People are not always rational when they have to choose an answer or make a decision. Rather, they tend to use an effort-reducing method of processing information called heuristics (see also Kahneman & Tversky, 2013). The term “heuristics” is derived from the Greek word “evrisko”, which means to find or to discover. Heuristics are widely used in decision-making problems because they help to speed up the decision-making or problem-solving processes (Emiliano, 2015). According to Shah and Oppenheimer (2008), heuristics facilitate effort reduction by, (a) considering fewer cues, (b) reducing the effort to retrieve cue values, (c) reducing the weighting of cues, (d) combining less information, and (e) exploring fewer alternatives.

To illustrate, a heuristic method is a quick way to solve a problem with a solution that is not the best but the most likely one, like a “mental shortcut” (Kahneman, 2011). This has the advantage of saving time and effort while simultaneously ending up (very often) making the right decision in the end. However, the disadvantage of these mental shortcuts and effort-reducing solutions is that they can sometimes lead to errors and biased judgments or logical fallacies (Liberali et al., 2012). They are statements that seem to be logical, but when examined using the rules of logic, turn out to be biased. Logical fallacies are often associated with a high feeling of confidence in an incorrect answer (Kahneman & Tversky, 2013). Metacognitive feeling of confidence has been found to be associated with affective reactions (Nerantzaki & Efklides, 2019; Vogl et al., 2019).

In general, many theorists have acknowledged the important role of affect in decision-making processes (Garcés & Finkel, 2019). For instance, the “affect heuristic” (Slovic et al., 2007) refers to a mental shortcut in which individuals make quick decisions primarily guided by their emotions. While the important role of affective processes in the formulation of judgments or decision making is well established (Garcés & Finkel, 2019; Schwarz & Clore, 2007), little is known about affective responses following the recognition of biases in our judgments. The present study aims to fill this gap by providing empirical evidence on affective reactions following biased heuristic errors. One category of affective reactions that is closely related to problem-solving activities (D’Mello & Graesser, 2014) and arises from unexpected and complex information (Chevrier et al., 2019; Vogl et al., 2019) are the so-called epistemic emotions.

Epistemic emotions were initially described by philosophers as affective states that can motivate critical reflection, knowledge acquisition, and exploration (see Brun et al., 2008; Morton, 2010). From a philosophical perspective, epistemic emotions drive knowledge acquisition and knowledge inquiry (Muis et al., 2018). From an educational psychology perspective, epistemic emotions are categorized within the broader domain of academic emotions (Pekrun, 2024). Academic emotions, which are specific to educational contexts, are closely associated with learning processes and academic achievement (Pekrun, 2006, 2016). Pekrun and Stephens (2012) defined epistemic emotions as emotions related to knowledge and knowledge-generating activities. Any emotion associated with knowledge can be characterized as epistemic (Morton, 2010; Muis et al., 2018). However, as Brun et al. (2008) suggest, curiosity, surprise, and confusion are epistemic emotions per se, whereas other affective states can fall into different emotional categories depending on their object focus. For instance, frustration or anxiety, arising from the inability to solve a mathematics problem, may be considered epistemic emotions if they stem from cognitive incongruity and the unresolved problem itself. However, if they are primarily focused on personal failure and the perceived inability to arrive at the correct solution, they should be classified as achievement emotions (Muis et al., 2015).

Specifically, epistemic emotions are triggered by a lack of prior knowledge, conflicting information, or a mismatch between new information and existing knowledge schemas (Muis et al., 2021; Nerantzaki et al., 2021; Vogl et al., 2019). A recent analysis of think-aloud protocols identified three key factors that precede epistemic emotions: assessments of epistemic (in)congruence (the congruence or incongruence between prior knowledge and new information), assessments of the novelty or conflicting information, and judgments about the achievement of epistemic goals (Chevrier et al., 2019). Within this context, it is important to distinguish between cognitive conflict and schema incongruence, as both concepts are closely related to epistemic emotions. Cognitive conflict arises when learners encounter contradictory information, in tasks involving conflicting responses or in texts presenting opposing perspectives on a topic (D’Mello & Graesser, 2014; Nerantzaki et al., 2024). Schema incongruence occurs when new information does not align with existing cognitive frameworks, prompting cognitive adjustments to resolve the mismatch (Reisenzein et al., 2019).

Schema incongruence also occurs when task-related feedback reveals that one’s beliefs are incorrect, such as in “high-confidence errors” (Marshall & Brown, 2006; Vogl et al., 2019). “High-confidence errors” refer to incorrect responses that individuals strongly believe to be correct. Vogl and collaborators (2019, 2020) examined those errors in the context of feedback provided after trivia questions were answered incorrectly. Their findings support propositions that the mismatch between expected and actual outcomes is a prime driver of epistemic emotions. This phenomenon can be further explained through the concept of cognitive dissonance. Cognitive dissonance refers to the psychological discomfort that arises when newly acquired information contradicts prior beliefs, expectations, or knowledge (Festinger, 1957). This tension creates a motivational drive to resolve the inconsistency, by adjusting one’s beliefs, seeking additional information, or reinterpreting the conflicting evidence (Harmon-Jones & Mills, 2019). In the context of negative feedback, this process can trigger epistemic emotions such as confusion, surprise, or curiosity as individuals attempt to reconcile the discrepancy and restore cognitive consistency (Stergiadou et al., 2023; Vogl et al., 2020).

Despite the similarities, it has been proposed that surprise, confusion and curiosity are triggered by different cognitive states (Nerantzaki et al., 2021). Surprise is a leading epistemic emotion that occurs when a stimulus is unexpected, discrepant, or inconsistent with a previously activated cognitive schema (Reisenzein et al., 2019; Topolinski & Strack, 2015). Confusion occurs when people encounter information that they cannot easily understand. It is generally associated with challenging tasks in the presence of impasse, complexity, conflict, contradiction, or incongruity (D’Mello & Graesser, 2012; Silvia, 2009). Curiosity represents a driving force or desire to seek new information that motivates the exploration and acquisition of knowledge (Kashdan & Silvia, 2009). Curiosity is aroused by unexpected events, unsolved problems, ambiguous ideas, conceptual puzzles, or information that makes one aware of gaps in one’s knowledge. It is a complex emotion, and its arousal involves both positive feelings (e.g., the satisfaction of learning something new, and negative feelings (e.g., impatience due to lack of knowledge) (Ainley, 2019).

In summary, the role of epistemic emotions in problem-solving activities is only beginning to be explored. A recent study examined the role of external feedback and found that the intensity of these emotions increased in proportion to the confidence participants had in their incorrect answers (Vogl et al., 2019). The focus of the present study was to extend these findings by examining how different types of feedback influence epistemic emotions in decision-making tasks involving high-confidence errors. For this purpose, two widely recognized scenarios were used. The first scenario was the conjunction fallacy, where participants were asked to estimate the probability of two events occurring together, often resulting in an overestimation of the combined probability (Liberali et al., 2012). The second scenario is a cognitive reflection task that tests participants' ability to override an intuitive response (Brañas-Garza et al., 2019). The choice of these scenarios was deliberate. The scenarios are not inherently difficult when approached with logical reasoning. However, many people tend to give incorrect answers, mainly due to an overestimation of their own intuitive judgement (Kahneman, 2011). We assumed that these scenarios would induce a high level of confidence in the participants’ answers, even though these answers ultimately were incorrect. In line with this, both scenarios are anticipated to evoke epistemic emotions, as they involve the evaluation of epistemic incongruence (Chevrier et al., 2019; Vogl et al., 2019).

The present study aimed to investigate epistemic emotions (surprise, confusion, curiosity) following different types of feedback after participants engaged in a faulty heuristic solution. This work bridges two different research traditions: the heuristics and biases tradition, which focuses on decision-making under uncertainty (Kahneman & Tversky, 2013) and the epistemic emotions tradition in learning and decision-making situations (Muis et al., 2015; Vogl et al., 2019). While prior research has explored epistemic emotions in response to schema-discrepant information (e.g., Chevrier et al., 2019; Muis et al., 2015; Vogl et al., 2019), these emotions have not been specifically examined in the context of heuristic errors that generate logical inconsistencies.

To address this gap, we provided participants with two types of feedback: “low-informative” feedback that merely indicated whether an answer was correct and “high-informative” feedback that included detailed justifications for the correct response. We hypothesized that high-informative feedback would facilitate the integration of new information into existing cognitive schemas (Muis et al., 2015), potentially reducing the intensity of epistemic emotions as cognitive dissonance is resolved (Nerantzaki et al., 2021).

Based on these considerations, two hypotheses were proposed:

The percentage of correct answers and biased heuristic answers was calculated for each task to create two groups: those who found the correct answer and those who did not. Descriptive statistics for all the variables in each scenario are presented in Table 1. Specifically, in the Job Scenario, 241 (40.5%) of the participants chose the correct answer. In the Disease Scenario, 279 (46.7%) participants selected the correct response.

The present study investigated how different types of feedback influence emotional reactions in decision-making tasks involving high-confidence errors. Participants completed two decision-making scenarios that activate biased heuristic solutions and received either low informative feedback (indicating correctness) or high informative feedback (offering detailed explanations of correct answers). They reported their confidence levels and epistemic emotions of surprise, confusion, and curiosity. Participants’ performance and the type of external feedback on their performance significantly affected all epistemic emotions. The results confirm previous evidence for the important role of high-confidence errors in eliciting epistemic emotions but also extend the role of external feedback. To illustrate, negative external feedback following an incorrect answer led to significantly higher levels of surprise, confusion, and curiosity in both tasks compared to those emotions following the correct answers.

Epistemic emotions were also negatively associated with performance in both tasks. The findings are consistent with the notion that cognitive incongruity serves as a key trigger for epistemic emotions (D’Mello & Graesser, 2012; Vogl et al., 2019). It seems that an incongruity between an activated schema (response) and newly received information (feedback) activated epistemic emotions. Participants were forced to construct a new schema by assimilating new information (Efklides, 2017). The aroused epistemic emotions reflect the activation of such revision processes (Vogl et al., 2019). Correlations between performance and epistemic emotions were stronger following the first feedback, as the second feedback provided all the necessary information to finalize the accommodation of the new schema. In contrast, participants who answered correctly experienced feedback that aligned with their existing cognitive schemas, resulting in low-intensity epistemic emotions from the outset.

Furthermore, one of the most interesting findings of the present study was the significant decrease in epistemic emotions after receiving highly informative feedback. The first external feedback activated epistemic emotions and induced a cognitive imbalance (Muis et al., 2015). Indeed, the intensity of these emotions was shaped by participants’ confidence in their incorrect answers, indicating a mismatch between their prior beliefs and the correct answer (Vogl et al., 2019). The cognitive imbalance was restored after the second feedback, which provided all the explanations necessary to assimilate the new information into one's existing cognitive schemas (Nerantzaki et al., 2021). Vogl et al. (2019) proposed that epistemic emotions are inherently situation-dependent, changing over time in response to variations in task information. This is indicative of the dynamic nature of epistemic emotions in learning or decision-making situations (Nerantzaki, et al., 2024; Pekrun, 2024).

However, there were also some differences in the arousal of epistemic emotions. Although all epistemic emotions followed a similar pattern of activation, the feedback effect was lower for curiosity and surprise, and higher for confusion. The differences in effect sizes for surprise, curiosity, and confusion can be interpreted through the distinct cognitive mechanisms underlying each epistemic emotion (Nerantzaki et al., 2021; Vogl et al., 2021). Surprise is often triggered by unexpected outcomes or violations of expectations (Reisenzein et al., 2019). Curiosity arises from an intrinsic desire for knowledge and information-seeking behavior, often stimulated by uncertainty or gaps in understanding (Litman, 2008). Both emotions strongly depend on cognitive evaluations of performance. In contrast, confusion demonstrated a higher effect size for feedback, but also of performance, indicating a stronger link with difficulty in processing information (D’Mello et al., 2014; Nerantzaki et al., 2024). Confusion arises during complex learning activities that involve challenging tasks, particularly when encountering discrepancies that disrupts understanding (D’Mello & Graesser, 2014). That means that the nature of the task plays a crucial role in determining which epistemic emotions are elicited (Vogl et al., 2021).

In addition, confusion was the emotion that showed the most drastic decrease in the second measurement for those who chose the biased heuristic answer. Informative feedback seems to be important in reducing confusion caused by information that is incongruent with previous beliefs. Confusion is triggered by complex information that is not fluently understood (D’Mello & Graesser, 2012; Silvia, 2009). Unlike surprise, which primarily refers to unexpected events (Reisenzein et al., 2019), confusion can also occur under so-called “mismatch conditions”, such as incongruence, dissonance, or conflict (D’Mello et al., 2014). In fact, confusion is not simply the detection of a mismatch, but an intense emotional state characterized by significant cognitive conflict, leading to uncertainty about how to proceed (D’Mello et al., 2014). Furthermore, D’Mello and Graesser (2014) suggested that confusion persists until the cognitive conflict is effectively resolved.

Overall, decision-making processes are often influenced by affective states (Garcés & Finkel, 2019). Our findings provide support for the proposition of Vogl et al. (2019) that epistemic emotions arise during high confidence errors, but we also found that the type of external feedback matters. The first negative feedback activated epistemic emotions and cognitive imbalance, which were restored after the second feedback, which presented the logical principle underlying the correct answer. In what follows, we consider limitations and directions for future work. We also discuss educational implications for research on epistemic emotions.