Neural activity when processing emotional cues (e.g., as depicted through facial expressions) and situations has also been examined in association with substance use, given the developmental shifts in emotionality during adolescence that coincide with brain maturation (Guyer et al., 2016), and the link between emotionality and substance use in adults (Hussong and Chassin, 1994). Effective social and emotional functioning depend on the ability to identify and be sensitive to the emotions of other people, but facial emotion recognition is impaired in adults who heavily drink alcohol and/or use cannabis (Castellano et al., 2015). Adults who heavily and frequently use substances typically struggle with maintaining interpersonal relationships, and are more sensitive to the effects of social stress (Hanlon et al., 2019, Sahani et al., 2022). There is also evidence that substance use and addiction are associated with activation of both reward-control regions and social-emotional regions (Courtney et al., 2019, Cousijn et al., 2018). Use of alcohol and cannabis is associated with disrupted socioemotional function, including processing of emotional faces, in adults (Miller et al., 2015) and adolescents (Leiker et al., 2019).

A larger body of literature (compared to social reward) has focused on the extent to which the brain’s response to emotion (e.g., facial expressions) predicts substance use, especially when considering stress and internalizing symptoms (Nikolova et al., 2016, Spechler et al., 2015, Spechler et al., 2019, Swartz et al., 2017). Findings suggest that relatively reduced social threat-related reactivity of the amygdala (i.e., to fearful faces) may increase risk of alcohol use or AUDs. At the same time, heightened amygdala response might buffer against VS hyper-activity and risk of stress-related problem drinking in young adults, such that impulsivity-related drinking driven by the VS is dampened down by emotional sensitivity driven by the amygdala (Gilman et al., 2008, Gorka et al., 2013, Sripada et al., 2011). Other regions of interest to be investigated from an emotional sensitivity perspective, especially given the reward literature, include the ACC, NAcc, and mOFC (Blair et al., 2019, Chaplin et al., 2019, Hardee et al., 2017, Milivojevic and Sinha, 2018, Wilcox et al., 2016).

Several studies have used a modified version of a faces-matching task (Hariri et al., 2002), which involves passive viewing of emotional faces to measure whether fear or anger elicits heightened response via the amygdala. Elsayed et al. (2018) found that adolescents who were later classified as early initiators (n = 21) demonstrated higher neural activity to fearful facial expressions, when contrasted with a group classified as late initiators (n = 231). In other words, the early versus late group had higher amygdala activity to implicit environmental signals of threat. These findings are consistent with other work showing that higher amygdala activity to threat predicted future problem drinking in response to stressful life events in college students (Nikolova et al., 2012). One study of 170 young adults examined both VS response to social-emotional reward (e.g., viewing a happy face) and amygdala response to social-emotional threat (e.g., viewing an angry face), finding one pattern in which the combined effects of high VS response to reward and low amygdala response to social threat contributed to greater risk for stress-related drinking, and a second pattern wherein low VS response to social reward and high amygdala response to social threat was also associated with stress-related drinking (Nikolova et al., 2016). Additionally, Swartz et al. (2017) found that greater amygdala activity to fearful faces was linked with higher peer-reported conscientiousness, as well as lower extraversion, in male college students. Conscientiousness in turn predicted problem drinking one year later; and amygdala response to fearful faces was indirectly associated with drinking via this pathway. Taken together, these patterns imply that heightened amygdala response to social threat may be a neural risk factor for later substance use, especially when considering stress-related motivations for use. Since facial emotion recognition is impaired in adults who heavily drink alcohol (Castellano et al., 2015), it is important to consider the role of threat-processing brain regions in the development of substance use, considering adolescents face many new stressors that may be threatening (e.g., rejection by romantic partners) and lead them to engage in substance use.

Regarding cannabis use as compared to alcohol, Leiker et al. (2019) investigated the relation between symptoms of AUD and CUD and neural processing of emotional faces in 123 adolescents in a separate task, as they identified whether a face was male or female in photos involving neutral, happy, and fearful expressions. Higher AUD scores were associated with less response to faces in the vmPFC and lingual gyrus, whereas higher CUD scores were related to less response in a region of the rostromedial PFC, which included the ACC. No differences by emotion emerged for CUD symptoms; however, AUD symptoms were positively associated with response to fearful faces in the inferior parietal lobule (iPL). Additionally, Blair et al. (2019) used a “looming task” with emotional faces (angry or neutral) that appeared to either loom toward the participant by increasing in size or recede away by decreasing in size, with the increase simulating the social threat of being approached by someone with an angry face. Adolescents’ scores on the Cannabis Use Disorders Identification Test (CUDIT) scores were positively associated with response to looming stimuli (both negative and neutral) in the rostral mPFC, left fusiform gyrus, and the cerebellum. Regardless of expression (i.e., angry versus neutral faces), neural activity was evoked when faces moved toward the participant, suggesting that someone else approaching can be perceived as threatening even if they do not display anger. Overall, use of alcohol and cannabis seems to differentially relate to deficits in neural systems involved in social stress.

While most work to date has relied upon self-reported substance use, the acute effects of substances on brain function can also be tested among adults by administering substances directly. In one experiment, 12 young adult heavy social drinkers were given either alcohol (0.8 g/kg; 16 % volume) or placebo before completing an emotional faces task (Gorka et al., 2013). Drinking alcohol led to decreased coupling between the amygdala and right OFC to emotional faces, both threatening and happy; and to decreased coupling between amygdala and left OFC to happy faces. Additionally, the same research group showed that connectivity between amygdala and OFC may mediate alcohol’s effects on social behavior. In a follow-up experiment, 16 young adults who had used cannabis at least 10 times in their lives (but were not daily/heavy users) were given either THC (7.5 mg; Marinol) or placebo before completing the faces task (Gorka et al., 2015). Consuming THC led to increased coupling of activity between the amygdala and the mPFC and rostral ACC while viewing threatening faces, compared to neutral shapes. Thus, THC might reduce threat perception and/or enhance socio-emotional regulation.

Beyond the brain’s response to emotional faces, alcohol may also acutely affect response to other types of threats, such as being evaluated by others while performing a difficult task. One study administered alcohol (1 ml/kg of 95 % alcohol) to 48 young adults who were classified either as high anxiety or high sensation-seeking, before they completed two fMRI tasks: a faces task, and the Montreal Imaging Stress Task (MIST) in which individuals performed arithmetic under time pressure, viewed a visual “performance scale,” and received negative verbal feedback from the researchers throughout the task (Shakra et al., 2017). Alcohol use was predicted by specific neural activations, such that anxious individuals showed attenuated amygdala response to angry faces, reversing their sober pattern of greater amygdala response. In the MIST, anxious individuals showed increased mOFC, pgACC and NAcc activity during social stress; and sensation-seeking men showed decreased mOFC response, which predicted problem drinking. These findings support the idea of distinct risk pathways to alcohol use, as well as sex differences given that only men showed differences by sensation-seeking. These three studies experimentally administered substances in a double-blind paradigm, providing valuable insight about short-term effects of alcohol and THC on social cognition; and more evidence that substance use both shapes, and is shaped by, neural processing of emotional stimuli.

Focusing on emotional faces and cannabis use among adolescents in the IMAGEN study, a large multi-site study as part of the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) Consortium, Spechler et al. (2015) tested how cannabis-experimenting adolescents reacted to video clips of faces turning angry, compared to shapes as a control. Cannabis-experimenting adolescents demonstrated greater activity to angry faces in the amygdala than non-users, whereas non-users had higher activity to neutral faces in the bilateral dlPFC and right TPJ. For cannabis users, there was no difference in TPJ activity between angry and neutral faces, while non-users showed lower activity to angry faces. Thus, cannabis use during early adolescence is linked to hyper-reactivity to negative affect in the amygdala, and hypo-reactivity in frontal cortical regions such as the dlPFC. A follow-up study (Spechler et al., 2020) examined whether amygdala reactivity at age 14 predicted use of cannabis, alcohol, and cigarettes by age 19. Heightened amygdala response to angry faces predicted greater use of cannabis, and the lowest amygdala activation was seen in adolescents who remained abstinent in a dose-dependent pattern. This pattern was specific to cannabis use, as amygdala reactivity did not predict alcohol or cigarette use. Amygdala activation at age 19 did not differ between cannabis users and non-users, suggestive that early adolescent amygdala response might be more predictive of later cannabis use. Although adult studies have shown differences between chronic cannabis users and non-users such as decreased amygdala response (Gruber et al., 2009), developmental differences in amygdala reactivity may also exist wherein early amygdala hyper-sensitivity may predict more cannabis use but then chronic use over time dampens amygdala response. Indeed, it may be that THC and other cannabinoids have an anxiolytic role in mechanisms of fear (Phan et al., 2008), such that highly-reactive individuals seek out cannabis to regulate emotional processes.

Given the theorized role for emotionally charged experiences involving threat in driving and/or maintaining substance use (Chaplin et al., 2018, Courtney et al., 2019), some work has used emotional scenes to examine associations between neural response to threat and substance use. Aloi et al. (2018) compared neural response between 47 adolescents receiving inpatient SUD treatment and 35 adolescents from the community during an affective Stroop task, requiring participants to count the quantity of numbers on the screen next to an emotionally-charged photograph (e.g., a tarantula crawling on a man’s shoulder). Results indicated differential impacts of alcohol and cannabis abuse on the adolescent brain’s processing of threatening stimuli, such that cannabis use was positively correlated with PCC, iPL, and precuneus, whereas alcohol use was positively correlated with amygdala but negatively correlated with dACC, dlPFC, and precuneus. For adolescents who used both substances, activation to negative emotional stimuli was negatively correlated with alcohol use only in those with low cannabis use, whereas activity was positively correlated with alcohol use for those with high cannabis use. Taken together, alcohol might have a stronger impact on threat-processing regions, while cannabis might affect attention-related regions. Considering the inconsistency of these findings with the face-processing results from 140 adolescents in Spechler et al., 2015, Spechler et al., 2020, despite small but sufficient sample sizes in each study, further work is needed to disentangle how amygdala response to emotional scenes is associated with substance use over time.

In addition to faces and other visual stimuli, the link between substance use and processing negative affective information has been assessed with emotionally-charged words. Heitzeg et al. (2015) used an emotion-arousal task with negative words (e.g., danger) and positive words (e.g., hope) with 40 young adults who either never used cannabis (n = 20) or used it heavily, finding that users had less activation to negative words in the right insula, PFC, and occipital cortex. Prefrontal response to negative words also mediated the link between adolescent cannabis use and young adult negative emotionality. One key takeaway from these results is that heavily using cannabis as an adolescent might impair later emotional functioning, but through a cortical rather than subcortical pathway.

Finally, some research has considered the effect of emotional scenes on cognitive control, given that substances impair regulatory abilities such as inhibiting responses and pursuing goal-directed behaviors especially in emotionally-charged situations. Using a Go/No-Go task with emotionally-arousing background images, Cohen-Gilbert et al. (2017) revealed that binge drinking in 23 college students was associated with less neural response to negative images in the dlPFC, dmPFC, and ACC, suggesting an inability to effectively use regulatory regions to tune out emotional distractors. Thus, reduced activity during inhibition with emotionally-charged backgrounds might reflect a failure to regulate emotion, with implications for future self-regulation problems linked to drinking. In a separate study of 60 college freshmen, problematic drinking was linked to less neural recruitment of attention regions during negative emotion No-Go trials, and greater recruitment of frontoparietal regions (Cohen-Gilbert et al., 2022). Other work on response inhibition in 30 adults (mean age 23 years) found that greater binge drinking was associated with reduced neural response in frontal and parietal regions during inhibition against background images of fearful faces (Herman et al., 2019). Sullivan et al. (2022) also reported that cannabis-using young adults (n = 34) demonstrated less rostral ACC activity while successfully inhibiting against a background of fearful faces, compared to non-users (n = 32; mean age 21 years across both groups). Men who used cannabis also displayed more connectivity between the right rACC and right cerebellum, when the background included calm faces. Finally, another study focused on the neural correlates of the ability to delay receipt of a reward, showing that greater binge drinking was linked to a steeper decrease in frontal pole activation when making a delayed choice against a background of fearful faces; again, demonstrating the influence that emotion has on cognitive control particularly among those who engaged in extreme drinking behaviors. Overall, alcohol use impacts – and is impacted by – altered function of neural regions implicated in the top-down control of emotion.

Social relationships are rich with emotional cues of various forms and serving various functions, and adolescents draw on such cues from both their parents and peers. Information from these social contexts has been linked to the neural basis of adolescent risk taking (e.g., van Hoorn et al., 2018), which sometimes includes substance use in composite measures. With regard to substance use specifically, parental relationships are an important context particularly given the association between negative emotional experiences and substance use. For example, negative parenting behaviors such as criticism and harsh vocal tone have been found to predict longitudinal increases in substance use across adolescence (Pettit et al., 2001). Emerging work has extended understanding of this behavioral pattern by focusing on neural mechanisms. Specifically, Chaplin et al. (2019) examined how 66 young adolescents’ conversations with a parent about previous conflicts were associated with neural response to negative photos (e.g., a gun in a mouth), as well as how neural response was linked to substance use. Maternal harshness was associated with more right ACC activity in girls while viewing negative images, and less anterior insula and left ACC activity in boys. Greater activity in the insula predicted more substance use in girls, but not in boys. For girls, insular response may reflect greater internalizing of negative emotional experiences, and this high emotional arousal could lead to more substance use (i.e., coping with emotional distress). Overall, these findings suggest that the adolescent brain’s response to negative emotionally-charged images is relevant for substance use, modulated by parental behaviors such as low warmth; but with girls following an internalizing pathway, similar to studies of reward response by sex revealing different pathways (Swartz et al., 2020).

Peer relationships take on heightened salience in adolescence (Nelson et al., 2016), and represent the most common contexts in which substance use begins. One core feature of strong friendships is emotional closeness (Flores and Berenbaum, 2014), as feeling close during conversations creates more positive affect among friends. While not directly testing substance use, two studies explored how adolescents’ risky behaviors (a composite including substance use) were associated with neural response while viewing their best friend’s positive affect during a conversation about past experiences (Ambrosia et al., 2018, Flores et al., 2018). Adolescents’ (N = 50) neural response in the vlPFC to a friend’s positive affect was linked to higher engagement in risk-taking behaviors (Ambrosia et al., 2018); however, substance use independent of other domains of risk-taking behaviors (e.g., sexual risk-taking) was not tested directly, limiting conclusions about the role of substance use specifically. Nonetheless, the task used in these studies is a good example of how research can simulate adolescents’ real-life experiences and explore connections to their health behaviors.