Adolescents' Cognitive Capacity Reaches Adult Levels Prior to Psychosocial Maturity (PDF)

Law and Human Behavior © 2019 American Psychological Association 2019, Vol. 43, No....

Law and Human Behavior © 2019 American Psychological Association 2019, Vol. 43, No. 1, 69 – 85 0147-7307/19/$12.00 http://dx.doi.org/10.1037/lhb0000315 Adolescents’ Cognitive Capacity Reaches Adult Levels Prior to Their Psychosocial Maturity: Evidence for a “Maturity Gap” in a Multinational, Cross-Sectional Sample Grace Icenogle, Laurence Steinberg, Natasha Duell, Lei Chang and Jason Chein University of Macau Temple University This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Nandita Chaudhary Laura Di Giunta This document is copyrighted by the American Psychological Association or one of its allied publishers. University of Delhi Università di Roma “La Sapienza” Kenneth A. Dodge Kostas A. Fanti Duke University University of Cyprus Jennifer E. Lansford Paul Oburu Duke University Maseno University Concetta Pastorelli Ann T. Skinner Università di Roma “La Sapienza” Duke University Emma Sorbring Sombat Tapanya University West Chiang Mai University Liliana M. Uribe Tirado Liane P. Alampay Universidad de San Buenaventura Ateneo de Manila University Suha M. Al-Hassan Hanan M. S. Takash Hashemite University and Emirates College for Hashemite University Advanced Education Dario Bacchini University of Naples “Federico II” All countries distinguish between minors and adults for various legal purposes. Recent U.S. Supreme Court cases concerning the legal status of juveniles have consulted psychological science to decide where to draw these boundaries. However, little is known about the robustness of the relevant research, because it has been conducted largely in the U.S. and other Western countries. To the extent that lawmakers look to research to guide their decisions, it is important to know how generalizable the scientific conclusions are. The present study examines 2 psychological phenomena relevant to legal questions about adolescent maturity: cognitive capacity, which undergirds logical thinking, and psychosocial maturity, which comprises individuals’ ability to restrain themselves in the face of emotional, exciting, or risky stimuli. Age patterns of these constructs were assessed in 5,227 individuals (50.7% female), ages 10 –30 (M ⫽ 17.05, SD ⫽ 5.91) from 11 countries. Importantly, whereas cognitive capacity reached adult levels around age 16, psychosocial maturity reached adult levels beyond age 18, creating a “maturity gap” between cognitive and psychosocial development. Juveniles may be capable of deliberative decision Grace Icenogle, Laurence Steinberg, Natasha Duell, and Jason Chein, sity of Delhi; Laura Di Giunta, Department of Psychology, Università di Department of Psychology, Temple University; Lei Chang, Department of Roma “La Sapienza”; Kenneth A. Dodge, Center for Child and Family Psychology, University of Macau; Nandita Chaudhary, Department of Policy, Duke University; Kostas A. Fanti, Department of Psychology, Human Development and Childhood Studies, Lady Irwin College, Univer- University of Cyprus; Jennifer E. Lansford, Center for Child and Family continued 69  70 ICENOGLE ET AL. making by age 16, but even young adults may demonstrate “immature” decision making in arousing situations. We argue it is therefore reasonable to have different age boundaries for different legal purposes: 1 for matters in which cognitive capacity predominates, and a later 1 for matters in which psychosocial maturity plays a substantial role. Public Significance Statement Cognitive capacity—the basic cognitive functions that serve as the foundation for higher-level, complex thinking processes—reaches adult levels during adolescence (around 16). In contrast, psychosocial maturity— one’s ability to exercise self-restraint in emotional situations—reaches adult levels during the 20s. Importantly, in a study of over 5,200 participants, these distinct age patterns emerge across 11 diverse countries around the world. Thus, having two legal age boundaries that distinguish adolescence and adulthood— one for decisions typically made with deliberation and another for decisions typically made in emotionally charged situations—may be more sensible than having just one. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. This document is copyrighted by the American Psychological Association or one of its allied publishers. Keywords: adolescence, maturity, law, age of majority, cross-national Supplemental materials: http://dx.doi.org/10.1037/lhb0000315.supp All developed societies draw chronological age boundaries be- protection in criminal matters: Children suffer from deficient tween minors and adults for legal purposes, among them, deter- decision-making abilities, which makes them less responsible for mining who is permitted to vote, drive, purchase alcohol, and make their bad acts; children are more amenable to rehabilitation, so they autonomous medical decisions and, if arrested, who is tried as an should be reformed, not punished (Davis, Scott, Wadlington, & adult. In many countries, age 18 is used for most purposes with Whitebread, 2009; Woolard & Scott, 2009). some exceptions (e.g., consent to research or medical treatment; The delineation of a specific age-boundary that separates chil- World Health Organization, 2014). In others, such as the United dren from adults has often resulted from practical considerations, States, different ages are used for different matters. For example, without reference to relevant empirical and theoretical foundations although the presumptive age of majority in the U.S. is 18, eligi- (Scott, 2000). For example, initially there were no age restrictions bility for driver’s licensing is generally granted at a younger age, for driving. As traffic safety became a concern, states began setting whereas the minimum legal purchase age for alcohol is 21. a minimum driving age, typically 18 (Mayhew, Fields, & Simpson, The idea that young people lack certain capacities or abilities 2000). In the 1920s and 1930s, many states lowered the driving necessary to assume the responsibilities or enjoy the privileges of age from 18 to 16 to allow minors to work in occupations requiring adulthood is undoubtedly part of the logic behind differentiating a vehicle. As teen driving fatalities increased, many states adopted between adults and minors in the law (Woolard & Scott, 2009). “graduated driver-licensing” in the 1990s, which lets 16-year-olds For example, the infancy doctrine, which allows minors who enter drive, but only under certain circumstances (e.g., no other teen into contracts to void them at their discretion, was fashioned to passengers in the car; Williams, 1999). protect minors from their immature judgment as well as adults who In other instances, political considerations led legislators to might capitalize on youths’ lack of understanding of the conse- draw or change legal boundaries. At the height of the Vietnam quences of the contract (Preston & Crowther, 2012). In the early War, when the military draft age was 18, the voting age was 21. 20th century, legislators established a separate justice system in the Many politicians argued that it was unfair to send 18-year-olds into U.S. for juveniles (Scott & Steinberg, 2008) based on similar logic, battle but prohibit them from voting, and Congress amended the namely, that children differ from adults in ways that require special Constitution in 1971 to lower the voting age to 18. In response, Policy, Duke University; Paul Oburu, Department of Educational Psy- The data reported here are drawn from a larger study of decision chology, Maseno University; Concetta Pastorelli, Department of Psychol- making in everyday life and an ongoing longitudinal study of parenting ogy, Università di Roma “La Sapienza”; Ann T. Skinner, Center for Child across cultures (funded by the Eunice Kennedy Shriver National Insti- and Family Policy, Duke University; Emma Sorbring, Centre for Child and tute of Child Health and Human Development, grant RO1-HD054805). Youth Studies, University West; Sombat Tapanya, Department of Psychi- Previous publications and parallel conference presentations have drawn atry, Chiang Mai University; Liliana M. Uribe Tirado, Consultorio Psi- on portions of the data described in the present study (e.g., Duell et al., cológico Popular, Universidad San Buenaventura; Liane P. Alampay, De- partment of Psychology, Ateneo de Manila University; Suha M. Al- 2017; Icenogle et al., 2016; Steinberg et al., 2017). This research was Hassan, Queen Rania Faculty for Childhood, Hashemite University, and supported by a grant to Laurence Steinberg from the Klaus J. Jacobs Emirates College for Advanced Education; Hanan M. S. Takash, Queen Foundation. Rania Faculty for Childhood, Hashemite University; Dario Bacchini, De- Correspondence concerning this article should be addressed to Grace partment of Humanistic Studies, University of Naples “Federico II”. Icenogle, who is now at Department of Psychological Science, University Natasha Duell is now at the Center for Developmental Science and Department of California, Irvine, 4201 Social and Behavioral Sciences Gateway, Ir- of Psychology and Neuroscience, University of North Carolina, Chapel Hill. vine, CA 92697. E-mail: [email protected]  ADOLESCENT MATURITY GAP 71 some states lowered their legal drinking age from 21 to 18 or 19 conditions but more poorly—and similarly with younger teenag- (Cook & Tauchen, 1984). Because not all states did this, young ers—when tested under arousing ones (Cohen et al., 2016). people living in places with higher drinking ages would drive Legal issues pertaining to cold cognition include voting, grant- across state lines to purchase and consume alcoholic beverages— ing consent for research participation, and making autonomous and then drive back home intoxicated. In 1984, under pressure medical decisions, where the presence of adult consultants and the from the federal government, all states raised the minimum legal absence of time pressure impose sufficient external control to drinking age back to 21. minimize the dangers of impulsive decision making (Grisso et al., Until recently, developmental psychology has not been an ex- 2003; Hein et al., 2015). Issues related to hot cognition include plicit force in determining specific legal age boundaries—when driving, consuming alcohol, and criminal behavior; it is easy to legislators lowered the driving age or raised the drinking age, no make impulsive choices when emotions are aroused, such as when one asked whether research on psychological development sup- behind the wheel, intoxicated, or committing a crime, behaviors ported either change. However, developmental science has grad- that often occur in the presence of peers during adolescence ually become more influential (Steinberg, 2017). In both legal and (Albert & Steinberg, 2011). This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. nonlegal venues, experts have weighed in on whether, and at what Importantly, the developmental trajectories of cold and hot This document is copyrighted by the American Psychological Association or one of its allied publishers. age, the law should distinguish between adolescents and adults. cognition differ. In studies using cold cognitive tasks, performance But considerable controversy has arisen because scientists have increases dramatically from childhood to early- or midadolescence answered this question in different ways, depending on the legal and then plateaus. This pattern is demonstrated on tasks of re- issue involved. sponse inhibition (e.g., Huizinga, Dolan, & van der Molen, 2006; In 2005, when the American Psychological Association (APA) Luna, Garver, Urban, Lazar, & Sweeney, 2004), cognitive flexi- submitted an amicus brief in Roper v. Simmons (2005) the U.S. bility (e.g., Crone, Ridderinkhof, Worm, Somsen, & van der Mo- Supreme Court case that abolished the juvenile death penalty for len, 2004), and working memory (e.g., Huizinga et al., 2006; Luna 16- and 17-year-olds, the APA argued that people younger than 18 et al., 2004; Prencipe et al., 2011). A similar developmental pattern lacked the psychological maturity necessary to be held fully re- appears in the psycholegal literature as well. For example, younger sponsible for their crimes and, therefore, that they should not be juveniles (11–15 years) are impaired at significantly higher rates than older adolescents (16 –17 years) on measures of understand- eligible for capital punishment (American Psychological Associ- ing and appreciation of Miranda rights and on other measures of ation [APA], 2004). Justice Antonin Scalia, in his dissenting adjudicative competence (Viljoen, Zapf, & Roesch, 2007). But by opinion, criticized the APA, because in an earlier case its experts the time they are 16- to 17-years-old, adolescents and adults score had opined that teenagers should have the right to make decisions comparably on abilities relevant to competence to stand trial, about abortion without involving their parents, on the grounds that (Grisso et al., 2003; Redlich & Shteynberg, 2016). their decision-making abilities were just as mature as adults’ In contrast, facets of hot cognition, including sensation seeking (APA, 1989). (or lack thereof), impulse control, future orientation, and resistance A group of psychologists argued that this apparent logical to peer influence, follow a protracted development into adulthood. inconsistency was actually in keeping with developmental science Sensation seeking, which peaks during adolescence (Steinberg et (Steinberg, Cauffman, Woolard, Graham, & Banich, 2009a). They al., 2008), decreases into the early- to mid-20s (Harden & Tucker- contended that because different abilities mature along different Drob, 2011; Shulman, Harden, Chein, & Steinberg, 2014). On the timetables, adolescents of a given age could be adult-like in some other hand, impulse control, future orientation, and resistance to respects but not others. Based on analyses of data from over 900 peer influence improve into adulthood in studies employing either individuals between the ages of 10 and 30, they noted that cogni- self-report assessments (e.g., Harden & Tucker-Drob, 2011; Ro- tive capacity—the basic cognitive processes supporting the ability mer, Duckworth, Sznitman, & Park, 2010; Steinberg et al., 2009c; to reason logically—matures by 16, whereas psychosocial matu- Steinberg & Monahan, 2007) or behavioral tasks (e.g., Chein, rity—the capacity to exercise self-restraint, especially in emotion- Albert, O’Brien, Uckert, & Steinberg, 2011; Cohen et al., 2016; ally arousing contexts— does not fully mature until several years Steinberg et al., 2009c). Similarly, relative to adults, adolescents later. Steinberg, Cauffman, Woolard, Graham, and Banich (2009a) demonstrate impaired decision making in emotionally arousing argued that these patterns justify having a lower age boundary for contexts, such as when being interrogated by police (e.g., Malloy, legal decisions that allow deliberation and a higher age boundary Shulman, & Cauffman, 2014). To the extent that these legal for matters pertaining to acts typically made under emotionally contexts become emotionally arousing as a consequence of exter- arousing circumstances (Scott & Steinberg, 2008). nal pressures— by friends, family, police, or the adolescent’s own One way to think about the difference between these capacities lawyer—adolescents’ decision making lags behind adults’. Thus, and abilities is to distinguish between “cold” cognition and “hot” if our laws were more closely aligned with developmental science, cognition. Cold cognition refers to mental processes (such as age boundaries for matters involving cold cognition might be working memory or response inhibition) employed in situations lower than those involving hot cognition, because effective hot calling for deliberation in the absence of high levels of emotion cognition requires both cognitive capacity and psychosocial ma- (e.g., Figner, Mackinlay, Wilkening, & Weber, 2009). Hot cogni- turity (e.g., self-restraint). tion involves mental processes in affectively charged situations There is no universally agreed-upon way to measure cognitive where deliberation is unlikely or difficult. Recent research has capacities or psychosocial maturity. Steinberg et al. (2009a) mea- borne out this distinction, showing that on response inhibition sured cognitive capacity using tests of short-term memory (STM), tasks, young adults (aged 18 –21) perform comparably with some- working memory, and verbal fluency. These measures of executive what older individuals when tested under emotionally neutral functioning undergird goal-directed behavior and higher-order log-  72 ICENOGLE ET AL. ical reasoning (Diamond, 2013). Their measure of psychosocial (from age 18 to 25) in China, Germany, Italy, Spain, and Switzer- maturity was based on self-reports of traits such as impulse con- land, though impulsivity did not. trol, sensation seeking, future orientation, and resistance to peer The paucity of cross-cultural work in this area is regrettable influence. These measures tap individuals’ ability to restrain them- because there is reason to expect cultural variation in development, selves in the face of temptations to pursue rewarding, immediately particularly with respect to psychosocial maturity. Cross-cultural gratifying, socially encouraged, or risky activities. These measures studies of self-regulation focus on the socialization of appropriate capture some, but not all, important aspects of cognitive and behaviors (i.e., emotional displays), especially in social interac- psychosocial functioning that are relevant to decision making (e.g., tions. That is, to the extent that emotions motivate or precipitate neither measure included an index of morality or perspective behavior (e.g., feeling anger may lead to an act of aggression), taking; see Fischer, Stein, & Heikkinen, 2009 and Steinberg, emotion regulation is necessary to comply with the behavioral Cauffman, Woolard, Graham, & Banich, 2009b for a response). It norms of a culture (LeCuyer & Zhang, 2015; Matsumoto et al., is notable that Steinberg et al. (2009a) measure of cognitive 2008). In this view, parents and peers shape self-regulation by capacity comprised all behavioral tasks, whereas their measure of encouraging culturally appropriate behaviors and discouraging in- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. psychosocial maturity comprised all self-report measures. These appropriate behaviors (Chen & French, 2008). Cultures valuing This document is copyrighted by the American Psychological Association or one of its allied publishers. limitations notwithstanding, both constructs include measures that individuality (typically Western cultures) encourage autonomy and are essential to decision making competence in legal contexts— self-assertion, whereas more collectivistic cultures stress the im- executive functions facilitate flexible, optimized decision making portance of suppressing one’s desires to benefit the group (Chen & (Diamond, 2013) whereas elements of psychosocial maturity are French, 2008). strongly tied to what some writers have referred to as “judgment” Cultural variations in self-regulation also extend to the manage- (e.g., Scott, Reppucci, & Woolard, 1995; Steinberg & Cauffman, ment of positive emotions. Americans, for example, are less likely 1996). to dampen positive emotions than are people from East Asian countries (Ma, Tamir, & Miyamoto, 2018; Miyamoto & Ma, 2011). This cultural difference has particular implications for Overview and Rationale of Present Study sensation seeking, which often involves a lack of regulation of The present study replicates Steinberg et al. (2009a) in a large positive affect (e.g., the thrill or excitement of doing something international sample. Such a replication is warranted for several fun, but dangerous, with friends). Indeed, according to Hofstede reasons. First, because most of the relevant research has been (2011), cultures vary considerably along a dimension of conducted in Western countries, it is not known to what degree “indulgence-restraint,” the degree to which societal norms encour- conclusions drawn from these countries extend to non-Western age hedonic satisfaction rather than the strict regulation of im- societies. Further, of the few cross-national studies of cognitive pulses. In summary, in East Asian and collectivistic cultures, capacity or psychosocial development that do exist, most suffer at children must learn to suppress (regulate) undesirable behaviors least one major limitation. Studies typically examine only a single (LeCuyer & Zhang, 2015). Thus, self-regulation may develop developmental period (e.g., middle adolescence, excluding the earlier in these contexts than in Western/individualistic societies transition into adulthood). Furthermore, these studies often exam- for both positive and negative affect (Lamm et al., 2017). Com- ine mean differences between cultures and not age trends (e.g., parative cross-cultural research on the development of self- Thorell, Veleiro, Siu, & Mohammadi, 2013). In addition, apart regulation is limited and equivocal. Some studies indicate similar from a few studies (e.g., Matsumoto et al., 2008; Vazsonyi & developmental patterns cross-culturally, but other research exam- Ksinan, 2017), most cross-cultural work examines only a few ining the socialization of self-regulation suggests divergent devel- countries or cultures at a time. Although some studies have col- opmental trajectories. lected data from a large number of participants from many coun- A second reason for the present replication concerns the mea- tries (e.g., Rossier et al., 2007), such studies are limited to adult surement of psychosocial maturity. Steinberg et al. (2009a) em- samples and rely on self-report measures, which typically do not ployed behavioral tasks to assess cognitive capacity and self-report assess executive functions. To our knowledge, no cross-cultural measures to assess psychosocial maturity. One outstanding ques- study has measured multiple elements of both cognitive and psy- tion is whether previously documented differences in age patterns chosocial development within a single, multiage sample. To the of cognitive capacity and psychosocial maturity are an artifact of extent that lawmakers look to science to guide their decisions, it is differences in methodology. For example, whereas self-report important to know how robust the scientific conclusions are. This measures tap the individual’s subjective assessment of their be- is especially important with respect to laws within the U.S., with havior, behavioral tasks provide a brief snapshot of behavior while its ever-growing population of immigrants (Migration Policy In- controlling for context, an important consideration in cross-national stitute, 2018). studies. In the present study, we employ a measure of psychosocial Despite the limitations of most of the published cross-cultural maturity that is based mainly on behavioral assessments, which allows research, some studies have examined age patterns of legally us to more directly compare its growth with a measure of cognitive relevant psychological phenomenon. For example, cognitive ca- capacity that is also based on behavioral assessments. pacity improves with age across childhood and into adolescence in Finally, since its 2005 decision on the juvenile death penalty, the Kenya (Alcock, Holding, Mung’ala-Odera, & Newton, 2008), Supreme Court has heard several other cases in which develop- Japan (Imada, Carlson, & Itakura, 2013), and Hong Kong (Wang, mental science was applied. During the past decade, the Court has Devine, Wong, & Hughes, 2016). Using self-report indices of decided cases on the constitutionality of sentencing juveniles to facets of psychosocial maturity, Rossier et al. (2007) found that life without the possibility of parole (Graham v. Florida, 2010; sensation seeking decreased with age during young adulthood Miller v. Alabama, 2012) and on the admissibility of the results of  ADOLESCENT MATURITY GAP 73 interrogations of adolescents in situations in which they may be we rely largely on behavioral measures of psychosocial maturity. inadvertently encouraged to confess to crimes (JDB v. North Like the original measure of psychosocial maturity, these behav- Carolina, 2011). And, in light of new evidence that brain matu- ioral tasks tap various manifestations of self-restraint. Here, we ration continues into adulthood (Casey, 2015), a number of experts include behavioral measures of sensation seeking (using the “stop- have asked whether these findings support raising the age of light game”), future orientation (using a delay discounting task), majority under criminal law and processing young adult offenders and impulse control (using the Tower of London task). We also in the juvenile justice system (Schiraldi, Western, & Bradner, measure cost sensitivity, or one’s ability to learn from negative 2015). Importantly, the relevant research on psychological devel- outcomes, using a modified version of the Iowa gambling task (IGT). opment in young adulthood— especially comparing young adults This measure has not been used before as an indicator of psychosocial with people in their mid- and late 20s—is very limited (Scott, maturity, but it can be used to assess self-restraint. Specifically, Bonnie, & Steinberg, 2016). successful performance on the IGT requires that one learn to resist In the present study, we compare two facets of development potential rewards that also carry high costs. Lastly, because a behav- relevant to the treatment of young people under the law— cogni- ioral measure was unavailable, we rely on a self-report measure of This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tive capacity (the predominant influence on cold cognition) and resistance to peer influence. In its opinion in Roper v. Simmons This document is copyrighted by the American Psychological Association or one of its allied publishers. psychosocial maturity (the predominant influence on hot cogni- (2005), the Supreme Court explicitly cited teens’ greater susceptibility tion)— using some of the same tasks as Steinberg et al. (2009a), to others as a mitigating factor for their bad behavior. Thus, given its but in an 11-country sample of more than 5,200 individuals be- important place in questions of maturity, self-reported resistance to tween the ages of 10 and 30. Countries vary in how they socialize peer influence was retained in the current study. youth (Chen & French, 2008; Matsumoto et al., 2008), but the question of where to draw a boundary between adolescence and Hypotheses adulthood is one that all societies face. Accordingly, we examine the second two decades of life to determine whether and in what We hypothesized that cognitive capacity would reach adult ways age differences in cognitive capacity and psychosocial ma- levels prior to age 18 and plateau in midadolescence, but that turity are evinced in a diverse group of countries. The countries in psychosocial maturity would not reach adult levels until after 18, this sample—China, Colombia, Cyprus, Jordan, Kenya, India, into the 20s. These hypotheses are consistent with the idea that Italy, the Philippines, Sweden, Thailand, and the U.S.—are diverse deliberative, “cold” decision making matures prior to “hot” deci- geographically, economically, and culturally, including on dimen- sion making. Given the substantial cultural variation in expecta- sions of individualism/collectivism and indulgence/restraint (Hof- tions for self-regulation (which likely influences each component stede, 2011). For example, Columbia and China rank as some of of self-restraint within the psychosocial maturity composite), we the world’s most collectivistic cultures, whereas Italy and the U.S. anticipated that patterns of age differences in psychosocial matu- are some of the most individualistic. Likewise, China and India rity would vary notably across countries (Chen & French, 2008; greatly value restraint, whereas Sweden and Columbia are highly Matsumoto et al., 2008). In contrast, we expected the pattern of indulgent (see Table 1 for details on country-level attributes and age differences in cognitive capacity to generally be more consis- the online supplemental materials for details on legal age bound- tent across countries. aries by country). Consistent with Steinberg et al. (2009a), we examine age dif- Method ferences using composite measures of psychosocial maturity and cognitive capacity. These composite variables allow us to capture Participants multiple facets of an overarching construct (executive functions in the case of cognitive capacity and self-restraint in the case of We recruited nine of the 11 countries of the present sample from psychosocial maturity). However, unlike Steinberg et al. (2009a), an ongoing longitudinal study of parenting across cultures (PAC; Table 1 Country-Level Attributes GDP per capita Country Individualism/Collectivism Indulgence/Restraint (PPP) 2014 (USD) China 20 24 13,200 Columbia 13 83 13,500 Cyprus — 70 30,900 India 48 26 5,800 Italy 76 30 35,100 Jordan 30 43 12,000 Kenya 25 — 3,100 Philippines 32 42 7,000 Sweden 71 78 46,200 Thailand 20 45 15,600 U.S. 91 68 54,400 Note. The Individualism/Collectivism Scale and the Indulgence/Restraint Scale range from 0 –100 (Hofstede, 2011). GDP per capita is given in U.S. dollars (Central Intelligence Agency, 2018).  74 ICENOGLE ET AL. Lansford & Bornstein, 2011), which has been described elsewhere questions or issues. A central coordinating center received and (Steinberg et al., 2017). The PAC study originally selected these checked all incoming data each week. nine countries because they differ in how children are parented and Participants completed a session that lasted 2 hr. Staff members disciplined, which is the focus of that study. In particular, these tested participants individually in their homes, schools, or other nine countries differ on several levels: (a) individualism versus locations designated by the participants. Parental consent and collectivism; (b) religious affiliation; and (c) laws governing par- adolescent assent were obtained at all sites for all youth under 18 enting behaviors (e.g., the one-child policy in China that was in except Sweden, where parental consent is not required for partic- effect at the time of data collection; Lansford & Bornstein, 2011). ipants over 15. Participants completed computerized versions of all The current study has a different focus, but collaborating with the measures including self-report measures, behavioral tasks, an in- PAC group allowed us to build on their cross-national infrastruc- telligence assessment, and a demographic questionnaire. ture. In addition to the PAC countries, the current study included In order to maintain participants’ interest and motivation, they Cyprus and India. Cyprus participated in the current study, but not were told they would receive a base payment ($30 in the U.S.) for the PAC project, because PAC data collection had already begun participating in the study, but that they could earn a bonus (equal This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. prior to Cyprus’ involvement. India was unable to join the PAC to 50% of the base payment) based on their performance on the This document is copyrighted by the American Psychological Association or one of its allied publishers. group because the Indian Council of Medical Research did not computer tasks. In actuality, all participants received this bonus. approve the PAC study given that the sample was not nationally Research staff debriefed participants regarding this deception in representative of India. Because the National Institutes of Health, countries where local Institutional Review Boards (IRBs) deemed which funds the PAC study, required this approval to fund data such disclosure necessary. Local investigators set base payments collection, India could not participate in the PAC study. However, so as not to be coercive; the participating university in Sweden because the Jacobs Foundation funded the current study, such prohibits paying research participants, so these participants re- approval was not required, thus allowing India to participate. ceived a base payment of two movie tickets and a bonus of one The sample (N ⫽ 5,404) comprises individuals between 10 and additional ticket. Local IRBs approved all procedures. 30 years of age in 11 countries: China (n ⫽ 493), Colombia (n ⫽ 513), Cyprus (n ⫽ 407), India (n ⫽ 425), Italy (n ⫽ 561), Jordan Measures (n ⫽ 506), Kenya (n ⫽ 488), the Philippines (n ⫽ 512), Sweden (n ⫽ 425), Thailand (n ⫽ 504), and the U.S. (n ⫽ 570; see Table Analyses focused on a demographic questionnaire, a measure of S2 for a breakdown of participants by age in each country). We intellectual ability, three measures of cognitive capacity, and five balanced the proportion of males and females in the full sample measures of psychosocial maturity. (50.8% female, n ⫽ 2,746), within each country (ranging between Demographics. Participants reported their age, sex, and the 48.9%–53.8% female) and across age group (ranging between level of education of each of their parents, as a proxy for socio- 48.7%–52.0% female). Each site recruited participants from urban economic status. Due to small but significant differences among centers in each participating country using flyers posted in neigh- age groups, we added average parental education as a covariate in borhoods, advertisements in newspapers, and word of mouth. All all analyses. sites attempted to recruit a minimum of 60 participants (50% Intellectual ability. We used the Matrix Reasoning subtest of female) for each of seven age groups: 10 –11 years, 12–13 years, the Wechsler Abbreviated Scale of Intelligence (WASI; Psycho- 14 –15 years, 16 –17 years, 18 –21 years, 22–25 years, and 26 –30 logical Corporation, 1999), administered on a laptop, to estimate years. (Because the 10- to 11-year-old group comprised PAC nonverbal intellectual ability. (Verbal subscales of the WASI were participants, within PAC countries, the number of participants in excluded due to the variability in language across sites in the this group generally exceeded those in other age groups.) Partici- sample.) The WASI has been normed for individuals between the pants came from households with similar levels of parental edu- ages of 6 and 89 years; an age-normed score (i.e., t-score) was cation (average ⫽ “some college”) and reflected the majority computed for each participant (see Icenogle et al., 2016 for more ethnicity of the country (except in the U.S., where we tried to details about this measure). We included intellectual ability, which recruit equal numbers of Black, Latino, and White participants, may influence task performance, as a covariate in all analyses to and in Kenya, where participants were of the Luo ethnic group). control for small but significant age differences in intellectual functioning in some of the countries. Cognitive capacity. We computed a measure of cognitive Procedure capacity by averaging within-country standardized scores from At each data collection site, research staff received identical digit span, working memory, and verbal fluency tasks. training for administering the test battery. Measures were admin- Digit span. Participants recalled strings of digits beginning istered in the predominant language at each site, following with two digits and increasing to eight. The outcome of interest is forward- and back-translation and a process of cultural adaptation the highest number of digits recalled in reverse order (DSB). DSB (Erkut, 2010). Translators were fluent in both English and the taps working memory because it requires individuals to hold and target language, and identified any items that translated poorly, manipulate information held in memory (Diamond, 2013). were culturally insensitive or inappropriate, or may have multiple Working memory. Participants saw four probe letters on the meanings. Site coordinators and translators then modified items as screen, followed by a target letter (Thompson-Schill et al., 2002). appropriate. During data collection, investigators from each site Participants indicated whether the target letter was among the four attended an annual in-person meeting to resolve any questions, probes. On difficult trials, two of the four probes had appeared on concerns, or obstacles, and to review study procedures. In addition, the previous trial; on easy trials, none of the four probes had sites regularly used e-mail and Skype calls to resolve ongoing appeared in the two previous trials. Participants completed two  ADOLESCENT MATURITY GAP 75 blocks of 32 trials. The outcome of interest is average accuracy 4-point scale in all countries, we used dichotomized responses across all trials. (i.e., which of the two statements was selected)—which yield Verbal fluency. Participants generated as many words as pos- better model fit indices—to compute scales scores. Further, psy- sible belonging to a given category in 1 min. Three categories were chometric properties of the RPI (based on fit indices from CFA) used: fruits, vegetables, and animals. The number of valid words were improved when using only seven of the original 10 items. (i.e., those that were not proper nouns, repeats, or different forms Thus, we used only dichotomized responses from these seven of the same word) generated within each category were averaged items. (See the online supplemental materials for details of this to create an overall fluency score. Kenyan participants scored procedure.) Reliabilities ranged from ␣ ⫽.43 (Philippines) to.79 notably lower than any other country (e.g., two thirds of the sample (India; see supplemental materials for reliabilities in each country did not produce one example of a fruit). A discussion with the and the full 10-item scale). principal investigator in Kenya revealed that these low scores may The stoplight task. To obtain a behavioral index of sensation be the result of low exposure to a variety of fruits, vegetables, and seeking, participants completed a computerized driving task called animals, or to performance-related anxiety. Further, Kenyans who the “stoplight task” (Steinberg et al., 2008). In this task, partici- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. have attended school are often multilingual, and a language mix-up pants approached a series of 20 intersections at which they decided This document is copyrighted by the American Psychological Association or one of its allied publishers. could have contributed to poor performance. Accordingly, the whether to run a stoplight as it turned yellow, or to stop safely. If cognitive capacity variable for Kenyan participants excluded ver- the participant chose to stop, he or she must wait 3 s before bal fluency. restarting. If the participant ran the light, he or she either passed Psychosocial maturity. We computed a composite measure through successfully (resulting in no loss of time) or crashed into of psychosocial maturity by averaging within-country standardized another car (resulting in a loss of 6 s). Performance on this task is scores from five measures: self-reported resistance to peer influ- associated with self-reported sensation seeking (Chein et al., 2011; ence, the stoplight task, delay discounting, the modified Iowa Steinberg et al., 2008). We defined sensation seeking as the pro- gambling task, and the Tower of London task. Although Steinberg portion of lights run (regardless of whether the participant passed et al. (2009a) also included a measure of risk perception, this safely through the intersection). Z-scores for this measure were measure was excluded from the current analyses (although it was reversed so that higher values indicate greater restraint (i.e., less administered). Unlike the other psychosocial measures, the risk sensation seeking). perception questionnaire asks about specific risky behaviors (e.g., Delay discounting. We employed a computerized delay dis- riding in a car with a drunk driver), perceptions of which may counting task to assess individuals’ future orientation (see Stein- differ across countries simply because the behavior is more or less berg et al., 2009a for details). In this task, participants made normative or common (e.g., sanctions against alcohol use in Jordan hypothetical decisions between an immediate but smaller reward give intoxicated driving a different meaning than in the United and a delayed but larger reward. The value of the delayed reward States). was held constant at 1,000 units of local currency. The starting These specific measures were chosen both because they tap value of the immediate reward was randomly determined for each aspects of development frequently cited in legal debates, and participant to be 200, 500, or 800 units of currency. Our version of because they are behavioral, not self-report, assessments. By using the task uses six delay periods: 1 day, 1 week, 1 month, 3 months, behavioral measures of both psychosocial maturity and cognitive 6 months, and 1 year (e.g., “Would you rather have 200 euros capacity, we are able to determine whether previously reported today or 1,000 euros in 6 months?”). The size of the immediate divergent age patterns of the two are merely a reflection of mea- reward was adjusted after each offer to converge at a value surement type. Although our measures, like those used by Stein- reflecting the subjective value of the delayed reward if it were berg et al. (2009a), reflect multiple domains of psychosocial func- offered immediately, referred to as the “indifference point” tioning, we create an aggregate for two main reasons. (Ohmura, Takahashi, Kitamura, & Wehr, 2006). As the delay Conceptually, these measures are related; each is undergirded by period lengthens, one must have a stronger sense of future orien- self-restraint (which manifests in different ways). Second, it is tation to forgo the immediate reward. That is, when waiting only important to produce scientific research that meaningfully guides 1 day or 1 week to receive a reward, it is relatively easy to forgo the legal field. Practically speaking, the courts and policymakers the immediate option. Longer delays, then, may better inform our require simple explanations to complicated questions (Steinberg, understanding of future orientation because they require projection Cauffman, Woolard, Graham, & Banich, 2009b). Thus, we con- into the extended future. Accordingly, indifference points for the solidate across measures to make our study more directly applica- three longest delay intervals (3 months, 6 months, and 1 year) were ble to legal settings. averaged and used as a measure of future orientation. A higher Resistance to peer influence. We used the Resistance to Peer value indicates a stronger willingness to forgo an immediate Influence Scale (RPI; Steinberg & Monahan, 2007), a 10-item smaller reward for a more valuable reward in the future (i.e., questionnaire that uses a two-stage response format (see online