Table 1.

Summary of Research on Social Cognitive and Affective Constructs Relationship with Substance Use in Adolescence (k=28).

Study	Sample	Key methods	Summary of key results	Association	Social Construct	Results
Empathy
Anh et al., (2011)	N=498 African American adolescents. Age= 11-15. 43% male.	Correlational study. Self-report empathy, drug refusal efficacy, drug use frequency (alcohol, marijuana, and cigarette use). Structural equation modeling.	General empathy has an indirect negative effect on drug use via drug refusal efficacy.	Concurrent	General (does not distinguish)	Moderation model Empathy a Drug refusal β= .11^* Drug refusal a use β= −.29^* Indirect effect of empathy on drug refusal on drug use .11 ^(−.29) = −.032^
Ferreira, Simoes et al. (2012)	N= 3,436 community sample adolescent students. Mean age 15. 47% male.	Correlational study. Self-report empathy (social and emotional competence) and frequency of substance use (tobacco use, drunkenness, and illicit drugs use). ANOVA and moderation.	Authors found a significant mean difference on empathy in youth who used either tobacco or illegal substances. Specifically, those who used had lower self-reported empathy when compared to youth who abstained.	Concurrent	General (does not distinguish)	Empathy on use Tobacco: F= 7.69^ Alcohol: ns Drugs: F= 32.20^*
Laghi et al. (2019)	N= 188 community samples of Italian high school students. Age= 15-19. 71.8% male.	Correlational study. Self-report of cognitive and affective empathy, number of binge drinking episodes in past 2 weeks, and self-efficacy in resisting peer pressure to drink. Moderation and hierarchical regression analysis.	Binge drinking was negatively correlated with empathetic concern. Perspective taking positively associates with an adolescent’s resistance to peer pressure. Direct findings did not hold in regression, but cognitive empathy moderated the association between resistance to peer pressure and drinking.	Concurrent	Cognitive & affective	Binge drinking associating with Affective: r= −0.16^* Cognitive: r= −0.06 Resistance to peer pressure associating with Affective: r= .12 Cognitive: r= .18^* Drinking regressed on Affective: β = −.06 Cognitive: β = −.02 Resistance X cognitive β= .28^*
Luengo et al. (1994)	N= 1,144 youth involved in the juvenile justice system (n= 103) and community sample (n= 1,041). Age= 14-18. 100% male.	Group compare study. Self-report empathy, impulsivity, sensation-seeking scale, and drug consumption. Pearson correlation and stepwise multiple regression.	Pearson correlations: cognitive and affective empathy had a significant negative association with drug use. (most associated with sensation seeking). Multiple regression: in controls, drug taking was predicted by affective empathy, self-esteem, and sensation seeking with affective empathy accounting for the most variance. In incarcerated samples it was predicted by sensation-seeking and impulsivity.	Concurrent	Cognitive & affective	Drug-use associating with Correlation whole sample Cognitive: r= −.10^ Affective: r= −.09^* Regression with controls Cognitive: ns Affective: β= −.423^** Regression with incarcerated people Sensation-seeking: .382^* Impulsivity: .425^*
Winters et al. (2020)	N=826 adolescents in outpatient substance use treatment. Age= 12-18. 18.68% male.	Cohort study (4 time points). Self-reported empathy and substance use over prior 3 months (alcohol, marijuana, cocaine, or a hallucinogen). Growth curve modeling.	Growth curve modeling: affective empathy consistently predicted reductions in substance use and cognitive empathy did not.	Prospective	Cognitive & affective	Predicting substance use Affective: β= −.04^** Cognitive: ns
Callous-Unemotional (CU) Traits
Andershed et al. (2018)⁽²⁾	N= 996 community sample of 12-year-old adolescents; 48% male.	Cohort study (3 time points). Self-report CU traits and substance use frequency. Correlation, multiple & logistic regression.	Baseline CU traits significantly correlated with substance use at 1st, 2^nd, and 3rd year follow up for boys and year 3 for girls. In logistic regression neither conduct or CU added prediction to the stability of substance use.	Prospective	Affective deficits	CU correlation with use Boys (girls) Y1: r= .14^*(ns) Y2: r= .11^(ns) Y3: r= .10^(.09^) CP correlation with use = ns Stability of use over time CP only = ns CU only = ns
Anderson et al. (2018)⁽²⁾	N= 753 children (n= 367 high-risk for behavior problems, n= 386 controls). Assessed from grade 7 to	Cohort study (8 time points). Self-report CU traits and parent-report CD symptoms. Self-report	CU traits associated with mean increases in substance and cigarette use over the 8 time points but not the change in slope.	Prospective	Affective deficits	CU with substance use: Intercept = 1.15^** Slope = ns
Muratori et al. (2016)⁽²⁾	N= 59 youth with disruptive behavior disorders. Age = 9 at start of study. 95% male.	Cohort study (from age 9 to 15; 4 time points) parent-report CU traits and self-report of substance use frequency (alcohol, tobacco, marijuana, and overall substances) in the past month. Latent growth curve analysis.	Youth with a high and relatively consistent CU trait level from childhood to adolescence were at an increased risk for behavioral problems and substance use in adolescence. Also, no baseline environmental and clinical factors associated with levels of CU traits.	Prospective	Affective deficits	CU correlating with use T1-T4: r= .17^-.66^ CU growth curve Substance Use: Slope β= 1.04^* Intercept β= .48^*
Pechorro et al. (2016) ^a2	N= 1,003, Portuguese public-school sample. Age= 12-20, (mean= 15.87). 52.6% male.	Group compare study. Self-report psychopathic traits, CU traits, empathy, alcohol abuse, and drug use. Correlations.	Psychopathic traits, notably CU traits, had a positive correlation with alcohol and cannabis use.	Concurrent	Affective deficits	Youth Psychopathic Trait Inventory short version CU Alcohol (total sample): 0.15^* Cannabis (total sample): 0.11^
Pechorro et al. (2017) ^{a2 (1)}	N= 1,003, Portuguese public-school sample. Age= 12-20 (mean= 15.87). 52.6% male.	Group compare study. Self-report psychopathic traits, CU traits, empathy, alcohol abuse, and drug use. Correlations.	Psychopathic traits, notably CU traits, had a positive correlation with alcohol and cannabis use.	Concurrent	Affective deficits	Youth Psychopathic Trait Inventory short version CU Alcohol (total sample): 0.18^* Cannabis (total sample): 0.13^*
Pechorro et al. (2017)^{a3 (1)}	N= 377 forensic females (n= 274) and controls (n= 103). Age= 14-19. 0% male.	Group compare study. Self-report CU traits, empathy, and substance use frequency (alcohol, cannabis, and cocaine/heroin use). Pearson correlation.	Callousness was significantly negatively correlated with empathy (stronger negative relationship with affective than cognitive empathy) and was positively correlated with use of alcohol, cannabis, cocaine, and heroin.	Concurrent	Affective deficits	CU total score associating with Alcohol r= .45^* Cannabis: r= .46^* Cocaine/heroin: r= .40^***
Pechorro et al. (2017)^a3	N= 377 forensic females (n= 274) and controls (n= 103). Age= 14-19. 0% male.	Group compare study. Self-report CU traits, empathy, and substance use frequency (alcohol, cannabis, and cocaine/heroin use). Pearson correlation.	Callousness was significantly negatively correlated with empathy (stronger negative relationship with affective than cognitive empathy) and was positively correlated with use of alcohol cannabis, cocaine, and heroin.	Concurrent	Affective deficits	CU total score associating with Alcohol: r= .31^* Cannabis: r= .34^* Cocaine/heroin: r= .31^***
Pechorro et al. (2016) ⁽¹⁾	N= 221 forensic males. Age 13-20. 100% male.	Within subject study. Self-report CU traits. Frequency of substance use. Pearson correlation	CU traits positively associated with substance use (alcohol, cannabis, and cocaine/heroin).	Concurrent	Affective deficits	CU total score associating with Alcohol: r= .26^* Cannabis: r= .32^* Cocaine/heroin: r= .17^***
Ray et al. (2016)⁽¹⁾	N= 1,216 adolescents involved with the juvenile justice system (from Crossroads Study). Age= 13-17. 100% male.	Correlational study. Self-report CU traits, impulse control, and frequency of substance use across lifetime. Latent class analysis and logistic regression.	CU traits added to the predictive probability between substance use groups.	Concurrent	Affective deficits	Abstainer & soft drug users: CU: OR= 1.03 Abstainer & hard drug users: CU: OR = 1.06^*** Soft vs hard drug users: CU: OR = 1.03^*
Thogersen et al. (2020) ⁽¹⁾	N=160 Norwegian adolescents with behavioral problems. Age= 11-19. 53.7% male.	Correlational study. Self-, parent-, teacher-report of CU traits and self-report frequency of problematic alcohol use (AUDIT). Logistic regression.	Self- and teacher- reported CU traits were significantly associated with problematic alcohol use. Parent-reported CU traits of adolescents failed to show a significant association with problematic alcohol use.	Concurrent	Affective deficits	Reported CU traits and problematic alcohol use Self-reported OR= 1.09^ Teacher-reported OR= 1.11^ Parent-reported OR = 1.04, p= 0.173
Thornton et al. (2019)	N= 1,216 male adolescents from juvenile justice system. Age= 13-17. 100% male.	Cohort study (4 time points). Self-report of CU traits and variety of substance use (13 substances surveyed at 2 time points by SU/AI). Zero-order correlations.	CU traits associated with substance use. Additionally, CU traits had an indirect effect on risky sexual behavior through adolescent’s substance use (regardless of the type of substance used).	Prospective	Affective deficits	CU traits with substance use r= 0.29^***
Waller et al. (2018)^{a1 (1)}	N= 1170 male forensic youth. Age= 14-18. 100% male.	Cohort and case control study (9 time points). Self-report CU traits, substance use (alcohol, and substance dependence) in the past 6 months. MANCOVA.	Youth with high and stable CU traits were associated with higher substance use, as well as harsh parenting, and contextual risk of violence exposure. Effect sizes (eta²) reported, and all are nominal to small (<= .03).	Prospective	Affective deficits	CU trait groups on drug F= 11.67^* High > moderate^* High > low^*** CU trait groups on alcohol F= 4.57^* High > moderate^* High < low^*
Wymbs et al. (2012) ⁽²⁾	N= 521 mixed sample community youth in 6^th grade followed up at 9^th grade. Age = 11-13.6 at start of study. 43% male.	Cohort study (2 time points). Self-report CU traits and frequency of alcohol and marijuana use in past 6 months. Hierarchal and logistic regression.	Self-reported CU traits at 6^th grade predicted onset, reoccurrence, and impairment of substance use by 9th grade. Females with higher CD and CU had higher odds of endorsing alcohol or marijuana.	Prospective	Affective deficits	Onset of impairment Self-report (parent report) CU: ns (.20^) CD:.51^(ns) CU&CD: −.11^(ns) Gender&CU: ns (ns) Gender&CD: −.33^*(ns) Genders&CU&CD: ns(ns)* Recurrent impaired use CU: ns(ns) CD: .19^(ns) CU&CD: −.06^(ns) Gender&CU: ns(ns) Gender&CD: −.15^*(ns) Genders&CU&CD: .08^(ns)
Theory of Mind (ToM)
Lannoy et al. (2020)	N=202 French secondary school adolescents. Age= 13-20. 37.6% male.	Correlational study. Self-report empathy, depressive/anxiety symptoms, and alcohol consumption (AUD1T-C, assessing frequency and intensity of drinking) and binge drinking score. ToM assessed with Yoni’s task. Correlational analysis and hierarchical linear models.	Binge drinking and alcohol abuse is negatively associated with second-order affective ToM. Empathy did not associate statistically.	Concurrent	Cognitive	Alcohol Cognitive: ns Affective: ns ToM cognitive 1^st: ns ToM affective 1^st: ns ToM cognitive 2^nd: ns ToM affective 2^nd:−.19^* Binge drinking Cognitive: ns Affective: ns ToM cognitive 1^st: ns ToM affective 1^st: ns ToM cognitive 2^nd: ns ToM affective 2^nd: −.22^*
Social Cognition
Fluharty et al. (2018)	N= 3,058 youth age 8 (prospective) and N= 3,613 adolescents age 18 (retrospective).	Cohort and case control study (2 time points). Parent report social communication disorders (age 8). Self-report frequency, age at first use, and variety of use for alcohol, cigarette, and cannabis use (age 15 & 18). Self-report social reciprocity (age 18). Logistic regression.	Temporally going forward: age 8 poor non-verbal communication decreased odds of substance use in adolescence. Retrospectively after substance use: Poor social communication and reciprocity increased the odds of adolescent substance use. Sex-stratified analysis indicated no differences.	Prospective & retrospective	General (does not distinguish)	Social communication Prospective Alcohol: ns Tobacco: OR= 1.95^*** Cannabis: ns Social reciprocity Retroactive Alcohol: OR= 1.44^ Tobacco: OR= 1.92^* Cannabis: OR= 1.44^**
Kirisci et al. (2004)	N= 215 youth, starting at ages= 10-12 and followed until 19. 100% Male.	Cohort study (4 time points). Self-report social cognition, beliefs about substance use, frequency of use (alcohol and marijuana consumption) in month prior to study, and risk of lifetime SUD. Social cognitive distortions report. Path analysis.	Inaccurate social cognition, significantly predicted by childhood SUD risk (measured by neurobehavioral disinhibition), predicted marijuana use at age 16 that led to prodromal SUD at age 19.	Prospective	Cognitive	SUD risk (neurobehavioral disinhibiden) 10yo on social cognitive distortions 12yo: β= −31^ (negative means more) Social cognitive 12-14yo distortions on marijuana use 16yo: β= −.13^ (negative means more) Prodromal marijuana use on SUD 19yo: β= .31^*

CU= callous-unemotional, CD= conduct disorder, CP= conduct problem, ToM= theory of mind, ANOVA= analysis of variance.

#= studies using the same sample (same number indicates which studies use the same sample), + = elevated, □ = moderate, − = low

= reported on the model without executive functioning

⁽¹⁾

= included in cross-sectional meta-analysis

⁽²⁾

= included in longitudinal meta-analysis.

ns = not significant

= p<.05

^**

= p<.01

^***

= p<.001.