Urge: Negative urgency Premed: Premeditation Persev: Perseverance Senseek: Sensation seeking BART: Average number of pumps per balloon on the Balloon Analogue Risk Task SLG: Percent of intersections where a participant ran through a yellow light on the Stoplight Game IGT: Learned avoidance of disadvantageous card decks by the end of the Iowa Gambling Task TOL: Average number of excess moves made on each trial of the Tower of London

A Multidimensional Model of Adolescent Risky Decision-Making: Integrating Behavioral and Self Report Measures

Natalie Kretsch, M.A.1, K. Paige Harden, Ph.D.1, Laurence D. Steinberg, Ph.D.,2 & Elliot M. Tucker-Drob, Ph.D.1 1The University of Texas at Austin 2Temple University

Methods Participants: 338 adolescents (52% male), ages 14-19 (Mage = 15.89, SD = 1.56), recruited from the Texas Twin Project (Harden, Tucker-Drob, & Tackett, 2013). Sixty-one percent were non-Hispanic White, 23% were Hispanic/Latino, 11% were African American, and 5% were of other race/ethnicity. Procedure: Adolescents completed behavioral and self-report measures of impulsivity, sensation-seeking, and reward sensitivity as well as an IQ test during a 3 hour lab session. Measures: • Self-reported impulsivity was assessed with the UPPS Impulsivity Scale (Whiteside & Lynam, 2001), a self-report measure that distinguishes four facets of impulsivity: sensation-seeking, perseverance, premeditation (failure to plan ahead), and negative urgency (the tendency to act without thinking when experiencing negative affect). • Behavioral measures included the Tower of London (Shallice, 1982), a test of strategic planning; the Iowa Gambling Task (Bechara, 2007), a measure of affective decision-making; the Balloon-Analogue Risk Task (Lejuez et al., 2002), a test of risky decision-making; and the Stoplight Game (Gardner & Steinberg, 2005), a simulated driving task that measures risky decision-making. • Intelligence (IQ) was assessed with the Wechsler Abbreviated Scale of Intelligence (WASI-IV). • Delinquency was assessed with a 36 item self-report survey. Adolescents were asked if they had ever engaged in a number of risky behaviors (e.g., including speeding, trespassing, runaway, theft, cheating, vandalism, and drunk driving). Items were rated on a 3-point scale (0=Never, 1=Once, 2=More than once) and responses were averaged to compute a mean delinquency score.

Background Understanding what makes adolescence a period of unique risk—for some more than others—is of considerable public health relevance, given that morbidity and mortality during this stage of life are largely the result of health risk behavior. Stemming from research in developmental neuroscience (e.g., Steinberg, 2008; Casey, Getz, & Galvan, 2008), theoretical work in this area has crystallized on two dimensions that are thought to contribute to adolescents’ greater propensity for risk-taking: reward sensitivity and impulse control. To assess these constructs, researchers have drawn on both self-report (survey) and behavioral (laboratory) measures. However, it is unclear whether different measurement modalities indeed tap the same underlying constructs. Few studies have examined how laboratory tasks of decision-making overlap with each other, with self-report measures, and with general cognitive ability. The current study examined two research questions: • What is the phenotypic overlap of behavioral and self-report measures of impulsivity, sensation-seeking, and reward sensitivity? • How do these constructs predict “real world” delinquent behavior?

Behavioral Measures

Discussion Consistent with a growing body of work supporting the dual

systems model of risky decision-making (Steinberg, 2008; Casey et al., 2008; Ellingson, 2013; Harden & Tucker-Drob, 2011), a reward seeking factor, indicated by performance on two risky decision-making tasks and by self-reported sensation-seeking, was independent from both behavioral and self-report measures of cognitive control and ability. Impulsivity and reward seeking predict delinquency in a manner consistent with the dual systems model. However, their interactive effects were suppressed among adolescents with high cognitive ability. Future research should examine the genetic and/or environmental underpinnings of these associations. In addition, future research may reveal whether these findings generalize to a higher risk sample.

These results illustrate the multidimensional nature of behavioral inhibition, a trait with relevance to developmental, clinical, cognitive, and personality psychologists. Laboratory measures of behavioral disinhibition show a great deal of task-specific variance and tap multiple cognitive and affective processes. Studies examining can benefit from using a multivariate measurement approach that combines self-report and behavioral assessments.

Results

*

*

*

*

What is the phenotypic overlap of behavioral and self-report measures of impulsivity, sensation-seeking, and reward sensitivity? Phenotypic correlations (not shown) between self-report measures and performance on behavioral tasks were modest. Exploratory factor analyses revealed three factors that were measured by this battery of assessments: Cognitive ability, self-reported impulse control (factor 2), and reward seeking. The three factor model fit the data well (RMSEA = .03; CFI = .97; TLI = .95; χ 2 = 28.45, df = 21, p = .13). How do these constructs predict self-reported risk behavior? •  Main effects: Low impulse control and high reward-seeking predicted

higher delinquency. •  Two-way interactions: Association between impulse control and

delinquency was stronger for adolescents with high reward-seeking and low cognitive ability. High impulse control attenuated the link between cognitive ability and reward-seeking.

•  Three-way interaction: As illustrated below, for adolescents with high cognitive ability, neither reward-seeking nor impulse control predicted delinquency.

Acknowledgements: Thank you to Rachel Polk and Jonathon Flores for their assistance with data collection and to Kaitlyn Breiner for assistance in data analysis. This research was supported by NIH grant R21-AA020588-02 awarded to Dr. Harden and Dr. Tucker-Drob at the University of Texas at Austin and Dr. Steinberg at Temple University. References: Steinberg (2008). A social neuroscience perspective on adolescent risk-taking. Developmental Review, 28, 78 – 106 � Casey, Getz & Galvan (2008). The adolescent brain. Developmental Review 28, 62-77 � Harden, Tucker-Drob, & Tackett (2013). The Texas Twin Project. Twin Research and Human Genetics, 16, 385-390 � Whiteside & Lynam (2001). The five factor model and impulsivity: using a structural model of personality to understand impulsivity. Personality and Individual Differences, 30(4), 669–689 � Shallice, (1982). Specific impairments in planning. Philosophical Transactions of the Royal Society of London. Biological Sciences 298, 199-209. � Bechara, (2007). Iowa Gambling Task professional manual. Lutz, FL: Psychological Assessment Resources. � Lejuez et al. (2002). Evaluation of a behavioral measure of risk taking: the Balloon Analogue Risk Task (BART). Journal of Experimental Psychology: Applied, 8, 75. � Gardner & Steinberg (2005). Peer influence on risk taking, risk preference, and risky decision making in adolescence and adulthood: An experimental study. Developmental Psychology, 41, 625– 635. � Ellingson et al. (2013). Are bottom-up and top-down traits in dual-systems models of risky behavior genetically distinct? Behavior Genetics, 43, 480-490. � Harden & Tucker-Drob (2011). Individual differences in the development of sensation seeking and impulsivity during adolescence: further evidence for a dual systems model. Developmental Psychology, 47, 739–46.

APPENDIX 7

APPENDIX 7

Iowa Gambling Task (IGT) Tower of London (TOL)

Balloon Analogue Risk Task (BART) The Stoplight Game (SLG)

APPENDIX 7

Tower of London

The object of this puzzle is to move the beads on the left so that they are in the goal arrangement. That is, the beads should end up like the arrangement shown on your right. You can only move one bead at a time. Both the location and color arrangement shown on the right are the goal you are trying to achieve. Try to achieve this goal in as few moves as possible. Click the mouse button when the bar is positioned below the bead you want to move. Position the bar under the peg on which you wish to place the bead and click the mouse button the second time. The bead will drop onto the peg. Do you have any questions? Screenshot:

Senseek

Persev

Premed

Urge

IGT

BART

SLG

TOL

IQ

1

1

1

.41

.56

-.44

-.30

-.32

.59

.61

.25

.68

.26

.13

Self-Reported Delinquency

Age

Cogni&ve  Ability    (F1)  

Self-­‐Reported  Impulse  Control  (F2)  

Reward  Seeking  (F3)  

Female

Results

.35 – .18(F1) - .25(F2) - .61(F1)(F2)

-.37 - .40(F1) + .25(F3) - .61(F1)(F3)

-.25 -.40(F2) +.18(F3) + .61(F2)(F3)

-2 -1 0 1 2

-0.5

0.0

0.5

1.0

1.5

Low Impulse Control

Reward Seeking

Del

inqu

ency

(mod

el-p

redi

cted

)

High Cognitive AbilityAvg Cognitive AbilityLow Cognitive Ability

-2 -1 0 1 2

-0.5

0.0

0.5

1.0

1.5

Avg Impulse Control

Reward Seeking

Del

inqu

ency

(mod

el-p

redi

cted

)

High Cognitive AbilityAvg Cognitive AbilityLow Cognitive Ability

-2 -1 0 1 2

-0.5

0.0

0.5

1.0

1.5

High Impulse Control

Reward Seeking

Del

inqu

ency

(mod

el-p

redi

cted

)

High Cognitive AbilityAvg Cognitive AbilityLow Cognitive Ability

.32

-.33

.32

-.21

Values are standardized loadings and correlations, and residual variances. Bolded values are significant at p < .05. Italicized values show model constraints. All analyses were conducted in Mplus (Muthen & Muthen, 1998-2012), using the “CENSORED” option was used to controlled for the skewed distribution of mean delinquency scores and the CLUSTER option to control for nonindependence of twin observations.