Ninety-eight Reichman University students (average age=28.15years; SD=10.42; range=20–67; 70 females) took part in this study for a monetary payoff. All participants were native Hebrew speakers. Participants were recruited via ads posted in students’ forums. Participation was voluntary and contingent upon signing a consent form. The sample size was predetermined by a power analysis (Faul et al., 2009) to allow 80% power to detect a moderate effect size (95 participants were needed).

The experimental procedure had three stages. First, participants were randomly assigned to one of two between-subject conditions that manipulated mode-of-thought. Based on Usher et al.’s (2011) “Declared” procedure, mode-of-thought was manipulated by informing participants about the “proven benefits” of decisions that are based on a specific thinking mode. Participants were told that “Research has shown that the best decisions are the ones made using intuition [logic and analytical thought]” and were encouraged to base their preferred choice on their “gut feeling” and general impressions [think carefully and logically about their choices]. We used response time to reinforce the manipulation (Hochman et al., 2015). Participants in the intuitive condition were asked to respond quickly without thinking (in 5s or less). By contrast, in the deliberative condition, participants were asked to respond after carefully considering all the information (for at least 30s). To ensure these instructions were followed, participants in both conditions were told that the computer would monitor their response time and alert the experimenter if they spent too much [too little] time before responding.

Then, participants completed a financial task composed of two blocks. In the first block, three economic scenarios were presented, followed by two or three questions for each scenario to examine the participants’ financial decisions. For example:

The second block was composed of 10 trials of lottery choices, as developed by Holt and Laury (2002), which measures risk tolerance and utility maximization in decisions under risk. In each trial, participants are asked to choose between two lottery options that differ in risk level and expected value. For instance:

Alternative A:

0.8 chance of getting 19.25 NIS

0.2 chance of getting 0.5 NIS

Alternative B:

0.8 chance of getting 10 NIS

0.2 chance of getting 8 NIS

The choice was made by pressing the button labeled “Alternative A” or “Alternative B” positioned at the top of each description. The dependent variable was the proportion of selections from the alternative yielding the higher expected value [the High-EV option, Option A in the example described above (Yechiam and Hochman, 2013)]. To motivate their choices, participants were told that the computer would randomly select one of the trials at the end of the study, and the chosen trial would be played for an actual payoff. In actuality, all subjects received 10 NIS (approx. 2.5 USD) regardless of their choices.

Finally, participants completed the REI questionnaire (Pacini and Epstein, 1999), a 24 item self-reported inventory that assesses individuals’ tendencies to include analytical and experiential considerations in the decision-making process. The REI consists of two unipolar scales (12 items each), which rank participants on two dimensions of thinking style. The first scale measures engagement and favorability of cognitive activities and corresponds to an analytical (deliberative) thinking style (e.g., I have a logical mind). The second scale measures engagement and favorability of automatic activities and corresponds to an experiential (intuitive) thinking style (e.g., When it comes to trusting people, I can usually rely on my gut feeling). Previous research has shown that the internal consistency reliability coefficient for each scale is high (above 0.85), whereas the correlation between them is small and negligible (Pacini and Epstein, 1999). Thus, the REI is assumed to support Epstein’s (1994) claim of two independent information processing systems. Participants were required to state how true each statement was for them on a scale from 1 (definitely not true of myself) to 5 (definitely true of myself).

The study was designed in VB 6.0. Participants were invited to the lab and were seated comfortably in front of a computer screen. Then, the experimenter provided instructions about the task. Each participant engaged in the task in a private setting. Participants were randomly assigned to either the intuitive mode-of-thought (n=50) or the deliberative mode-of-thought (n=48) condition. After completing the study, the participants were paid, thanked, and debriefed. The complete materials of this study (as well as the other studies in this manuscript and all the data) can be found at https://osf.io/34duc/. The Ethics Committee of Reichman University approved this study.

Before testing the hypotheses, we assessed the reliability of the REI Hebrew translation. The internal consistency of the REI was adequate for both the Analytical scale (Cronbach’s α=0.83) and the Experiential scale (Cronbach’s α=0.83). In addition, as predicted, the correlation between the two scales was negligible and insignificant (r=−0.078, p=n.s).

Next, we examined participants’ judgment quality. To do so, each financial decision that maximized utility (i.e., had a higher expected value) was coded “1” (“0” otherwise). For questions with more than two possible alternatives on the economic scenarios task, the normative solution was coded as 1, and the other options were coded as 0. The decision quality was measured as the mean maximization score where a higher score indicates a more economically rational choice. The participants’ judgment quality in the economic scenarios was 0.76 (SD=0.20) and 0.65 (SD=0.19) in the lottery choices.

To test the effect of mode-of-thought on decision-making quality, we calculated the mean EV maximization score for each condition. In the intuitive condition (n=50), the mean score was 0.71 (SD=0.20), and 0.81 (SD=0.18) in the deliberative condition (n=48). An independent samples t-test revealed a significant difference between the two conditions (t(96)=−2.49, p<0.05). Thus, participants who were encouraged to use in-depth evaluation processes exhibited more economically rational choices than participants who were instructed to employ intuitive processes.

A multiple (stepwise) regression analysis was conducted to examine the effect of mode-of-thought manipulation (deliberative mode=1), individual differences in the two processing styles, and their interaction term on decision-making quality. The predictors were entered into the model in a single block using the stepwise method (with a probability for F of p=0.10 as the exclusion criterion and p=0.08 as the inclusion criterion). The results indicated that the best model included the interaction between the analytical and experiential scales (β=0.24, p=0.017) and the mode-of-thought manipulation (β=0.19, p=0.052; F(2, 95)=6.22, p<0.01, R²=11.6%). The analytical (t=0.022, p=0.982) and experiential (t=0.176, p=0.861) scales were not significant predictors. The results of this analysis are summarized in Table 1.

Since the deliberative mode-of-thought was coded as 1, these results suggest that greater deliberative considerations resulted in more rational decisions on this task. However, in line with Ayal et al. (2011), the interaction between the analytical and experiential scales suggests that participants who rely on both processes are less prone to judgmental biases.

To examine the effect of repetition, we tested the participants’ performance over time. A 6 (question order – within-subject variable)×2 (mode-of-thought manipulation – between-subject variable) repeated measures ANOVA was conducted to examine decision quality. This analysis revealed a significant main effect for order [F (5, 92)=5.98, p<0.001], indicating that all the participants improved over time. In addition, there was a main effect for mode-of-thought [F(1,96)=6.24, p<0.05], as well as a marginally significant order × mode-of-thought interaction [F (5,92)=2.31, p=0.051]. This pattern of results indicates that participants in the deliberative condition exhibited more economically rational choices than participants in the intuitive condition. However, the quality of the decisions improved with time in both modes-of-thought (albeit more rapidly in the deliberative one).

To examine judgment quality on the lottery (Holt and Laury, 2002) task, the proportion of selecting the alternative yielding the higher expected value was calculated for each participant (Yechiam and Hochman, 2013). In the deliberative condition, the average rational choice was 0.74 (SD=0.17), compared to 0.56 (SD=0.16) in the intuitive condition. An independent samples t-test revealed that this difference was significant (t(96)=−5.02, p<0.001). Thus, the deliberative mode-of-thought led to more economically rational choices than the intuitive mode.

A multiple (stepwise method) regression analysis was conducted to examine the effect of mode-of-thought and processing style on decision quality on the lottery task. The results showed that the best model included the mode-of-thought manipulation (β=0.43, p<0.001) and the analytical scale (β=0.3, p=0.002; F(2,95)=18.6, p<0.001, R²=28.1%). These results are summarized in Table 2. The experiential scale was not significant (p=0.77). Thus, similar to the previous findings, the results suggested that in-depth considerations resulted in more rational decisions (in terms of EV) than intuitive responses. In line with previous research (Ayal et al., 2011, 2012, 2015), there was a positive correlation between the analytical scale and decision quality, but at the same time, no negative connection between rational decision-making and experiential thinking style.

Study 1 reveals the clear benefit of the deliberative mode-of-thought in making financial decisions. While decision quality increased as participants answered more questions on the economic scenarios, this effect was more pronounced for the deliberative mode-of-thought. This pattern of results is consistent with the dual-system approach, which holds that rational behavior stems from a deliberative thought process (e.g., Kahneman, 2003, 2011). Nevertheless, as in previous studies (Ayal et al., 2011, 2012, 2015), we found no correlation between decision quality and experiential thinking style. In addition, the interaction between the analytical and experiential scales was a better predictor of rational choice behavior (Ayal et al., 2011, 2012). In line with Ayal et al.’s (2011) claim, these results suggest that deliberative thinking may benefit from the additional input of the intuitive system in certain situations. Thus, the combination of highly deliberative and intuitive considerations seems like an important factor in normative reasoning.

One hundred Israelis took part in the study. Forty-four were males, and 49 were females (seven participants did not indicate their gender). The average age was 39.4years (SD=13.73; range=20–74). All participants were native Hebrew speakers. Participants were recruited through posts on social media. Participation was voluntary and contingent upon signing a consent form. The sample size was predetermined by a power analysis (Faul et al., 2009) to allow 80% power to detect a moderate effect size (based on the effect size of Study 1, 94 participants were needed).

The first stage of the experiment consisted of the mode-of-thought manipulation, as in Study 1. Next, participants were presented with a Qualtrics web-based questionnaire composed of two blocks. The first block was a modified version of the debt management game (Amar et al., 2011). In this game, participants are given multiple debt accounts that vary in amounts and annual interest rates. The game has a total of six rounds. At the beginning of each round, participants receive a hypothetical sum of money they can use as a down payment on one of the two open debt accounts. Each round presents a pair of different debts. For instance,

The first round has Debt 1 vs. 2, the second round has Debt 2 vs. 3, and so on. All six rounds were presented in the same fashion to all participants. In each round, the participants were asked to choose the debt they wanted to reduce by allocating their money. A financially optimal player should allocate the available sum to the open debt with the highest interest rate in each round. By contrast, a non-rational player would focus on the amount and allocate the available money to the smallest open debt to further reduce it (“debt account aversion”; Amar et al., 2011). The debt management game is a financial task that includes complex information that is difficult to process (such as interest rate, the debt amount, and the available money). This study aimed to assess potential differences in choice quality motivated by deliberative vs. intuitive considerations.

The second block was composed of four prototypical questions (the Bias Thinking questionnaire adapted from Ayal et al., 2011) that measure adherence to biased judgments. Specifically, Question 1 taps irrational diversification (Ayal and Zakay, 2009), i.e., a false evaluation of diversity based on the perceived rather than the normative value. Question 2 examines the ratio bias, i.e., the tendency to judge a low probability event as more likely when presented as a ratio with large numbers than a ratio with small numbers (e.g., 9/100 vs. 1/10; Kirkpatrick and Epstein, 1992). Question 3 tests the availability heuristic, i.e., individuals’ tendency to estimate the frequency of an event “by the ease with which instances or associations come to mind” (Tversky and Kahneman, 1973, p.208). Finally, Question 4 examines the conjunction fallacy (Tversky and Kahneman, 1983), i.e., people’s tendency to ignore the statistical cut-off law and estimate that the probability of the conjunction between two events together is greater than the probability of a single event alone. The order of presentation was identical for all participants.

A link to the experimental questionnaire was sent to participants via email. Participants were randomly assigned to either the intuitive (n=58) or the deliberative mode-of-thought (n=42) condition. The study lasted about 10min. After completing the questionnaire, the participants were debriefed and were given the experimenter’s contact information for any questions. The Ethics Committee at Reichman University approved this study.

One hundred and thirty-six Israelis took part in the study. Seventy-one were males, and 65 were females. The average age was 41.43years (SD=14; range=21–69). All participants were native Hebrew speakers. These participants also took part in Study 3. Participants were recruited through posts on social media. Participation was voluntary and contingent upon signing a consent form. The sample size was predetermined using a power analysis (Faul et al., 2009) to allow 80% power to detect a moderate effect size similar to Study 1 (94 participants were needed).

Participants were presented with a Qualtrics web-based questionnaire. First, participants completed the REI questionnaire (Pacini and Epstein, 1999), as in Study 1. Then, participants were presented with the mode-of-thought manipulation (deliberative vs. intuitive mode-of-thought) as in the previous studies. Finally, they were asked to complete the same debt management game (Amar et al., 2011) as in Study 2a, except that the available amount of money given to participants at the beginning of each round was enough to eliminate the debts, not just reduce them. The game had seven rounds, which were presented in the same order for all participants.

A link to the experimental questionnaire was sent to participants via email. Participants were randomly assigned to one of two between-subject conditions [intuitive mode-of-thought (n=70) and deliberative mode-of-thought (n=66)]. After completing the questionnaire, the participants were debriefed and were sent the link to Study 3. The Ethics Committee at Reichman University approved this study.

The reliability of the REI scale indicated high internal consistency for the analytical (Cronbach’s α=0.84) and the experiential (Cronbach’s α=0.87) scale. In addition, as predicted, no correlation was found between the two scales (r=0.102, p=0.24).

Next, we examined the mean response time in the two experimental conditions (mode-of-thought manipulation). The mean response time was 9.34m (SD=3.27) in the time-pressure condition (i.e., intuitive mode-of-thought) and 27m (SD=14.63) in the no-time-pressure condition (i.e., deliberative mode-of-though). An independent samples t-test revealed that this difference was highly significant [t(71.14)=9.578, p<0.0001]. These results validated our manipulation and suggested that the participants understood and adhered to the instructions to follow their gut feelings or examine all the available information (depending on the experimental condition; Hochman et al., 2015).

Each financial decision that was based on the utility maximizing principle was coded “1” (and “0” otherwise). The decision quality was measured by calculating the mean of each participant’s choices. Across participants and conditions, the mean score for rational choice in the debt management game was 0.65 (SD=0.23).

The mean rational choice score in the debt management game was 0.67 (SD=0.22) in the intuitive condition, compared to 0.63 (SD=0.24) in the deliberative one. As in Study 2a, this pattern of results hinted at a slight advantage (of 4%) for intuitive processes relative to deliberative ones. However, an independent samples t-test revealed that this difference was not significant (t (134)=−1.039, p=0.302).

As in the previous studies, we also tested the participants’ performance over time. A 7×2 repeated measures ANOVA was conducted to predict decision quality with block (1 to 7) as a within-subject factor and mode-of-thought (intuitive vs. deliberative) as a between-subjects factor. This analysis revealed a significant main effect for block number [F (6, 129)=7.02, p<0.0001], and a marginally significant block × mode-of-thought interaction [F(6,129)=2.06, p=0.06]. By contrast, no effect was found for mode-of-thought [F (1, 134)=1.08, p=0.301, n.s.]. These results are presented in Figure 3 and show a significant improvement in decision quality with time in both conditions. However, there was more improvement in the intuitive condition than in the deliberative one. Thus, similar to Study 2a, repeated experience with financial decisions improved decision quality, and this effect was more pronounced in the intuitive condition.

We conducted a multiple stepwise regression analysis to test the relationship between mode-of-thought manipulation, information processing style, and decision quality. The results showed that the best model only included the analytical scale (β=0.3, p=0.0003; F(1,134)=13.6, p<0.001, R²=9.3%). These results are summarized in Table 3. The experiential scale was not significant (p=0.73). Thus, consistent with previous research, the analytical scale was a crucial predictor in optimal decision-making (Ayal et al., 2011, 2012).

Overall, Studies 2a and 2b provide converging evidence that deliberative considerations do not outperform automatic evaluations under situations of repeated financial selections. In the debt management game, participants needed to select the account with the highest interest rate rather than the lowest nominal value (Amar et al., 2011). Both studies showed that the ability to identify the correct debt to repay improved significantly with experience. Importantly, there was greater improvement when participants based their decisions on intuition rather than deliberation. In addition, similar to previous findings (Ayal et al., 2011; Usher et al., 2011), no difference was found for biased decisions between deliberative and intuitive mode-of-thought. These findings extend previous research showing that intuitive reasoning may be associated with highly normative judgments (Davis et al., 1993; Glöckner, 2007; Ayal et al., 2011).

Our results suggest that deliberative processes that include irrelevant information (e.g., the number of open debts) might impede decisions (Ayal et al., 2011). At the same time, intuitive processes might be less affected by it, as they are more oriented toward self-interest (Hochman et al., 2015). Furthermore, experience and familiarity with choice alternatives which ease negative emotions (Zandstra et al., 2000; Yechiam et al., 2019) lead people to make better financial decisions. However, these studies focused on description-based decisions, in which all information is provided. Study 3 focused on repeated financial decisions that could be learned from experience to examine the effect of mode-of-thought on decision quality in this kind of situational condition (i.e., learning from experience).

One hundred and thirty-six Israelis participated in the study (65 women). The sample size was predetermined by a power analysis (Faul et al., 2009) to allow 80% power to detect a moderate size effect (112 participants were needed). Study 3 was administered directly after Study 2b to the same participants.

Participants were presented with a Qualtrics web-based questionnaire composed of two blocks. In the first block, participants filled in the REI questionnaire (Pacini and Epstein, 1999) as in Study 1. Next, participants were presented with the mode-of-thought manipulation of the previous studies (deliberative vs. intuitive mode-of-thought).

The final block consisted of a repeated saving task adapted from Thaler et al. (1997). Subjects were required to manage a small savings plan in the amount of 20,000 NIS for a five-year period. Each year involved four trials, which represented quarters. Thus, the entire task comprised 20 trials in total. The bank clerk suggested two different savings plans (Plan A and Plan B): Plan A with a variable interest of prime plus 1.5% and plan B with a fixed interest of 2.5%. Each quarter, participants were allowed to switch plans or to stick with the previously selected plan. According to the rational model, optimal players should allocate their money to the plan with the highest interest rate on each trial, since this plan would yield the highest expected return. Plan A had high variance but also high-profit expectancy vs. Plan B, which was more secure but with a lower profit expectancy. Participants had to learn about the risks and returns of the two plans through experience (Thaler et al., 1997).

Participants were randomly assigned to one of two feedback conditions, full or partial. In the full feedback condition, participants received information about the “prime” interest rate at the end of each quarter (savings period). In the partial feedback condition, the prime rate information was only provided on every other trial, starting from Trial 1. This served to examine participants’ financial preferences over time and whether learning from experience (under full and partial feedback) affected their preferences.

A link to the experimental questionnaire was sent to participants via email. Participants were randomly assigned to one of four between-subject conditions: intuitive mode-of-thought with full feedback (n=31), deliberative mode-of-thought with full feedback (n=37), intuitive mode-of-thought with partial feedback (n=39), and deliberative mode-of-thought with partial feedback (n=29). The study lasted approximately 20min. After completing the questionnaire, the participants were debriefed and were given the experimenter’s contact information for any questions. The Ethics Committee at Reichman University approved this study.

To test the effect of mode-of-thought and feedback on decision quality on the savings task, we calculated the mean rational choice for each condition. This analysis revealed that in the intuitive condition, the mean score was 0.54 (SD=0.26) under full feedback and 0.56 (SD=0.21) under partial feedback. In the deliberative condition, the mean score was 0.71 (SD=0.25) under full feedback and 0.57 (SD=0.25) under partial feedback. These results are presented in Figure 4.

A two-way ANOVA revealed a significant effect for the mode-of-thought condition [F (1,132)=4.72, p<0.05] and a marginally significant interaction effect [F(1,132)=3.29, p=0.072]. No main effect was found for feedback [F (1,132)=1.68, p=0.197, n.s.]. This pattern of results suggests that asking participants to make decisions based on deliberative processes resulted in higher financial decision quality than the intuitive responses. However, the deliberative thinking scores were only noticeably better in the full feedback condition.

We conducted a moderation analysis to examine whether the analytical information processing style moderated the effect of feedback (full vs. partial) on decision quality in the repeated savings task. Feedback (as a binary variable, full=1; partial=2) and the analytical scale (categorical variable based on the median split; low coded as 0 and high coded as 1) were simultaneously entered as predictor variables. Following Aiken and West (1990), the predictor variables were centered before conducting the regression analyses. All analyses were performed using SPSS20 with the PROCESS macro (Hayes, 2012) to address the moderation hypotheses.

The analysis revealed a significant marginal main effect for the analytical thinking style (β=0.15, B=0.07, t=1.78, p=0.07), indicating that high analytical participants made better decisions in the repeated savings task. In addition, the interaction between feedback and the analytical scale was significant (β=0.17, B=0.17, t=2.01, p=0.04). No main effect for the feedback condition was found (β=−0.13, B=−0.06, t=−1.6, p=0.11, n.s.).

To understand the nature of this interaction effect, we followed Aiken and West’s procedure (1990) and estimated two regression lines of feedback condition on decision quality at two levels (high and low) of the analytical scale. Simple slope tests indicated that the feedback condition was only significantly associated with financial decision quality among participants low on the analytical scale (β=−0.34, B=−0.17, t=−2.57, p<0.05). This interaction effect is presented in Figure 5.

As shown in the figure, there was no difference in the decision quality between the high and low analytical participants in the full feedback condition. However, high analytical participants were more rational than low analytical participants under the partial feedback condition. That is, in cases where the information was not presented to the decision-maker in full, not relying on analytical thinking may have hampered the quality of the participants’ financial decisions.

As in Study 1, the results of Study 3 point to a clear advantage for deliberative processes in financial decision-making when all the information required for the decision is presented to the decision-maker. In contrast, there was no difference in decision quality between the deliberative and intuitive modes-of-thought conditions when information was lacking. Moreover, individual differences in the analytical style did not influence decision quality when full feedback was provided to the participants. However, when information was partial, low analytical participants were more prone to biased economic decisions. Thus, low analytical thinking (but not high experiential one) may impede rationality in situations where full information is unavailable.