Emotions are complex mental states that involve several components, including subjective experiences, behavioral responses, and physiological changes (Lane et al., 2000). The component process model posits that categorical emotions, such as fear or joy, arise from integrating these components (Sander et al., 2018). While much research has focused on these components in adults (for reviews, see, e.g., Kragel & LaBar, 2016; Nummenmaa & Saarimäki, 2019), the developmental trajectory of emotions from childhood to adulthood remains unmapped.
Subjective experiences, commonly referred as feelings, reflect the conscious aspects of emotions (Anderson & Adolphs, 2014; Barrett et al., 2007). These are often assessed through self-report ratings of some aspect of emotional experience, such as the intensity of discrete emotions (Cowen & Keltner, 2017; Goldin et al., 2005; Horikawa et al., 2020), or valence and arousal (Chan et al., 2020; Nummenmaa et al., 2012). However, the reliability of self-reports in young children remains contentious. Children as young as three years of age can provide self-reports of their emotions, which are aligned with their vocal, bodily, and facial behaviors (Durbin, 2010; Gabel et al., 2019). Self-reports of bodily changes associated with basic emotions are already distinct in six-year-old children and become increasingly similar to those reported by adults during maturation (Hietanen et al., 2016). Nevertheless, young preschool children tend to provide extreme responses when using Likert-type scales to rate their emotional states (Chambers, 2002) and rate the intensity higher compared to external coders (Gabel et al., 2019). Additionally, children may lack emotional vocabulary or awareness of their emotional states (Zeman et al., 2007). A more comprehensive understanding of children’s emotional states may be gained by using self-report and behavioral measures in combination with psychophysiological measures (Sohn et al., 2001; Wilhelm et al., 2006; Zeman et al., 2007).
Emotional behaviors can be manifested in facial movements (Barrett et al., 2019), body posture and movements (Witkower & Tracy, 2019), vocal prosody (Cowen et al., 2019), blink rates (Skaramagkas et al., 2021) or other gaze behaviors (Adams & Kleck, 2005). For instance, Maffei and Angrilli (2019) found that blink rates decrease during compassionate and scenic scenes and increase during fear-inducing clips. Similarly, neural networks can infer emotional states from blink data, with the highest accuracy observed during fear-inducing clips (Goshvarpour & Goshvarpour, 2021). Significant differences in blink modulation were found in children when they viewed unpleasant pictures compared to neutral and pleasant ones (McManis et al., 2001; Waters et al., 2005), potentially due to the high arousal of unpleasant stimuli (see, e.g., Cuthbert et al., 1996).
Physiological indicators of emotional states include heart rate (Golland et al., 2014; Wallentin et al., 2011), respiration (Brouwer et al., 2015; Nummenmaa et al., 2014), skin conductance (Eisenbarth et al., 2016), muscle tension (Scheer et al., 2021), and hormone levels (Joseph et al., 2021). While adult emotional physiology is well-documented (for meta-analyses, see Kreibig, 2010; Siegel et al., 2018), fewer studies focus on children. Sacrey et al. (2021) reviewed studies using cardiac activity to measure physiological responses to emotions in children from newborns to 4-year-olds and found similar patterns to adults, such as increased HR and decreased HRV in fear, anger, or anxiety (for a meta-analysis, see Kreibig, 2010). Several studies (Koch & Pollatos, 2014; Leventon et al., 2014; McManis et al., 2001) have demonstrated that school-aged children display comparable but weaker physiological responses to emotional stimuli than adults. In general, cardiac measures are considered reliable for assessing emotional states in children (Wilhelm et al., 2006). However, when assessing age differences in physiological responses to affective stimuli, developmental changes in cardiac functioning must be considered (Harteveld et al., 2021).
These three components of emotions – subjective experiences, behavioral responses, and physiological changes – are interconnected (Mauss et al., 2005; Scherer & Moors, 2019). However, research on the development of children’s emotions has often focused on isolated single components (Fiskum et al., 2019; Gilissen et al., 2008; von Leupoldt et al., 2007). While some studies have explored combinations of two components (Davis et al., 2016; De Wied et al., 2009; Gabel et al., 2019), studies involving multicomponent approach with more than two components are rare. Notably, Leventon et al. (2014) combined self-reports, physiological measures (heart rate, heart rate variability, and event-related potentials), and behavioral memory performance, while McManis et al. (2001) assessed physiological responses (heart rate), self-report, and behavioral responses (EMG and blinks). Both studies used affective pictures to induce emotional states, although recent research suggests that dynamic, naturalistic stimuli are more effective in eliciting emotional responses (Lebert et al., 2020; Sun et al., 2020), thus increasing the ecological validity (Saarimäki, 2021).
This study investigated age-related changes in affective responses to emotionally salient stimuli, aiming to determine the extent of alignment between children's and adults' emotional reactions. This research addresses a gap in the current understanding of emotional development during middle childhood and adolescence, as most existing studies have primarily focused on adults and often limited their scope to just one or two emotional components. In contrast, our study employed a multicomponent approach, examining three modalities of emotional reactions – self-reported, physiological, and behavioral responses – in 8–15-year-olds compared to those in adults when exposed to emotional videos. We used linear mixed-effects models (LMMs) to examine whether emotional responses associated with different emotion categories vary across ages and representational similarity analysis (RSA) to investigate whether specific emotion categories are associated with distinct physiological and behavioral response patterns and how these similarities emerge across development.