| Abstract (english) | Introduction: The development of a gambling habit is undeniably one of the most intriguing phenomena within the realm of cognitive psychology and decision-making. This curiosity stems from the paradoxical nature of gambling, succinctly captured by the phrase "The house always wins." This paradox is reflected in the fact that most gamblers are acutely aware that, from an economic standpoint, gambling carries a negative expected value, rendering it a highly risky behavior. Consequently, we're compelled to ask: What cognitive mechanisms drive the risktaking inclinations of frequent gamblers? This inquiry holds particular significance when we consider that persistent gambling can lead to a host of negative psychosocial consequences, potentially meeting the clinical criteria for problem gambling (Tremblay et al., 2010). In this context, it's vital to explore whether individuals who engage in frequent gambling consistently make riskier decisions compared to their non-gambling counterparts. Do they retain the capacity for sound decision-making, and if so, what cognitive mechanisms underlie this capacity? This same line of questioning applies to individuals who have experienced adverse psychosocial consequences due to excessive gambling. Is their ability to make better and less risky decisions potentially more compromised than those who gamble frequently but have not developed clinical symptoms of problem gambling? Traditionally, most studies on the determinants of risky decision-making within the realm of gambling have adopted a classical cognitive perspective. This perspective asserts that gamblers persist in their behavior, making riskier choices, because they hold distorted beliefs about gambling, leading them to overestimate their chances of winning (Ladouceur and Walker, 1996). However, this approach has faced criticism due to inconsistencies in findings. Research has shown that distorted beliefs about gambling do not necessarily correlate with the level of gambling-related problems. Even individuals who are not regular gamblers or only gamble occasionally exhibit the same biases as problem gamblers (Delfabbro, 2004). In addition to the classical cognitive paradigm, some studies have embraced a psycho-biological perspective. They seek to identify differences in various aspects of brain functioning between groups of participants with and without gambling problems. Thus, various neurochemical studies, functional brain imaging studies, and neuropsychological studies were conducted within this perspective. When it comes to neuropsychological studies on risky decision-making in adult problem gamblers, such as the work of Forbush et al. (2008), they have uncovered deficits reminiscent of those found in individuals with damage to the ventromedial prefrontal cortex (vmPFC). These individuals often grapple with real-life challenges in financial decision-making. However, there has been a dearth of neuropsychological studies focused on adolescent samples. One rare study by Ciccarelli et al. (2016) did find that adolescents who made riskier decisions also experienced more pronounced psychosocial consequences related to gambling, along with distorted beliefs about gambling. This suggests that certain dysfunctions in the vmPFC may also serve as significant determinants of risk-taking behavior in adolescents who gamble. Nevertheless, before we hastily conclude that adolescents who gamble frequently or face higher levels of gambling-related problems consistently make irrational and riskier decisions, we should consider the development of theories concerning adolescent risky decision-making. The assumptions made by earlier normative models, which posited that lower rationality and heightened cognitive deficits in adolescents were the primary drivers of their inclination toward risk, have been challenged. Research has demonstrated that, in contrast to adults, adolescents are not inherently less capable of discerning the risk associated with certain behaviors or evaluating their negative consequences (Reyna and Farley, 2006). The perplexing propensity for risky behaviors among adolescents, despite their capacity for rational decision-making, has given rise to the so-called dual-process models of adolescent risky behavior. These models emphasize that age-related differences in risk-taking cannot be solely attributed to the development of analytical ability. Instead, they suggest an imbalance in two information processing systems: the rational/analytical system and the reactive system. This imbalance is linked to the development of two neural regions in adolescence: the prefrontal cortex and the amygdalo-striatal system (Casey et al., 2008). The amygdalo-striatal system matures earlier and governs reactive decision-making, leading to heightened sensitivity to rewards and potentially riskier choices. In contrast, the prefrontal cortex, responsible for rational decisionmaking, matures later, thus inhibiting cognitive control and delaying gratification. In essence, adolescents engage in risky behaviors not because they lack the ability to think rationally, but because they are driven by emotional reactions. Consequently, poorer functional connectivity between the prefrontal cortex and the amygdalo-striatal system also may serve as a significant determinant of the proclivity for risk-taking in frequent or problematic adolescent gamblers. To test these assumptions rigorously, we require neuropsychological tasks capable of robustly activating both systems. With this objective in mind, Figner et al. (2009) developed the Colombian card task, featuring two parallel versions. The critical distinction lies in the type of cognition each version engages: "cold" cognition for rational decision-making, which strongly activates the prefrontal cortex, and "hot" cognition for reactive decision-making, which strongly activates the amygdalo-striatal system. Utilizing this measure allows us to gain a more comprehensive understanding of adolescent risky decision-making. In both versions of the task, we can assess not only the degree of risk propensity but also determine which of the three available information parameters (loss probability, gain amount, and loss amount) adolescents, on average, prioritize or overlook when making risky decisions. Surprisingly, despite its potential, this measure has not yet been applied in studies investigating the cognitive aspects of adolescent gambling. Its adoption could enhance our comprehension of the irrationality observed in adolescent gamblers, especially when considering variations in decision-making quality among adolescents with different gambling frequencies and varying levels of gamblingrelated problems. Importantly, adolescents who have experienced a range of psychosocial consequences due to excessive gambling might exhibit further impairments in decisionmaking. According to dual-process theories, we could hypothesize that adolescents who gamble more frequently or face higher levels of gambling-related problems possess heightened reward sensitivity. Consequently, they should exhibit riskier decision-making when the reactive pathway is activated. Moreover, employing this task enables us to delve deeper into the role of distorted gambling beliefs during the activation of both decision-making pathways. Clark (2010) suggests that these distorted beliefs likely originate at the neural level and may become more pronounced during reactive decision-making situations. This could potentially exacerbate the inclination toward risky decisions, particularly among adolescents who gamble more frequently or have experienced more significant psychosocial consequences due to gambling. Lastly, by leveraging this task, we can explore which information adolescents rely on when making risky decisions, contingent on their gambling frequency and gambling-related problems. Figner et al. (2009) found that, unlike adults, adolescents only consider information about loss probability and gain amount when making reactive decisions. In contrast, in rational decision-making scenarios, they take into account all three available information parameters. Consequently, it would be intriguing to investigate whether these findings hold true when comparing adolescents with differing levels of gambling frequency and gambling-related problems. It is plausible that adolescents who gamble more frequently or face higher levels of gambling-related problems possess heightened reward sensitivity, potentially limiting their use of all available information during reactive decision-making. The aim and the research methodology: The primary aim of this research was to examine the assumptions of dual-process theories concerning risky decision-making in adolescents. We sought to explore potential distinctions among adolescents based on their varying levels of gambling frequency and the extent of their gambling-related problems. Specifically, we wanted to investigate whether these differences influenced their propensity for risky decision-making depending on the type of decisionmaking pathway activated. Furthermore, we aimed to determine if these effects were contingent on the presence of distorted beliefs about gambling. We also set out to understand which types of available information adolescents with varying levels of gambling frequency and gambling-related problems relied on during reactive and rational decision-making. Our study involved a total of 378 adolescent boys, with an average age of 17 years (M = 17.03, SD = 1.296). The inclusion of girls was limited due to the infrequent occurrence of gambling among them. To effectively analyze the data, we divided the participants into two distinct subsamples: one consisting of adolescents with different gambling-related problems and the other comprising adolescents with varying gambling frequencies. For the sub-sample of adolescents with different gambling-related problems, we rigorously classified individuals who met the criteria for problematic gambling (demonstrating a high degree of psychosocial consequences resulting from gambling) and risky gambling (exhibiting a moderate level of psychosocial consequences due to gambling) on both the South Oaks Gambling Screen (SOGS-RA; Winters et al., 1993) and the Canadian Adolescent Gambling Inventory (CAGI; Tremblay et al., 2010). We identified a total of 34 such adolescents in each of the two groups. To maintain balance, we also included 34 adolescents in the non-risky category (displaying a low level of psychosocial consequences related to gambling) using a random selection method. This led to a total of 102 adolescents in this sub-sample. To avoid any overlap between the effects of gambling-related problems and gambling frequency, we included only those adolescents who did not meet the criteria for risky or problematic gambling on the aforementioned instruments. We then sorted them based on the frequency of their gambling, utilizing the Gambling Activities Questionnaire (Ricijaš et al., 2016). We identified 26 adolescents who fell into the occasional and frequent gambling categories. Additionally, we included 26 adolescents with infrequent gambling frequency through random selection, given their larger representation in the overall sample. Consequently, this sub-sample comprised a total of 78 adolescents. Our research took place in secondary schools and student dormitories within the areas of the City of Zagreb, Karlovac, and Varaždin Counties. We employed a convenient sampling procedure, and students participated in small groups (1-4 students) within separate rooms during regular school hours. This was coordinated with the school's professional staff or dormitory personnel. Before the measurement process, each student drew a unique short code consisting of a random sequence of numbers and letters (6 characters). This code was used to match the data collected from various measures, including questionnaires and experimental tasks, as all respondents completed all the measures. The experimental task used in our study was a shortened version of the Columbia Card Task (CCT; Figner et al., 2009). This task allowed us to assess two forms of information processing during decision-making: "cold" cognition (involving rational and intentional information processing) and "hot" cognition (involving reactive and affective information processing). To mitigate the transfer effect, we interspersed part of the questionnaire measures between the two experimental tasks, as participants needed to complete both versions of the task, with the experimental conditions rotated among them. In addition to measures of gambling frequency and gambling-related problems, we used the Gambling Belief Scale (Ricijaš et al., 2011) to gauge irrational beliefs about gambling. We also incorporated control measures for inter-individual personality differences, including the Need for Arousal Questionnaire (Figner et al., 2009) to assess situationally non-specific need for arousal, the disinhibition subscale of the Sensation Seeking Scale (SSS - form V; Zuckerman et al., 1978) to measure cognitive disinhibition, the Barratt Impulsiveness Scale-Version 11 (BIS-11; Patton et al., 1995), and the emotional stability subscale of the IPIP 50S questionnaire (Mlačić and Goldberg, 2007). All data were collected using laptops, with experimental tasks conducted in a specialized local programming environment designed for experiments (Psychopy). Questionnaire measures were administered via the Google Forms platform. On average, each data collection session with a single respondent lasted approximately 45 minutes. Results: The findings revealed that adolescents who frequently engage in gambling tend to make riskier decisions compared to those who gamble occasionally and those who rarely gamble, particularly when the reactive decision-making process is at play. However, there was no discernible difference between the latter two groups in this context. In contrast, when it comes to rational decision-making, the three aforementioned groups didn't exhibit an equal propensity for making risky decisions. Specifically, adolescents who frequently gamble, even when the rational decision-making pathway was activated, made riskier choices compared to those who rarely gamble. However, they did not differ significantly from adolescents who gamble occasionally. Moreover, no significant distinction was observed between the latter two groups either. A similar trend was also identified among adolescents with varying levels of gamblingrelated problems. Problematic gamblers tended to make riskier decisions than both risky and non-problem adolescent gamblers when the reactive decision-making pathway was engaged. Additionally, in this condition, risky adolescent gamblers also made significantly riskier decisions compared to their non-problem peers. The pattern of differences remained consistent during the activation of the rational decision-making pathway: problematic adolescent gamblers made riskier decisions than both risky and non-problematic adolescents, while risky gamblers made riskier decisions than non-problematic ones. The results further indicated that there was no interaction effect between irrational beliefs about gambling, the activated decision pathway, and gambling frequency, or gambling-related problems on the degree of risk propensity. In other words, a higher level of irrational beliefs about gambling did not lead to an increase in the riskiness of decisions made by adolescents who gamble frequently or those with higher levels of gambling-related problems in the context of reactive decision-making. Regarding the types of information adolescents employ when making decisions, the results from a sub-sample of adolescents with varying gambling frequencies showed that, during the reactive decision-making task, adolescents who frequently gamble primarily relied on information related to loss probability, while neglecting information about the gain amount. Adolescents who gambled occasionally also considered information about the loss probability but additionally factored in information about the loss amount, while disregarding information about the gain amount. Conversely, adolescents who rarely gambled considered all three types of available information parameters in their decision-making process. In the rational decisionmaking task, a similar pattern emerged. Adolescents who rarely gambled continued to utilize all three types of information at their disposal, whereas adolescents who gambled occasionally and frequently focused solely on information concerning loss probability. A comparable pattern was observed in a sub-sample of adolescents with different levels of gambling-related problems. During the reactive decision-making task, only adolescents with lower levels of gambling-related problems considered all three types of information, while those with moderate and high levels of gambling-related problems exclusively relied on information about loss probability. In the rational decision-making task, only non-risky adolescents utilized all three types of information, while risky and problematic adolescents took into account both loss probability and loss amount. Discussion: Given the pattern of interaction effects observed, the results only partially align with the initial expectations. Specifically, concerning the differences observed in the inclination to make risky decisions in the context of rational decision-making, it appears that adolescents who frequently gamble, including those with gambling problems, might have a compromised ability to make sounder (more rational) decisions. One plausible explanation for this diminished capacity to make better decisions might be attributed to potential impairment of the ventromedial prefrontal cortex in these adolescent groups. A study by Figner et al. (2009) demonstrated that, during rational decision-making, both adults and adolescents made relatively comparable, nonrisky decisions. However, in situations where the reactive decision-making process was engaged, adolescents tended to make riskier decisions than adults. These findings supported the core tenets of dual-process theories and suggested that adolescents, when faced with scenarios that strongly activate the prefrontal cortex, enabling better inhibitory control, exhibit the capacity to make more mature and less risky decisions similar to adults. Given that participants in their study were drawn from the general adolescent population, there is a possibility that, in adolescents who gamble frequently or have higher levels of gambling-related problems, cognitive control ability may be further constrained and impaired beyond the developmental deficits commonly associated with adolescence. Research focused on adult problem gamblers employing the IOWA risky decision-making task, among others, has consistently revealed that prefrontal cortex pathophysiology is a reliable marker of problem gambling. Some functional brain imaging studies have also detected reduced activity in the prefrontal cortex of problem gamblers compared to controls. Intriguingly, some of these studies also recorded diminished activity in the striatum/ventral striatum, providing evidence in support of the "insufficient reward" hypothesis. This hypothesis posits that problem gamblers engage in excessive gambling to stimulate an inadequately activated neural reward system due to reduced ventral striatum activation, while reduced prefrontal cortex activity concurrently impairs cognitive control and inhibitory mechanisms. Remarkably, certain studies conducted on the general adolescent population have identified evidence of reduced activity in this neural circuit, indicative of hyposensitivity. However, it's essential to note that most of these studies have reported the opposite pattern – hypersensitivity in the striatal system during reward anticipation. Thus, it's plausible that adolescents who gamble frequently, as well as those with higher levels of gambling-related problems, exhibit reduced activity in the striatal system compared to their counterparts in the general population, i.e., those who rarely gamble or have lower levels of gambling-related problems. In essence, the potential striatal system hyposensitivity in these adolescent groups could elucidate why they make riskier decisions, even during the rational decision-making task. This could be attributed to their need to stimulate an insufficiently activated reward system due to reduced ventral striatum activity, coupled with the possible additional impairment of cognitive control resulting from reduced prefrontal cortex activity, as evidenced in adult gamblers. Another potential explanation for the heightened propensity for risky decision-making in both frequent and problematic adolescent gamblers during the rational decision-making task may lie in recent neuroscience findings. These findings suggest the existence of an additional (third) system – the neural basis of which is the insular cortex or simply, the insula. Classic dual-process models of judgment and decision-making emphasize the presence of two distinct systems that compete when determining decision outcomes: one favoring immediate concrete rewards and the other considering abstract rewards in the future. However, the insula plays a role in decision-making by representing the subjective experience of craving and maintaining homeostasis, an aspect not entirely captured by standard dual theories. Cravings emerge as a result of the brain and body's combined efforts to correct a homeostatic imbalance, modulating the activity of both the ventromedial prefrontal cortex and the amygdala to restore internal biological equilibrium. This homeostasis establishment activates the insula, which is perceived as a craving sensation. Current bodily needs tend to outweigh long-term abstract goals, potentially leading the insula to inhibit the reflective system (ventromedial prefrontal cortex) while amplifying the impulsive system (amygdalo-striatal). Therefore, it's conceivable that, even during the rational decisionmaking pathway, frequent and problematic adolescent gamblers in this study, owing to increased insula activity, may have still experienced a sense of craving, impairing their cognitive control abilities. This sensation may have disrupted their decision-making abilities, causing them to make riskier choices. Furthermore, the absence of a significant interaction effect between irrational beliefs about gambling, the activated decision-making pathway, and gambling frequency, as well as gambling-related problems, on the degree of risk propensity, did not align with the initial expectations. The reason why higher levels of irrational beliefs about gambling did not intensify the degree of decision riskiness in adolescents, especially those who gamble more frequently or have higher levels of gambling-related problems during the reactive decision-making task, could potentially be due to methodological factors. Prior studies (Clark et al., 2009), which suggested that irrational beliefs about gambling could influence the activity of the reward neural circuit, employed carefully designed tasks aimed at simulating structural characteristics of games, such as the "near miss" and personal control effects. These characteristics are believed to manipulate the player's perception, leading to the formation of irrational beliefs that foster an illusion of control, subsequently increasing risktaking behavior and persistence in gambling. This study, however, did not incorporate such game elements. In other words, the Columbia Card Task (Figner et al., 2009) did not induce the "near miss" effect, which previous research (Clark et al., 2009) demonstrated to enhance the activity of the ventral striatum and anterior insula. This neural response corresponds to the one reported during monetary wins. Consequently, it remains uncertain whether, during the reactive decision-making task, which structurally activates the striatal regions of the neural reward circuit, adolescents who frequently gamble and those with higher levels of gamblingrelated problems would make even riskier decisions if this task simulated the "near miss" effect, as it is known to promote activation of similar reward circuit regions. To sum up, this study examined the types of information adolescents with different gambling frequencies and varying gambling-related problems use in rational and reactive decision-making tasks. The results, encompassing both sub-samples, suggest two general conclusions: Firstly, all groups of adolescents primarily rely on information concerning loss probability when making decisions, both in the reactive and rational versions of the task. Secondly, adolescents who gamble occasionally and frequently, along with those with medium to high levels of gambling-related problems, employ significantly less information than expected during the rational version of the task. This finding implies a potential impaired ability among these adolescent groups to make better and more rational decisions, beyond the normative expectations. It seems that their prefrontal cortex inhibitory control capacity is notably diminished compared to the general population of adolescents, even in situations (tasks) that should theoretically bolster it, ultimately leading to their failure to consider all available information that could contribute to better decision-making. |
