. Additional researchendeavoring to understand better how psychological factors influ-ence the biological impact of food is particularly warranted. Per-haps if we can begin to approach even the healthiest foods with amindset of indulgence, we will experience the physiological sat-isfaction of having had our cake and eaten it too.

That we obtained these results independent of the intrinsicproperties of food challenges the typical assumption with respectto the physiology of food intake and, in so doing, may provide amissing link in the inexact science of weight and metabolic main-tenance. Although the caloric balance equation seems simple andstraightforward (Ayyad & Andersen, 2000), these findings suggestthat the psychological mindset of sensibility while eating mayactually dampen the effect of ghrelin. Although the effect of suchpsychologically mediated differences on subsequent consumptionor long-term alterations in weight were not measured in thisparticular study, future research on the impact of this phenomenonon metabolic maintenance is warranted. Elevated ghrelin levelscan cause increased body weight and fat gain through increasedcaloric consumption, decreased energy expenditure, and a shiftaway from the metabolic utilization of adiposity as an energysource (Murphy et al., 2006; Zigman & Elmquist, 2003). Ghrelinantagonists produce the opposite effect: increasing energy expen-diture, decreasing food intake, and producing long term fat reduc-tion and weight loss (Castan ̃eda et al., 2010). The relatively flatghrelin profiles in response to consuming the shake in a sensiblemindset may be placing participants in a psychologically challeng-ing state marked by increased appetite and decreased metabol

When participants drank the indulgent shake, they had a signif-icantly steeper decline in ghrelin than when they drank the sensibleshake. The observed pattern of ghrelin response is consistent withwhat one might observe if participants actually consumed bever-ages with differing caloric contents (i.e., high vs. low energyintake; Taheri, Lin, Austin, Young, & Mignot, 2004). However, inthis case the distinctive ghrelin profiles were psychologicallymediated; they were dependent on the perceived expectancies ofthe milkshakes’ nutritional contents as opposed to objective nutri-tional differences.

model analysis of variance (ANOVA) was conducted with shaketype (indulgent, sensi-shake), restrained eating (nonrestrained, re-strained), and order (Session 1, Session 2) included as factors inthe model. For perceived healthiness, these analyses revealed asignificant main effect for type of shake,F(1, 89)
42.50,p.01,2
.33, with no significant main or interaction effects forrestrained eating or for the order in which the shakes were con-sumed. Simple effects tests suggested that participants rated thesensi-shake as significantly healthier than the indulgent shake,t(44)
15.61,p.01. These differences are illustrated in Figure3. There were no significant main or interaction effects of shake,order, or restrained eating on perceived tastiness of the shake.To test the effect of shake condition on ghrelin and hunger, thedata were first assessed using mixed-model ANOVA with time(baseline, anticipatory, postconsumption), shake type (indulgent,sensi-shake) and order (Session 1, Session 2) as fixed factors.Because the model failed to identify significant main effects orinteractions involving order (suggesting that the ordering of thesessions did not elicit any reliable differences), the data werecollapsed over order and analyzed using a 2 (shake type: indulgent,sensi-shake)3 (time: baseline, anticipatory, postconsumption)repeated-measures general linear model (GLM) with restrainedeating as a between-subjects variable. Reflecting the differentpatterns of response hypothesized during the anticipatory andpostconsumption phases, we expected a quadratic shaketimeinteraction effect.For ghrelin, the 2 (shake type: indulgent, sensi-shake)3(time: baseline, anticipatory, postconsumption) repeated-measuresGLM produced a reliable quadratic effect,F(1, 44)
4.36,p
.04,2
.091. To be specific, participants exhibited a steeper risein ghrelin in anticipation of the indulgent shake, followed by asignificantly steeper reduction in this biological marker of hungerafter consuming the shake. When drinking the shake in an indul-gent mindset, participants’ levels of ghrelin reflected a moderatelevel of physiological craving followed by a significant level ofphysiological satiety. On the other hand, when drinking the shakein a sensible mindset, participants exhibited flat or slightly in-creased levels of ghrelin over the course of consumption suggest-ing that, despite consuming the same nutrient contents, they werenot physiologically satisfied. The 2 (restraint: nonrestrained, re-strained)2 (shake type: indulgent, sensi-shake)3 (time:baseline, anticipatory, postconsumption) interaction was not sig-nificant nor was there a significant between-subjects effect ofrestrained eating.Figure 4 provides a graphic representation of the effects ofshake label on ghrelin as a function of mindset. To understandfurther the differences in ghrelin levels between the two types ofshakes at the anticipatory and postconsumption time points, theinteractions were decomposed by conducting separate analyses forthe anticipatory and postconsumption intervals. These analysessuggest that the primary driver of the quadratic effect was theresponse to consuming the shake rather than anticipation of it, thatis, the 2 (shake type: indulgent, sensi-shake)2 (time: anticipa-tory, postprandial) effect was significant,F(1, 44)
5.75,p
.02,2
.12, whereas the 2 (shake type: indulgent, sensi-shake)2(time: baseline, anticipatory) effect was not,F(1, 44)
0.94,p
.34,2
.02. For the measure of hunger, these analyses producedno significant main or interaction effects as a function of shake,time, or restrained eating.

Ghrelin.Ghrelin was measured using a double antibody RIA(GHRT-89HK) with intra-assay variability of 4 to 10%, and inter-assay variability of 4.8 to 12.8% (Millipore; St. Charles, MO).Samples were kept on ice during the collection period after whichthey were spun and plasma was stored at70 °C until they werebatch analyzed. Total amount of blood collected was 90 ccs (45 ccsper visit).Taste ratings.During consumption, participants were askedto comment on various aspects of the milkshake including smell,appearance, and taste as well as enjoyment and healthiness. Re-sponses to these questions were assessed via 100-mm visual ana-logue scales ranging from 0 (not at all) to 100 (extremely).Hunger ratings.Ten minutes prior to each ghrelin measure-ment, participants were asked to rate their subjective feelings ofhunger. Responses to these questions were assessed via 100-mmvisual analogue scales ranging from 0 (not at all) to 100 (ex-tremely).Restrained eating.The Dutch Eating Behavior Question-naire (DEBQ; Van Strien, Frijters, Bergers, & Defares, 1986) wasused to assess dietary restraint. Although this measure also has

Participants were scheduled for two, 2.5-hr sessions at the YaleClinical Research Center Hospital Research Unit (HRU). Thesesessions were exactly 1 week apart, either at 8:00 a.m. or 8:20 a.m.after an overnight fast. At the first session, participants were toldthat the metabolic kitchen at the Yale Center for Clinical Investi-gation was working on designing two different milkshakes withdifferent nutrient contents and that they would taste one milkshakein the first week and another milkshake the following week. Theywere told that the goal of the study was to evaluate whether themilkshakes tasted similar and to examine the body’s reaction to thedifferent nutrients (high vs. low fat, high vs. low sugar, etc.).Unknown to the participants, the contents of the two milkshakeswere identical. However, the labels depicting these beverages differedfrom Time 1 to Time 2: The indulgent condition presented themilkshake as a high fat, high calorie “indulgent” shake; the sensi-shake condition touted the milkshake as a low fat, low calorie “sen-sible” shake. The corresponding labels are presented in Figures 1 and2. Please visit www.GibbsGraphicsArt.com/Crum.html to view thecolor versions of these designs.At each session, an intravenous catheter was placed for blooddrawing, and after a 20-min rest period, the first blood sample wasdrawn, followed by samples taken at 60 and 90 min. During the firstinterval (between 20 and 60 min) participants were asked to view andrate the label of the shake. During the second interval (between 60 and90 min) participants were asked to drink and rate the milkshake. Tocontrol for speed of consumption, participants were instructed toconsume the shake in its entirety within the first 10 min of thisinterval. Order of presentation of the two milkshakes was counterbal-anced so that approximately half (45%) of the participants receivedthe sensi-shake in the first session and half (55%) of the participantsreceived the indulgent shake in the first session.

Participants were recruited through fliers presenting the oppor-tunity to participate in a “Shake Tasting Study” at the Yale Centerfor Clinical Investigation in exchange for $75 for the two 2.5-hrsessions. These fliers were posted around the New Haven com-munity in both on- and off-campus locations in an attempt torecruit a diverse sample. Participants were between the ages of 18and 35, within a normal to overweight range of body mass index(BMI;M
22.5,SD
4.04), and were prescreened for diabetes,pregnancy, chronic medical or psychiatric conditions, and foodallergies to lactose or eggs. Fifty-three participants were recruited;however, two participants did not attend the second session, andfive participants did not complete the preliminary survey. Datawere analyzed using the 46 participants who completed all com-ponents of the study (65% women, 78% student, 22% member ofcommunity; 56% White, 12% African American, 11% AsianAmerican, 10% Hispanic/Latino, and 11% other).

In the past decade, investigators have improved our understand-ing of the molecular mechanisms that control food intake and bodyweight. Central to this line of research has been the identificationand characterization of metabolic signals that serve as fundamentalindexes of energy balance. A hormone that has proved to beparticularly influential is the gut peptide ghrelin. Identified in1999, ghrelin is an essential indicator of energy insufficiency.When energy intake is low or the stomach is empty, ghrelin issecreted from the endocrine cells of the stomach and transported inthe bloodstream to the brain, where it binds with receptors in thearcuate nucleus and the ventromedial hypothalamus to produce thesensation of hunger and motivate consumption. As energy intakeincreases and nutrients are detected in the gastrointestinal tract,ghrelin levels are suppressed, thereby signaling to the brain vianeural and endocrine mechanisms to reduce appetite and increasefeelings of satiety (Baynes, Dhillo, & Bloom, 2006; Murphy,Dhillo, & Bloom, 2006).In principle, the rise and fall of ghrelin occur systematically andin proportion to calories consumed to achieve a healthy metabolicbalance (Zigman & Elmquist, 2003). However the communicationbetween the metabolic and neurological systems is complex. Evensubtle changes can have profound implications for health andhomeostasis (Murphy et al., 2006). For example, among obeseindividuals, the usual postprandial reduction in ghrelin is absent orattenuated suggesting that abnormalities in the gut hormone sig-naling system may be associated with weight gain and obesity(Cummings, 2006). Peripheral or intracereroventricular adminis-tration of ghrelin in both humans and rodents has been shown topromote food intake and body weight gain (e.g., Theander-Carilloet al., 2006; Wren et al., 2001, as cited in Castan ̃eda et al., 2010).In light of the power of beliefs and expectations in affecting otherphysiological processes, we sought to determine whether subtlechanges in the mindset associated with eating might influence therelease of ghrelin in response to food consumption. Considering themoderating influence of restraint on the psychological effects ofeating behavior and satiety (e.g., Heatherton, Polivy, & Herman,1989) and the influence of restrained eating on the ghrelin response(Schur, Cummings, Callahan, & Foster-Schubert, 2008), restrainedeating was also included in the analyses.

tions) is a key component in various domains of health.The mere expectation to heal even in the absence of active phar-maceutical or chemical substances enhances the effect of medica-tion (e.g., Brody, 1980; Price, Finniss, Benedetti, 2008); one’sinterpretation of events despite their objective characteristics de-termines the impact of stress and illness on the body (e.g., Cohen& Williamson, 1991; Park, 2006); and identifying housework as agood source of exercise can elicit corresponding physiologicalbenefits without any changes in actual activity (Crum & Langer,2007). Evidence continues to point to the idea that one’s state ofmind influences the body, and we cannot easily separate theinterdependence of mind and body (Langer, 2009).

Yale UniversityObjective:To test whether physiological satiation as measured by the gut peptide ghrelin may varydepending on the mindset in which one approaches consumption of food.Methods:On 2 separateoccasions, participants (n
46) consumed a 380-calorie milkshake under the pretense that it was eithera 620-calorie “indulgent” shake or a 140-calorie “sensible” shake. Ghrelin was measured via intravenousblood samples at 3 time points: baseline (20 min), anticipatory (60 min), and postconsumption (90 min).During the first interval (between 20 and 60 min) participants were asked to view and rate the(misleading) label of the shake. During the second interval (between 60 and 90 min) participants wereasked to drink and rate the milkshake.Results:The mindset of indulgence produced a dramaticallysteeper decline in ghrelin after consuming the shake, whereas the mindset of sensibility produced arelatively flat ghrelin response. Participants’ satiety was consistent with what they believed they wereconsuming rather than the actual nutritional value of what they consumed.Conclusions:The effect offood consumption on ghrelin may be psychologically mediated, and mindset meaningfully affectsphysiological responses to food.

iven that a stress-is-enhancing mindset mitigates the development of depressive and anxietysymptoms in college students faced with stress, stress mindset interventions could be a viableresource in the transition to college. Short stress mindset interventions have been effective for indu-cing a stress-is-enhancing mindset and achieving improvements in health, well-being, and perform-ance (Crum et al.,2013, 2017;Jamieson et al.,2018). Additionally, Goyer et al. (2018) found that astress-is-enhancing mindset intervention given to students in the summer prior to beginning6N. A. HUEBSCHMANN AND E. S. SHEETScollege predicted higher positive affect during spring exam periods of theirfirst two years. Neverthe-less, more research is needed to assess the long-term effectiveness of stress-mindset interventions incollege settings, and their effects on mental health outcomes. Stress mindset appears to be an impor-tant factor in the stress response, and having a stress-is-enhancing mindset may be broadly beneficialfor emerging adults

DiscussionThe

Anxiety symptomsAlthough the interaction term was not significant for baseline anxiety (p= .123), a similar trend was seenwith a stress-is-debilitating mindset (β= .40,p< .001) predicting greater response to perceived stressthan a stress-is-enhancing mindset, (β= .25,p= .001). Significant moderation was found again foranxiety at follow-up (p=.025); perceived stress increased anxiety, but more so for those with a stress-is-debilitating (β=.56,p< .001) rather than stress-is-enhancing mindset (β= .31,p< .001). Stressmindset also remained a significant moderator of stress and anxiety at follow-up when controlling forbaseline anxiety (p= .034; stress-is-debilitating:β=.43,p< .001; stress-is-enhancing:β= .21,p= .002).

Depressive symptomsStress mindset significantly moderated the relationship between perceived stress and depressivesymptoms at both baseline (p= .026) and follow-up (p= .008; seeTable 2). To interpret thesefindings, simple slopes were tested at 1 SD below the mean, indicating a more stress-is-debilitatingmindset, and 1 SD above the mean, indicating a more stress-is-enhancing mindset. Across analyses,there was a consistent pattern in which higher stress was associated with higher mental health con-cerns, but the risk was greater for those with a stress-is-debilitating mindset (seeFigure 1). At base-line, perceived stress increased depression, but more so for those with a stress-is-debilitating (β= .54,p< .001) than stress-is-enhancing mindset (β= .34,p< .001). The same pattern was demonstrated atfollow-up (stress-is-debilitating:β= .63,p< .001; stress-is-enhancing:β= .34,p< .001). Stress mindset

Stress Mindset Measure-GeneralThe Stress Mindset Measure-General (SMM-G; Crum et al.,2013) is an 8-item measure that was admi-nistered at baseline to assess the extent to which individuals believe the effects of stress are enhan-cing or debilitating. Participants rate the extent to which they agree with a statement about stress(e.g., The effects of stress are positive and should be utilized.) on a scale from 0 (strongly disagree)to 4 (strongly agree). The measure had good internal consistency in the present sample (Cronbach’sα= .83). Prior research has demonstrated both discriminant and criterion validity (Crum et al.,2013).Perceived Stress ScaleThe Perceived Stress Scale (PSS; Cohen et al.,1983) is a 14-item measure that was administered atboth baseline and follow-up to assess stress over the previous month. Participants responded howoften they have felt or thought a certain way during the last month (e.g., In the last month, howoften have you felt nervous or“stressed”?) on a scale from 0 (never)to 4 (very often). The measuredemonstrated internal consistency with the present sample (Cronbach’sα= .67 at baseline; Cron-bach’sα= .63 at 1-month follow-up). Previous research has summarized concurrent and predictivevalidity for the measure (Cohen et al.,1983; Mitchell et al.,2008).Beck Depression Inventory IIThe Beck Depression Inventory II (BDI-II; Beck et al.,1996) is a 21-item measure that was administeredat both baseline and follow-up to assess depressive symptoms over the previous two weeks. Items(e.g., sadness) are rated by selecting one of four statements of increasing severity, and items arescored from 0 to 3. The measure had excellent reliability in this sample (Cronbach’sα= .93 at baseline;Cronbach’sα= .91 at 1-month follow-up). Across a variety of studies, the BDI-II has been shown tohave discriminant, concurrent, content, and structural validity (Wang & Gorenstein,2013).State-Trait Anxiety Inventory–StateThe State subscale of the State-Trait Anxiety Inventory (STAI-State; Spielberger,1983) is a 20-itemmeasure that was administered at both baseline and follow-up to assess current levels of anxiety. Par-ticipants rate the extent to which they feel a certain way (e.g., I feel nervous.) on a 4-point scale fromnot at alltovery much so. The measure had excellent internal consistency in the present study (Cron-bach’sα= .94 at baseline; Cronbach’sα= .93 at 1-month follow-up). The STAI has been shown to haveconstruct and concurrent validity (Spielberger,1989).Statistical analysesBasic descriptive statistics are presented inTable 1. Due to outliers, a 95% windsorization was per-formed on the PSS, BDI-II, and STAI-State data at both the baseline and follow-up.1Pearson’s corre-lations at both study sessions were calculated to examine the bivariate associations betweenperceived stress and the other variables of interest. Moderated linear regression analyses were per-formed to examine stress mindset as a moderator of the relationships of perceived stress and mentalhealth outcomes. First, perceived stress and stress mindset were mean-centered. For each regressionmodel, perceived stress, stress mindset, and a perceived stress X stress mindset interaction term wer

Participants were 293 students (69.3% female, Mage= 19.94 ± 1.53 years) at Colby College, a liberalarts college in the northeastern United States. Participants reported the following racial identities:Caucasian/White (70.99%), Asian (10.58%), Multiracial (7.85%), African-American/Black (5.12%),Latinx (3.75%) and Other (1.71%). Participants were recruited into a larger project examiningcampus climate and health behaviors. This project was formally approved by the college’s Insti-tutional Review Board. Participants were compensated $10 or 1 psychology research credit foreach baseline and follow-up session. Data collection was terminated after 4 semesters.

Taken together, there is evidence that stress mindset has positive effects on well-being, and that itmay act as a moderator to mitigate negative effects of stressful experiences. Prior work has examinedstress mindset in the workplace (e.g., Casper et al.,2017; Crum et al.,2017) or in the lives of youngeradolescents (Jiang et al.,2019; Park et al.,2018). However, to the best of our knowledge, stressmindset has not been closely examined in the college setting, despite the fact that college studentsare a population for which stress is common and mental health concerns are rising. In a recentnational survey of college students, 45.1% of students reported more than average stress and12.7% reported tremendous stress within the last 12 months (ACHA,2018). Among undergraduatestudents at an American university, 15% of students reported severe or extremely severe levels ofanxiety and 11% reported severe or extremely severe depression (Beiter et al.,2015). Furthermore,stress is strongly associated with a greater likelihood of mental health diagnoses and suicide attemptsin college students (Liu et al.,2019). The purpose of the present study was to examine stress mindsetamong college students, and to examine how stress mindset may moderate the relationship betweenperceived stress and mental health symptoms. We would expect higher perceived stress to be associ-ated with worse mental health outcomes. However, it has not yet been established if the associationbetween perceived stress and mental health outcomes varies based on one’s stress mindset.

In addition to the independent influence that stress mindset has on the stress response, stressmindset may act as a moderator of the associations between stressful experiences and well-being.Park et al. (2018) followed a group of socioeconomically and ethnically diverse adolescents acrossa full school year

Short interventions to induce a stress-is-enhancingmindset have positive outcomes as well, including decreases in anxiety and depressive symptoms,better performance at work, and physiological thriving in stressful situations regardless of appraisa

obal ratings of stress have been conceptualized as the interaction of events with appraisals of pre-dictability, control, and coping resources related to these events, or expected future events (Cohenet al.,1983). Stress mindset is defined as the extent to which one believes stress has enhancing ordebilitating effects on stress-related outcomes such as performance and productivity, health andwell-being, and learning and growth (Crum et al.,2013). Individuals who generally believe stresscan have positive effects have a stress-is-enhancing mindset, whereas individuals who believestress has negative effects have a stress-is-debilitating mindset. Upon developing the StressMindset Measure-General (SMM-G), Crum et al. (2013) demonstrated the construct’s discriminant val-idity such that it correlated in the expected direction with other measures of stress amount, appraisal,and coping, but only with weak to moderate strength. Preliminary work on stress mindset indicatesthat the way individuals conceptualize stress and its effects is malleable, and may be quite impactful.

ABSTRACTBackground and Objectives:Having a stress-is-enhancing mindset–theextent to which one believes the effects of stress are enhancing ratherthan debilitating–is associated with greater health and well-being, andmay mitigate negative outcomes associated with experiencing stress. Thepresent study aimed to examine stress mindset and whether it moderatesthe association between perceived stress and mental health outcomes.Design and Methods:Participants (N= 293) completed questionnairesassessing their stress mindset, perceived stress level, and currentdepressive and anxiety symptoms. Perceived stress and mental healthwere assessed again at a follow-up session one month later.Results:Across moderated linear regression analyses, there was a consistentpattern in which higher perceived stress was associated with higher mentalhealth concerns, but the risk was greater for those with a stress-is-debilitatingmindset. Stress mindset moderated the perceived stress-depressionassociation at baseline (p= .026), at follow-up (p= .008), and whenfocusing on change in depressive symptoms from baseline to follow-up (p=.006).Conclusions:These results suggest that a stress-is-enhancing mindsetmitigates the development of depressive and anxiety symptoms in collegestudents faced with high levels of stress. The benefits of introducingemerging adults to an adaptive stress mindset are discussed.

heCOVID19pandemichashada majornegativeimpactonmentalhealth.ItseffectwasfirstdocumentedinChinabutrecentlyhasbeendocumentedmoreglobally[1,2].AccordingtoanApril2020survey,morethana quarterofAmericanadultsmeetcriteriafor“seriousmentaldistress,”aneightfoldincreasesince2018[3].DatafromtheHealthlineMentalHealthIndexindicatesthatupto45%ofUSadultshaveelevatedlevelsofdepressionandanxiety[4].Notsurprisingly,collegestudentsmaybeparticularlyvulnerabletothementalhealtheffectsoftheCOVIDpandemicgiventhemagnitudeofthestressorduringa periodcharacterizedbystilldevelopingemotionalregulatorynetworksinthebrain[5,6].A recentstudythatsurveyedstu-dentsacross195U.S.collegesfoundthat71%ofthesamplereportedincreasedanxietyduetothepandemic[7].

Personalitytraitshavebeenfoundto berelatedto a varietyof healthoutcomes.Theaimofthisstudywasto examinehowpersonalitytraitswereassociatedwithadjustmentto theCOVIDpandemicin collegestudents.Thesampleincluded484first-yearuniversitystu-dents(76%female)attendinga northeasternuniversitywhocompletedtheBigFiveInven-tory(BFI)personalityassessmentat thebeginningof a semesterthatwasdisruptedbytheCOVIDpandemic.Usinga phone-basedapp,studentscompleteddailyratingsof mood,perceivedstresslevels,andengagementin a numberof healthpromotionactivities(exer-cise,mindfulness,adequatesleep,etc.)throughoutthesemesterbothbeforeandaftertheonsetof thepandemic(e.g.,a within-personlongitudinaldesign).Results,asexpected,showedthatmoodandwellnessindicesgenerallydeclinedduringtheCOVIDperiod,althoughstresslevelsactuallydecreased.Further,irrespectiveof COVID,improvedmood,lessperceivedstressandgreaterparticipationin healthpromotionactivitiesweresignifi-cantlyassociatedwitha numberof personalitytraitsincludingneuroticism(lower),extraver-sion(higher),agreeableness(higher),andconscientiousness(higher).Ofprimaryinterest,mixed-effectsmodelswereusedto testhowmajorpersonalitytraitsinteractedwithanychangesin dailyratingsfromthepre-COVIDto COVIDperiod.Significantinteractionstermswerefoundsuggestingdifferentialimpactsof theCOVIDepidemicforstudentswithlowver-sushighlevelsof particulartraits.Higherlevelsof extraversion, forexample,werefoundtoberelatedto decreasesin moodasthepandemicprogressedin contrastto thosewithlowerextraversion,forwhomtherewasa slightincreasein moodovertime.Thesedatasupporttheconclusionthatpersonalitytraitsarerelatedto mentalhealthandcanplaya rolein a per-son’sabilityto copewithmajorstressfulevents.Differenttraitsmayalsobemoreadaptiveto differenttypesof stressors.

behavior of the rest of the group. Thus, our results show how gossip, when paired with a mechanism for partner selection, can foster and sustain high levels of coopera­tion even in noniterated interactions.Our results may seem at odds with research showing that gossip alone produces cooperation, because contri­butions in the gossip game gradually decreased. This past research, however, showed that the threat of gossip, relative to a control, deters selfish behavior, a finding our results replicated: Participants cooperated more in the gossip game than in the basic game. Gossip alone likely promotes cooperation because gossiping and knowing that others could gossip about you makes reputation salient, which tends to foster prosociality (Willer, Feinberg, Irwin, Schultz, & Simpson, 2010), and because defecting when future partners will know what you did will lead these partners to not cooperate with you, which reduces the incentive to defect in the first place. However, these forces, over the long run, were insufficient to main­tain high levels of cooperation. This may have been the case because exposure to gossip about a low contributor from a prior round also stimulates fears of exploitation, which could result in reduced contribution to avoid exploitation (Kuwabara, 2005).Our findings fit well with models of biological markets, which argue that individuals will choose partners based on others’ reputation or “market value” when partner selection is possible (Barclay, 2013; Noë & Hammerstein, 1995). As exemplified in the present research, reputational information obtained through gossip greatly expands the breadth of individuals’ knowledge of others’ past behavior. Further, models of biological markets contend that indi­viduals often compete to demonstrate their value as a part­ner. In the present research, because having the lowest market value of the group led to the highest likelihood of being ostracized, participants likely engaged in such “com­petitive altruism” (Roberts, 1998), vying to be more proso­cial than the other group members to avoid exclusion. In such a dynamic, the standard for avoiding ostracism escalates, which further explains why contributions con­tinually increased across the rounds of the gossip­with­ostracism game. Moreover, these competitive pressures to cooperate would likely have been even greater had we allowed group members not only to exclude individuals, but also to select partners for inclusion—an important topic for future research.Finally, our results add to the literature on how indi­viduals respond to ostracism. Whereas some research has shown that ostracized individuals respond to exclu­sion with decreased prosociality (e.g., Mulder, van Dijk, De Cremer, & Wilke, 2006; Twenge, Baumeister, DeWall, Ciarocco, & Bartels, 2007), our finding that ostracized individuals behaved as cooperatively as everyone else upon returning to their groups fits well with a social­dilemmas perspective of responses to ostracism (Balliet & Ferris, 2013; Joireman, Daniels, George­Falvy, & Kamdar, 2006). This perspective holds that after exclu­sion, individuals face competing incentives: They are tempted in the short run to respond negatively, possibly by behaving more selfishly out of spite. But, in the long run, they benefit most by conforming to group expecta­tions, especially when punishment has significant reper­cussions (Tenbrunsel & Messick, 1999; van Lange, Joireman, Parks, & van Dijk, 2013). In the present study, because ostracized participants faced heavy punishment—earning nothing at all—the dominant incentive was to withhold retaliation and, instead, cooperate.

ghest individual­level earnings by Round 5, omnibus F(2, 159) = 45.47, p < .001, η2 = .36, pairwise t(159)s > 4.36, ps < .001, ds > 0.69, and the highest group­level earnings by Round 6, omnibus F(2, 159) > 22.71, p < .001, η2 = .22, pairwise t(159)s > 2.41, ps < .05, ds > 0.38. Further, an examination of earnings in the gossip­with­ostracism game from the second round onward revealed a significant upward linear trend, Fs(1, 53) > 42.32, ps < .001, η2s > .44, which shows that earnings levels were increasing when the game concluded.

(gossip game: βs < −0.49, ps < .001, R2s > .23; gossip­with­ostracism game: βs < −0.24, ps < .001, R2s > .06),

9.08.07.06.05.04.03.02.01.010.0Round 1Round 2Round 3Round 4Round 5Round 6Gossip-With-OstracismGossip NeutralFig. 2.

408075706560555045Round 1Round 2Round 3Round 4Round 5Round 6Average Individual Earnings101817161514131211Round 1Round 2Round 3Round 4Round 5Round 6abBasicGossip

Our central hypothesis was that groups in contexts that featured both gossip and a means for exclusion would achieve higher levels of cooperation. First, we compared the total amount participants contributed to their group fund, aggregated across all six rounds (possible range = 0 to 60 points) for each of the three experimental games. In the gossip­with­ostracism game, whenever partici­pants were ostracized, we coded their contribution for that round as zero. A within­subjects analysis of variance (ANOVA) yielded a significant omnibus difference across games, F(2, 430) = 249.89, p < .001, η2 = .54.1 Comparisons between games revealed that participants contributed significantly more when playing in the gossip game (M = 29.79, SD = 16.54) than they did when playing in the basic game (M = 17.54, SD = 16.28), F(1, 215) = 195.04, p < .001, η2 = .48. This finding captures the unique effect of having one’s behavior potentially communicated to future interaction partners (Barclay, 2004).More relevant to our hypothesis, further comparisons revealed that when participants played in the gossip­with­ostracism game (M = 42.89, SD = 14.79), they con­tributed significantly more than they did when playing in either the basic game, F(1, 215) = 417.06, p < .001, η2 = .66, or the gossip game, F(1, 215) = 110.80, p < .001, η2 = .34. Even in the first round, participants contributed sig­nificantly more when in the gossip­with­ostracism game (M = 6.80, SD = 3.17) than they did when in either the basic game (M = 4.91, SD = 3.56), F(1, 215) = 49.66, p < .001, η2 = .19, or the gossip game (M = 6.01, SD = 3.31), F(1, 215) = 8.83, p < .01, η2 = .04, which suggests that simply knowing about the potential to be gossiped about and ostracized by future group members was enough to engender an increase in cooperation. Importantly, we found differences in total contributions across the six rounds between the gossip­with­ostracism game and the two other games, even though in 15% of the rounds of the gossip­with­ostracism game, partici­pants were excluded and could not contribute anything. This result points to the significant role gossip plays in fostering cooperation, especially when it can be used for partner­selection purposes.To analyze whether contributions tended to increase or decrease as rounds progressed, we conducted a two­way within­subjects Game × Round Number ANOVA. This analysis yielded a significant omnibus interaction, F(10, 2150) = 22.92, p < .001, ηp2 = .10. Analyses examining potential linear­trend differences across the six rounds for each game revealed significant differences between the gossip­with­ostracism game and both the basic game, F(1, 215) = 132.91, p < .001, ηp2 = .38, and the gossip game, F(1, 215) = 62.34, p < .001, ηp2 = .23. Additionally, there was a significant linear­trend difference between the basic game and the gossip game, F(1, 215) = 10.23, p < .01. ηp2 = .05. Separate within­game linear­trend analyses revealed that there was a decrease in contributions as rounds pro­gressed in both the basic game, F(1, 215) = 162.43, p < .001, ηp2 = .43, and the gossip game, F(1, 215) = 54.44, p < .001, ηp2 = .20, a common finding in public­goods studies (Ledyard, 1995). In the gossip­with­ostracism game, however, contributions increased as rounds pro­gressed, F(1, 215) = 15.29, p < .001, ηp2 = .07 (Fig. 2).

Overall, participants played 18 total rounds of the public­goods exercise—six rounds for each game (see Fig. 1). Once participants completed the final round of the last game, the experimenter informed them that the study was over. The experimenter then debriefed the par­ticipants, paid them the amount of money they had earned, and dismissed them from the study.

BasicGossipGossip With OstracismRound 1Round 2ContributeResultsContributeResultsNote OpportunityContributeResultsNote OpportunityContributeResultsContributeResultsNote OpportunityReceive Note(s)Vote to ExcludeReceive Note(s)SelfExcludedOtherExcludedNo OneExcludedPlay With2 OthersNoPlayPlay WithAll 3Note OpportunityResultsRounds3, 4, 5, 6Fig. 1.Schematic showing the timeline of the experimental procedure. In the basic game, all partici­pants received an allotment of 10 points at the beginning of each round and determined how many of the points they wished to contribute to a group fund and how many they wished to keep for them­selves. At the end of each round, the 4 participants in each group learned how much each member had contributed and earned. Participants were then assigned to a new group and the process was repeated. In the gossip game and the gossip­with­ostracism game, after learning the results of each round, partici­pants were given the opportunity to send a note to the upcoming game partners of 1 of the participants they just played the game with. At the beginning of each round in the gossip­with­ostracism game, after receiving the gossip notes (if any were sent), participants could anonymously vote to exclude 1 participant from playing in the upcoming round; if a participant was excluded by receiving two or more exclusion votes, the remaining 3 participants played without him or her.

After learning how to play in the public­goods exercise, all participants completed a five­question comprehension check. The computer displayed a mes­sage informing participants which questions they had missed (if any) and re­presented each of these ques­tions until they selected the correct answer. Once par­ticipants had successfully completed the comprehension check, they were provided with a practice round of the public­goods exercise to familiarize them with the com­puter interface.

ch participant received an allot­ment of 10 points at the beginning of each round of the exercise. Each point was worth 2.5¢. All 4 participants then determined how many of their 10 points they wished to contribute to a group fund and how many they wished to keep for themselves. Whatever number of points all 4 participants contributed to the group fund as a whole was then doubled and redistributed equally to each group member. Researchers commonly use this public­goods exercise to examine social dilemmas because indi­vidual participants will benefit the most by selfishly free riding off of everyone else’s contributions while contrib­uting nothing themselves

The study involved nine separate groups of 24 partici­pants each and was run in a behavioral­economics labo­ratory at a large public university. The experimenter seated all participants at separate computer stations and requested that they not verbally communicate with any­one else at any time during the study. The experimenter then informed participants that the study would be con­ducted using the computer and directed them to follow the directions presented on the computer in front of them.

2 male, 134 female; mean age = 20.4 years) took part in this study in return for a flat payment of $5 and the opportunity to earn an addi­tional payment ranging from approximately $2 to $12.

The existence of cooperation poses a puzzle to the bio­logical and social sciences because each person faces strong incentives to exploit the cooperative tendencies of others (Dawes, 1980; Komorita & Parks, 1996). What, then, can explain the existence of widespread coopera­tion observed in human societies? Recent research on reputation systems has provided one potential explana­tion: The knowledge that one’s behavior will be known by many others reduces an individual’s incentive to behave selfishly, thereby promoting cooperation (Hardy & van Vugt, 2006; Simpson & Willer, 2008; Willer, 2009). In addition, knowing which of one’s potential partners has a reputation for cooperation or defection helps sus­tain cooperation in at least two ways. First, such informa­tion assists individuals in determining whom to selectively interact with by allowing them to choose to pair with more cooperative partners (Barclay & Willer, 2007). Further, within relationships, such information can be used to guide how much trust individuals should invest in a given partner (Barclay, 2004). But although scholars have identified reputation systems as valuable in foster­ing cooperation (Milinski, Semmann, & Krambeck, 2002; Wedekind & Milinski, 2000), little research has explored how these systems emerge. When do individuals share information on the past behavior of others, and what effects does it have? Here, we examine how the spread of reputational information through gossip facilitates coop­eration and limits defection in groups.Although gossip is often considered trivial or anti­social, many positive social functions of gossip have been proposed (Foster, 2004). One prominent theory views gossip as a policing mechanism that helps indi­viduals track those who have exploited other group members, even when such exploitation was not directly observed (Dunbar, 2004). Consistent with this argument, 510184PSSXXX10.1177/0956797613510184Feinberg et al.Gossip and Ostracism Promote Cooperationresearch-article2014Corresponding Author:Matthew Feinberg, Center for Compassion and Altruism Research and Education, Stanford University, 306 Jordan Hall, Stanford, CA 94305 E­mail: mfeinber@stanford.eduGossip and Ostracism Promote Cooperation in GroupsMatthew Feinberg1,2, Robb Willer3, and Michael Schultz41Center for Compassion and Altruism Research and Education, Stanford University; 2Graduate School of Business, Stanford University; 3Department of Sociology, Stanford University; and 4Department of Sociology, University of California, BerkeleyAbstractThe widespread existence of cooperation is difficult to explain because individuals face strong incentives to exploit the cooperative tendencies of others. In the research reported here, we examined how the spread of reputational information through gossip promotes cooperation in mixed­motive settings. Results showed that individuals readily communicated reputational information about others, and recipients used this information to selectively interact with cooperative individuals and ostracize those who had behaved selfishly, which enabled group members to contribute to the public good with reduced threat of exploitation. Additionally, ostracized individuals responded to exclusion by subsequently cooperating at levels comparable to those who were not ostracized. These results suggest that the spread of reputational information through gossip can mitigate egoistic behavior by facilitating partner selection, thereby helping to solve the problem of cooperation even in noniterated interactions.Keywordssocial dilemmas, cooperation, gossip, reputation, altruism, ostracism, partner choice, social behavior, social interaction, punishmentReceived 2/14/13; Revision accepted 9/20/13Gossip and Ostracism Promote Cooperation 657ethnographic evidence suggests that group members readily spread reputational information about and stig­matize those who do not conform to normative levels of cooperation (Wilson, Wilczynski, Wells, & Weiser, 2000). In addition, experimental studies have investigated the link between gossip and cooperation, finding that gossip can facilitate indirect reciprocity (Sommerfeld, Krambeck, Semmann, & Milinski, 2007) and deter exploitative behav­ior in groups (Beersma & van Kleef, 2011; Feinberg, Willer, Stellar, & Keltner, 2012; Piazza & Bering, 2008).In the research reported here, we explored the possi­bility that gossip promotes cooperation by facilitating partner selection. We hypothesize that gossip fosters and sustains high levels of cooperation when paired with a means for social exclusion. Specifically, if individuals are made aware of others’ past behavior through gossip, they will use this information as a guide for selectively inter­acting with only those people known to be cooperative, ostracizing those known to be defectors. As a result, we expect such reputational­information sharing to promote cooperation in groups by allowing more cooperative individuals to exclude free riders and thus reap the ben­efits of group efforts while avoiding exploitation.Additionally, ostracism should serve as a powerful tool for mitigating free riding. Social exclusion is an effective means of social and economic punishment. Ostracized individuals cannot reap the benefits of group efforts (Ouwerkerk, Kerr, Gallucci, & van Lange, 2005; Spoor & Williams, 2007; Williams, 2007), which makes the threat of expulsion a strong disincentive to defection. Beyond its economic effects, research has shown that social exclusion activates neurological responses analogous to pain responses associated with physical injury (Eisenberger, Lieberman, & Williams, 2003; MacDonald & Leary, 2005). Thus, it is likely that people will find that the costs of ostra­cism outweigh the potential benefits for selfish behavior, which will lead ostracized individuals to cooperate at higher levels in subsequent group settings. We therefore expect that gossip and ostracism will work especially well in tandem, because gossip facilitates diffusion of informa­tion about formerly exploitative group members and ostra­cism provides a means for partner selection.To test these claims, we conducted a large­scale group­interaction study. In each round of the study, participants decided whether or not to make a costly contribution that would benefit their group before moving onto the next round, in which they interacted with an entirely dif­ferent group. The study featured two treatment condi­tions and one control condition. In both treatment conditions, prior to the beginning of a subsequent round, group members could relay reputational information about one of their current group members to that per­son’s future interaction partners. Additionally, in one treatment condition, recipients of this reputational infor­mation could use it as a means for partner selection by excluding one of the prospective group members. We hypothesized that when participants could relate reputa­tional information and recipients could act on the infor­mation they received by ostracizing a suspect group member, groups would achieve significantly higher levels of cooperation.

The widespread existence of cooperation is difficult to explain because individuals face strong incentives to exploit the cooperative tendencies of others. In the research reported here, we examined how the spread of reputational information through gossip promotes cooperation in mixed­motive settings. Results showed that individuals readily communicated reputational information about others, and recipients used this information to selectively interact with cooperative individuals and ostracize those who had behaved selfishly, which enabled group members to contribute to the public good with reduced threat of exploitation. Additionally, ostracized individuals responded to exclusion by subsequently cooperating at levels comparable to those who were not ostracized. These results suggest that the spread of reputational information through gossip can mitigate egoistic behavior by facilitating partner selection, thereby helping to solve the problem of cooperation even in noniterated interactions.

In the present research we directly replicatedCialdini et al.’s(1975)door-in-the-face technique. The results are almost identicalto those obtained in the original article and support the assumptionthat reciprocity norms play a crucial role in sequential persuasivestrategies. The finding shows that a fundamental finding in socialpsychology can be reproducible in a close replication almost halfa century later in a different country on a different continent

Second, a limitation of our results concerns the assessed sample.As our experiment was a direct replication, we assessed, in linewithCialdini et al. (1975), only students, who were predominantlyfemale. Moreover, similar toCialdini et al. (1975), we did notemphasize diversity or inclusiveness. Hence, our findings leaveopen to which degree they can be generalized to other samples.Nevertheless, our experiment extends the findings obtained byCialdini et al. (1975)concerning the sample properties. That is, 45years after publication of the original finding, we were able toreplicate the DITF effect in another country (i.e., Germany) withinanother culture. In line with previous research (Dillard et al., 1984;Feeley et al., 2012;Fern et al., 1986;O’Keefe & Hale, 1998), thissuggests that the DITF technique works across cultures and dif-ferent samples.Third, another limitation of our replication as well as the orig-inal DITF research concerns the potential influence of self-selection. That is, only passersby who actually talked to the ex-perimenters could be included in the experiment. Thus, one mightargue that only participants with a certain prosocial orientationtook part in the experiments, which might have influenced theirprosocial response (i.e., acting as a chaperone for a group ofjuvenile delinquents on a 2-hr trip to the zoo). However, asprevious research indicates that the influence of self-selection onthe results of experiments involving prosocial behaviors are neg-ligible (Abeler & Nosenzo, 2015;Falk, Meier, & Zehnder, 2013),we regard the influence of self-selection in our experiment asrather small. Nevertheless, future research may aim at testing towhich degree self-selection influences the effectiveness of DITFeffects.

Cialdini et al. (1975)assumed that the DITF technique’s successis based on reciprocity norms (Gouldner, 1960). Accordingly, ashift from an extreme request to a smaller request is perceived asa concession from the requester. After rejecting the extreme re-quest, participants may feel the need to reciprocate to this conces-sion by accepting the subsequent smaller request (Diekmann,2004;Hale & Laliker, 1999;Reeves et al., 1991;Turner et al.,2007). This theoretical assumption has been challenged by previ-ous research (Tusing & Dillard, 2000; see alsoTurner et al., 2007).According to this critique, saying “no” to the extreme requestelicits negative feelings in participants, leading them to avoidsaying “no” a second time. Consequently, individuals’ motivationto comply with the second request is to avoid two successiverejections. If this is indeed the case, one would predict that par-ticipants are more likely to agree to the small request when theyhave rejected another small request beforehand. However, we donot find this effect as participants were less likely to agree with thesecond request when they had rejected another similarly smallrequest beforehand. Thus, our data support the idea that DITFeffects are driven by reciprocity norms (Gouldner, 1960).Besides this implication, our findings are also important in lightof the current debate on the crisis of confidence in psychologicalresearch. Past research has reported difficulties in replicating some(social) psychological findings (e.g.,Caruso, Shapira, & Landy,2017;Flore, Mulder, & Wicherts, 2018;Hagger et al., 2016;OpenScience Collaboration, 2015;Wagenmakers et al., 2016), whichled to the widespread assumption that classical (social) psycho-logical textbook findings are not robust (e.g.,Schimmack, 2018).However, as previous research has mostly focused on replicatingfindings published within the last few decades, a rich area of thepsychological literature has been ignored. We would like to arguethat in order to get a more complete and accurate picture of thereplicability of the field, it is essential to replicate classical find-ings published earlier in the last century as well.By replicating one of the landmark findings in the literature ofsocial influence and attitude formation, our research is a first stepin this direction and indicates that classical studies can be repli-cable. This contrasts the view that social psychological findingsand theories are limited to a particular time, place, culture, andpopulation (Gergen, 1973;Van Bavel et al., 2016). To test towhich degree this is indeed the case, future replication efforts mayfurther aim at replicating classical textbook findings that wereoriginally published in the 1970s and earlier. Moreover, to accu-This document is copyrighted by the American Psychological Association or one of its allied publishers.This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.e5DOOR-IN-THE-FACE REPLICATIONrately depict the actual replication rate within (social) psychologyit might be worthwhile considering not only replicating a prioriassumed weak effects, but also potentially strong effects. Th

In a first series of preregistered analyses, we ran the exact samechi-squared tests as performed byCialdini et al. (1975). The firstanalysis tested for a statistical difference between the two controlconditions (i.e., smaller request only condition vs. equivalent re-quest control condition). As in theCialdini et al. (1975)study, wedid not find a significant difference between these conditions,22.11,p.146. The second analysis tested for a difference betweenthe rejection-moderation condition and the combined control con-ditions (i.e., smaller request only condition & equivalent requestcontrol condition). Similar toCialdini et al. (1975), participants inthe rejection-moderation condition (51.28%) complied more oftenwith the smaller request than participants in the combined controlconditions (33.58%),210.82,p.001.In a second series of preregistered analyses, we performedadditional analyses that were not conducted byCialdini et al.(1975). First, we compared participants’ compliance rate with the

Interestingly, a very promising, but widely forgotten test of thisalternative explanation has been offered byCialdini et al.’s (1975)Study 3. In this experiment, the authors implemented threebetween-subjects conditions. In therejection-moderation condi-tion, student research assistants approached participants on theuniversity campus and initially made an extreme request. Specif-ically, they introduced themselves as being with the County YouthCounseling Program and then asked whether participants would bewilling to work voluntarily as a nonpaid counselor at the CountyJuvenile Detention Center. The position would require 2 hr of theirtime per week for a minimum of 2 years. After participants deniedthis extreme request, the experimenters brought forward a smallerrequest. That is, they asked whether participants would be willingto act as chaperones for a group of juvenile delinquents on a 2-hrtrip to the zoo. In thesmaller-request only control condition,experimenters asked for the small request only (i.e., 2-hr trip to thezoo). In theequivalent request control condition,the researcherstested whether participants’ likelihood to agree to the small requestwould increase if they rejected another similar small request be-forehand. In this condition, the experimenters first asked partici-pants whether they would act as chaperones for a group of juveniledelinquents on a 2-hr trip to the city museum and then broughtforward the same small request (i.e., 2-hr trip to the zoo) as in theother conditions. The reciprocal concessions account predicts alarger amount of agreement with the second request in therejection-moderation conditionthan in theequivalent request con-trol condition,because participants should perceive a concessionin therejection-moderation condition,but not in theequivalentrequest control condition. In contrast, if individuals’ motivation tocomply with the second request is to avoid saying no twiceaccounts for the effect, one would assume that theequivalentrequest control conditionwould lead to the same amount ofagreements with the second request as therejection-moderationcondition, because in both conditions participants should be mo-tivated to avoid saying no twice. Interestingly, this is not whatCialdini et al. (1975)found. Rather opposite to this prediction,participants in thesmaller-request only control conditionandparticipants in theequivalent request control conditionwere bothless likely to agree with the small request as compared to therejection-moderation condition. It is important to note, however,that with22.88,p.09 this result did not reach conventionallevels of significance. Moreover, to the best of our knowledge,Cialdini et al.’s (1975)Study 3 has never been directly replicatedwithin a large sample. Thus, it remains unclear whether a high-powered replication of this experiment would actually supportCialdini et al.’s (1975)reciprocal concessions theory.The Present StudyThe fact that most previous replication efforts in social psychol-ogy have focused on recent publications, coupled with the claimthat social psychological findings are limited to a particular time,place, culture, and population (Gergen, 1973;Van Bavel, et al.,2016), motivated us to directly replicate a classical finding pub-lished nearly half a century ago in another country (i.e., Germany)on another continent (i.e., Europe). Given the ongoing debate onthe underlying mechanisms and the fact thatCialdini et al.’s (1975)critical test of the alternative explanation that individuals avoidsaying no twice was not statistically significant, we chose todirectly replicateCialdini et al.’s (1975)Study 3 with more pre-cision allowed by a larger sample. We preregistered our experi-ment at aspredicted.org (https://aspredicted.org/jq5fe.pdf) andmade all our materials and data openly accessible at the OpenScience Framework (OSF;https://osf.io/t6zaw/).

Hence, some scholars have suggested that social psychologicalfindings and theories do not persist over time

In-deed, recent investigations suggest that the success rate of repli-cations in social psychology strongly depends on contextual fac-tors, such as time, culture, location, and populatio

Taken together, previous replications cast doubt on the repro-ducibility of a large body of psychological findings. However, thisimplication is largely based on replications of studies originallypublished within the last two decades thereby ignoring classicalsocial psychological studies published much earlier. We believethat in order to get a more accurate picture about the state of(social) psychology, researchers need to strive for replications ofearlier studies as well. Moreover, given that these landmark studiesmake up the foundations of the field—relevant for theory building,hypothesis generation, research planning, and teaching we regardit as particularly important to replicate these classical findings.

un to focus morestrongly on replications of previous findings. A substantial pro-portion of these replications failed to support the original findings

In the research reported here, we tested the hypothesis that sustained engagement in learning new skills that activated working memory, episodic memory, and reasoning over a period of 3 months would enhance cognitive function in older adults. In three conditions with high cognitive demands, participants learned to quilt, learned digital photography, or engaged in both activities for an average of 16.51 hr a week for 3 months. Results at posttest indicated that episodic memory was enhanced in these productive-engagement conditions relative to receptive-engagement conditions, in which participants either engaged in nonintellectual activities with a social group or performed low-demand cognitive tasks with no social contact. The findings suggest that sustained engagement in cognitively demanding, novel activities enhances memory function in older adulthood, but, somewhat surprisingly, we found limited cognitive benefits of sustained engagement in social activities.

A pattern of early cognitive exploration also makessense from an evolutionary perspective. Across manyspecies, flexibility, brain size, and intelligence are associ-ated with a long, protected period of immaturity - a longchildhood. Human beings have the largest brains, themost flexible intelligence, and the longest childhood ofany species. One explanation for this distinctive life his-tory is that an early protected period allows young organ-isms to explore possibilities in an unconstrained way.This early exploratory learning, in turn, allows learners toact more effectively when they grow older (Buchsbaum,Bridgers, Weisberg, & Gopnik, 2012). Childhood may beevolution's way of performing simulated annealing.Adults may sometimes be better at the tried and true,while children are more likely to discover the weird andwonderful. This may be because as we get older, we bothknow more and explore less.

Another factor may be that as children grow older, there arechanges in the way they learn that make them intrinsicallyless flexible and less able to attend to unusual possibilities.There are complementary computational, neuroscientific,and evolutionary reasons for thinking this might be true.

First, the very fact that older learners know more may makeit more difficult for them to learn something new. Once alearner has inferred a general principle (e.g., that people actbecause of their traits, or that individual objects, rather thancombinations of objects or relations between them, havecausal powers), that principle may constrain his or her inter-pretation of new data. Causal relationships that conflict withthat principle may then be more difficult to learn.Probabilistic-model-based approaches to cognitivedevelopment can provide a more precise version of thisidea (for more on such approaches see, e.g., Gopnik &Tenenbaum, 2007; Gopnik, 2012; Gopnik & Wellman,2012; Kushnir & Xu, 2012; Tenenbaum, Kemp, Griffiths,& Goodman, 2011). A Bayesian learner assesses howlikely various hypotheses are, given a pattern of newdata. Learners do this by using Bayes's rule to combinetwo probabilities. One is the prior probability of any par-ticular hypothesis - how likely the hypothesis was beforethe learner saw the data. The other is the likelihood -how likely it was that that hypothesis would have gener-ated the new data.As a result, if the prior probability distribution stronglyfavors one hypothesis - that is, if the learner initiallythinks that hypothesis A is much more likely than B -then the learner will need more evidence to overturn Aand accept B instead. If the prior is "flat" - that is, if thelearner initially thinks that A and B are equally likely -then the learner will require less evidence to accept B.In an extension of this idea, called hierarchicalBayesian learning (Griffiths & Tenenbaum, 2007), data ata more specific level, like the relations between stomach-aches and food, can be used to learn a higher-level prin-ciple - in this case, the overhypothesis that illnesses havebiological causes. This kind of learning might explain thecounterintuitive pattern in our studies.

We gave 18- to 30-month-olds a causal higher-orderrelation problem (Walker & Gopnik, 2013, 2014). Amachine played music when an experimenter put twosimilar blocks on it but not when she put two differentblocks on, or vice versa. Toddlers then had to choosebetween two novel pairs of blocks - one pair of two simi-lar blocks and one pair of two different blocks - to acti-vate the machine.Surprisingly, these toddlers were adept at the task, incontrast to the older children in previous studies. Thenwe gave 3-year-olds exactly the same task as the toddlers.They performed at chance level. Further studies showedthat this was because they assumed that the individualobjects, rather than the relations between them, wouldactivate the machine, in spite of the data.

Preschoolers correctly learned both the individual andcombination principles from the unambiguous examplesand used them to interpret the ambiguous new data anddesign the right action (Fig. 2a). T

a b■ Single object■ Multiple objectsChildren Adults25 25■■■i10o L ^ ^Combination Individual Combination IndividualPrinciple Principle

In the kind of studies we report here, learners see aseries of events and choose between two abstract hypoth-eses, A and B, that could explain those events. HypothesisA initially seems less likely than B, at least from the adultperspective, but it is better supported by the evidence thelearner has seen. Younger learners turn out to be morelikely to infer A than older learners, who, despite thedata, are more likely to stick with B.

■ 6-year-olds■ 4-year-oldsIkPerson Control Situation

Many studies have shown that children as young as 15months old can learn specific cause-effect relationshipsfrom statistical data (Gopnik et al., 2004; Gopnik &Schulz, 2007; Gopnik & Wellman, 2012; Gweon & Schulz,2011). These studies have shown the typical develop-mental pattern - either younger and older children per-form similarly, or older children do better. In the newstudies we describe here, we investigated whether chil-dren can use patterns of data to infer more abstract, gen-eral causal principles, or overhypotheses - that is,

We suggest that the apparent limitations in children'sknowledge and cognitive abilities may actually some-times make them better learners. Empirically, we haverecently found a similar pattern across different problemsand age ranges. Younger learners are, surprisingly, betterthan older ones at inferring unlikely or unusual abstractcausal hypotheses from evidence.

We describe a surprising developmental pattern we found in studies involving three different kinds of problems andage ranges. Younger learners are better than older ones at learning unusual abstract causal principles from evidence.We explore two factors that might contribute to this counterintuitive result. The first is that as our knowledge grows,we become less open to new ideas. The second is that younger minds and brains are intrinsically more flexible andexploratory, although they are also less efficient as a result.

ethods

In other words, our study reframes the discussion in terms of(false) remembering (statements about events that are code-termined by memory information and the context of remem-bering, e.g., ref. 61) rather than (false) memories (supposedstable entities within people’s minds). This, again, is not meant tobelittle the importance of false memories in any way (e.g., bysuggesting that they are inherently transitory); quite to thecontrary, social context will often help to sustain them (e.g., refs.70, 79, 82–84), and there is no doubt whatsoever that false re-membering/memories can have devastating real-life conse-quences. It does suggest, however, that—given the situationaldependency of false remembering demonstrated here—focusingon the mere prevalence of false memories (as if it were a con-stant) in attempts to resolve the false memory debate leads to anunproductive dead end. Future research should rather seek(similar to the system factors approach in eyewitness identifica-tion research; e.g., refs. 85–87) to further our understanding ofsituational factors under the potential control of investigatorsthat minimize the risk of producing/retrieving false memoriesand maximize veridical remembering.

that the interview procedures at follow-up differed slightly fromthe original interviews (inevitably, seeMethods), which is why werefrain from making direct statistical comparisons with the originalstudy. Still, in absolute terms, there is very little endorsement of thefalse memories at follow-up. As shown further in Fig. 2, the over-whelming majority (74% across suggestion conditions) of participantseither rejected the false event (mostly straight away—e.g.,“Well thiswas one of the events that don’texist”) or had no memory of it. Theremaining events were mostly just accepted (i.e., merely believed tohave happened), and in only a handful of cases (n=4) in the massivesuggestion condition one could still speak of false memories.Statistical comparisons within the follow-up are still appro-priate and show, unsurprisingly, a massive difference in memoryquality between true and false events (F(1, 37)=126.02,P<0.001,ηp2=0.77). More interesting, and different from what weexpected, there was no statistical difference between the two sug-gestion conditions (F(1, 37)=0.21,P=0.648), unlike in the originalinterviews where massive suggestion resulted in clearly higher levelsof false memory, even after source and false memory sensitization.That is, the effects of massive suggestion in the original interviewsdid not persist over time.These findings were further validated by additional ratings(i.e., their self-reported belief in occurrence and recollection ofevents) participants provided in the follow-up. True events scoredhigher on both dimensions (belief: M=6.34, SE=0.17; recol-lection: M=5.48, SE=0.21) than false events (belief: M=2.54,SE=0.23; recollection: M=2.77, SE=0.19), with the differencesbeing highly significant (belief:F(1, 37)=139.18,P<0.001,ηp2=0.79; recollection:F(1, 37)=77.47,P<0.001,ηp2=0.68). Fur-thermore, paralleling the coding-based findings above, there wereno significant effects of the amount of suggestion (minimal versusmassive;Fs<2.4,Ps>0.128

Our study demonstrates that successfully induced and consoli-dated (over three successive interviews) false memories can besubstantially reversed again after the fact, using two ecologicallyrealistic sensitization strategies that reshape the way intervieweesapproach the retrieval task (even before full debriefing). At afollow-up 1 y later, endorsement of the false memories furtherdeclined to a very low level of 5% overall. Several aspects ofthese findings are noteworthy and deserve further comment.

Reversal or Only Reduction?

ong-Term Effects?False information provided in the context ofpsychological research may have long-term effects despite thor-ough debriefing (64), and false memories have partly survivedtheir discrediting in other research (15, 40, 65, 66). To explorethe persistence of false event memories in our young adult par-ticipants after source sensitization, false memory sensitization, anddebriefing, we invited them back for a follow-up interview morethan a year (mean [M]=390.71 d; range=277 to 485; SD=63.28) after the original study (see below for details). Not alloriginal participants attended, but the subsample who did (seeMethods) was statistically indistinguishable from those who did not(as measured by their scores at the end of the original study,Ps>0.227).

To examine if thesource sensitization and false memory sensitization proceduresled to a reversal of false memories, we compared (false) memoryquality across the last three measurement occasions (Interview 3,source sensitization, false memory sensitization) and found thepredicted decrease across measurements (F(1.740, 102)=44.732,P<0.001,ηp2=0.467).

Did We Effectively Induce False Memories?We first tested a nec-essary precondition for false memory reversal—that false mem-ories had been induced and consolidated in the initial interviews.We analyzed memory reports for false events across Interviews 1through 3, rated by two coders on a memory quality scale used inprevious research (29) but slightly adapted for the present pur-poses (see below). As displayed in Fig. 1 (orange lines), memoryquality increased across the first three interviews (F(1.238, 102)=25.327,P<0.001,ηp2=0.332), from means around 3 (repre-senting mere belief in/acceptance of the event) to means closer to4 (representing at least partial false memory). Furthermore, usingmassive suggestion led to higher overall levels of false memory(F(1, 51)=9.859,P=0.003,ηp2=0.162) but not to a steeperincrease in (false) memory quality over time (F(1.518, 102)=2.698,P=0.086 for the interaction).When true events were included in the analysis, a main effectof event type (true or false) indicated that memory quality wasgenerally (and unsurprisingly) higher for true events (F(1, 51)=114.170,P<0.001,ηp2=0.691, green lines in Fig. 1). Moreinteresting, a significant interaction of event type, measurementoccasion, and suggestion emerged (F(1.730, 102)=5.762,P=0.006,ηp2=0.102). A breakdown showed that true and falseevents were similar in their development over time in the mini-mal suggestion interviewing condition (F(1.554, 102)=0.987,P=0.358), but this trajectory differed significantly in the massivesuggestion condition (F(1.758, 102)=15.829,P<0.001,ηp2=0.237); false events increased significantly in memory quality acrossinterviews (F(1.492, 102)=22.977,P<0.001,ηp2=0.311), but trueevents remained rather stable (F(1.662, 102)=2.608,P=0.088).That is, massive suggestion produced additional effects over timecompared with the mere assertion that the events happened.Breaking down (false) memory quality into distinct levels in aseparate analysis (Fig. 2), we found that 27% and 56% of par-ticipants in the minimal and massive suggestion conditions, re-spectively, developed false memories (partial to robust) for thefalse event after 2 wk (Interview 3). Furthermore, coded falsememory quality was generally (i.e., across interview conditions andtimes) meaningfully related to recollective experience (e.g., par-ticipants’self-rated amount of sensory information,rs=0.427 to0.645,Ps<0.003; clarity/vividness of the recollection,rs=0.530 to0.631,Ps<0.001; participants’confidence,rs=0.529 to 0.640,Ps<0.001; there were only two exceptions, both in the first minimalsuggestion interview: clarity/vividness,r=0.217,P=0.192, andconfidence,r=0.233,P=0.158).

rationale behind this strategy was to 1) raise participants’aware-ness of the possibility of false memories, 2) urge them to criticallyreflect on their recollections, 3) facilitate false memory identifi-cation by reducing the uncertainty the initial suggestion had eli-cited, 4) strengthen participants’trust in their own perspective,and—through the new interviewer—5) lower the social costs as-sociated with memory retraction (61–63). As preregistered (seeMethods), we expected both strategies to significantly reduce falsememories without exerting the same influence on true memories.

Our reversal attempts consisted of two sensitization strategiesthat were ecologically valid in the sense that they 1) can princi-pally be implemented in the real world and 2) do not requireknowledge about the truth status of the memories.

he second strategy (false memory sensitization) was inspiredby the enlightenment technique (59), a type of postwarning thathas been successfully used to reverse the eyewitness misinfor-mation effect (39, 45) and stereotype effects on memory (60) andinvolved alerting participants to the possibility of false memorieswhile trying to avoid a general response bias.

Here, we present a systematic attempt to undo false memoriesof autobiographical events. As this requires the antecedent cre-ation of false memories, the first part of this study consisted ofinterviews designed to elicit false memories. Following previousprocedures (27), we recruited participants for a study on“child-hood memories”and sent their parents questionnaires detailingthe purpose of the study but asking them not to tell their children,who were likewise instructed not to talk about the study with theirparents. Parents indicated for a number of negative events (seehttps://osf.io/ser63/for the full list) whether or not their child hadexperienced them and additionally suggested two negative eventsthat did definitely not happen to their children but were plausible.Participants were then invited to three successive memory inter-views (with a lag of 1 wk between each interview) and asked torecollect four events from their childhood—the two plausible butnot experienced (false) events (e.g., getting lost, running away,being involved in a car accident, injuries, being the perpetrator orvictim of material damage) and two actually experienced (true)events (seehttps://osf.io/ser63/for the full interview protocol).

SignificanceHuman memory is fallible and malleable. In forensic settings inparticular, this poses a challenge because people may falselyremember events with legal implications that never actuallyhappened. Despite an urgent need for remedies, however,research on whether and how rich false autobiographicalmemories can be reversed under realistic conditions (i.e., usingreversal strategies that can be applied in real-world settings) isvirtually nonexistent. The present study therefore not onlyreplicates and extends previous demonstrations of falsememories but, crucially, documents their reversibility after thefact: Employing two ecologically valid strategies, we show thatrich but false autobiographical memories can mostly be un-done. Importantly, reversal was specific to false memories(i.e., did not occur for true memories).

While remedies are urgently needed, systematic research onhow to undo or reverse implanted false autobiographical mem-ories has been scarce. Specifically, previous attempts at reversingfalse memories have been limited in two important respects:First, they involved mostly small and peripheral details of ob-served events [i.e., using the eyewitness misinformation para-digm (39)] or laboratory-produced mini-events [such as clickingone’s fingers (40)], both of which lack the embeddedness ofautobiographical memories within the person’s life story. Sec-ond, almost all of these studies used reversal techniques that areimpossible to implement or highly implausible in the real world,such as positively telling participants that some events had merelybeen suggested or even identifying those events (15, 16, 39–45).This not only requires privileged knowledge on the part of theinvestigator but also creates a highly targeted response set inparticipants (i.e., trying to positively identify the false events) thatis unlikely to be transferable to real-world settings. Lastly, existingreal-world evidence stemming from individuals who later retract

False memories of autobiographical events can create enormousproblems in forensic settings (e.g., false accusations). While mul-tiple studies succeeded in inducing false memories in interviewsettings, we present research trying to reverse this effect (andthereby reduce the potential damage) by means of two ecologi-cally valid strategies. We first successfully implanted false mem-ories for two plausible autobiographical events (suggested by thestudents’parents, alongside two true events). Over three repeatedinterviews, participants developed false memories (measured bystate-of-the-art coding) of the suggested events under minimallysuggestive conditions (27%) and even more so using massive sug-gestion (56%). We then used two techniques to reduce false mem-ory endorsement, source sensitization (alerting interviewees topossible external sources of the memories, e.g., family narratives)and false memory sensitization (raising the possibility of falsememories being inadvertently created in memory interviews, de-livered by a new interviewer). This reversed the false memorybuild-up over the first three interviews, returning false memoryrates in both suggestion conditions to the baseline levels of thefirst interview (i.e., to∼15% and∼25%, respectively). By compar-ison, true event memories were endorsed at a higher level overalland less affected by either the repeated interviews or the sensiti-zation techniques. In a 1-y follow-up (after the original interviewsand debriefing), false memory rates further dropped to 5%, andparticipants overwhelmingly rejected the false events. One strongpractical implication is that false memories can be substantiallyreduced by easy-to-implement techniques without causing collat-eral damage to true memories.

Our results pose questions with both theoretical and practicalimpact. The current results show that infants readily learn fromforeign-language exposure, which raises a question about thehuman capacity to acquire more than one language. Because thephonetic contrasts of different languages often require differentperceptual groupings (in Japanese, for example,randlbelongto the same category, whereas in English they are separate),bilingually raised infants must learn two different ways ofclassifying speech sounds. Are there limits on infants’ abilities tolearn the phonetic cues of different languages? Infants in thecurrent tests readily acquired phonetic information from alanguage they had not previously heard, suggesting that, at leastat this age, infants can learn from exposure to two distinct lan-guages. A second question is the resilience of early phoneticlearning; that is, does short-term exposure during a sensitive periodfor learning have lasting effects? In our tests, the ability to discrim-inate the Mandarin contrast was unaffected after an almost 2-wkdelay. Follow-up studies on these infants are now underway toassess the long-term impact of early foreign-language exposure.Early language learning may find a useful biological frameworkin Greenough and Alcantara’s ‘‘experience-expectant’’ learning(51). Experience-expectant learning has two principal components:neural development that occurs in anticipation of the opportunityto learn and environmental information that is reliably present atthat time. Infants’ avid language learning in the second half-year oflife, shown both in the present study and in previous studies (3, 19),could be indicative of neurological development that enables in-fants to neurally code the properties of language. The environmentreliably provides language in a social context. Neural developmenton a maturational timetable and critical environmental informationin a social setting may combine to provide an example of experi-ence-expectant learning for language acquisition.Traditional theories describe phonetic learning as an innate‘‘modular’’ process in infants, one specific to speech and encap-sulated in a way that isolates it from general systems (52, 53). Analternative view, buttressed by data on categorical perceptiontests in infants and animals, argues that infants’ initial capacitiesare based on more general perceptual systems (54). The presentdata support the view that language acquisition initially draws ona broad set of perceptual, cognitive, and social abilities. Acorollary argument, addressing evolutionary origins, suggeststhat language evolved to capitalize on preexisting general sys-tems and then went beyond them (55, 56). The characteristicsthat facilitate language acquisition in children may thus haveinfluenced the nature of language itself, ensuring that languagewas learnable by infants in natural settings.

The experiments reported here examine the effects of foreign-language intervention on the decline normally observed ininfants’ perception of foreign-language phonetic units at the endof the first year of life. Experiment 1 asked whether first-timeforeign-language exposure between 9 and 10 mo of age inducesphonetic learning. Experiment 2 asked whether learning fromforeign-language exposure at this age occurs in the absence ofsocial interaction.

The attention scores for infants in the AV and A groupsrevealed that they visually attended less than infants in the liveexposure sessions of Experiment 1. For infants in the AV group,visual attention ratings were 2.79 (range2.14–3.33); for infantsin the A group, the attention scores were 1.48 (range1.03–2.06). Across the two experiments, attention scores for thefour groups differed significantly,F(3, 45)101.76,P0.001.Post hoc tests revealed that the AV and A group attention scoresdiffered significantly from each other, and that both AV and Agroup attention scores differed significantly from the two liveconditions (live Mandarin3.53; live English3.56) (P0.001for all comparisons).

Methods and Materials.Participants.The participants were 32full-term infants, assigned randomly to one of two groups, 16(eight boys) in each. Both groups were exposed to MandarinChinese, one via A and the other via AV input. Criteria for infantparticipants were identical to Experiment 1. The mean age ofinfant participants when the language exposure sessions beganwas 9.29 mo (278.94 days, range272–287 days) for the AVgroup and 9.30 mo (279.06 days, range273–287 days) for theA group. Of the 32 infants who began the exposure sessions, 28completed all of the exposure sessions and the behavioral testing(15 in the AV group and 13 in the A group).Language exposure sessions.As in Experiment 1, each infant tookpart in 12 language exposure sessions, each 25 min in duration,scheduled over a 4-wk period. During these sessions, 16 infantsin the AV group received foreign-language exposure from DVDmovies on a 17-in Panasonic (Seacaucus, NJ) television withboth an audio and video signal. The 16 infants in the A groupreceived foreign-language exposure via A exposure to the DVD

Experiment 2 examined the role of social interaction in phoneticlearning from a foreign language. If phonetic learning is trig-gered solely by linguistic input, then exposure to languagematerial via DVD (AV or A) should result in learning. Alter-natively, if learning from complex natural language exposure inhumans is enhanced by social interaction, as it is in song learningfor certain avian species, then exposure without human inter-action may not be sufficient to induce phonetic learning.

Mandarin Chinese test stimuli.The identical Mandarin Chinesephonetic stimuli used in Experiment 1 were used to test infantsin Experiment 2 (Fig. 1).Phonetic perception test.The identical testing procedure, equipment,and testers used in Experiment 1 were used in Experiment 2.Attention scores.As in Experiment 1, visual attention during the12 exposure sessions was scored on a five-point scale by anassistant observing the infants via video monitor outside thesoundproof room. The rater coded the degree to which infantsfocused their attention on the video screen during the AV andA sessions.

Discussion.The results of Experiment 1 confirmed the hypothesisthat foreign-language intervention at 9 mo of age would alterphonetic perception for a phonetic contrast of that language.Exposure to infant-directed speech by four speakers of Mandarinin 12 sessions between 9 and 10 mo of age was sufficient toreverse the decline seen in the English control group, a declinetypically observed in the absence of experience for foreign-language phonetic contrasts. The American infants exposed toMandarin Chinese performed at a level comparable to that ofinfants raised in Taiwan and significantly better than infants inthe English control group who were not exposed to Mandarin.The finding suggests that short-term exposure to a foreignlanguage is sufficient to induce learning at 9 mo of age.

Additional analyses reveal that learning in the current exper-iments was not short-lived. There was a substantial range in thedelay between an individual infant’s last language exposuresession and the test of Mandarin phonetic discrimination. Infantsreturned to the laboratory for their discrimination tests between2 and 12 days after the final exposure session, with a median of6 days intervening between final exposure and test. This allowedus to examine whether longer delays resulted in poorer discrim-ination performance. Infants were divided into two groups onthe basis of whether their test took place before or after themedian delay in days between exposure and test. The resultsshowed that there were no significant differences betweendiscrimination performance based on the length of time betweenexposure and test (Mann–WhitneyUtest6.50,P

s.The results demonstrate that live exposure to MandarinChinese at 9 mo of age reverses the decline typically seen inforeign-language speech perception, a decline experimentallyverified by the English control group. The results demonstratethe effects of foreign-language intervention (Fig. 2A).

Attention scores.Each infant’s visual attention during the 12exposure sessions was scored on a five-point scale (1inatten-tive, 5very attentive) by an assistant observing the infants viavideo monitor outside the soundproof booth. The rater codedthe degree to which the infant focused attention on the foreign-language speaker or the books or toys the speaker used.

The HT procedure consisted of two phases: conditioning andtest. In the conditioning phase, only change trials were run, sothat infants learned the association between a change in thesound and the presentation of the visual reinforcer. Duringconditioning, the target sound was initially presented at a louderlevel than the background sound (4 dB sound pressure level)to alert infants to the sound change, and the reinforcer wasautomatically activated after two target sounds were presentedso the infant learned to associate a change in the sound with thereinforcer. After two consecutive correct HTs that anticipate thepresentation of the toy, the intensity cue was removed. Afterthree consecutive anticipatory HTs with no intensity cue, the testphase began.

Phonetic perception test.A head-turn (HT) conditioning procedurefrequently used in tests of infant speech perception (31) was usedto test infants’ Mandarin speech discrimination. To ensure thatboth groups of infants were equally good at discriminating anative phonetic contrast, HT tests were also conducted using anAmerican Englishra-lacontrast.

Language exposure sessions.Infants took part in 12 languagesessions, each 25 min in duration, scheduled over a 4-wk period.During these sessions, native speakers of Mandarin (or English)read from children’s books for 10 min and played with toys for15 min using prescribed materials. Children’s storybooks weretranslated into Mandarin for the reading period, and various toys(puppets, a train, and ring stacks) were provided for the adultsto use during the play period. The same materials were used tointeract with infants during each session. Four native Mandarin(or English) speakers (both male and female) took turns con-ducting the language sessions so that infants were exposed to avariety of speaking styles over the 12 sessions.

The participants were 32full-term infants, assigned randomly to one of two groups: 16(eight boys) to the Mandarin exposure group and 16 (eight boys)to the English control group. Criteria for infant participantsincluded: (i) English as the only language spoken in the house-hold; (ii) no known physical, sensory, or mental handicap; (iii)gestational age at birth at 403 weeks; and (iv) birth weightbetween 5.5 and 10 lb. The mean age of infant participants whenthe language exposure sessions began was 9.3 mo (282.71 days)(range275–314 days) for the Mandarin exposure group and9.3 mo (284.59 days) (range265–319 days) for the Englishcontrol group. Of the 32 infants who began the exposuresessions, 21 completed all 12 of the exposure sessions and thebehavioral testing (10 in the Mandarin exposure group and 11 inthe control group).

A second issue investigated in the present experiments iswhether phonetic learning during this period is enhanced bysocial interaction

Exposure to a particular language has another consequence:it reduces sensitivity to foreign-language speech

Moreover, there is increasing evidence that infantlearning relies on sensitivity to the statistical properties con-tained in language input.

Infants acquire language with remarkable speed, although little isknown about the mechanisms that underlie the acquisition pro-cess. Studies of the phonetic units of language have shown thatearly in life, infants are capable of discerning differences amongthe phonetic units of all languages, including native- and foreign-language sounds. Between 6 and 12 mo of age, the ability todiscriminate foreign-language phonetic units sharply declines. Intwo studies, we investigate the necessary and sufficient conditionsfor reversing this decline in foreign-language phonetic perception.In Experiment 1, 9-mo-old American infants were exposed tonative Mandarin Chinese speakers in 12 laboratory sessions. Acontrol group also participated in 12 language sessions but heardonly English. Subsequent tests of Mandarin speech perceptiondemonstrated that exposure to Mandarin reversed the declineseen in the English control group. In Experiment 2, infants wereexposed to the same foreign-language speakers and materials viaaudiovisual or audio-only recordings. The results demonstratedthat exposure to recorded Mandarin, without interpersonal inter-action, had no effect. Between 9 and 10 mo of age, infants showphonetic learning from live, but not prerecorded, exposure to aforeign language, suggesting a learning process that does notrequire long-term listening and is enhanced by social interaction.

OUT-OF-BODY EXPERIENCE2. The subject sees the hand prodding towards the camera as he feels his chest being prodded. He also sees his body from behind. This creates a vivid sense that his real body is oating behind the one he sees.

Not everyone succumbs. Ehrsson sus-pects that people who can expertly localize their limbs without sight, such as dancers or musicians, would be less susceptible than the students with whom he normally works.

Henrik’s work speaks to the idea that there is no such thing as a soul or a self that’s independent of the brain

u can be certain of in this world, it’s that your hand is your hand,” says Ehrsson. Yet Ehrsson’s illusions have shown that such cer-tainties, built on a lifetime of experience, can be disrupted with just ten seconds of visual and tactile deception. This surprising malle-ability suggests that the brain continuously constructs its feeling of body ownership using information from the senses — a finding that has earned Ehrsson publications in Scienceand other top journals, along with the atten-tion of other neuroscientists.

hey have swapped bodies with another person1, gained a third arm2, shrunk to the size of a doll or grown to giant proportions3.

Henrik Ehrsson, a neuroscientist at the Karolinska Institute in Stockholm, who uses illusions to probe, stretch and displace people’s sense of self

Ehrsson is now working to turn his illusions towards a practical goal: developing better arti-ficial limbs

Accord-ing to textbook wisdom, people build up a perception of their bod-ies using ‘proprioception’ — signals from the skin, muscles and joints that indicate the relative position of body parts.

Participants spent a median of 37 min per class browsing the Internet for non-class-related purposes with their laptops. They spent the most time using social media, followed by reading e-mail,2 shopping, watching videos, chatting, reading news, and playing games (Table 1). Social media sites also had the highest number of HTTP requests, but thereafter the order differed: shopping, watching videos, reading e-mail, chatting, reading news, and playing games. Note that the minutes and requests do not add up to the total Internet time. The remaining time and requests reflect Internet use that did not fall into one of the seven categories. These URLs were related to checking background certificates for a Web site; Google-provided services such as calendar, maps, and analytics; visits to university sites (e.g., the registrar); and advertisements. Students also browsed the class-related Web sites for 4 min and approximately 5 requests per class session.

Five hundred seven students enrolled in an introductory psychology class in fall 2014 were invited to participate in this experiment for course credit. One hundred twenty-seven students consented to participate. The final sample consisted of the 84 participants who checked into the proxy server during more than half of the 15 sessions and logged in an average of 12.7 times. Participants who logged in for less than half of the sessions were excluded from analysis.The majority of participants were freshmen (56.5%) and sophomores (33.9%); there were a few juniors (5.7%) and seniors (4.0%). These percentages were similar to the percentages across the entire class of 507 students (47.1%, 34.8%, 12.7%, and 5.4%, respectively), although more freshmen and fewer juniors participated. Participants did slightly better on the final exam (M = 81.4%, SD = 10.54) compared with the class average of 78.3% (SD = 11.7). This most likely reflects the better attendance of partici-pants, given that they were required to attend class to participate in the study. In fact, participants attended more lectures (83.8%) than the class average (80.3%).

It is also possible, however, that phone use could play a positive role insocial interactions. When a conversation lags or turns to dull topics,smartphones could provide reliable access to an array of brief engagingactivities. Researchers have theorized that engaging one's attention withdesired stimuli should decrease boredom, speed the passage of time, andpromote a sense of agency (for a review, seeEastwood, Frischen, Fenske, &Smilek, 2012). By allowing us to engage our attention with an array ofstimuli on demand, therefore, phones may decrease boredom, make timepass more quickly, and give us a greater sense of control.

In Study 2, we surveyed people 5 times a day for one week, askingthem to report how they had been feeling and what they had beendoing over the past 15 min