Introduction
Metacognition refers to the capacity to reflect on, monitor, and control cognitive processes, such as decision-making, memory, and perception (Flavell, 1979). Efficient metacognition commonly represents better self-assessment for one ability. While measuring metacognition ability may be achieved by assigning self-report confidence rating in a given task, efficiency in metacognition is commonly achieved by studying the difference in CR between correct and incorrect performance (Polyanskaya, 2023). Over decades, researchers have been motivated to learn the extensive role metacognition plays in executing adaptive human behaviors, mainly focusing on education and learning aspects (Baer and Kidd, 2022), with relatively less interest in emotional perception (Laukka et al., 2021; Rahnev, 2023). However, studying efficiency in metacognition is still scarce. Given the fundamental role metacognition took in emotion perception as a perceptional process, a better understanding of measuring efficiency in metacognition of emotion recognition is much needed. To fill this gap in the literature, in this preliminary study we aimed to investigate metacognitive efficiency in emotion recognition focusing on voice communication, across different emotions category.
Metacognitive efficiency
Efficient metacognition generally reflects a better introspection of cognitive process. Specifically, effective metacognition is suggested to promote effectiveness in noticing relevant cues and to seek additional information when necessary (Drigas and Mitsea, 2020; Brus et al., 2021). Therefore, metacognition is considered foundational to flexible and adaptive behavior in a range of settings (Baer and Kidd, 2022; Katyal and Fleming, 2024). Furthermore, promoting efficient metacognition by guided intervention had a positive impact on discrimination, working memory, and spatial tasks (Carpenter et al., 2019; McWilliams et al., 2023). A widely used method to study metacognition is to have observers do a challenging task followed by an explicit confidence judgment regarding their task performance (McWilliams et al., 2023). This retrospective CR is further used as a measure of monitoring effectiveness, when a larger difference in CRs assigned to correct and incorrect performances reflects the ability to estimate the likelihood of the observer making an error (Rahnev, 2021; Polyanskaya, 2023). Based on CR measurement, metacognitive efficiency is sometimes considered domain-specific, meaning that one's ability to effectively evaluate their correctness in performance is specific to the object under scrutiny. Hence, individuals can also have high awareness of their performance in one cognitive domain while having limited awareness of their performance in another (Rouault et al., 2018). As such, CRs for correct performance were lower in the perceptual task (i.e., visuospatial task) than in the memory task (Pennington et al., 2021). However, less studies posited a domain-general metacognitive competence based on audio, visual and audio visual modalities (Ais et al., 2016; Carpenter et al., 2019). Together, inconsistency in findings highlights the importance of independent metacognitive efficiency study for a range of domains and skills. Yet, most research remains focused on general knowledge and learning domains, with less interest in social domains (Saoud and Al-Marzouqi, 2020).
Metacognitive efficiency in emotion recognition
As metacognition contributes to accurate performance in other perception tasks, few researchers sought to gain further insights into its potential contribution to accurate perception of others' emotions (Jeckeln et al., 2022; Michel, 2023). Emotion perception competence refers to the ability to perceive and interpret (i.e., decode) relevant social–emotional cues expressed by others (i.e., to encode; Bänziger and Scherer, 2007). Accurate perception of others' feeling and thinking is a key skill associated with achieving successful interpersonal relationships (Mitsea et al., 2022). In the visual channel, for example, authors revealed that higher CR was associated with higher accuracy in the facial emotional recognition task, underscoring the fundamental role of CR in the adjustments of social behavior (Bègue et al., 2019; Garcia-Cordero et al., 2021; Folz et al., 2022; Katyal and Fleming, 2024), whereas metacognitive dysfunction has been implicated as a key source of maladaptive behavior regarding emotional perception (Arslanova et al., 2023).
When it comes to voice communication, studying efficiency in metacognition of emotional perception is relatively small in amount. Voice expressions can be either verbal (e.g., words and sentences) or nonverbal (e.g., screaming, laughing), where all have proven effective in conveying emotions (Eyben et al., 2015). In general, voice comprises unique audible features called prosody. Prosody is an important feature of language, mainly represented by intonation (i.e., how low or high a voice is perceived), loudness (i.e., how loud a sound is perceived), and tempo (i.e., how quickly an utterance is produced; Mitchell et al., 2003). Prosody also conveys important information about a speaker's sex, age, attitude, and mental state (Schwartz and Pell, 2012). Lausen and Hammerschmidt (2020) conducted one of the few studies examining CR's role in vocal emotion perception. They found that participants' confidence judgments increased with increasing correctness in emotion recognition tasks that included verbal and nonverbal stimuli. As a result, they speculated that specific characteristics of voiced stimuli affect CRs. They also found that meaning played a significant role in performance confidence as a higher confidence rating was assigned to stimuli containing lexical-semantic content (e.g., nouns and lexical sentences) rather than stimuli without meaning (e.g., pseudonouns and pseudosentences). More recently, Laukka et al. (2021) extended the concept of CRs to include multimodal analysis. On the basis of visual, audio, and visual–audio channels, the researchers also found a strong correlation between emotional recognition accuracies and CRs, in each channel (Laukka et al., 2021).
While the above studies demonstrate the importance of metacognitive accuracy, based on CR measurement, as far as we know, no study focused on metacognitive efficiency. In the current study, we deal for the first time with efficiency in metacognition concerning voice emotion recognition. Specifically, we asked whether correct performance in emotion recognition test would yield significant higher CR, compared to CR assigned to incorrect performance? As we intend to focus on voice communication, we adhere to previous suggestions that gender roles should be studied within a given societal context, particularly language (Keshtiari and Kuhlmann, 2016; Hall et al., 2021). As a result, we used an emotional target originally designed in Hebrew (Sinvani and Sapir, 2022) in the form of repeated lexical sentences, reflecting discrete emotions (anger, sadness, happiness, fear, and surprise) as well as neutral expression. In this way, we intended to focus on emotional prosodic cues separately from semantic cues. Earlier studies have discussed such methodologies (Paulmann and Kotz, 2008; Sinvani and Sapir, 2022), with particular recommendations in the diagnostic analysis of nonverbal accuracy 2 (DANVA-2), emphasizing the independent role of prosodic cues in emotion recognition (Nowicki and Duke, 2008). To answer our question, we used performance-based tests for emotion recognition by voice, followed by CR reported trial by trial. For the purpose of metacognitive efficiency detection, we set the mean CR scores for correct to incorrect performances. We hypothesis to find significant difference between CR assigned to correct and incorrect performance, where higher CR would be assigned to the correct performance. By hypothesizing that, we actually expect to illustrate here efficient in metacognition across all tested emotions, perceived by voice.
Discussion
In the current preliminary study, we investigated the metacognition efficiency of an emotion recognition task performance, conveyed by voice. To this end, we analyzed differences in CRs between correct and incorrect performance in the emotion recognition task, based on identical vocal lexical sentences portrayed into emotional prosodies of anger, happiness, sadness, surprise and fear, and neutral expression. Following the literature dealing with the global concept of metacognitive efficiency, we expected to demonstrate a greater difference between CRs assigned to correct and incorrect performances, thus illustrating a convenient representation of metacognitive efficacy. As far as we know, this is the first study conducted to examine efficiency in metacognition skills concerning emotional recognition by voice.
As predicted, our results indicate a consistent difference in CRs between correct and incorrect performance, where higher CRs were significantly assigned for correct performances. In accordance with previous studies, our findings support the general established finding for metacognition efficiency in diverse cognitive tasks (Ais et al., 2016; Carpenter et al., 2019). To our knowledge, only two studies have been conducted to examine metacognitive ability based on retrospective confidence rating for vocal emotion recognition (Lausen and Hammerschmidt, 2020; Laukka et al., 2021). Although these studies based their examination for metacognitive skill on the measure of CR in a given response, representing metacognitive accuracy, they ignore the difference in CR between correct to incorrect performance, comprising the metacognitive efficiency.
Following extensive study focusing on brain mechanism underlying correctness in emotion recognition tasks, one of the worth noting question concerns the neural selectivity is if it is determined by the emotion presented by encoder or by the emotion perceived by a decoder (Wang et al., 2014). Based on face stimuli, Wang et al. (2014) concluded that neurons in the human amygdala encode the perceived judgment of emotions rather than the stimulus features themselves. Recently, Zhang et al. (2022) further showed that responses of anterior insula and amygdala were driven by the perceived vocal emotions. However, they revealed that responses of Heschl's gyrus and posterior insula (i.e., both involved in auditory perception) were determined by the presented emotion. Together, it was speculated that extensive connectivity of the amygdala with other brain regions allows it to modulate a few functions essential in adaptive and socio-emotional behaviors (Whitehead, 2021; Atzil et al., 2023). Given the fundamental role metacognition (i.e., confidence) plays in adaptive and socio-emotional behaviors, our results highlight the further need for incorporating neuroscience study into neural mechanism underlying self-confidence to distinguish correct and incorrect performance in emotion recognition task. This study has theoretical and practical significance. From the theoretic point of view, our results confirm earlier studies concerning various cognitive processes by establishing that individuals are capable of efficiently assessing their performance in voice emotion recognition (Cho, 2017; Skewes et al., 2021; Polyanskaya, 2023). Moreover, given the significant association that was indicated across all emotions and neutral expressions, we speculated metacognitive efficiency of being individual “fingerprint”, rather than domain-specific (Wang et al., 2014; Ais et al., 2016; Zhang et al., 2022). Additional studies should incorporate neuroscience methodologies in diverse channels of emotion communication to specify the mechanism of efficiently introspecting vocal emotion recognition to gain support for this claim (Wang et al., 2014; Whitehead, 2021; Zhang et al., 2022; Atzil et al., 2023).
From the practice point of view, in light of previous results showing the increase in performance found among students after they participate in guided intervention for metacognitive awareness (Nunaki et al., 2019), our findings should be taken one step further regarding designing interventions focused on emotion recognition through voice. Unlike vocal channel which deserved much less attention, several studies demonstrated the effectiveness of practicing facial emotion recognition skill as beneficial to social interactions (Chaidi and Drigas, 2020; Dantas and do Nascimento, 2022a,b). Importantly, those works highlight the flexibility of facial emotion recognition skill which in our opinion justify comprehensive research in the voice channel. In addition, our findings may consider valuable by diagnosis means, as extensive research dealt with metacognitive maladaptive in diverse subclinical and clinical populations (Hoven et al., 2019).
Although our study offers a valuable contribution to the current understanding of metacognition in social communication of emotions, some limitations should be noted. First, to investigate emotional recognition, we used a lexical sentence as a repeated item. Therefore, one can argue that prosodic cues are more important than semantics. However, lexical sentences can still affect perception and confidence by contributing meaning (Lausen and Hammerschmidt, 2020). Therefore, researchers should consider various types of stimuli for a better generalization of findings regarding efficiency in metacognition of emotion recognition by voice. Second, it is important to note that our investigation was blinded to acoustical parameters, which have been documented as contributing to the CR discussed elsewhere (Lausen and Hammerschmidt, 2020). Spectral parameters and prosodic contours (Shaqra et al., 2019) would be valuable tools for future research, allowing researchers to detect physiological changes in voice and emotional expressions.