5.2.1.1. Ellsberg’s paradox: highlighting ambiguity aversion

For a long time, normative models of decision-making have considered that the individual does not behave differently in ambiguous situations than in risky situations. For example, according to the theory of subjective utility expectation proposed by Savage (1954), when faced with an ambiguous alternative, the individual would attribute subjective probabilities to the various events that could occur, based on their representation of the situation (the hypothesis of the attribution of subjective probabilities to ambiguous situations).

Through a set of ingenious situations involving a choice between two urns whose contents are known to a varying degree (Box 5.1), Ellsberg (1961) challenged this idea of subjective probability assignment. He demonstrated that the theory proposed by Savage was not able to account for the existence of paradoxical decision-making behavior. Indeed, an individual’s preferences are not dictated by probabilities, but by the amount of information available regarding each option. Individuals thus systematically choose the option associated with the highest level of information, regardless of its advantageous nature.

The idea of subjective probability allocation is challenged, in favor of an ambiguity aversion effect leading individuals to systematically prefer a risky situation to an ambiguous one, despite its advantageous or disadvantageous nature on the mathematical level.

Thus, aversion to the lack of information can lead the individual to make erroneous probability judgments, even if the choice involved is basic in terms of the logical-mathematical skills involved.

5.2.1.2. The origin of the ambiguity aversion phenomenon

Although its robustness has been widely demonstrated, the nature and origin of ambiguity aversion are still debated. Several explanations have been put forward to account for this phenomenon.

For Einhorn and Hogarth (1985), ambiguity aversion is a purely cognitive process. Faced with an ambiguous prospect, the individual initiates a provisional estimate of the probabilities, which they then adjust upward or downward on the basis of the simulation of all conceivable distributions.

The hypothesis of an emotional origin, linked to the physiological state triggered by uncertainty, is put forward by Pulford and Colman (2007) and appears today as the most convincing hypothesis (Rubaltelli et al. 2010). In line with the dual process models (Evans and Over 1996; Evans 2011), ambiguity aversion could be an emotional bias that is part of a competition between two types of strategies: the rational analysis of the situation (known as type 2) and the heuristic strategy (or automatic, known as type 1), which consists of considering a lack of information as dangerous and systematically rejecting an ambiguous option in favor of a safe or risky option that may be less advantageous. Two neuroimaging studies have confirmed the emotional nature of ambiguity aversion by highlighting the involvement of an emotional network, underpinned by the amygdala and orbitofrontal cortex, in the detection of ambiguity (Hsu et al. 2005; Levy et al. 2010). When young adults were presented with ambiguous options, Hsu et al. (2005) found a bilateral activation peak in the amygdala and orbitofrontal regions, which are respectively involved in the reaction to and integration of emotional information. Furthermore, the superiority of activation of the orbitofrontal cortex in the ambiguous condition compared to a risky condition appears to be strongly related to the strength of participants’ ambiguity aversion. These neural patterns suggest the existence of a vigilance system, implying a quick assesment of the level of uncertainty, which involves the orbitofrontal cortex and the amygdala. This system becomes increasingly engaged as the ambiguity of the situation increases. It could play an adaptive role by allowing the organism to be alerted to the lack of information in order to avoid a situation associated with an excessive level of uncertainty and to mobilize the cognitive and behavioral resources necessary to acquire additional information in the environment.

5.2.1.3. Is greater tolerance of ambiguity an explanation for adolescent risk-taking?

The hypothesis of reduced ambiguity aversion in adolescents has recently been put forward to explain their tendency to engage in risky behavior. In fact, the risk-taking observed in adolescence most often occurs in the context of ambiguous decision-making, in the absence of information allowing for a real assessment of the risk level. For example, it is unlikely that adolescents would know the exact probability of being involved in an accident if they decided to drive a scooter under the influence of alcohol. Some authors suggest that an adolescent’s engagement in risky behavior reflects a greater tolerance of situations with ambiguous consequences rather than a preference for risk per se.

A study by Tymula et al. (2012) supports this hypothesis by offering adolescents and young adults a series of choices between a sure option (e.g. a certain payoff of $5) and an option that may be either risky (e.g. a 50% chance of winning nothing and a 50% chance of winning $50) or ambiguous (e.g. unknown probabilities of winning nothing or winning $50). Adolescents were less likely to accept the risky option compared to the sure option and were more risk averse than adults.

Moreover, they more readily accept an ambiguous option compared to a sure one, indicating a less marked aversion to ambiguity than adults. In addition, adolescents’ tendency to accept ambiguous options increases linearly from the age of 10 years to 25 years and appears to be correlated with risky behaviors in daily life (Blankenstein et al. 2016).

Thus, the tolerance for ambiguous situations could explain some of the risk taking that is observed during adolescents, but may be an adaptive trait during this developmental period. Indeed, a greater tolerance for ambiguity may create opportunities for children and adolescents to learn from their environment. Imagine a 4-year-old child climbing the shelves of his or her parents’ closet, trying to open a box hidden on the top. According to Tymula et al. (2012), such behavior will be perceived as curiosity offering the opportunity of discovering their environment and not as a real attempt to take risks. Similarly, a greater tolerance for the unknown encourages exploration of the environment and promotes an essential learning mechanism during the adolescent period. Risk-taking activities may allow adolescents to learn more about themselves and to see their future in a different way (e.g. taking the risk of discovering new activities or making new acquaintances outside of one’s group of friends may allow one to discover new passions by stepping out of one’s comfort zone).

In conclusion, the emotional phenomenon of ambiguity aversion seems to be less robust in adolescents. We can suppose that this developmental window would be more favorable to the exploration of the environment and the unknown, which would allow adolescents to learn and discover new things. We will now see that in situations of complete ambiguity, our emotional states provide clues to learn from our previous choices, in order to choose advantageously.

5.2.2.1. The origins of the somatic marker hypothesis

It is common to think that a good decision is free from any emotional influence. As we saw earlier, this idea is supported by in cognitive science, which attributes an emotional nature to various several studies biases, as is the case for the phenomenon of ambiguity aversion (Pulford and Colman 2007).

Breaking with this ancient dualism between emotion and reason (or between body and mind), the neurologist Antonio Damasio (1994) developed the hypothesis that emotions could participate in adapted decision-making in ambiguous situations. It was the in-depth neuropsychological examination of patients with a lesion of the ventromedial prefrontal cortex that enabled Damasio to draw up a clinical profile mainly characterized by difficulties in decision-making and socially maladjusted behavior (Box 5.2). As this region is known to handle emotional processes, he hypothesized that emotions are an integral part of adaptive decision-making. The environment in which we live is so complex that it is impossible to weigh up the benefits and costs of each option available to us, especially since we often have little information about the associated issues and their probabilities.

The somatic marker hypothesis (SMH) suggests that the retrieval of past emotional experiences unconsciously guides us in our choices, allowing us to learn from our past decisions to improve our future ones. This hypothesis refers to somatic markers because our emotional reactions are necessarily anchored in our body (heart rate acceleration, muscle tension, sweating, etc.). They are perceived as emotional signals that would involve two critical regions: on the one hand, the amygdala, for an immediate emotional reactivity (e.g. emotion related to immediate gains and losses), and, on the other hand, the ventromedial prefrontal cortex for the anticipation of a future emotional reactivity (e.g. related to long-term consequences).

5.2.2.2. Empirical evidence in favor of the somatic marker hypothesis

To test this hypothesis, numerous studies have assessed adults’ ability to learn to choose advantageously in an ambiguous financial decision-making game: the Iowa Gambling Task (IGT) (Bechara et al. 1994); for a review, see Dunn et al. (2006). Participants are instructed to choose from four piles of cards with the goal of winning the most money, knowing that some piles are more advantageous than others. Participants do not have explicit information about the riskiness of their choices because they do not know the probabilities, the amounts of gains and losses associated with each option, or even the number of trials that will be made (i.e. 100).

In fact, the feedback given to participants after each choice combines gains and losses of different magnitudes and frequencies, so that only two of the four options proposed are advantageous in the long-term (they lead to modest gains but remain greater than the losses in the long-term). While limited attentional capacities do not allow for precise monitoring of the benefit/cost ratio of each of these options, healthy adults gradually manage to disengage from the attractive but disadvantageous options in the long-term, and preferentially choose the less attractive but advantageous options in the long-term (Figure 5.3).

Consistent with the SMH, groups of patients with lesions to the ventromedial prefrontal cortex or the amygdala were unable to anticipate the future consequences of their choices. Analysis of the skin conductance responses (SCRs) of these different groups of participants supports the SMH. Healthy adults showed an emotional reactivity in responses to the complex feedback such that it distinguished net gains from net losses. More surprisingly, they also progressively developed an anticipatory emotional response that distinguished disadvantageous from advantageous options (Bechara et al. 1999), a response which appears to be correlated with the ability to choose advantageously (Carter and Pasqualini 2004). The behavioral difficulties of the patient groups can be explained by different phenomena: a lack of immediate emotional reactivity in patients with amygdala lesions and a lack of anticipatory emotional reactivity in patients with ventromedial prefrontal cortex lesions. This work thus shows that emotional reactions may well allow us to learn from our past choices, and support advantageous decision-making in ambiguous situations (Bechara et al. 2005). Most critics of the SMH question the influence of other cognitive mechanisms for learning under ambiguity, such as the role of executive functions (Fellows and Farah 2005) or conceptual knowledge (Maia and McClelland 2004), or question the validity of a physiological measurement such as the SCRs (Tomb et al. 2002). Yet, two decades later, new experimental protocols continue to prove the existence of an emotional guidance at the heart of the learning process in decisions under ambiguity (see Aïte et al. (2013, 2014)).

5.2.2.3. The development of decision-making under ambiguity

Analysis of these decision-making abilities throughout development was done using child-friendly adaptions of the IGT (e.g with smaller sums) and suggests that this emotional learning remains ineffective in children and adolescents (Crone and van der Molen 2004, 2007; Kerr and Zelazo 2004; Cassotti et al. 2011). Indeed, results show a linear development: children fail to disengage from attractive but disadvantageous options in the long run, showing an inability to anticipate the future consequences of their choices, while adolescents show an intermediate profile, better than children but lower than adults (Crone and van der Molen 2004, 2007; Cassotti et al. 2011). Both children and adolescents show emotional reactions immediately after feedback has been received but only adolescents develop an anticipatory emotional reaction. On the other hand, unlike adults, this anticipatory signal focuses on less adaptive indicators because it distinguishes options according to the frequency of financial losses, regardless of their long-term advantageous nature. In summary, adolescents seem to develop an anticipatory emotional reactivity that guides them in an inappropriate way toward the goal of the game (i.e. to win as much money as possible). One of the reasons for these learning difficulties in a complex environment has to do with the ability to adjust one’s behavior immediately after feedback (Cassotti et al. 2011). Unlike adults, children and adolescents more systematically change options after receiving negative feedback, which may prevent them from learning more finely the advantageous, or non-advantageous, nature of the options presented (Aïte et al. 2012). It should be noted that all options lead to gains and losses in the IGT. Because of this, to better understand the advantageous or disadvantageous properties of each option, it is best to choose them repeatedly, despite the losses that may occur. Another mechanism that may be lacking in younger children is cognitive control to enable greater tolerance for the occurrence of negative feedback (Cassotti et al. 2014). Studies using neuroimaging support the idea that children have more difficulty learning on the basis of negative feedback than on the basis of positive feedback (Van Leijenhorst et al. 2008). Taken together, this work shows that emotions can serve learning under uncertainty but that this emotional learning remains inefficient in children and adolescents and may require the intervention of additional regulatory processes to become efficient (Cassotti et al. 2014).

5.2.2.4. Risk-taking in adolescence as a learning difficulty under ambiguity

Beyond a choice between a risky and a safer option, everyday risk-taking can result in a gradual risky behavior. Imagine yourself driving to a job interview for which you are already running late. The importance of this interview will probably lead you to drive faster than usual. At what point does the risk to your life and the lives of others outweigh the benefit of your punctuality at the interview? In such circumstances, the individual’s choice would not simply be between two options of different mathematical expectation, but to decide on a limit to their increasing involvement in risky behavior, re-evaluating at each stage the relative weight of the positive and negative consequences. The Balloon Analogue Risk Task (BART), originally proposed by Lejuez et al. (2002), measures a risk-taking behavior in which, as in most everyday situations, risk is rewarded up to a certain point but is accompanied by negative consequences beyond that point. Participants are asked to blow up balloons that may explode randomly, in order to win as much money as possible. Each inflation results in an increase in the size of the balloon and the reward. The participant can then inflate each balloon as much as they want to accumulate rewards, knowing that explosion will be accompanied by the loss of all the rewards accumulated on this balloon. They can then decide at any time to save the money in a final pool and move on to the next balloon. The number of inflations on the balloons and the number of explosions thus reflect the individual’s commitment to risk-taking (i.e. taking the risk of exploding the balloon to maximize the reward).

Using an adaptation of the BART, Osmont et al. (2017) suggest that the difficulty of adjusting one’s behavior, following learning based on the receipt of feedback, can be advanced to account for adolescents’ engagement in ambiguous risky behaviors. This adaptation involves manipulating the level of available information by contrasting two conditions: an “uninformed” ambiguity condition requiring the inference of the balloons’ resistance level based on the feedback received and an “informed” condition for which the balloons’ resistance level is directly accessible. The results show that the availability of information about the level of risk is sufficient for adolescents to perform as well as adults at the end of the game. However, adolescents aged 14–16 years show difficulties in adjusting their risk-taking to the resistance of the balloons in uninformed conditions: they take more risk than adults for low resistance balloons (when the risk is disadvantageous) and on the contrary less risk than adults for the most resistant balloons (when the risk is advantageous). Thus, the risk-taking of adolescents reflects more a learning difficulty based on the processing of feedback received, rather than a general propensity for risk.

5.3.1.1. Neurobiological modeling of risk-taking during development: Casey’s model

Recent neurocognitive modeling of adolescence underlines that emotions account for the individual’s massively irrational behaviors in the areas of probability judgment and decision-making. The model by Casey et al. (2008) suggests that an individual’s decision-making behaviors reflect the existence of a competition between two systems:

• – an emotional system, that involves subcortical limbic regions (e.g. the nucleus accumbens), responding to emotionally salient cues in the environment;
• – a top-down control system, that involves the prefrontal regions, responsible for the regulation of these emotions.

This model offers an explanatory framework that can account for the increase in risky behavior during adolescence, which would reflect a greater maturation gap between these two systems at this time. As shown in Figure 5.3, these systems have distinct developmental trajectories. While the prefrontal regions mature linearly and relatively late, the emotional system matures earlier. The adolescents’ risky behaviors could be explained by a hyper-sensitivity to emotional stimuli in the environment, associated with an immaturity of the top-down control processes necessary for the regulation of these emotions.

Casey’s model is based on three behavioral and neurofunctional studies showing slow maturation of the prefrontal cortex and emotional hypersensitivity in adolescence (Galvan et al. 2006; Hare et al. 2008; Somerville et al. 2011). When confronted with a reward-receiving task in functional magnetic resonance imaging (fMRI) (Galvan et al. 2006), adolescents aged 13–17 years show stronger activation of the nucleus accumbens (part of the ventral striatum), involved in reward sensitivity, compared to children and adults. In contrast, there is a progressive decrease with age in the level of activation of regions involved in cognitive control (i.e. orbitofrontal cortex), as well as more diffuse activation in children and adolescents compared to adults, indicating that these regions become more and more efficient with age.

In addition, two fMRI studies have highlighted the specificity of emotional reactivity and emotional regulation difficulties in adolescents thanks to an emotional adaptation of the Go/NoGo task, originally designed to measure the inhibition abilities of a motor response (Hare et al. 2008; Somerville et al. 2011). In this adaptation, participants were asked to provide the fastest possible motor response to a given category of stimuli (i.e. Go trials) while not responding to another category of less frequent stimuli (i.e. NoGo trials). The lack of response to NoGo trials then required the inhibition of a motor routine that results from the overrepresentation of Go trials. Three categories of faces from the NimStim Set of Facial Expressions (Tottenham et al. 2009) expressing emotion (fear, joy and neutral) were used to form a set of Go and NoGo trials. Hare et al. (2008) demonstrated a higher activity of the amygdala after the presentation of fearful faces in adolescents compared to children and adults. They thus confirmed the specificity of emotional reactivity in adolescence. Furthermore, the link between the strength of fronto amygdala connectivity and the habituation of the amygdala to repeated presentation of emotional stimuli (i.e. the difference between first vs. last trials) constitutes evidence for the involvement of frontal regions in the regulation of these emotions.

Finally, the study by Somerville et al. (2011) complements these results by highlighting a greater number of false alarms (i.e. lack of inhibition in response to a NoGo trial) for happy faces than for neutral faces in adolescents, whereas the number of false alarms was not modulated by the type of emotional face in children and adults. The authors found greater activation of the ventral striatum in adolescents than in adults and children in response to emotional faces, as well as a linear decrease with age in the recruitment of the right inferior frontal gyrus (IFG) during the inhibition of the motor response for NoGo trials.

All these behavioral and neuroimaging data confirm the existence in adolescence of a peak in the activation of regions involved in the reward system (i.e. the nucleus accumbens) as well as a later functional development of top-down control regions (i.e. the orbitofrontal cortex). This shift would explain a difficulty in the specific control of emotional stimuli in adolescents, compared to children or adults. This model is a part of dual models, in the sense that they predict behavior based on the maturity of an emotional system on the one hand and a cognitive control system on the other hand (Shulman et al. 2016). Part of the scientific community is enthusiastic for these models that account for the specificity of adolescence, but several authors call for caution as they have yet to be specified (See for example Pfeifer and Allen 2012, 2016).

5.3.1.2. Sensitivity to rewards in adolescence

In addition to the above-mentioned studies, developmental neuroscience has aroused considerable interest in the question of the development of sensitivity to rewards as an explanatory factor for risk-taking in adolescents. A series of studies on the neurofunctional development of the brain regions underlying the reward system has emerged. Behaviorally, multiple adaptations of the iconic marshmallow task (Mischel and Metzner 1962; Mischel et al. 1989) – which involves choosing between an immediate reward (e.g. a marshmallow now) and a larger delayed reward with a variable time delay (e.g. three marshmallows in 20 minutes) – have shown an increase with age in the ability to resist immediate rewards (Mischel et al. 1989; Steelandt et al. 2012). Focusing on the adolescent period and through the use of financial rewards, Steinberg et al.’s (2009) study found that young adolescents exhibit less significant future orientation and accept more weak immediate rewards compared to more significant delayed rewards than late adolescents (over the age of 16 years) and young adults.

Neuroimaging results seem to be more contrasted: some studies attest to a hypersensitivity of the reward system in adolescence and others suggest, on the contrary, a reduced sensitivity to the rewards of a behavior (for a review, see Galvan (2010)). In line with Galvan et al. (2006), several recent studies provide evidence for hypersensitivity to rewards with tasks outside the decision-making domain (Padmanabhan et al. 2011; Smith et al. 2011). Indeed, associating a reward with a trial in an antisaccade task leads to increased performance only in preadolescents (8–13 years) and adolescents (14–17 years), who otherwise show higher reward-related neural activity compared to adults (Padmanabhan et al. 2011). Similarly, in a sustained attention task associated with the receipt of rewards, the Rewarded Continuous Performance task, adolescents aged 10–17 years show greater sensitivity to the effect of rewards than adults: their response times decrease for rewarded targets compared to unrewarded targets. However, this study does not isolate clear specifics of adolescent neural responsiveness (Smith et al. 2011). On the other hand, conflicting results suggest decreased recruitment of the right striatum and insula during anticipation of a potential payoff in adolescents aged 12–17 years compared to adults (Bjork et al. 2004). According to these authors, adolescents engage massively in risky behaviors to satisfy an exacerbated search for rewards, in compensation for the under-activation of regions linked to the anticipation of potential gains.

Several hypotheses can be put forward to account for these discrepancies: the age of the sample, the existence and characteristics of the comparison group, the experimental design or the type of processing and statistical analyses used. These differences are also reflected in the fundamental distinction between two components linked to the notion of reward, wanting and liking, proposed by Berridge et al. (2009). While wanting refers to the desire to obtain the reward (anticipation of rewards), liking corresponds to the pleasure associated with obtaining the reward. Adolescents’ hypersensitivity to rewards could be specifically related to obtaining a gain and not to its anticipation. Using the Slot Machine Task paradigm, Van Leijenhorst et al. (2010) investigated the functional development of regions involved in anticipation versus receipt of non-predictable rewards unrelated to the participant’s responses. On each trial, participants aged 10–12, 14–15 and 18–23 years activated a slot machine consisting of three wheels stopping successively on a picture. They received a reward if the machine stopped on three identical images (X-X-X configurations). The authors then defined two activation contrasts allowing them to isolate the processes of anticipation and reception of a reward. The anticipation contrast corresponds to the difference in activation between trials with two identical first images (X-X-Y configurations) and trials with two different first images (X-Y-Z configurations). The reward reception contrast corresponds to the difference between trials with three identical images (X-X-X configurations) and trials with the first two identical images (X-X-Y configurations). The results of this study indicate an adolescent hypersensitivity restricted to the reward reception phase, showing a quadratic evolution of ventral striatum activation in adolescence following reward reception, with a peak around 14–15 years of age. In contrast, the authors noted a linear decrease in anterior insula recruitment with age during reward anticipation. The specificity of emotional reactivity in adolescence can thus be qualified: adolescents show a superior neural response to the receipt of a reward, but on the contrary an attenuated neural response during its anticipation, which leads them to seek the potential rewards of a risky behavior more (Spear 2011).

In conclusion, dual models of decision-making broadly consider that emotions lead to greater risk-taking in adolescence (Shulman et al. 2016). However, risk-taking is not necessarily maladaptive, unregulated or impulsive. It could be planned in some contexts. Indeed, adolescents may deliberately engage in risk-taking behaviors in order to gain social rewards, such as interpersonal acceptance (Rawn and Vohs 2011). For some, planned risk-taking may even require a form of control, in order to overcome risk aversion, such as a distaste for alcohol or fear of the negative consequences associated with psychoactive substance use, which may be valued highly socially (e.g. group acceptance).

5.3.1.3. Taking into account sensitivity to negative consequences: the Ernst model

Contrary to the hypothesis of adolescents’ hypersensitivity to the reception of rewards, their reactivity to losses and associated negative emotions remains little studied and is often excluded from theoretical models such as that proposed by Casey et al. (2008). However, decision-making processes do not only involve the positive consequences associated with our choices but also the potential negative outcomes, which are fundamental to the definition of risk-taking. Some studies suggest a greater emotional impact of negative feedback in early adolescence. According to these studies, preadolescents and adolescents show greater activation of the amygdala, orbitofrontal cortex and anterior cingulate cortex in response to negative stimuli, compared to adults (Monk et al. 2003; van Leijenhorst et al. 2006; Guyer et al. 2008). Conversely, Ernst et al. (2004) confirm a hypersensitivity to rewards in adolescents and suggest a lower reactivity to negative stimuli. The proposed task consists of making a choice between two options, presented in the form of a wheel of fortune divided into two equal portions (pink and blue). If the wheel of fortune stops on the chosen color, the participant wins a gain of varying magnitude. Adolescents show greater satisfaction after receiving a reward than adults do. Furthermore, the receipt of feedback results in a lower engagement of the amygdala (i.e. involved in the avoidance of negative stimuli) and a higher contribution of the nucleus accumbens (involved in the reward system) in adolescents.

These results led Monique Ernst et al. to propose the triadic model of motivated behavior, which allows for the integration of three systems in the understanding of decision-making processes (Ernst 2014; Ernst et al. 2006, 2009) (Figure 5.4):

• – an approach system engaged in reward seeking: the dorsolateral prefrontal cortex and ventral striatum (nucleus accumbens);
• – an avoidance system defined as a “behavioral brake” engaged in the avoidance of aversive stimuli: the amygdala;
• – a top-down regulation system, which ensures the modulation of the participation of the two other systems: the prefrontal cortex.

Applied to adolescence, this model postulates the existence of an imbalance between the approach and avoidance systems in favor of the approach system, associated with an immaturity of the system involved in the regulation of the balance between sensitivity to rewards and sensitivity to aversive stimuli. It thus makes it possible to account for exacerbated risk-taking in adolescence, by clarifying the emotional issues present in the Casey et al. (2008) model, through the integration of reactivity to negative feedback and emotions.

5.3.2.1. How can regret be studied experimentally?

The feeling of regret and its influence on decision-making are typically studied using computerized tasks that offer a choice between two gambles represented by wheels of fortune (Mellers et al. 1999; Camille et al. 2004). Each gamble presents two possible outcomes (e.g. +50 vs. −200 or −200 vs. −50) with varying probabilities of gain and loss (Figure 5.6). Thus, these are mostly decision-making under risk situations. Once participants have made their choice, the outcome of the chosen gamble (factual outcome – partial feedback) and the outcome of the unchosen gamble (counterfactual outcome – complete feedback) are presented on the screen. In this way, the participants have the opportunity to compare the factual result with the counterfactual result. To follow up the feedback, participants indicate their emotional state on a Likert scale (e.g. a scale from −50 to +50 (Camille et al. 2004)). The partial feedback aims to investigate the feeling of disappointment (in case of a loss) and satisfaction (in case of a gain), whereas the complete feedback aims to investigate the feeling of regret (in case of an unfavorable counterfactual comparison) or relief (in case of a favorable counterfactual comparison).

COMMENT ON FIGURE 5.6. – When the participant is given the opportunity to compare the outcome they have obtained with the outcome they could have obtained if they had made a different choice, their emotional response is modulated by the alternative outcome. A positive outcome will be perceived as less pleasant if the alternative outcome is higher (regret effect), whereas a negative outcome will be perceived as less unpleasant if the alternative outcome is lower (relief effect – adapted from Camille et al. (2004)))

Numerous studies indicate that emotions reported by the participants about the outcome obtained are modulated by the value of the outcome of the unchosen gamble, thus leading to regret or relief (Camille et al. 2004; Bault et al. 2008; Habib et al. 2012). For example, participants reported negative emotional feelings when they lost $8 (the monetary unit varying by country) and discovered that they could have gained $200 by choosing the alternative gamble. Similarly, participants reported slightly negative emotional feelings when they won $50 but found that they could have won $200 by choosing the alternative gamble. This effect illustrates the regret felt by participants as a result of the counterfactual comparison, which leads them to evaluate a win as a subjectively negative outcome. Conversely, participants reported an average emotional feeling close to 0 (on an emotional scale ranging from −50 to +50) when they lost $50, but discovered that they could have lost $200 by choosing the alternative gamble. This effect illustrates the relief participants feel as a result of the counterfactual comparison, which leads them to evaluate a loss as an emotionally “neutral” outcome. These results also highlight that gains and losses do not have an objective value, but a subjective value determined by a reference point, constituted here by the outcome of the unchosen gamble (Coricelli et al. 2007; Habib and Cassotti 2015).

5.3.2.2. The development of regret and its influence on the decision-making of children and adolescents

Research on the development of regret has sought to determine the age at which regret can be experienced and its influence on decision-making in children and adolescents. A similar paradigm to the one used in adults was developed to study the development of the feeling of regret in children (Weisberg and Beck 2010, 2012). Children chose between two boxes, before discovering their own outcome (partial feedback) and the outcome of the unchosen box (full feedback). The latter could be higher or lower than their own result. After each feedback, children were asked to indicate their emotional state by selecting one of five different faces, ranging from an unhappy face to a happy face. Children reported being more unhappy when they discovered a more favorable counterfactual outcome (full feedback) around the age of 5 years (Weisberg and Beck 2012). There is variability in the age at which children may first experience regret, but the majority of children appear to experience this emotion by the age of 6 years (O’Connor et al. 2012). However, the feeling continues to develop from childhood to adulthood (Habib et al. 2012; Feeney et al. 2018). Moreover, the notion of responsibility for choice is necessary for the feeling of regret. Indeed, children more often reported being sadder when they themselves made the choice leading to an unfavorable outcome, rather than when the choice was made by the experimenter or as a result of rolling a die (Weisberg and Beck 2012). The word “regret”, meanwhile, is thought to be understood around the age of 9–10 years (Guttentag and Ferrell 2004; Baron-Cohen et al. 2010), but little research currently exists on the topic of understanding and labeling this complex emotion.

As regret is defined as an adaptive emotion that allows us to learn from our past mistakes, its role in the willingness to reconsider a choice and its actual reconsideration have been explored in children (Habib et al. 2012; O’Connor et al. 2014). Habib et al. (2012) asked children (10–11 years-old), adolescents (14–15 years-old) and young adults (19–24 years-old) how much they were willing to reconsider their choice in situations leading to regret (after obtaining a less favorable outcome than the counterfactual one). When children have won a small amount of money but could have won a larger amount, they do not wish to reconsider their choice, unlike adults and adolescents. However, they do want to reconsider their choice in situations of regret leading to an actual loss. Thus, among 10–11 year-olds, the feeling of regret does not systematically lead to a desire to reconsider their choice in order to obtain a more favorable outcome, unlike adolescents and adults. O’Connor et al. (2014) studied the changes in choice following the feeling of regret (adaptive switch) in 5, 7 and 9 year-olds. Children had to choose between two boxes of different colors. By means of a trick, they were first confronted with a situation in which they won one token, whereas they could have won 10 by choosing the other box. The majority of the 7 and 9 year-olds expressed regret in this situation. The next day, the children were faced with the same choice. The 7 and 9 year-olds who felt regret then chose the box that had the most tokens the day before. Thus, the children would change their choice after experiencing regret. The association between the experience of regret and adaptive choice switching is confirmed in two similar experiments (see experiments 2 and 3 (O’Connor et al. 2014)).

Furthermore, in some decision-making situations in everyday life, it may be relevant to resist immediate rewards in order to wait for a delayed but more beneficial reward for the individual (e.g. resisting the immediate euphoria caused by drunk driving to preserve one’s own health, that of one’s passengers, and that of potential pedestrians). Regret is thought to play a role in the ability to delay gratification, that is resist a small immediate reward in order to choose a larger delayed reward (McCormack et al. 2019). To study this link, children aged 6–7 years were confronted with two boxes: one that unlocked after 30 seconds and another that unlocked after 10 minutes. The children had to choose to open one of the boxes and were not informed that the reward associated with the late opening box was greater than the reward for the early opening one (four vs. two candies). Unaware of this fact, the majority of children chose to open the box that unlocked faster, only to learn that it contained a smaller reward than the second box. The next day, the children performed the same task. It was found that children who felt regret after opening the box containing an immediate small reward were more likely to change their choice the next day, to wait for the box containing a larger, albeit delayed, reward to unlock. The results of these studies indicate that the feeling of regret allows us to learn from our past mistakes, in order to adapt our choices to a similar situation in the future, with a view to improving our decisions and the benefits associated with them. This would be possible from the age of 7 years, with an improvement of this capacity during childhood and adolescence.

Current research on regret indicates that this emotion can be felt retrospectively (after the choice has been made) and can also be anticipated. Indeed, it is an emotion that one may feel before making a choice, for fear of regretting it. Anticipatory regret can be defined as a prediction made about the emotion one will feel when imagining that an alternative choice might lead to a better outcome (Mellers et al. 1999). To date, two studies have sought to determine the age at which children can anticipate regret, using a choice between boxes procedure (Guttentag and Ferrell 2004; McCormack and Feeney 2015). McCormack and Feeney’s (2015) study replicated Guttentag and Ferrell’s (2004) methodology, while overcoming some methodological biases. Children were told that they would play a game in which they could earn tokens that they would exchange for stickers of their choice. After selecting one of the boxes, the experimenter showed them the result (a token). They then reported their emotional feelings. They were then asked “how they thought they would feel if there were more chips in the unselected box” (anticipatory regret) or “if there were no chips in the unselected box” (anticipatory worry). The results indicate that children are able to anticipate the regret they would feel from the age of 8 years onward.

According to studies conducted in developmental psychology to date, the feeling of regret (but not its anticipation) can lead to emotional learning before the age of 8 years. This emotional learning consists of learning from the emotions felt as a result of our past mistakes to improve our future choices. After the age of 8 years, we can expect that the anticipation of regret will also contribute to this emotional learning, as it does in adults. However, this point remains to be explored further upstream.

The analysis of regret is also informative in order to better understand adolescents’ specific behaviors. Indeed, anticipatory regret could induce an emotionally motivated learning process in adolescents, leading to a decrease in risky behaviors: adolescents who are better able to anticipate the negative emotions associated with the consequences of their actions are more likely to avoid risky behaviors (Galvan et al. 2007). In addition, encouraging adolescents to anticipate the regrets associated with risky behavior contributes to significantly reducing engagement in this type of behavior (Richard et al. 1996; Connor et al. 2006). Finally, the more adolescents are motivated to avoid negative emotional consequences resulting from unfavorable future outcomes, the more likely they are to avoid risky behaviors such as drug use (Caffray and Schneider 2000), alcohol use (Richard et al. 1996) or unsafe sexual behavior (van der Pligt and Richard 1994; Richard et al. 1996).

5.3.2.3. The role of the feeling of regret and its anticipation in learning during decision-making in adults

Camille et al. (2004) provide evidence for the role of regret in the ability to choose advantageously in adults, in order to minimize the experience of regret. In a wheel of fortune task, participants experiencing regret learned to choose the most advantageous gamble as the task progressed, leading to greater gains at the end of the task (Camille et al. 2004). This was not the case for patients with orbitofrontal cortex lesions, for whom the emotional feeling was not modulated by the counterfactual outcome, and who therefore did not feel regret. These participants were less successful at choosing advantageously. Thus, it would seem that a choice made at a point in time influences both our emotional state in the moment and our subsequent choices (Coricelli and Rustichini 2010).

Regret avoidance also influences individuals’ risk-taking. Zeelenberg et al. (1996) confronted participants with a choice between a risky gamble (with a 35% probability of winning) and an advantageous gamble (with a 65% probability of winning). The type of feedback given to participants varied across conditions. In the first condition, participants always receive feedback on the risky gamble, but receive feedback on the advantageous gamble only if they chose it. Thus, choosing the risky option leads to an absence of counterfactual feedback, because participants do not get feedback on the advantageous option they did not choose. In this case, participants would more often choose the risky option. In another condition, participants receive systematic feedback on the advantageous option, but only receive feedback on the risky option if they choose it. This time, choosing the advantageous option leads to an absence of counterfactual feedback. In this condition, participants were more likely to choose the advantageous gamble.

These studies show that individuals tend to avoid choices that lead to counterfactual feedback, which can lead to regret. They illustrate that anticipating regret influences our choices to the point of leading us to risk-taking or risk-averse behaviors in order to avoid the experience of regret (Zeelenberg and Beattie 1997). The role of regret experience in subsequent decisions has also been studied using a decision-making task involving the opening of doors to win treasure by accumulating gold coins (Brassen et al. 2012). Eight doors were presented to the participant, which they could open sequentially to earn gold coins. The risk-taking lies in the fact that a devil was hiding behind one of these doors. The discovery of the devil led to the loss of the treasure. The more doors the player opened, the greater the risk of running into the devil and losing the treasure. If the participant stopped before encountering the devil, they pocketed the accumulated gold coins and were informed of the missed opportunity (the number of additional doors they could have opened). The results indicate that the degree of missed opportunity predicts subsequent risk-taking behavior: the higher the degree of missed opportunity, the more risk the participant takes on the next trial. However, the authors did not directly study the regret felt by the participants, nor did they study its relationship with changes in subsequent choices. This series of studies highlights that the desire to minimize the feeling of regret can sometimes lead to a reduction in risk-taking and sometimes to an increase in risk-taking. However, these studies indicate that decision-making can be adapted as the task progresses, when it achieves the expected goal of maximizing gains.

One may ask in what precise way regret plays a role in the decision-making process. To answer this question, Bault et al. (2016) explored the attentional pattern of regret through eye-tracking and using a task of choosing between wheels of fortune leading to counterfactual feedback. Combined with behavioral measures, the eye-tracking technique assumed that visual attention plays an active role in shaping emotional experiences and decisions (Bault et al. 2008). It allows for the study of participants’ visual fixation times before the choice and after the full feedback (outcome of the unchosen gamble), which enables the identification of the parameters that are involved in the decision, and whether participants make a counterfactual comparison. The results indicate that participants primarily compared the expected values of the wheels of fortune presented on the screen in order to make their choices: they looked at all the information on one of the gambles and then moved on to the second gamble. However, when the difference between the expected values of the two gambles was small, participants made more comparisons between the two gambles. They then chose the one with minimal regret and the lowest expected value (i.e. the gamble with the lowest loss, regardless of the associated gain). Following the counterfactual feedback (in which the outcomes of both gambles were provided), participants spent considerable time reviewing the outcome of the unchosen gamble and ignored the unrealized outcome of the chosen gamble. This study provides evidence for an attentional pattern related to regret in adult decision-making. These observations confirm that the counterfactual information (in this case, the outcome of the unchosen gamble) is an informative comparison point and that regret is embedded in the decision-making process. Finally, participants looked more closely at the counterfactual outcome after a loss, confirming the regret theory hypothesis that comparing the outcome of a decision with the outcome of the alternatives is of major value after a loss. Thus, this study indicates that regret modulates the attentional process during a decision. Regret anticipation could be an effective tool for reconsidering a decision to act or for resisting impulsive actions and immediate rewards.

Research on the role of regret in decision-making emphasizes that a “hot” process, namely, an emotion, can help improve the decision-making process. Its feeling and its anticipation develop according to different patterns. However, they contribute to modifying our choices from childhood, through the feeling of regret, and then during adolescence, through its feeling and anticipation. We will see later in this chapter that other socio-emotional factors come into play in learning in a decision-making situation.

5.4.1.1. Steinberg’s model: the socio-emotional context as a decisive factor of adolescent risk-taking

Adolescents engage in daily risky behavior primarily when they are with their peers and rarely when they are alone. Research on everyday risk-taking shows, for example, that young drivers are at greater risk of serious crashes when accompanied by a peer passenger (Simons-Morton et al. 2005, 2011; Rhodes et al. 2015). Also, risky peer behaviors appear to be an excellent predictor of adolescents’ engagement in risky behaviors (Maxwell 2002; Varela and Pritchard 2011), including alcohol use (Beal et al. 2001; Jackson et al. 2014), unsafe driving (Carter et al. 2014) or engagement in deviant behaviors (Dahl and van Zalk 2014).

Steinberg’s model (Steinberg 2007, 2008) suggests that the absence of risk-taking, often observed in adolescents in laboratory situations, is explained by the emotional neutrality of this context compared to everyday situations involving the salience of socio-emotional factors. The exacerbated risk-taking of adolescents is specific to a salient socio-emotional context, such as the presence of peers, via an increase in sensitivity to the emotional cues of the situation. This model is in line with the hypothesis of emotional hypersensitivity in adolescents associated with the immaturity of regulatory processes (Casey et al. 2008), but it integrates social context as a modulator of this emotional sensitivity.

To validate this hypothesis, Gardner and Steinberg (2005) designed an original experimental procedure that enabled the simulation of a peer socio-emotional context in the laboratory and was later replicated in fMRI (Chein et al. 2011). In this study, adolescents aged 14–18 years, young adults aged 19–22 years and adults aged 24–29 years were asked to complete, alone or in the presence of two friends of the same sex and age, a risk-taking task on a driving simulator, the Stoplight task (Figure 5.7). The participants’ goal was to drive along a road with 20 intersections as quickly as possible. At each intersection, the participant was confronted with a traffic signal, which could be yellow. They could then choose to stop for 3 seconds (safe choice) or to pass to avoid the delay (risky choice). This second choice could randomly result in a positive outcome (i.e. passing without time loss) or a negative outcome (i.e. an accident resulting in an additional 6-second delay). When performing the task individually, adolescents showed the same risk-taking as adults. In contrast, the number of risky choices increased massively in the presence of peers, especially among adolescents. The behavioral results of this study thus highlight a sensitivity to the presence of peers specific to adolescents. Also, the fMRI results support that the increase in risk-taking specific to the adolescent group in the presence of peers is accompanied by an increase in the activation of emotional regions (ventral striatum and orbitofrontal cortex). Activation of cognitive control regions (lateral prefrontal cortex) is characterized by a progressive increase in the recruitment of these regions with age, but is not modulated by the socio-emotional context.

These studies raise the question of the special status of peers for adolescents. Does the increase in risk-taking with the presence of an observer depend on the identity of this observer? A recently published study confirms the specific impact of the peer group in the increase of risky behavior (Telzer et al. 2015), using a similar procedure, with one exception: the presence of the adolescent’s mother was substituted for the presence of peers. In contrast to the previous results, the presence of the mother led to a decrease in adolescent risk-taking, accompanied by a reversed modulation of brain activations compared to the presence of peers: the mother’s presence was accompanied by an increase in the activation of control regions during safe choices (ventrolateral prefrontal cortex) and a decrease in the activation of emotional regions (ventral striatum and amygdala) following a rewarded risk-taking.

It is clear that the work on the presence of peers using the Stoplight task is situated within the framework of risk-taking in an ambiguous situation, thus confirming the impact of the socio-emotional context on adolescents’ decision-making when they do not have the necessary information to assess the level of risk. But what about situations in which the adolescent has this information? A study explored this question (Smith et al. 2014a) and showed that the fictitious presence of an anonymous peer is sufficient to increase risk-taking by 15–17 year-olds when the stakes and probabilities associated with each option are available.

5.4.1.2. From presence to peer influence

Think back to your first cigarette. It is likely that you were not alone but with friends, after school or perhaps at a party. But what role did your friends play in that moment? Was it their mere presence that caused you to light up that cigarette or was it their prompting? In most everyday situations, peers are not simply present but advise, encourage or relate their own experiences about risky behavior. Adolescents may observe a friend with a bottle of hard liquor and encourage him with a challenge or a drinking game. While the research presented above attests to the impact of the mere presence of peers on risk-taking, it is important to note that opacity regarding the possibility of peer intervention exists in some of these studies.

In the context of an ambiguous risk-taking task (the BART), Reynolds et al.’s (2013) study highlights the critical role of risk elicitation as a social-emotional context. One week after completing the BART in a neutral context, the 18- to 20-year-old participants were assigned to three experimental conditions: a control condition, a physical peer presence condition and an influence condition. In the presence condition, peers (two close friends of the same age and gender) were encouraged to observe their peer’s responses without providing verbal or non-verbal feedback. In the influence condition, peers were instead asked to encourage their peer to blow up the balloons as much as possible and were informed that their reward would increase with the participant’s risk-taking. However, one limitation of this study is that although this procedure ensured that peers were motivated to provide risk incentives, it was difficult to control the frequency of intervention and the investment and persuasiveness of the peers. The increase in risk-taking between the two sessions was much greater in the influence condition than in the control and peer conditions, which did not differ from each other. Thus, this study confirms the impact of peer influence on decision-making and suggests that the presence of peers alone is not sufficient to increase risk-taking in late adolescence.

Beyond this distinction between peer presence and peer influence, different types of interventions need to be considered. Is the impact of simple encouragement by a peer on an adolescent’s tendency to drink heavily really comparable to the impact of recounting one’s own consumption or of inducing a drinking game, in which the largest amount of alcohol ingested will determine the winner? Other types of social influence have therefore been introduced into the literature.

For example, Habib et al. (2015) focused on the development of the feeling of regret and relief in a competitive socio-emotional context. Faced with a competitor losing more money than them, adolescents felt relieved more intensely than children and adults, whereas they did not show such a specific pattern in a neutral context (Habib et al. 2012). Regarding negative emotions, the authors highlight a decrease in the feeling of regret in a competitive context, which they attribute to a down-regulation of this emotion, in response to adolescents’ difficulty in tolerating losses in a competitive context. This modulation of emotional feeling (Habib et al. 2015), and also of the neural response of regions involved in evaluating a reward in the face of an opponent (Fareri and Delgado 2014), suggests that competition constitutes a socio-emotional context that may impact risk-taking during development.

5.4.1.3. The possibility of a positive impact on decision-making from peer influence

To date, the majority of studies have focused on the negative role of the socio-emotional context. However, if the influence of peers can push adolescents to take risks, it is possible that peers can help reduce risk-taking and carry a preventive message in everyday situations.

Imagine yourself in the following situation: you are a student and volunteer to participate in an experiment conducted by researchers at your university. As soon as you arrive at the research laboratory, the person in charge of the study informs you and another participant that the experiment will take place in a building at the other end of the campus. A car will take you there in less than 5 minutes. Now imagine that when the car arrives, there are obvious signs that the driver is obviously drunk: the car has stopped, the music is very loud, the driver is having difficulty speaking and there are empty beer bottles on the floor of the car. How willing would you be to get into the car? Using this experimental procedure, Powell and Drucker (1997) were able to demonstrate the role of peer positioning in the tendency to get into a car with a drunk driver. Surprisingly and disturbingly, 100% of the participants agreed to ride with a drunk driver when they were alone or when the second participant (i.e. the experimenter’s accomplice) agreed to ride. Participants refuse to comply only when the accomplice refuses to get into the car. Although this alarming finding must be qualified in light of the considerable progress made over the past 15 years in the perception of the risks associated with drinking and driving, this study illustrates the possibility of a positive influence of the position of a peer.

Only a few recent studies have looked at the positive role of peers, but they lead to contradictory results. By applying the experimental method in the context of daily risky behavior, two studies were able to show that adolescents confronted with an anti-alcohol norm judged more negatively and identified less with regular consumers than adolescents confronted with a pro-alcohol norm. The introduction of an anti-alcohol norm, conveyed by popular peers, led to a decrease in adolescents’ willingness to drink (Teunissen et al. 2012, 2013). Furthermore, Knoll et al. (2015) evaluated the impact of social influence on risk perception during development. More than 500 participants divided into five age groups (pre-adolescents, young adolescents, adolescents, young adults and adults) were asked to rate the level of risk associated with different scenarios involving potential danger (crossing the street while texting, riding a bike without a helmet or climbing on a roof) at two time points. Between their evaluations, participants received the result of the evaluation of the scenario by a group of peers or by a group of adults in the influence condition, and their own evaluation in the control condition. The authors’ conclusion suggests the existence of an influence, both positive and negative, of the peer evaluation on the perception of the level of risk. Indeed, the initial risk perception was re-evaluated in the direction of that given by the peers whatever the age, and this effect decreased linearly over the course of development. Furthermore, young adolescents (12–14 years) were more influenced by the responses given by other adolescents than by those of adults, while this effect was reversed in the two adult groups.

However, other studies fail to show a positive impact of peers on adolescents. Haddad et al. (2014) proposed a risk-taking task to adolescents aged 11–18 years and adults in which they had to choose between a risky option (associated with a greater gain but a lower probability) and a safer option (associated with a moderate gain but a higher probability), according to four experimental conditions manipulated within subjects: the absence of a socio-emotional context, observation, the receipt of risky advice and the receipt of prudent advice from three unknown peers of the same age and sex, represented by a photograph. The risky influence of peers led all participants to choose the risky option more often, whereas the prudent advice seemed to have a differentiated effect according to age: only adults followed the advice of their peers conveying an incentive to be prudent. According to the authors, this lack of impact of a precautionary message among adolescents was based on the specificity of their perception of social interactions. Perceiving their peers’ advice as motivated by a competitive rather than a collaborative strategy, adolescents were less inclined to follow their cautionary advice, whose real objective could have been to minimize their performance.

5.4.2.1. Presence of peers and sensitivity to rewards

According to Laurence Steinberg and her team, the presence of peers could increase adolescents’ sensitivity to immediate rewards, to the detriment of consideration of potentially negative long-term consequences. Take the example of dangerous driving: the presence of peers in the car would pre-activate the reward system and encourage a search for pleasure and strong sensations associated with speed (immediate rewards) without considering the risks incurred regarding a possible withdrawal of license and serious or even fatal accident.

Several behavioral and brain arguments support this hypothesis. In Gardner and Steinberg’s (2005) study, adolescents and young adults were asked to evaluate the cost-benefit ratio of hypothetical scenarios involving risky behaviors (unprotected sex, being a passenger of a drunk driver, testing a new drug without any information about it, etc.), individually or in the presence of two friends. The results support the hypothesis of an impact of the socio-emotional context on sensitivity to rewards, since adolescents attributed more weight to the benefits of risky behavior when they completed this scale in the presence of peers. To confirm this hypothesis, O’Brien et al. (2011) asked young adults aged 18–20 years to complete a delayed reward task, the Delay Discounting Task, either alone or in the presence of two friends, which involved choosing between an immediate reward and a larger reward delayed in time (e.g. $600 immediately or $1,000 in 6 months). Adolescents showed a greater preference for immediate rewards in the presence of peers than when performing the task individually; and simply manipulating the presence of peers caused 18- to 20-year-old participants to show a preference for immediate rewards similar to that of 14–15-year-olds in a neutral context in a previous study (Steinberg et al. 2009).

Finally, the hypothesis that the reward system is involved in increased risk-taking in the presence of peers is supported by neuroimaging work that reveals increased striatum activation in the presence of peers, specific to adolescents (Smith et al. 2014b). Twenty adolescents aged 14–19 years and 20 adults completed the High/Low Card Guessing task in fMRI, a guessing game, alone or in the presence of two friends observing their choices in an adjacent room. After receiving a clue about the size of a potential reward, participants were asked to guess whether the number on the back of the card presented to them was less than or greater than the number 5. Positive or negative feedback was then provided to indicate the outcome of the bet and conditioned on receiving the reward. During receipt of the reward, adolescents showed greater activation of the nucleus accumbens than adults, only in the presence of peers.

5.4.2.2. Peer influence as an opportunity for social gratification

Behavioral and neuroimaging data converge on the hypothesis of increased salience of potential rewards in the presence of peers, but do not rule out the involvement of other psychosocial mechanisms in modulating risk-taking in salient social-emotional contexts. Social context may increase the salience of the potential implications of an individual’s choice for determining future social status, rather than the salience of immediate rewards per se. As group-valued behaviors vary with age, the presence of peers may have a distinct influence on decision-making, despite a more cross-sectional sensitivity to social-emotional context. In fact, the peak in risk-taking observed in the presence of peers also reflects an evolution with age in the norms conveyed and the behaviors valued by the group, whose risk-taking is a sign of the adolescent period. The popularity of adolescents seems to be positively related to their involvement in risk-taking, such as drug use or aggressive behavior (Allen et al. 2005).

Cohen and Prinstein’s (2006) study highlights this social gratification issue in peer influence during adolescence. According to the authors, imitating popular peers represents an opportunity to maximize one’s own social status. To study this hypothesis, the authors tested the existence of a “social contagion effect” on the aggressive behaviors of 16 to 17 year-old adolescents according to the popularity of their peers. Confronted with a set of fictitious scenarios involving potentially risky and aggressive behavior, the participants were asked to indicate their own reaction to this type of situation. To answer this question, adolescents participated in an online discussion with three popular peers, during which the participant was the last to give their response. For most scenarios, the opinions conveyed by the peers consisted of a set of aggressive responses. Interestingly, peer status appeared to be a key determinant of the tendency of adolescents to follow peer advice. When confronted with the advice of popular peers, adolescents chose more risky or aggressive responses under social pressure than in control scenarios. In contrast, adolescents favored cautious, non-aggressive responses when confronted with social pressure from unpopular peers who themselves exhibited aggressive behavior. This paradox between a tendency to imitate popular peers and a behavioral distancing from unpopular peers seems to confirm the importance of the search for social value as a motivator of the social influence phenomenon in adolescence.