Attention-deficit/hyperactivity disorder (ADHD) is a common psychiatric disorder with a prevalence of 5% worldwide (Polanczyk et al., 2007). ADHD presents with heterogeneous symptomatology characterized by inattention, impulsivity, and hyperactivity. Patients are also significantly affected in several aspects of daily life including work performance, planning, decision-making (DM), and psychosocial interactions (Must et al., 2013; Polanczyk et al., 2014). A meta-analysis on longitudinal studies has shown that up to 50% of diagnosed children still meet partial ADHD-related symptoms by the age of 25 (Faraone et al., 2006). However, across the lifespan the main symptom of hypermotoric behavior shifts to an inner restlessness, inattention, and emotional dysregulation (Gibbins et al., 2010; Francx et al., 2015). Furthermore, it is noteworthy that adult ADHD has an increasing tendency to engage in risky behavior that derives from inadequate DM. Those heightened risk taking behaviors are seen particularly in situations of risky driving, risky sexual behavior and pathological gambling (Flory et al., 2006; Thompson et al., 2007; Faregh and Derevensky, 2011).

Investigating risky behavior in the context of disadvantageous decisions is challenging in an experimental setting and study results are often inconsistent and do not reflect the daily behavioral deficits of ADHD (Mowinckel et al., 2015; Dekkers et al., 2016). It is often not clear which neuropsychological functions are responsible during the performance of risk-taking paradigms. Furthermore, the wide variety of methodologies makes it difficult to compare different studies with each other. However, according to the Dual Pathway Model (DPM) it is known that two different signaling pathways are involved in the development of behavioral actions: including cognitive-analytic functions (cold functions) and the intuitive-affective functions (hot functions) (Sonuga-Barke, 2002). The most common decision-making paradigms; IOWA Gambling Task (IGT), Columbia Card Task (CCT-cold), and Game of Dice Task (GDT) involve cognitive-analytic functions, whereas the Balloon Analogue Risk Task (BART) and Colorado Balloon Game (CBG) can be associated with more affective-emotional driven behavior (Lejuez et al., 2002; Brand et al., 2007; Figner et al., 2009; Crowley et al., 2010; Mäntylä et al., 2012).

Due to the heterogeneity in study results regarding risky DM in ADHD, as well as the lack of focus on hot function driven behavioral impairments, further research on affective functioning is needed. Additionally, based on the prefrontal recovery hypothesis (Halperin and Schulz, 2006), it has been shown that developmental improvements in cognitive control functions may favor a reduction in behavioral disturbances in adult ADHD (Groen et al., 2013; Mowinckel et al., 2015). Conversely, significant increases in impaired emotional competence can be observed and are known to negatively affect the patients’ socio-emotional life (Shushakova et al., 2018; Materna et al., 2019; Beheshti et al., 2020). Moreover, these emotional difficulties are also a risk factor for the development of affective comorbidities, such as major depressive disorder (Bresner et al., 2009). It can thus be assumed that impairments of cold functions in adulthood diminish, whereas dysfunctions in emotional control and regulation become more prevalent (Shushakova et al., 2018; Materna et al., 2019; Beheshti et al., 2020; Mayer et al., 2021). This leads to the assumption that heightened risk-taking in adult ADHD that arises from deficits in DM ability may result from an impairment in hot functions.

Referring to the DPM, hot functions describe particular behavior that arise quickly and intuitively (Sonuga-Barke, 2002; Mäntylä et al., 2012; Shoham et al., 2016), in which perceived stimuli are transmitted directly from the sensory system to the amygdala (Damasio et al., 2000). Due to this activation of the limbic system, the autonomic nervous system is stimulated and induces physiological responses such as modulations in heart rate, sweating, breathing and eye blinking (Figner and Murphy, 2011; Bellato et al., 2020). These physiological changes, in turn, are perceived by the body in the form of a feedback loop which can then elicit a behavioral action (Christopoulos et al., 2019). Thus, the signaling pathway of hot functions is composed of an interconnection of perception of the stimulus, activation of the autonomic nervous system and physiological changes. Although these processes are not consciously experienced, this interconnection to a certain stimulus can be learned and stored as somatic markers in the brain, according to prominent theories (Damasio, 1996). Thus, the somatic response can also act as an early-warning-system that guides subconscious decisions from the learned connection of the stimulus and body reaction.

A promising measurement technique to detect somatic marker functioning and anticipatory physiologically changes, is the recording of skin conductance responses (SCR) (Starcke et al., 2009; Wright and Rakow, 2017; Christopoulos et al., 2019). When emotional arousal occurs the electrical property of the skin changes due to the increased sweat secretion. This can be detected by applying an external direct current with constant voltage to the skin (Boucsein, 2012; Uddin et al., 2017). It is important to determine at which timepoint during a DM process a change in skin conductance occurs. In the present study changes that follow a decision are described as reactive SCR (rSCR) whereas changes that precede a decision are described as anticipatory SCR (aSCR). Although SCRs represent emotional arousal, it cannot be clearly assessed whether an rSCR follows a positive or negative feedback (Figner and Murphy, 2011). It has only be shown that highly evaluated feedback (either being high rewards or high punishments) were associated with increased SCR amplitudes (Wilkes et al., 2010; Lole et al., 2012). In addition, risky behavior was found to decrease when rSCR was already elevated in the previous trial (Masunami et al., 2009; Wright and Rakow, 2017). Regarding aSCRs, it is well documented that disadvantageous DM behavior is associated with higher amplitudes (Garon et al., 2006; Starcke et al., 2009; Dawson et al., 2011; Wright and Rakow, 2017). Thus, it can be shown that when a high risk condition is present, an increased amplitude in aSCR represents an implicit perception of risk (Guillaume et al., 2009; Rogalsky et al., 2012; Wright and Rakow, 2017; Agren et al., 2019).

There are only a few studies which have investigated the relationship between electrodermal activity and risky DM behavior in adult ADHD. However, results indicated an altered psychophysiological activity in the context of behavioral impairment in adult ADHD (Lazzaro et al., 1999; Lackschewitz et al., 2008; Matthies et al., 2012; Ziegler et al., 2016). Furthermore, a recent systematic review investigating autonomic nervous system function in ADHD identified altered activity patterns at rest but also during task performance. Results were not consistent and showed both hypoactive and hyperactive patterns in ADHD compared to healthy controls (HC) (Bellato et al., 2020). To date, however, much of the research in this area has mainly focused on abnormalities in rSCR regarding feedback sensitivity. In these studies, results indicated greater risky DM in conditions of punishment (DeVito et al., 2008; Masunami et al., 2009), whereas rSCRs have been shown to decrease in response to an error feedback or the omission of rewards (Iaboni et al., 1997; O’Connell et al., 2004). In contrast, hyperactivity of the rSCR is partially observed after rewarding feedback (Masunami et al., 2009; Bellato et al., 2020). However, to the best of our knowledge, no study so far has investigated somatic responses as an anticipatory correlate that guides a quick and intuitive behavior in adult ADHD. Assuming dysfunctional affective processes in ADHD, it can therefore be suggested that (1) alterations of aSCR and rSCR do not occur during risky DM; or (2) increases in aSCRs are not linked to advantageous behavioral actions.

The aim of the present study was to investigate whether a dysfunction in affective pathways is associated with an altered risky DM behavior in adult ADHD. To address this, we analyzed SCRs as well as the relationship between aSCRs and task-behavior. Therefore, we used a modified version of the Balloon Analogue Risk Task (BART) to demand hot function-guided risky decision behavior (Hüpen et al., 2019). The BART has been demonstrated to induce naturalistic risky DM (Lejuez et al., 2002). Moreover, the modified version was designed to trigger emotional arousal and intuitive guided behavior, and to investigate DM depending on the reward magnitude (Hüpen et al., 2019, 2020). Additionally, questionnaires were used to investigate self-assessment of risky behavior and emotional competence.

The present study investigated whether there is a potential causal relationship of affective pathways and abnormal risky DM behavior in adult ADHD. Therefore, different phases of hot function-guided DM were analyzed and compared between adult ADHD and HC on the basis of behavior, physiological activity, and self-assessment.

Results showed higher emotional arousal in ADHD indicated by elevated SCRs preceding a DM. These findings are consistent with the observation of hyperactivity of the autonomic nervous system being increasingly associated with motoric hyperactivity and impulsive behavior in ADHD. These results were suggested to reflect an over-activation of autonomic functions (Wilbertz et al., 2012). It is discussed whether hyperactivity of somatic functions in ADHD serves to induce a stimulating environment in order to promote a certain stability of vigilance (Geissler et al., 2014). In addition, these hyperarousal activities can also be related to a reduced ability of downregulating excessive arousal. However, other findings showed that during the performance of monotonous tasks, quick exhaustion may occur, which is then characterized by hypoactivity of bodily responses (Kuntsi and Klein, 2012; Geissler et al., 2014). It can therefore be assumed that, depending on the demands and excitement of the task; emotional arousal can be either hyper-/or hypo-threshold.

Regarding the available effect of somatic markers at this point, high aSCRs would be expected to be associated with high risk behaviors, as the somatic response will increase with increased risk engagement in order to avoid disadvantageous behavior (Garon et al., 2006; Starcke et al., 2009; Dawson et al., 2011; Wright and Rakow, 2017). However, in the present study risky behavior could not be detected in ADHD. Instead, HCs with overall higher RTs, showed significantly greater risky DM than ADHD. At this point, it can be questioned whether the modified BART accurately reflects daily risky DM and whether it is a valid measurement under laboratory conditions. Moreover, the BART primarily examines risk engagement in the context of financial behaviors and therefore cannot represent all domains of potential daily risk taking behaviors. In addition, recent meta-analyses also highlight the fact that behavioral results of many studies represent rather a suboptimal DM behavior than a risk-seeking DM behavior in ADHD (Dekkers et al., 2021; Roberts et al., 2021). In this context, the underlying mechanism regarding the disentanglement of disadvantageous decisions and risk-seeking decisions reflect an additional important aspect in the field of DM and need to be considered in further studies (Dekkers et al., 2021). On the other hand, the behavior shown could also reflect an action that demonstrates shorter RT according to the Intolerance of Uncertainty that was found to be a transdiagnostic construct in psychiatric disorder (Gramszlo et al., 2018). Subsequently, not the risky DM behavior itself is an interesting outcome, but rather the relationship and interconnection of bodily response and DM.

Thus, in order to investigate how the somatic response influences behavior we analyzed the relation of aSCRs and RT. Results showed that risk engagement in HC coincides with increased aSCR, representing the correlation of disadvantageous behavior and higher SCRs (Garon et al., 2006; Starcke et al., 2009; Dawson et al., 2011; Wright and Rakow, 2017). In turn, increased risk engagement in ADHD was associated with lower aSCR. This negative correlation of RT and aSCR in ADHD confirms the hypothesis about an altered relation between affective state and behavior. However, the strongest affective responses were observed at low RTs. Thus, it is shown that as risk-engagement increases, affectivity decreases, however quick responses tend to elicit the greatest arousal. This suggests that the physical response shown may be related to additional factors than risk. Therefore, it can be assumed that in adult ADHD, there is a missing interconnection between bodily response and behavior regarding risky DM. Nevertheless, further emotionally arousing functions involved in this BART need to be examined.

To understand further aspects of hot function-guided DM, SCRs in response to feedback were also investigated. In this context, the somatic response reflects the evaluation and emotional arousal toward feedback. Results of the rSCRs indicate that feedback is evaluated more emotionally arousing for ADHD than to HC. We could also identify that the feedback of loss is more arousing than gain across both groups. The high rSCRs are also consistent with findings of previous studies on reward sensitivity in ADHD. It was shown that hyperarousal during feedback is associated with weaker inhibitory abilities (Iaboni et al., 1997; Masunami et al., 2009). Furthermore, also in HC it could already be shown that gains produce higher physiological activity (Lole et al., 2012). The impact of feedback, whether it appears as a gain or a loss, can moreover encourage behavioral adaptation. In their prospect theory, Kahneman and Tversky (1979) describe higher emotional arousal and motivation toward loss outcomes. This motivation is accompanied by a decreased propensity to risky DM and is particularly evident in mixed gain/loss prospects. However, it could also be shown that norepinephrine plays a significant role in risk appraisal and propensity (Trepel et al., 2005). Accordingly, it was found that a central norepinephrine blockade decreases the sensitivity to risk-taking significantly (Rogers et al., 2004). Relating this to the underactivity of norepinephrine in ADHD, it seems plausible that reduced risk-taking is caused by the norepinephrine deficiency. Reflecting the elevated rSCRs and the reduced risky DM behavior it can be assumed that the autonomic signal might not get properly transferred to the central nervous system (Critchley and Garfinkel, 2018).

Risk engagement was also investigated by self-assessment using the domain of probability of risk engagement of the DOSPERT questionnaire. Results showed no differences in the self-awareness of risky DM behavior. Contrary to the postulated daily behavior in ADHD, self-assessment does not seem to be perceived as an altered behavior. However, it should be noted that in the current study only the subdomain “probability of risk engagement” was used for a group comparison. Thus, a general evaluation of risky DM based on the DOSPERT results is limited. As questionnaires mostly comprise a subjective and consciously driven self-assessment, it can be assumed that on a conscious level, there are smaller differences in the risk-engagement between HC and ADHD. Consequently, it can be assumed that more unconscious, emotional-motivational driven processes, thus control the increased risky DM behavior in ADHD that is postulated in daily life. This is further supported by the self-assessment of emotional competence using the EKF questionnaire. Results indicated an impaired ability of emotional regulation, perception, understanding, and expression in ADHD that are important requirements for a proper hot function-guided decision (Sonuga-Barke, 2002). Moreover, risky DM is not only guided by previous experience of loss and gain, but also by the potential reward amount. Different studies have shown that with increasing magnitude of reward the risk avoidance also increases (Kahneman, 2003; Van Leijenhorst et al., 2010; Christopoulos et al., 2019; Hüpen et al., 2019). However, the present study indicates greater risk-engagement under high reward conditions, demonstrated by longer RTs, but not by changes in skin conductance. There was, however, no group difference and both ADHD and HC showed greater risky behavior under high reward conditions. Similar results were also shown in studies on ADHD and on Borderline Personality Disorder, arguing that the range between reward conditions was too narrow (Luman et al., 2008; Hüpen et al., 2020).

Overall, the measurement of SCRs has been shown to be a robust method to easily detect subconscious emotional arousal in anticipation and evaluation of DM. It yields a continuous measure that is related to activity in the sympathetic branch of the autonomic nervous system (Figner and Murphy, 2011). However, similar to many other indirect measurement methods, there are also some limitations. Subsequently, it must be taken into account that some participants can be “non-responders” (Figner and Murphy, 2011). In this study, we decided to include all measurements in the analyses as our statistical model corrected for individual differences in SCRs (Hüpen et al., 2019). Furthermore, there are contradictory recommendations on the pretreatment of the skin, whether the skin should be treated with water, oil or nothing at all before attaching the electrodes (Boucsein, 2012). We decided to follow the BIOPAC guidelines and used the saline paste supplied. Medication intake was also proven to affect activity of the autonomic nervous system (Bellato et al., 2020). Therefore, medication effect on behavioral results should be considered in future studies. Additionally, despite the potential individual lack of activity, there might also be environmental disturbances on the electrodermal recording that affect the signal. For instance, it was shown that electromagnetic noises such as overhead lights can disturb the signal and cause unrelated changes in the SCRs. We tried to avoid these artifacts by using low-pass filters and conducting the study in interference-free rooms, but a potential effect cannot be ruled out when considering the results. Furthermore, since the present study focuses on risky behavior in adults with ADHD it is important to consider that there may be limited comparability with child studies. Behavioral deficits in children may be deferred due to the delayed cortical maturation but may not persist in adulthood (Whelan and Mchugh, 2009; Dekkers et al., 2016; Koscielniak et al., 2016). In addition, a possible effect of comorbidities on the results must also be taken into account. For instance, patients with Antisocial Personality Disorder are also affected by deficits in affective functions (Glenn et al., 2013). Since the Mini-DIPS does not assess personality disorders, additional questionnaires should be included in future studies. Moreover, it should be taken into account that the present study investigated the specific functionality of emotional competence, but ADHD is particularly characterized by a heterogeneity of dysfunctions. However, the effectiveness of hot functions is still underrepresented in behavioral studies on ADHD. To gradually shed more light on this topic, future studies should specifically examine demographic effects such as age, gender, and education in relation to the ability of affective functioning in adult ADHD.

In conclusion, the present study is the first to investigate hot functions as underlying mechanisms for risky DM in adult ADHD. Results show significantly higher arousal before a DM and after feedback display in ADHD. However, ADHD participants were unable to use this physiological information to modify behavior. This finding is also confirmed by self-reports that showed a weaker ability in the perception of arousal and emotion in ADHD, whereas self-reported risk behavior was not altered compared to HC. Further research is needed to investigate how emotional processing can be influenced or improved in ADHD. However, our research underlines the importance of considering emotional therapeutic technics in the work with patients with ADHD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committee of the Medical Faculty of the University of Bonn. The patients/participants provided their written informed consent to participate in this study.

EH, SL, and PH contributed to the conception and design of the study. EH, FK, and AH acquired the data and performed data analyses. PH designed the paradigm and performed data analyses. EH, MB, BH, CD, SL, and AP contributed to data interpretation and discussion of results. BA recruited patients with ADHD for study participation. EH wrote the first draft of the manuscript. All authors read and approved the submitted version.