Open Access

Confidence judgments and decision making are part of everyday life. In an ideal world, people would assess their skills and knowledge accurately and base their decisions only on rational deliberations. Yet, this is often not the case. Confidence judgments in own skills or performance are often biased (e.g., Dunning, 2011; Moore & Healy, 2008; Moore & Schatz, 2017; Sanchez & Dunning, 2018; Pikulina, Renneboog, & Tobler, 2017; Michailova & Schmidt, 2016). Also, people tend to deviate from rational decision strategies (e.g., Achtziger & Alós-Ferrer, 2014; Alós-Ferrer, Hügelschäfer, & Li, 2016, 2017; Charness, Karni, & Levin, 2010; Erev, Shimonovich, Schurr, & Hertwig, 2008; Fiedler, Brinkmann, Betsch, & Wild, 2000; Tschan et al., 2009). Therefore, the research aim of the present dissertation was twofold. In the first chapter of the present dissertation I investigated confidence judgments in own skills and the confidence bias, the processes underlying these confidence judgments, and the influences of gender and monetary incentive on confidence judgments. The second aim was to investigate the influence of goal and implementation intentions on rational decision making and how this influence is reflected in the neural correlate of reinforcement learning. A common finding in research on confidence judgments is the confidence bias (e.g. Moore & Schatz, 2017; Moore & Healy, 2008; Pikulina et al., 2017; Sanchez & Dunning, 2018; Lebreton et al., 2018). In most cases, the confidence bias reflects overconfidence, which means that people’s subjective confidence exceeds their actual ability or performance (Fischhoff, Slovic, & Lichtenstein, 1977). In some cases, there is also evidence for underconfidence, suggesting that people underestimate their abilities (Kruger & Dunning, 1999; Kruger & Burrus, 2004). Gender is an important predictor of the confidence bias. Underconfidence is more prevalent in females, whereas males often display overconfidence (e.g., Barber & Odean, 2001; Hügelschäfer & Achtziger, 2014; Niederle & Vesterlund, 2007). In Study 1, I investigated the processes underlying confidence judgments and the confidence bias by means of response times, and I examined potential gender differences. Participants answered general knowledge questions and judged their confidence on the correctness of each answer. Participants had overall a good sense of whether their answer was correct or incorrect. This was reflected by higher confidence judgments on correct answers compared to incorrect ones. The analysis of response times on the confidence judgments revealed that male participants who took longer to judge their confidence were made more accurate judgments than males who responded quickly. This relationship was not found for females. In Study 2, half of the participants received a monetary incentive for good performance in the general knowledge test. The monetary incentive for performance increased the time invested in both tasks (the knowledge questions and the confidence judgments). However, this increased effort did not lead to better performance on the knowledge questions, nor did it yield more accurate confidence judgments. The response times suggested that males who invested more time in the confidence judgments were more accurate (as in Study 1). However, the opposite was true for females. The more time females invested in their judgment the more underconfident they were. This influence of the response time on the confidence bias was only found for incentivized participants. In Study 3, the accuracy of the confidence judgment was incentivized. Contrary to the expectations, the monetary incentive did not reduce the confidence bias but led both males and females to be overconfident. In this study, the response time on the confidence judgment did not predict the confidence bias. On the whole, the results demonstrate that (a) the processes of confidence judgments differ between females and males, and (b) the effectiveness of monetary incentives for improving the accuracy of confidence judgments depends strongly on the incentive being contingent on the performance in the task at hand. The second chapter of the present dissertation investigated the influence of goal and implementation intentions (P. M. Gollwitzer, 1999) on rational decision making (see also Hügelschäfer & Achtziger, 2017). The impact of intentions was examined by the neural correlate of reinforcement learning, i.e. the feedback-related negativity (FRN; Holroyd & Coles, 2002). Participants worked on a probability updating task in which the optimal strategy to maximize the expected payoff was to follow Bayes’ rule by integrating new information with prior probabilities (Bayes & Price, 1763). The optimal decision rule conflicted with a simpler suboptimal decision strategy, i.e. the reinforcement heuristic (see Achtziger & Alós-Ferrer, 2014; Charness & Levin, 2005). The goal and implementation intention manipulation was proposed to control the automatic process of the reinforcement heuristic and hence foster rational decision making. The results showed that the goal intention and the implementation intention had no influence on the number of reinforcement errors (in contrast to the findings of Hügelschäfer & Achtziger, 2017). However, both, the goal and implementation intentions increased the amplitude of the FRN which, on the neural level, indicated a stronger reliance on the reinforcement heuristic than in the control group. The findings shed some light on the impact of goal and implementation intentions on rational decision making. They demonstrate that careful consideration of the use of intentions as an intervention for improved decision making is required to avoid undesired side-effects. Taken together, the present dissertation provided new insights into the processes underlying confidence judgments, the confidence bias, rational decision making, and its neural correlates.