In the field of decision-making in risk situations, one normative theory has dominated the analyses performed for many years: the Expected Utility Theory (von Neumann & Morgenstern, Reference von Neumann and Morgenstern1944). According to this rational theory, the decision rule for risky choices involves maximizing the expected utility of wealth; that is, decision makers evaluate outcomes based on the utility of the final asset positions. Kahneman and Tversky (Reference Kahneman and Tversky1979), convinced of the error of this approach, developed an alternative model, which they called the Prospect Theory, to describe the way human beings make choices and act in risk situations. The prospect theory, in contrast to the expected utility theory, proposes that the value function depends on changes in total wealth, defined as deviations from the reference point (Kahneman, Reference Kahneman2003; Tversky & Kahneman, Reference Tversky and Kahneman1981). The only outcomes taken into account, according to the prospect theory, are those that affect this point of reference, that is, gains, losses or neutral results (Kahneman & Tversky, Reference Kahneman and Tversky1984). Since then, the idea of rational choice, a process that emphasizes the utility of the final states in selecting the best option, has been widely questioned (Kahneman, Reference Kahneman2012; Kahneman & Tversky, Reference Kahneman and Tversky2000).

To test their model, the authors designed a series of well-known experiments in which the participants had to choose between two options (problem 1):

1. a) Winning $300 for sure.

2. b) 80% chance to win $400 and 20% chance to win nothing.

The majority of people (80%) chose option “a”, showing an aversion to risk. This effect was referred to by Kahneman and Tversky (Reference Kahneman and Tversky1979) as the “certainty effect”, which points out that people give too much weight to outcomes that are considered certain, compared to outcomes that are merely probable.

Nonetheless, this occurs not only in gain situations. Reversing the previous alternatives, the following choices appear (problem 2):

1. a) Losing $300 for sure.

2. b) 80% chance to lose $400 and 20% chance to lose nothing.

In this case, 92% of participants chose option “b”, again showing that the certainty outcome is over-weighted. In other words, now participants seek risk to avoid a sure loss. This phenomenon was called the “reflection effect” by the authors, as the preferences described involved a mirror image of the gain situation.

The expected utility theory assumes that people have consistent risk preferences. In contrast, prospect theory clearly shows that this is not true (Frank, Reference Frank2009). According to the expected utility theory, the individuals who avoid risk in problem 1 should again choose the sure option in problem 2; however, the opposite occurs, showing a risk preference reversal pattern. These and other results have led different authors to argue that it is time to recognize that expected utility is an ex-hypothesis (Rabin, Reference Rabin2000).

Later, other studies have pointed out the effect of the emotions in the decision-making process. Thus, motivational needs and emotionally attractive stimuli produce specific effects that cannot be explained by the aforementioned theories. These relationships between emotion and decision have been addressed from different disciplines, such as economics or psychology. The disposition effect observed in stock market investors is explained based on the emotions evoked by different situations related to the decision to buy or sell (Summers & Duxbury, Reference Summers and Duxbury2012). Hsee and Kunreuther (Reference Hsee and Kunreuther2000) demonstrated that affect influences decisions about whether to purchase insurance. Tyszka and Przybyszewski (Reference Tyszka and Przybyszewski2006) concluded that emotional attachment to a currency can cause an increase in the price evaluation of a good by adding an emotion-based value to the purchasing value of the currency. Clinical psychology, on the other hand, has managed to establish new study models for decision-making in patients with altered emotional states like depression or anxiety (Paulus & Yu, Reference Paulus and Yu2012). Tabernero and Wood (Reference Tabernero and Wood2009) showed that subjects in the condition where positive affect was associated with possible errors showed more interest in the task they were doing and better scores, both individually and in groups. Likewise, Regueiro and León (Reference Regueiro and León2003) found that the more a decision was perceived as stressful in the first evaluation process, the greater the negative emotional potential associated with it. Along with these examples, it is also worth mentioning the identification of the neurophysiological structures that evaluate and react to the emotional significance of the stimuli and the effect these structures have on the decision-making process (Gupta, Koscik, Bechara, & Tranel, Reference Gupta, Koscik, Bechara and Tranel2011; Ohira, Reference Ohira2011; Simón, Reference Simón1997).

Over time, different methods have been used to study the majority of the prospect theory hypotheses based on their interactions with the emotions. It has been demonstrated that neither the measurement of the different cognitive systems (intuitive and reasoned), nor the value and weight functions described by Kahneman and Tversky (Reference Kahneman and Tversky1979), fit the response patterns produced in individuals when a certain emotional state is induced (Angie, Connelly, Waples, & Kligyte, Reference Angie, Connelly, Waples and Kligyte2011), or when their affective thinking style varies (Mukherjee, Reference Mukherjee2011).

In addition, Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001) pointed out that weighting the probability of the possible alternatives can alter the normal course of risk decision-making situations, but in this case when stimuli with different affective values are presented. Dividing a sample of 40 participants into two groups of 20, they offered the first group a choice between kissing their favorite movie star or receiving 50 dollars, while the second group had to choose between the same options, but with only a 1% chance of winning. Their results were quite surprising, to the extent that the number of people who chose the money was much greater in the condition of sure gain, and the opposite was true in the low probability condition. According to the authors, these results suggest that in situations of sure gains, affect-poor stimuli are preferred, while in the case of probabilities near zero, affect-rich stimuli are preferred. In the second experiment, one group of participants had to put a price on a coupon for a trip through Europe that could be exchanged for 500 dollars, whereas another group had to put a price on a coupon that could only be exchanged for partial payment of university fees. For each prize, two different conditions were used, a 1% or 99% chance of winning. Again, the results obtained corroborate their predictions, reformulating the weighting function originally proposed by Kahneman and Tversky (Reference Kahneman and Tversky1979).

The main conclusion reached by Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001) helped them to confirm their initial proposal, an idea that has been applied to other contexts (Tyszka & Zaleskiewicz, Reference Tyszka and Zaleskiewicz2012). Based on their data, the independence between the outcomes and the probabilities presented by the classical decision theories or by the prospect theory was not upheld. The probability of the alternatives has a weight (weighting function) determined by the type or nature of the outcomes (value function). That is, the probability and the outcomes cannot be separated when analyzing stimuli with different affective values.

Thus, the emotional load of the stimuli changes the alternative chosen in decision making. However, in the experiments by Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001), the equality of the probabilities was assumed when establishing the different experimental conditions, thus leaving out one of the main characteristics of the experiences designed by Kahneman and Tversky (Reference Kahneman and Tversky1979): the risk we have to face when making decisions. By always using high or low probabilities in both alternatives, the distinct evaluations people can make of the certain or probable options are left out.

Moreover, Dhar and Wertenbroch (Reference Dhar and Wertenbroch2000) introduce a parallel distinction to the contributions of Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001) by differentiating between hedonic goods and utilitarian goods. On the one hand, they define hedonic goods as those whose consumption is primarily characterized by an affective and sensory experience of aesthetic or sensual pleasure, fantasy and fun. On the other, utilitarian goods are those whose consumption is more cognitively driven, instrumental and goal oriented, fulfilling a functional or practical role. In other words, hedonic goods provide consumption that is more experiential, exciting, enjoyable and appealing to the senses, while utilitarian goods are primarily useful, functional and help to achieve a goal. Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001) note that hedonic goods may be thought of as affect-rich, and utilitarian goods as affect-poor. Dhar and Wertenbroch (Reference Dhar and Wertenbroch2000) demonstrate a preference asymmetry in the way people alternate between the hedonic and utilitarian dimensions in acquisition and forfeiture choices. They also show that this asymmetry in preferences can also be expressed in terms of differential loss aversion for hedonic and utilitarian attributes. However, unfortunately, the experiments carried out by these authors focus on riskless choices.

Taking this into account, the main objective of the present article is to complement the prospect theory by Kahneman and Tversky (Reference Kahneman and Tversky1979). More specifically, we intend to find out whether the certainty and reflection effects are fulfilled as predicted in this theory when stimuli with an emotional load are introduced. Based on the contributions made by Rottenstreich and Hsee (Reference Rottenstreich and Hsee2001) and Dhar and Wertenbroch (Reference Dhar and Wertenbroch2000), we will incorporate the importance of the emotive decisions into Kahneman and Tversky’s theory, but this time in risk conditions.

For this purpose, we have used two of the problems described by Kahneman and Tversky (Reference Kahneman and Tversky1979) as a model, in order to ensure that the alternatives meet the criteria for certainty-uncertainty, one in the domain of gains and the other in the domain of losses. Furthermore, we have added the corresponding affect-rich and affect-poor stimuli to the description of these problems. In the end, two situations were obtained, gain and loss, combined with two other conditions, rich and poor affect, resulting in four different experimental conditions.

The following hypotheses are proposed:

Therefore, we expect that, unlike what is predicted in the prospect theory, a greater proportion of affect-rich alternatives will be chosen, regardless of the assigned probability (sure or probable option) or whether the situation is one of gain or loss. We believe this precisely because of the effect of the emotive decisions on the decision-making process. In a situation of gain, the affect-rich alternatives (or hedonic goods) provide a differential value that makes them more experiential and produces more excitement, fantasy and fun, causing them to be preferred by many people, even though there is risk involved in getting them (even rejecting sure options based on the poor emotions they produce). In contrast, in loss situations this process becomes completely inverted. The affect-rich alternative is chosen because the affect-poor option (or utilitarian goods) is more valued. Faced with the mindset of having to lose, it is better to lose something rich in affect (merely influenced by emotional desires in search of pleasure) than something poor in affect but more functional, instrumental, and practical. For this reason, the participants will continue to take risks to avoid losing something useful, but they will be more reluctant to risk (in contrast to what is proposed by the prospect theory) when this risk involves the danger of losing something functional and practical.

Results

Gain situation

On the one hand, according to the data shown in Table 1, the affect-rich risk alternative (b) was chosen more in condition 2. In contrast, in condition 1, the greater percentage is observed in the affect-rich sure alternative (a). The first hypothesis predicts a greater tendency to choose the affect-rich alternatives in the gain situation. Therefore, we compared the data for the alternatives (a) in each condition (condition 1 and condition 2), by means of a Z test to contrast hypotheses about two independent proportions. The results suggest that there are statistically significant differences between the two alternatives (z = 3’53; p ≤ .01).

On the other hand, the binomial test was used to establish whether there were statistically significant differences between alternatives (a) and (b) in each condition. In condition 1, this was the case (p ≤ .05); while in condition 2 significant differences could not be determined (p > .05).

Table 1.

Frequency and percentage in gain situation

Loss situation

Table 2 reflects the responses chosen and their proportions in the loss situation. In condition 4, a greater percentage chose alternative (b), the one that implies a risky loss and an affect-rich stimulus. This same type of stimulus is found in alternative (a) in condition 3, which is the alternative chosen more, even though it is a sure option. The second hypothesis was tested by applying a Z test between the (a) alternatives in conditions 3 and 4. The results of this test yielded statistically significant differences between the percentages compared (z = 2’74; p < .05).

The binomial test behaved in a similar way to the previous gain situation. In condition 4, a statistically significant difference was obtained when comparing the percentages of the two alternatives (p ≤ .05), but this did not occur in condition 3 (p > .05).

Table 2.

Frequency and percentage in loss situation