Journal of Gambling Studies

, Volume 28, Issue 3, pp 393–403

The Effects of Sleep Debt on Risk Perception, Risk Attraction and Betting Behavior During a Blackjack Style Gambling Task

Authors

    • Department of PsychologyLondon South Bank University
Original Paper

DOI: 10.1007/s10899-011-9266-9

Cite this article as:
Frings, D. J Gambl Stud (2012) 28: 393. doi:10.1007/s10899-011-9266-9

Abstract

Gamblers often gamble while experiencing fatigue due to sleep deprivation or cumulative sleep debt. Such fatigue has been shown to make decision makers behave more riskily. The present study aimed to test the role of two cognitive processes, risk perception and risk attraction, in this effect. Two hundred and two participants played twelve hands of a black-jack style card game while either fatigued or reasonably alert. Findings showed that both fatigued and alert participants rated higher risk bets as more risky than lower risk bets, suggesting risk perception was unaffected by fatigue. However, fatigued participants did not rate higher risk bets as less attractive than lower risk bets, and reduced the size of their wager to a lesser extent when objective risk increased. These findings are discussed in relation to the effects of fatigue on motivated tasks and the need for gamblers to be aware of the effects of fatigue.

Keywords

GamblingFatigueRiskSleepBlackjackBetting

Introduction

Fatigue can be described a physiological and/or a mental state which impairs cognitive performance (see Noy et al. 2011; Rabinbach 1990). Factors contributing to cognitive or general mental fatigue include high mental workload, sustained work effort, circadian time of day effects, and especially sleep deficit and night work (Krueger 1989). The present study focuses upon fatigue due to sleep debt (cumulative sleep loss over a number of nights) sufficient to impair performance (Frings 2011; Van der Linden et al. 2003). Specifically, it investigates the effects of fatigue upon gamblers’ ability to assess risk, the degree to which risk is perceived as attractive and subsequent risky behaviors in a gambling context.

Gamblers are likely to gamble while fatigued for a number of reasons. Gambling is often a sustained activity, with casinos operating 24 h a day, 7 days a week. Gamblers, particularly problem gamblers, often play continuously for long sessions over several days (McBride and Derevensky 2009). During this period they often experience insomnia or sleep debt (Rosenthal and Lesieur 1992). Additionally, gamblers playing late at night will often be at a point in their circadian rhythms which makes them susceptible to fatigue (see Czeisler et al. 1980).

Fatigue and Risk

Relative to simpler tasks (such as vigilance and alertness) little evidence has looked at the effects of fatigue on ecologically valid/complex decision making tasks (Harrison and Horn 2000). However, fatigue has been shown to influence risk taking behavior. For instance, Killgore et al. (2006) had participants who had undergone 49.5 h continual wakefulness complete the Iowa Gambling Task. Participants were presented with multiple decks of cards, some of which were associated with winning over time, others with losses over time. This task measures participants’ ability to integrate feedback from successes into future decision making. Over time, reasonably alert participants altered their behavior and chose more frequently from decks associated with gains. In contrast, fatigued participants were unable to learn from experience and did not adjust their behavior. Venkatraman et al. (2007) has shown that, when presented with pairs of odds, fatigued participants (24 h continual wakefulness) had increased patterns of neural activation in the nucleus accumbens (associated with risk seeking behavior) and decreased activation in areas associated with the anterior insula (linked with processing losses). In summary, it appears that sleep deprived decision makers are more sensitive to gains, less sensitive to losses, and unable to learn from prior experience.

Existing studies on fatigue and risk use relatively acute levels of sleep deprivation, beyond those typically experienced by gamblers. The present study addresses this issue by looking at participants experiencing less acute sleep deprivation, incurred through a sleep debt rather than continual wakefulness. It also tests the role of cognitive factors which may mediate the fatigue-behavior link. Two fundamental cognitive processes, risk perception and risk attraction, are investigated.

Risk Perception

Risky decisions and behavior do not occur spontaneously, but are the result of various cognitive processes. Engagement in risky behavior is guided by perceptions of the degree of risk entailed. A gambling behavior with a set positive outcome is less likely to be engaged in the greater or more probable an associated loss is perceived to be. A key component in risk perception is the ability to differentiate between relative levels of risk and potential outcomes (e.g. be able to identify when one option carries more or less risk than an alternative). Risk perception has been shown to predict behaviors in a variety of contexts including drug use, financial decisions, health and safety choices and recreation (Byrnes et al. 1999; Weber et al. 2002).

Several factors affect how risky a given decision is perceived to be. For instance, motivational states such as challenge (a feeling that resources outweigh demands, e.g. Blascovich 2008) and promotion focus (a focus on maximizing gains rather than avoiding losses, see Higgins 1997) can reduce perceived risk. Similarly, it could be argued that observed effects of fatigue on risk taking may be due to changes in risk perception.

Risk Attraction

A common limitation in the fatigue literature is that risk perception and behavior are treated synonymously. In contrast, the wider risk behavior literature recognizes that the link between perception and behavior is not clear cut (Lichtenstein and Slovic 1971). When a behavior is undertaken, there is no way to tell if participants perceive the action as less risky, or if they do perceive increased risk but ignore this information (Tversky 1969). This presents alternative explanations for existing findings. For instance, Killgore, et als’ fatigued participants may have learnt that the advantageous decks were less risky, but not chosen those decks for other reasons. Alternatively they may have not perceived the less advantageous deck as entailing more risk, thus being unable to take this information into account when choosing how to behave.

One way to disentangle these issues is to not only consider risk perception, but also risk attraction. Risk attraction is the extent to which a risky decision entailing potential gains and losses is seen as aversive or attractive. Examining risk attraction independently of risk perception adds detail to existing explanations. The present study tests the novel hypothesis that when sleep deprived, fatigued gamblers will take more risks not simply because they do not perceive situations as risky, but also because they are more tolerant of risk.

There is considerable individual variation in the extent to which risk affects preparedness to undertake a behavior. Some individuals are prone to perceive recognized risks as attractive, while others find the same perceived level of risk as being aversive (Li and Liu 2008). More importantly, contextual factors have been shown to change how decision makers evaluate risk. These include framing situations in terms of losses or gains arising for behavior, or variation in resources available to cope with outcomes (e.g. Tversky and Kahneman 1974). Finally, decision makers can experience unrealistic optimism, the belief that one’s personal odds of achieving (or avoiding) an outcome are better than those of the population at large (e.g. Babad and Katz 1991; Weinstein 1980; cf. Harris and Hahn 2011).

There are a number of reasons to expect fatigue to affect risky behavior via risk attraction as well as, or instead of, risk perception. Ingestion of alcohol (which has similar cognitive effects to fatigue, see Krueger 1989) can polarize risk tolerance amongst decision makers, making them either more or less risk aversive depending on situational cues (Hopthrow et al. 2007). Also, fatigue still affects behavior during risky decision making even when objective probabilities of winning and losing are explicit, rather than inferred by participants on the basis of previous outcomes (e.g. Venkatraman et al. 2007). Finally, fatigued decision makers respond differentially to gains and losses; becoming risk averse when losses are framed, and risk tolerant when gains are presented (e.g. McKenna et al. 2007).

Considering the role of risk perception and risk attraction as two distinct factors allows the postulation of a number of cognitive mechanisms through which fatigue may affect risk taking behavior, in particular the decision to gamble on a given bet or not. One possibility is that fatigue affects risk perception but not risk attraction. In particular, fatigued decision makers may be unable to differentiate between bets which are higher risk and those which are lower risk. Experimentally, this can be tested by carrying out planned comparisons between higher and lower risk hands, amongst decision makers who are fatigued or reasonably alert. If these comparisons indicate that risk perception appears unaffected by fatigue, it is possible that more risky behavior is due solely to increased risk attraction. In this scenario, individuals may perceive that a particular bet is more risky, but this information fails to be factored into the decision as to whether to engage in betting behavior. Experimentally, this would be indicated by an increase in risk attraction amongst fatigued individuals, but no effect of fatigue upon perception. Specifically, it would be expected that both fatigued and alert individuals would have perceptions of risk which vary in line with objective probabilities. If this perception does not translate into changes in behavior, fatigued individuals would perceived both higher and lower risk bets as equally attractive, and bet equal amounts on them. In contrast, reasonably alert individuals would be expected to find higher risk bets less attractive than higher risk bets, and bet less upon them.

A final possibility is that both risk perception and attraction are affected. In this instance, fatigued decision makers should not change their perception of risk in line with objective probabilities, and should find all bets, regardless of perceived risk, equally attractive. In this instance, an effect of fatigue upon both risk perception and attraction would be expected amongst fatigued decision makers but not reasonably alert decision makers, with fatigued individuals not differentiating the risk associated with higher and lower risk bets. In addition, increased risk attractiveness would result in both higher and lower risk bets being perceived as more attractive.

To test between these three possibilities in the context of card gambling, participants who were either reasonably alert or fatigued were asked to rate a number of bets for risk and attractiveness, and then subsequently state how much they would like to wager on each bet. By testing planned comparisons (responses to higher vs. lower risk bets amongst fatigued or alert individuals) the relative effects of risk perception and attraction were assessed.

Method

Participants

Two hundred and two Army Officer Cadets were recruited from the University of London Officer Training Corps. Ages ranged from 18 to 24. Eighty-six percent of the sample was male. Participants were randomly allocated to fatigue condition. One hundred and four participants took part in the reasonably alert condition, 98 in the fatigued condition.

Design

A two-way design was used with a between subjects factor (Fatigue: Reasonably alert vs. fatigued) and a within subjects factor (Risk of Bet: Higher risk, lower risk). Dependent variables were risk perception, risk attraction and amount bet. Self-reported levels of fatigue were measured as a manipulation check.

Materials

Fatigue

Eight items from the Piper Fatigue Scale (Piper et al. 1998) were selected to rate the level of fatigue. This scale was selected because it measures fatigues across a number of domains. Specifically, it measures perceived degree of fatigue, fatigue related affect and also perceived levels of cognitive impairment attributed by the participant to fatigue (rather than to other contextual factors). Three items asked participants: ‘How would you describe the intensity of severity of the fatigue which you are experiencing now’, ‘To what degree is the fatigue you are now feeling interfering with your ability to complete tasks set for you?’ and ‘To what degree would you describe the fatigue you are experiencing now as being?’. All items were measured on seven point scales. The first two items were anchored a 1 (None) and 7 (A great deal). The latter item was anchored at 1 (Mild) to 7 (Severe). A further five items measured ‘To what extent are you now feeling’ followed by seven point scales anchored with the following oppositional pairs; Exhilarated/Depressed, Able to concentrate/Unable to concentrate, Strong/Weak, Able to remember/Unable to remember, Able to think clearly/Unable to think clearly. Internal reliability of all eight items was good (Cronbach’s α = 0.90) and a composite mean score was calculated. Higher scores indicate increased subjective feelings of fatigue.

Bets

Participants were presented with a betting task in which they were shown two cards they had been dealt and one the ‘dealer’ had been dealt. They were told the dealer would be dealt another card. The rules were similar to blackjack but simplified. If the total pip value of the dealer’s two cards (the one the participant could see, and the one the dealer would be given) was equal or higher to that of the players two cards, the player would lose whatever they chose to bet. Should the player’s pip value be higher, the player would receive their bet back, and win an additional equal amount. As well as indicating the amount they would bet, participants also rated how risky and attractive they perceived each hand they bet on to be (see below). Participants played 12 hands. The probability of the player winning each hand varied between 0.20 and 0.65. Hands were categorized as higher risk (less than 0.50 probability of player winning, n = 4) or lower risk (greater than 0.50 probability of player winning, n = 8). In the lower risk condition probabilities of winning (each individual hand) were; 0.65, 0.59, 0.59 and 0.53. In the higher risk condition; 0.49, 0.47, 0.47, 0.41, 0.35, 0.24, 0.20 and 0.20. Before the start of the study, trials were ordered randomly once. All participants then received the trials in this order.

Amount Bet

For each bet, participants were asked to indicate an amount between nothing and £5.00 that they would like to wager on the winning the hand. The mean amount bet for high and low risk bets were calculated separately (Cronbach’s αs > 0.68). If participants placed a bet, they were asked to rate the perceived risk and attractiveness associated with their bet.

Risk Perception

How risky each bet was perceived to be was measured using a single Likert type scale item for each bet. The item read ‘How risky was this hand?’. The scale was anchored at 1 (Not at all risky) and 7 (Very risky).

Risk Attractiveness

Risk attractiveness was measured using one item for each bet on a Likert type scale anchored at 1 (Not at all attractive) and 7 (Very attractive). The item was ‘How attractive was this hand?’.

Procedure

The study took place in two waves over the course of two winter training exercises in England. At the start of each exercise, all participants arrived in the exercise area on the evening of day zero of the exercise. They had not had any special interventions to their sleep patterns during the previous week, and had been told to be well rested before the exercise by their officers. It was made clear by the experimenter and officers present that participation in the study was voluntary, and there would be no penalty for nonparticipation, or for subsequently withdrawing. The task was undertaken in study rooms.

Participants in the reasonably alert condition completed the study during the first wave of the data collection, specifically upon arrival in the training exercise area on evening of the day before the exercise commenced (around 7 pm). Participants in the fatigued condition completed the task around 10:00–11:00 am 2 days later (1st wave of data collection) or 4 days later (2nd wave data collection). Differences in fatigue levels experienced by participants across the two waves of data collection exercises did not significantly affect the pattern of results or levels of significance when included in subsequent analysis. Between arrival and testing, as part of their scheduled training, participants in the fatigue condition were subjected to intensive teaching, vigorous exercise, and systematically disturbed sleep (sleep period duration of 5 h or less per night, and being woken up during this period for an hour of watch duty each night), all in cold, rainy, outdoor conditions. To minimize the possible effects of physiological arousal or short term motivational changes due to the training exercise (e.g. increased adrenaline or aggression) participants were given at least 1 h to relax before taking part in the study.

Upon arrival at the testing facility all participants gave consent and completed the fatigue scale. They then completed the gambling task (as outlined above) as part of a test battery. Once the tasks were completed participants were dismissed into the care of their officers. Debriefing followed after all participants had had an opportunity to nap.

Results

Manipulation Check

Seven participants failed to complete this measure. To ensure participants in the fatigued condition felt subjectively more fatigued than participants in the reasonably alert condition, a t-test was conducted upon the fatigue scale. Fatigued participants reported higher levels of fatigue (M = 4.86, SD = 1.13) than reasonably alert participants (M = 3.74, SD = 1.05), t(193) = 7.16, P < .001.

Main Analysis

Risk Perception

An ANOVA was conducted with risk perception as the dependent variable, with Fatigue condition (Fatigued vs. Reasonably Alert) as a between subjects factor and Risk of Bet (Higher Risk, Lower Risk) as a within subjects factor. A main effect of Risk of Bet was present, F(1, 198) = 33.59, P < .001, η2 = 0.15. Overall, participants perceived higher risk bets as more risky (M = 4.51, SD = 0.84) than lower risk bets (M = 3.97, SD = 1.18). There was no main effect of Fatigue, F(1, 198) < 1.00, P = .42, η2 < 0.001. The interaction between Fatigue condition and Risk of Bet was not significant, F(1, 198) < 1.00, P = .64, η2 < 0.001. Planned comparisons using simple effects revealed that both fatigued and reasonably alert participants perceived higher risk bets as being more risky, Ps < .001. In summary, both fatigued and reasonably alert participants recognized that higher risk hands were more risky than lower risk hands. The fatigue manipulation did not affect risk perception.

Risk Attraction

An ANOVA was conducted with risk attraction as the dependent variable with independent variables as above. Means for this ANOVA can be seen in Table 1. A main effect of Risk of Bet was observed, F(1, 199) = 10.37, P = .001, η2 = 0.05. Higher risk bets were seen as less attractive than lower risky bets. There was no main effect of Fatigue, F(1, 199) < 1.00, P = .462, η2 < 0.01. The interaction between Risk of Bet and Fatigue approached significance, F(1, 199) = 2.70, P = .10, η2 = 0.01. Planned comparisons between higher and lower risk bets were conducted by examining the simple effects of risk of bet in the reasonably alert and fatigued conditions. Perceived attraction of higher versus lower risk bets differed in the reasonably alert condition, F(1, 199) = 12.14, P < .001, η2 = 0.06, but not in the fatigued condition, F(1, 199) = 1.21, P = .27, η2 < 0.001. In summary, participants who were reasonably alert rated the attractiveness of bets in line with objective probabilities, finding bets they were more likely to win more attractive. In contrast, although there was a small trend towards fatigued participants rating lower risk bets as more attractive, this effect was smaller and ultimately statistically non-significant. This suggests that fatigued participants did not differentiate attraction towards higher or lower risk bets to the same degree as alert participants.
Table 1

Mean perceived attraction of lower and higher risk hands by fatigued and reasonably alert participants (Standard deviations in parentheses)

 

Lower risk hands

Higher risk hands

Reasonably alert

3.95 (1.00)

3.59 (0.77)

Fatigued

3.90 (0.80)

3.78 (1.01)

Amount Bet

An ANOVA was conducted with the amount bet as the dependent variable and independent variables as above. Means for this ANOVA can be seen in Fig. 1. A main effect of Risk of Bet was present, F(1, 200) = 63.14, P < .001, η2 = 0.24. Overall, participants bet less money on higher risk hands than lower risk hands. A main effect of Fatigue was also present, F(1, 200) = 4.53, P = .035, η2 = 0.02. Fatigued participants bet more money than reasonably alert participants. The expected interaction was not significant, F(1, 200) = 1.75, P = .19, η2 = 0.01. Again, planned comparisons using simple effects tested the effects of fatigue on betting behavior during higher and lower risk hands. The amount of money being bet on higher risk hands varied as a result of fatigue F(1, 200) = 7.30, P = .008, η2 = 0.04. However, fatigue did not affect significantly the amount of money bet upon lower risk hands F(1, 200) = 1.10, P = .60, η2 < 0.01. In summary, fatigued participants bet more money than reasonably alert participants. This effect was due to reasonably alert participants altering their behavior and betting less upon higher risk hands, while fatigued participants did not differentiate their behaviors to the same extent.
https://static-content.springer.com/image/art%3A10.1007%2Fs10899-011-9266-9/MediaObjects/10899_2011_9266_Fig1_HTML.gif
Fig. 1

Mean amount bet on lower and higher risk hands by fatigued and reasonably alert participants

Discussion

Existing research has linked fatigue due to sleep deficit with risky behavior. Gamblers are at risk of gambling while fatigued (Rosenthal and Lesieur 1992). The present study investigated how fatigue affected two cognitive factors related to risk taking behavior, risk perception and risk attraction. Existing research suggests that fatigue may affect gambling behavior by affecting either risk perception, risk attraction or both perception and attraction. The present study had participants play a number of hands of a blackjack type card game to test between these three possibilities.

Fatigue affected gambling behavior. Fatigued individuals bet significantly more over the course of twelve hands of cards than reasonably alert individuals. This finding confirms previous research linking fatigue to risk taking (e.g. Killgore et al. 2006; McKenna et al. 2007) but also demonstrates that this effect can be generalized to casino style card gambling. The present findings also extends this literature testing the role of specific deficits caused by fatigue in a gambling context. Closer inspection of betting behavior reveals that increased betting is driven by fatigued individuals not reducing their bets when odds were against them to the same extent as reasonably alert participants did. Importantly, this effect did not seem to be caused wholly by differences in risk perception; both fatigued and reasonably alert participants perception of risk varied in line with objective probabilities. If betting behavior was solely driven by risk perception, it would be expected that fatigued participants would always bet more than alert participants, but would still alter their behavior when presented with higher versus lower risk hands. This was not the case.

A second factor through which fatigue may affect gambling behavior is risk attractiveness (the degree to which a given level of risk is appraised as attractive or aversive). The present study showed that reasonably alert individuals found higher risk bets less attractive than lower risk bets, while fatigued individuals did not differentiate to the same extent—a pattern of results mirroring participants’ actual betting behavior. Taken together, these findings suggest that although fatigued decision makers do perceive changes in levels of risk in their environment, such changes do not significantly affect how attractive they perceive bets to be to the same extent as alert participants, or influence subsequent betting behavior. This distinction between risk perception and attractions provides novel insights into risk taking behavior under conditions of fatigue. It also suggests that future research should assess risk perception, risk attraction and behavior as separate (although possibly related) constructs. Such research should also investigate how participants themselves operationalize ‘risk perception’ and ‘risk attractiveness’. For instance, risk could be perceived as objective probabilities (the operationalization assumed for the present study) or could depend on other factors such as the size of bet or bet size proportionate to the decision maker’s total resources. Likewise, attractiveness could be related to enjoyment of experiencing risk, or only upon the possibility of wins and losses.

As well as providing a detailed explanation of how fatigue affects gambling behavior on a cognitive level, the present research reveals how little sleep debt needs to be accrued for cognitive functioning to be impaired. Previous researchers have suggested that complex or motivationally important tasks are unlikely to be affected by all but severe levels of fatigue (see Harrison and Horn 2000). The present research shows effects of fatigue upon gambling after only a low number of nights of interrupted sleep, adding to a growing body of literature which suggests individual decision makers may be more susceptible to low levels of fatigue than previously thought (e.g. Venkatraman et al. 2007).

As well as these theoretical implications, the present findings also have implications for gambling awareness campaigns. Such campaigns typically focus on implementation action tendency (e.g. ‘know when to stop before you start’) or the effects of alcohol on gambling behavior. While gamblers are likely to be aware of the risk of addiction, overspending and the link between alcohol and risk taking, they may be unaware of the link between gambling behavior and fatigue. The present study suggests that highlighting the increased risk of making poor judgments when fatigued may be as important to responsible gambling vendors as emphasizing the risk of overspend or alcohol.

One alternative explanation for the present results is that fatigue affected participants’ ability to learn the task (simplified blackjack), rather than affect their actual responses to risk. Although possible, the finding that both fatigued and reasonably alert participants had equal perceptions of risk suggests that fatigued participants learnt the game being played sufficiently well to recognize high and low risk hands when they were presented. Additionally, other research with also shows that participants are able to learn implicit rules present in systems when experiencing comparable levels of fatigue, even if they do not respond appropriately to them (e.g. Frings 2011).

The present study had a number of limitations and these provide avenues for further study. The number of hands played by participants in the present study (12) is relatively low compared to other studies, and to typical gambling sessions (e.g. McBride and Derevensky 2009). The effects of fatigue may change over the course of longer session sessions: indeed task fatigue (in contrast to fatigue related to sleep deprivation) has been shown to inhibit the ability to react to new information (Van der Linden et al. 2003). Moreover, during long gambling sessions, participants are likely to become more fatigued over the course of the session. Thus, it is possible that over longer games the effects of fatigue on behavior and risk attraction will become more pronounced. Additionally, it is also possible that, as fatigue increases, risk perception abilities will eventually become impaired. Future studies could address this be extending the number of trials, and examining blocks of trials at different time points.

As well as comprising a relatively low number of hands, prior experience playing blackjack was not controlled for in the present study. However, participants were randomly assigned to condition, and the rules were sufficiently simplified such that any advantage experienced blackjack players may have had over non-experienced participants should have been minimized.

An additional factor which was not controlled was the presence of circadian rhythm effects. Circadian rhythms influence body temperature, alertness and feelings of sleepiness levels independently of sleep deficit, an (see Czeisler et al. 1980). Circadian related sleepiness usually increases from 7 pm, with a trough in alertness observed around 4am (see Lavie 1991). A second lesser trough is experienced between 2 and 5 pm. In the present study, participants in the reasonably alert condition were tested in the evening, after the afternoon trough. In the fatigue condition they were tested mid-morning, when circadian induced sleepiness would least pronounced. Thus, while future research should consider the role of circadian rhythms, it can be argued that results in the present study reflect fatigue effects related to sleep deficit rather than circadian effects.

The present study considered the role of fatigue independently from other factors such as alcohol and caffeine, both commonly found in gambling contexts. Although controlled caffeine use may temporarily improve performance of fatigued decision makers, cessation of use can increase fatigue (see Roehrs and Roth 2008; Addicott and Laurienti 2009). Alcohol consumption has been shown to increase the extremity of risk taking behavior (Abrams et al. 2005; Steele and Josephs 1990). Alcohol also increases subjective levels of daytime sleepiness over time (Roehrs and Roth 2001). Thus, alcohol and fatigue may have additive effects on risk behavior, with the highest level of risk taking occurring amongst fatigued and intoxicated players.

In summary, the present research confirms that fatigue increases risky behavior, specifically the amount gamblers bet over a series of hands. Uniquely, it identified and tested two fundamental processes which may underlie this effect, risk perception and risk attraction. The findings demonstrate that risk perception is unaffected by moderate levels of fatigue. In contrast, fatigued decision makers were shown to be less able to differentiate between high and low risk bets in terms of attractiveness and actual betting behavior.

Acknowledgments

This research was supported by ESRC grant RES-000-22-3460 and complies with BPS and APA ethical guidelines. The author would like to thank the London University Officers Training Corps for assistance given.

Copyright information

© Springer Science+Business Media, LLC 2011