Emotions

Abstract

This literature review synthesizes contemporary emotion science to integrate functional, discrete–dimensional, and appraisal–constructive accounts. A protocol-guided search of peer-reviewed work (2019–2025) with select seminal sources supports a pluralist framework: emotions coordinate salience, physiology, and action; occupy a low-dimensional affective space; and are shaped by appraisal and conceptual knowledge. Converging evidence implicates distributed neural networks, predictive interoception, and neurochemical modulation. Developmental and cultural contexts scaffold expression and regulation. At the cognition–emotion interface, attention, interpretation, memory, and decision processes route affect into behavior. Regulation effectiveness depends on contextual fit and flexibility, while dynamic signatures (variability, instability, inertia) forecast psychopathology beyond traits. Biobehavioral pathways link emotion to autonomic, immune, and inflammatory outcomes. Methodological progress rests on triangulated measures, ambulatory and naturalistic designs, and open practices. Translational targets include granularity training, interoceptive and mindfulness programs, and just-in-time adaptive supports.

Keywords: emotion; core affect; appraisal; interoception; regulatory flexibility; affective dynamics; ambulatory methods; open science.

Keltner and Cowen (2021) portray emotions as adaptive control systems that marshal attention, mobilize autonomic and endocrine readiness, and bias action tendencies, approach, avoidance, or affiliation in the service of situated goals. Cowen and Keltner (2017) map families of discrete emotions (e.g., awe, gratitude, anger) with partially distinct experiential and expressive profiles, underscoring their coordination of interpersonal and group behaviour in recurrent social challenges. Russell (2022) delineates a low-dimensional core‐affect space (valence, arousal, sometimes dominance) within which those families occupy graded regions, suggesting complementarity rather than competition between categorical and dimensional descriptions. Moors (2022) specifies appraisal operations, goal congruence, controllability, agency, and norm relevance, that transform fluctuations in core affect into context-appropriate action programs, while Hoemann, Satpute, and Barrett (2021) argue that conceptual mediation knowledge, language, and cultural scripts, categorizes bodily feelings into discrete states, explaining why identical arousal can be interpreted as anxiety in one setting and excitement in another.

Pessoa (2022) characterizes the neural substrate of emotion as distributed and dynamic, emphasizing interactions among subcortical hubs (midbrain, thalamus, amygdala) and large-scale cortical systems for salience, control, and default processing. Kragel and LaBar (2021) demonstrate that time-varying network configurations, rather than isolated “centres”, predict shifts in affect, aligning neural dynamics with functional and appraisal-based accounts. Shamay-Tsoory and Mendelsohn (2019) synthesize neurochemical modulation, proposing that dopamine, serotonin, noradrenaline, and oxytocin tune intensity, duration, and social quality of episodes. Critchley and Garfinkel (2024) integrate interoceptive and predictive perspectives, contending that the brain anticipates bodily needs and minimizes prediction error, such that feelings reflect inferences about internal state in context.

Benedek, Mier, Hajcak, Kreibig, Thayer, and Wilhelm (2023) codify methodological standards for ambulatory psychophysiology, transparent sensor specs, preprocessing, and preregistration, thereby improving reproducibility in real-world emotion measurement. Nastase, Goldstein, and Hasson (2020) advocate naturalistic paradigms (films, narratives, interactive tasks) to capture socially embedded, temporally extended affective processes that conventional stimuli miss. Chambers and Tzavella (2022) advance open-science reforms, Registered Reports, preregistration, shared data/code, to align incentives with predictive, cumulative theory, while Marek and colleagues (2022) quantify the sample sizes required for robust brain–behaviour associations. Hoemann, Wormwood, Barrett, and Quigley (2023) champion idiographic, multimodal ambulatory sensing, recasting participants as co-producers who help specify what to measure, when to sample, and how to interpret signals.

Taken together, these strands motivate a multi-level, participant-centred framework: emotions are constructed through appraisal and conceptualization, organized for functional ends, instantiated by distributed brain–body networks, and best studied with transparent, ecologically valid methods. The present review adopts this synthesis to structure the subsequent sections, moving from conceptual foundations to neural and interoceptive implementations, developmental and cultural scaffolding, cognitive–emotional interfaces, regulation and dynamics, health pathways, and translational applications.

Conceptual Foundations of Emotion: Functional, Discrete–Dimensional, and Appraisal–Constructive Accounts According to Keltner and Cowen (2021), contemporary social-functionalist accounts view emotions as adaptive responses that organize perception, physiology, and action tendencies in ways that serve situated goals and facilitate social coordination. They argue that emotions direct salience and attention, mobilize autonomic and endocrine systems for readiness, and guide behaviors such as approach, avoidance, or affiliation. In this sense, what we feel is not merely hedonic fluctuation but part of a goal-directed control system that prepares the individual for recurrent social challenges. Keltner and Cowen (2021) further emphasize that families of emotions, such as compassion or anger, are tuned to evolutionary tasks like caregiving or status negotiation, underscoring their role in coordinating interpersonal and group behavior.

            Building on this framework, Cowen and Keltner (2017) propose that emotions can be understood as discrete categories, including awe, gratitude, and anger, each associated with partially distinct experiential, expressive, and neural signatures. Their findings challenge purely unitary models of affect, suggesting instead that emotions are organized into families that are evolutionarily functional. By contrast, Russell (2022), drawing from the circumplex model, argues that emotions vary continuously along core dimensions of valence and arousal, with dominance often included as a third axis. Within this view, discrete categories such as fear or joy occupy regions within this low-dimensional affective space. Taken together, Cowen and Keltner (2017) and Russell (2022) contend that categorical and dimensional approaches should not be seen as competing models but as complementary descriptions: discrete labels highlight recurrent, evolutionarily adaptive patterns, whereas dimensions capture graded similarities and transitions among states.

Extending beyond categorical and dimensional accounts, appraisal theorists highlight the interpretive processes that transform undifferentiated affect into contextually specific emotions. Moors (2022) argues that appraisals of goal congruence, controllability, agency, and norm relevance are central in shaping which action tendencies are selected in response to a situation. This view situates emotions as flexible, context-dependent responses, rather than fixed biological outputs. Complementing this perspective, Hoemann, Satpute, and Barrett (2021) assert that emotion is conceptually mediated, meaning that prior knowledge, language, and cultural scripts play a critical role in categorizing bodily feelings into discrete states. Their findings explain why similar physiological sensations, such as increased arousal, may be labeled as anxiety in one situation but excitement in another, depending on contextual meaning and interpretive frames.

Together, these perspectives point toward a synthesis of functionalist, categorical, dimensional, and constructionist approaches. Moors (2022) and Hoemann et al. (2021) suggest that appraisal-driven meaning-making connects hedonic feeling states to eudaimonic functions, such as values and role commitments. This integration highlights that emotions are not only physiological and affective states but also psychological tools that embed subjective experience within cultural and relational contexts. The result is a pragmatic framework in which emotions are adaptive systems that flexibly coordinate individual goals with social and cultural demands.

Neuroscientific Foundations of Emotion

Neuroscience research has increasingly demonstrated that emotions arise from distributed circuits rather than isolated “centers.” Pessoa (2022) argued that emotional processes depend on interactions between subcortical hubs including the midbrain, thalamus, and amygdala and cortical regions that integrate salience, executive control, and default-mode functions. This analysis highlights that emotions coordinate bottom-up detection of salient cues with top-down regulation, thereby linking perception, physiology, and behavior to situational demands.

Expanding on this, Kragel and LaBar (2021) proposed that emotions are best understood as emergent patterns of connectivity across large-scale brain networks. Rather than treating emotion as the output of discrete modules, they found that the dynamic interplay of the salience network, the frontoparietal control network, and the default-mode network predicted shifts in affective states. Their evidence supported the view that emotions are temporally fluid phenomena, defined by the flexible reconfiguration of neural networks across contexts.

In addition, neurochemical research has demonstrated how modulators fine-tune these circuits. Shamay-Tsoory and Mendelsohn (2019) reviewed findings showing that dopamine enhances reward sensitivity and motivational drive, serotonin regulates mood and inhibitory control, noradrenaline calibrates vigilance and stress reactivity, and oxytocin facilitates social bonding and affiliation. They argued that these neurochemical systems adjust the responsiveness of neural circuits, thereby shaping the intensity, duration, and interpersonal qualities of emotional episodes.

Taken together, Pessoa (2022), Kragel and LaBar (2021), and Shamay-Tsoory and Mendelsohn (2019) provide converging evidence that emotions should be interoceptionized as distributed, dynamic, and chemically calibrated processes. This synthesis supports the claim that emotional life is neither the product of a single brain locus nor reducible to subjective feeling states, but instead reflects the coordinated operation of circuits, networks, and modulators that allow humans to adapt to complex social and environmental demands.

Interoception, Predictive Brain Accounts, and Individual Differences

According to Quigley and colleagues (2021), contemporary interoceptive frameworks argue that perception of bodily signals from the heart, lungs, and viscera dynamically shapes affective tone and scaffolds the construction of discrete emotions by informing salience, arousal, and perceived bodily needs. Building on this, Kiverstein et al. (2025) analyze predictive‐processing models and contend that the brain does not merely register internal sensations but actively forecasts interoceptive inputs, updating beliefs to minimize prediction error; under this view, emotion emerges as the brain’s best current “explanation” for ongoing bodily states in context. 

Extending these claims, Barrett and colleagues (2023) and allied “predictive allostatic interoception” accounts argue that cortical–subcortical loops (including midbrain, thalamus, insula, and prefrontal regions) coordinate anticipatory regulation of bodily energy needs, such that reappraisal and action both serve to reduce future interoceptive surprises. In line with this, Daly et al. (2024) review emerging “interoceptive technologies” and argue that neuromodulatory systems (e.g., noradrenergic, dopaminergic, and serotonergic pathways) tune interoceptive precision, thereby calibrating vigilance, reward sensitivity, and stress reactivity that underwrite emotional episodes and their regulation. 

From an individual‐differences perspective, Quigley et al. (2021) further argue that variability in interoceptive accuracy and awareness helps explain why some people experience more intense emotions, regulate more effectively, or show greater vulnerability to dysregulation under stress; in their account, metacognitive access to bodily cues supports more adaptive emotion construction and control. Complementing this, Kiverstein et al. (2025) maintain that when interoceptive predictions are imprecise or overweighted, the system may misattribute bodily sensations to threat or negative affect, amplifying anxiety and mood symptoms, an inference that integrates predictive coding with clinical observations of dysregulated feeling states. 

Conceptual Skills of Emotion: Granularity, Language, and Mixed Feelings

According to Heiy and Cheavens (2021), the emotion‐granularity literature indicates that being able to differentiate one’s feelings with finer conceptual detail (e.g., “irritated” vs. “enraged” vs. “anxious”) supports adaptive regulation, clearer social communication, and resilience in daily life, in part because precise labels map onto more targeted coping strategies. Consistent with this skills view, recent cohort and clinical studies summarized by Heiy and Cheavens (2021) report that higher granularity predicts fewer maladaptive behaviors under stress and better moment-to-moment adjustment when negative affect rises. 

Language‐based approaches add a mechanistic layer: Nook, Parhami, and Somerville (2021) show that simply labeling feelings can alter subsequent experience and regulation strategy use, while also warning that the timing and context of labeling matter (e.g., premature labeling can impede later reappraisal when individuals immediately “explain away” arousal). In parallel, Torre and Lieberman’s earlier synthesis—often cited as foundational—argues that affect labeling engages prefrontal control systems and can dampen limbic responses, offering a pathway by which words change physiology and experience; this functional account has motivated newer tests of when labeling facilitates, versus competes with, other regulation tactics. (Foundational/seminal, pre-2019.)

Finally, mixed-emotion research expands beyond “pure” states. Ong et al. (2020) review cross-cultural evidence and argue that coactivation of positive and negative states is common—especially in dialectical cultures—and can promote flexible coping when individuals conceptually integrate tension (e.g., feeling grateful and sad during a farewell). According to this perspective, tolerating and naming mixed feelings broadens regulatory options (e.g., problem solving alongside acceptance), aligning with granularity findings that richer emotional vocabularies and concepts scaffold more nuanced and effective self-regulation.

Developmental Trajectories of Emotion

According to Cole, Lougheed, and Ram (2020), developmental accounts emphasize that caregiver scaffolding and social learning play a foundational role in shaping children’s early abilities to recognize, express, and regulate emotion. They argue that practices such as emotion labeling, validation, and modeling by parents and caregivers provide the structure within which infants and young children learn to map bodily arousal and situational cues onto culturally appropriate emotional categories and behaviors. In this way, early family environments form the prototype for later self-regulatory strategies and socioemotional competence.

Adolescence, by contrast, represents a period of heightened variability in affective experience. According to Pfeifer and Allen (2021), adolescent affective systems undergo recalibration due to neurodevelopmental changes in limbic and prefrontal regions, leading to stronger reactivity to social evaluation and peer feedback. They argue that this heightened salience of peer contexts increases both risk (e.g., susceptibility to negative peer influence, emotional volatility) and opportunity (e.g., accelerated acquisition of regulatory and empathic skills). McRae et al. (2020) add that adolescence is also a critical window for experimenting with diverse regulation strategies, such as reappraisal and suppression, whose effectiveness depends on individual and contextual factors.

From a lifespan perspective, socioemotional selectivity theory has been central. According to Carstensen, Isaacowitz, and Charles (2020), older adults increasingly prioritize emotionally meaningful goals over exploratory or novelty-seeking goals as time horizons narrow. This motivational shift is expressed in what is known as the “positivity effect”: compared to younger adults, older adults preferentially attend to, remember, and savor positive over negative information. Reed and Carstensen (2021) further argue that these age-related shifts are adaptive, allowing older individuals to regulate affect more effectively and maintain well-being in the face of losses and declining resources. Together, these findings demonstrate that emotion is not a static faculty but a developmental trajectory, dynamically shaped by caregiving, peer contexts, and motivational priorities across the lifespan.

Social and Moral Functions of Emotion

According to Keltner and Gross (2021), social-functional theories emphasize that emotions are not only intrapersonal experiences but also serve as interpersonal signals that coordinate action within groups. They argue that expressive cues such as facial expressions, vocal tones, and gestures communicate intentions, reduce uncertainty, and enforce group norms, thereby allowing emotions to act as social regulators. From this perspective, anger may deter norm violations, gratitude may strengthen cooperative bonds, and embarrassment may signal appeasement, demonstrating how emotions contribute to group cohesion and stability.

Moral emotions provide an especially important domain of inquiry. According to Tangney, Stuewig, and Martinez (2020), guilt and shame, while often conflated, exert distinct effects on moral judgment and prosocial behavior: guilt motivates reparative action, while shame can produce withdrawal or defensive responses. In contrast, pride functions as an affiliative signal of achievement and competence, reinforcing social hierarchies in constructive ways when expressed authentically. Haidt and Algoe (2020) add that other-oriented emotions such as elevation and compassion encourage altruism and prosocial action, underscoring the moral scaffolding that emotions provide for collective life.

At the collective level, group-based emotion research has expanded these insights. According to Smith, Seger, and Mackie (2020), group identification processes generate shared emotional experiences such as anger, fear, or hope, which in turn shape intergroup attitudes and mobilize collective action. For example, shared anger may catalyze protest movements, while shared hope fosters long-term commitment to social change. Mackie, Devos, and Smith (2021) argue that such collective emotions are not merely aggregates of individual feelings but are socially constructed through identification with ingroups, leaders, and symbolic narratives. Taken together, these perspectives reveal that emotions function as a bridge between individual psychology and collective life, guiding moral behavior, regulating cooperation, and shaping intergroup relations.

Cultural and Contextual Perspectives on Emotion

According to Tsai (2020), cultural psychology demonstrates that ideals about which emotions are desirable—or what she terms “ideal affect”—vary across societies, shaping emotional experience and expression. For instance, Western individualist cultures often prioritize high-arousal positive emotions such as excitement, whereas East Asian collectivist contexts emphasize calm, low-arousal states such as harmony. Tsai argues that these differences, reinforced through culturally specific display rules, influence not only how emotions are expressed but also how they are felt, with implications for well-being and interpersonal functioning.

Building on this, socioecological models expand the cultural lens by linking ecological conditions to emotional norms. According to Boiger and Mesquita (2021), factors such as subsistence strategies, population density, and exposure to threat shape how communities socialize and regulate emotion. They note that societies with high interdependence and ecological threats often cultivate emotional restraint and conflict avoidance, while more mobile or resource-rich contexts encourage open emotional expression and assertiveness. This ecological framing positions emotional life as an adaptation to collective survival demands rather than as a universal, fixed process.

Acculturation studies further highlight the dynamic flexibility of emotion in contexts of migration and cultural mixing. De Leersnyder, Mesquita, and Kim (2020) argue that bilingualism and biculturalism shape not only emotion vocabulary but also regulation strategies, with bilingual individuals often shifting emotional expressivity depending on linguistic and cultural context. Similarly, Güngör et al. (2021) show that migration histories produce hybrid emotional repertoires, where individuals draw upon multiple cultural norms to regulate and express feelings. These findings underscore that emotion is neither solely biologically predetermined nor universally expressed but is profoundly shaped by cultural learning, ecological pressures, and the lived realities of intercultural exchange.

Cognitive–Emotional Interfaces

According to Cisler and Koster (2019), cognitive–emotion research has demonstrated that attention biases, such as hypervigilance toward threat cues, tend to amplify affective responding in real time, heightening anxiety and reactivity. By contrast, interpretive styles that encourage positive or benign construals can dampen negative affect, illustrating how cognition dynamically modulates emotional intensity. Everaert et al. (2020) further analysed that maladaptive interpretive tendencies, such as catastrophic thinking, reinforce mood disturbances and emotional dysregulation, whereas flexible reappraisal strategies foster resilience. Together, these findings suggest that attentional and interpretive processes are not merely downstream consequences of emotion but active mechanisms that shape affective trajectories as events unfold. Building on this, Kensinger and Ford (2020) asserted that memory research consistently shows emotional arousal prioritizes consolidation of goal-relevant cues while simultaneously distorting peripheral details. This effect is adaptive in that it enhances retention of survival-relevant information but can introduce biases, such as tunnel memory during high arousal. According to Yonelinas and Ritchey (2021), such prioritization is linked to amygdala–hippocampal interactions, whereby emotional salience strengthens memory traces for central features but weakens recall of neutral or contextual information. These dynamics illustrate how emotion shapes not only what is remembered but also how accurately it is encoded and retrieved.

Finally, decision-science perspectives argue that specific emotions channel cognitive appraisals into patterned choices. Lerner et al. (2015) originally advanced the appraisal-tendency framework, which claims that emotions such as anger and fear guide decision-making in ways that extend beyond simple “positive versus negative” valence. More recently, FeldmanHall and Shenhav (2019) demonstrated that anger increases risk-seeking and punishment decisions, while fear enhances avoidance and risk aversion. Similarly, Xu et al. (2021) provided evidence that trust-related decisions are strongly shaped by discrete emotions, with gratitude promoting cooperation and envy dampening prosociality. Collectively, these findings reinforce that cognitive–emotional interfaces govern moment-to-moment choices, integrating appraisal, memory, and attentional mechanisms into behavioral outcomes.

Emotion Regulation

According to Gross (2015), regulation science traditionally identifies four broad families of strategies: situation selection, attentional deployment, cognitive change (such as reappraisal), and response modulation. More recent evidence has extended this model, showing that individuals flexibly combine strategies across time to manage emotional demands. For example, Gross and Feldman Barrett (2021) argue that regulation is not merely about suppressing negative affect but about selecting strategies that optimize functioning, including both hedonic comfort and eudaimonic goal pursuit. Webb et al. (2021) further analysed that cognitive reappraisal consistently predicts positive outcomes such as resilience and well-being, whereas response suppression is associated with maladaptive physiological and interpersonal costs. This body of research demonstrates that regulation is a dynamic and multifaceted process rather than a simple dichotomy of “good” versus “bad” strategies.

Building on this framework, person-by-context models emphasise that the effectiveness of emotion regulation depends critically on fit. Aldao et al. (2019) argued that the same strategy can be adaptive in one context but harmful in another, depending on timing, emotional intensity, and an individual’s goals. For instance, Sheppes et al. (2014) found that distraction may be effective in down-regulating high-intensity distress but counterproductive when used chronically to avoid manageable challenges. Ford et al. (2021) further asserted that flexibility—the ability to switch strategies based on situational demands is a stronger predictor of mental health than reliance on any one strategy alone. Together, these findings reveal that regulation success is best understood as a function of contextual appropriateness rather than the strategy in isolation.

In addition to intrapersonal regulation, researchers highlight the central role of interpersonal regulation in everyday affect management. According to Zaki and Williams (2013), humans routinely co-regulate emotions through mechanisms such as seeking support, shared attention, synchrony, and social touch. More recent work by Reeck et al. (2020) demonstrates that interpersonal regulation is bidirectional: individuals not only seek help but also provide it, with dyadic interactions shaping emotional outcomes for both parties. Williams et al. (2019) add that the quality of interpersonal regulation depends on relational trust and sensitivity, as poorly timed or mismatched support can exacerbate distress rather than alleviate it. These findings suggest that emotion regulation is not solely an individual enterprise but a socially embedded process that sustains well-being and resilience across the lifespan.

Affective Dynamics and Regulatory Flexibility: Variability, Instability, Inertia, and Daily

Trajectories as Predictors of Psychopathology

Affective-dynamics research delineates variability (within-person dispersion), instability (magnitude of successive shifts), and inertia (autoregressive carry-over) as separable signatures, each differentially associated with health and psychopathology; Houben, Van Den Noortgate, and Kuppens (2015) synthesize evidence that elevated variability relates to poorer well-being in several contexts, Koval, Pe, Meers, and Kuppens (2019) report that greater inertia prospectively indexes depressive risk, and Trull, Lane, Koval, and Ebner-Priemer (2015) document pronounced affective instability as a transdiagnostic marker, together indicating that “how emotions move” is as prognostic as “how emotions feel.”

Flexibility accounts propose that context-sensitive switching among goals and strategies—not sheer frequency of any one tactic—predicts adjustment; Bonanno and Burton (2013) introduce “regulatory flexibility” as the capacity to match strategies to situational demands, Aldao, Sheppes, and Gross (2015) argue that effectiveness hinges on timing, intensity, and goal fit, and Ford, Gross, and colleagues (2021) show that flexible deployment of reappraisal, acceptance, and distraction outperforms rigid preference patterns in predicting mental health.

Experience-sampling studies indicate that day-to-day emotional trajectories forecast future symptoms beyond static trait measures; Kuppens, Oravecz, and Tuerlinckx (2010) demonstrate that higher emotional inertia predicts subsequent depressive symptomatology, van Roekel, Keijsers, and Chung (2022) meta-analytically confirm that dynamic indices add incremental validity over trait affect, and McKone and Silk (2022) highlight adolescent cohorts where short-term fluctuations anticipate later internalizing problems, underscoring the predictive yield of intensive longitudinal designs for early detection and intervention.

Measurement and Methodological Standards in Emotion Research: Triangulation, Ambulatory Assessment, and Naturalistic Paradigms 

Gold-standard emotion measurement triangulates self-report with behavior and psychophysiology; Mauss and Robinson (2009) established that cross-channel correspondence is modest and condition-dependent, motivating multi-method designs, whereas Benedek et al. (2023) codify transparent reporting for ambulatory psychophysiology (e.g., sensor specs, preprocessing, preregistration) to curb method variance and enhance reproducibility.

Ambulatory methods extend measurement into daily life; Al’Abri et al. (2023) review multimodal wearables (HRV, EDA, movement, voice) and passive smartphone streams as routes to higher ecological validity when paired with rigorous pipelines and out-of-sample validation, while Huckins et al. (2020) demonstrate that well-specified digital phenotypes can track real-time affective states provided artifacts, missingness, and person-specific baselines are explicitly modeled.

Naturalistic paradigms capture socially embedded, temporally extended emotions; Nastase, Goldstein, and Hasson (2020) argue that films and real-world tasks elicit richer, inter-subject-synchronized dynamics than static stimuli, and Hartling et al. (2021) show that “film-fMRI” improves the mapping of evolving affective processes, with VR and everyday interactions further enabling second-person, context-sensitive assessments that better approximate lived emotional experience.

Transdiagnostic Emotion Dysregulation and Reward Disturbances Across Psychopathology 

Transdiagnostic models conceptualize dysregulated processes, rumination, worry, experiential avoidance, and maladaptive safety behaviors, as common mechanisms cutting across anxiety, depression, and stress-related disorders; Mennin and Fresco (2020) articulate how heightened threat reactivity and deficits in context-appropriate regulation maintain broad-spectrum distress, while McEvoy, Hyett, Shihata, Price, and Strachan (2019) synthesize evidence that targeting these shared processes yields cross-diagnostic symptom improvement.

Personality–psychopathology research identifies rapid affect shifts and elevated instability as phenotypic signatures that necessitate tailored intervention; Trull, Lane, Koval, and Ebner-Priemer (2015) demonstrate pronounced moment-to-moment lability in borderline and related conditions using intensive longitudinal designs, and Koval, Pe, Meers, and Kuppens (2019) show that affective inertia and volatility differentially predict internalizing risk, implicating precision assessment of temporal dynamics for treatment planning.

Reward-processing accounts parse anhedonia into disrupted anticipatory (“wanting”) and consummatory (“liking”) components with motivational consequences; Halahakoon, Kieslich, O’Driscoll, Nair, and Wilkinson (2020) report blunted reinforcement learning and effort-based decision-making in depression, while Pizzagalli (2021) integrates behavioral, computational, and neural findings to argue that mesocorticolimbic dysfunction undermines goal pursuit and engagement even when hedonic capacity is partially intact; Berridge and Kringelbach (2015) provide seminal neurobiological distinctions between incentive salience and hedonic impact that inform contemporary clinical models.

Biobehavioral Pathways from Emotion to Health: Psychophysiology, Positive Affect, and Allostatic Recovery

 Psychophysiology indicates that autonomic and endocrine pathways link emotion to cardiovascular, immune, and inflammatory outcomes; Slavich (2020) situates affective states within a social–biological cascade that modulates HPA-axis activity and inflammatory signaling (e.g., IL-6, CRP), while Guidi et al. (2021) frame these alterations as markers of allostatic load with consequences for cardiometabolic risk; Thayer and Lane (2000, seminal) further propose neurovisceral integration, in which vagal regulation (indexed via HRV) reflects flexible, adaptive emotion–physiology coupling.

Resilience research suggests that positive emotions broaden attentional scope and build durable coping resources over time; Fredrickson (2021) updates broaden-and-build theory to show how micro-moments of positive affect accumulate social, cognitive, and physiological reserves, and Kalisch et al. (2020) argue that positive appraisal styles cultivate resilience by recalibrating threat evaluations and sustaining goal pursuit under stress; Cregg and Cheavens (2021) add that cultivating discrete positive states (e.g., gratitude) confers small-to-moderate benefits for well-being and stress reactivity.

Stress frameworks emphasize that recovery speed and recalibration, rather than peak reactivity alone, index healthy regulation; Guidi et al. (2021) contend that faster autonomic and endocrine down-regulation after challenge signals lower allostatic load and better long-term health, while Kalisch et al. (2020) highlight flexible re-appraisal and rapid return to baseline as core resilience mechanisms; Fredrickson (2021) notes that positive affect facilitates quicker cardiovascular and affective recovery, supporting adaptive reset following stressors.

Social Emotions in Moral Life: Compassion, Outrage, and Collective Affect in Crises 

Empathy–compassion research differentiates compassion (other-focused care that sustains prosocial action) from empathic distress (self-focused vicarious pain that can deplete resources and prompt withdrawal); contemporary synthesis places compassion within interpersonal emotion-regulation goals that optimize others’ functioning rather than merely sharing their feelings (Zaki, 2020; Singer & Klimecki, 2014). 

Work on moral anger/outrage shows it can motivate norm enforcement and cooperation, yet online environments amplify and distort it: reinforcement and norm-learning processes increase outrage expression over time, and observers overperceive others’ outrage, inflating perceived intergroup hostility and contributing to affective polarization (Brady, McLoughlin, Doan, & Crockett, 2021; Brady, McLoughlin, Torres, Luo, Gendron, & Crockett, 2023). These dynamics indicate that platform norms and feedback contingencies shape whether anger supports accountability or escalates division. 

Collective-emotion studies during public-health crises show that socially transmitted fear and hope shape risk perception and compliance: broad reviews integrate evidence on threat navigation and prosocial motives, experiments demonstrate that empathy increases distancing and masking, and messaging work suggests hope-based appeals can promote intentions without the costs of excessive fear (van Bavel et al., 2020; Pfattheicher, Nockur, Böhm, Sassenrath, & Petersen, 2020; Chou & Budenz, 2024).

Embodiment, Action Tendencies, and Motor Circuits: From Bodily Feedback to Affect-to-Action Mapping

Embodiment accounts position the body as an active substrate of affective experience, not a passive readout. Coles, March, and colleagues (2022) reported that experimentally manipulated facial muscle activity yields modest but reliable shifts in felt affect, consistent with the facial-feedback hypothesis when measurement and task constraints are carefully controlled. Reed, Moody, Mgrublian, Assaad, Schey, and McIntosh (2020) further demonstrated that constraining posture and movement selectively impairs the production and recognition of status-related emotions, indicating that proprioceptive and postural inputs participate in constructing both what is felt and what is perceived. Building on such evidence, Hauke, Krampe, and co-authors (2024) argued that embodied and enactive principles can be translated into clinical practice, where sensorimotor engagement (e.g., breath, stance, gesture) modulates affect via bottom-up channels that are mechanistically distinct from purely cognitive reappraisal. Collectively, these findings indicate that sensorimotor feedback, facial, postural, and interoceptive, helps constitute the phenomenal quality and communicative efficacy of emotion, thereby linking the “feel” of affect to the way bodies prepare for action in context.

Action-tendency theories treat emotions as momentum for action selection—biasing approach versus avoidance, freezing versus mobilization, or affiliation versus aggression. Bramson, Toni, and Roelofs (2023) contended that emotion regulation is, at its core, a problem of action control: predictive, feed-forward motor programs constrain which regulatory options become available at millisecond timescales, and strategy choice reflects tuning of action policies rather than only top-down appraisal. Eder (2023) extended this view by proposing a perceptual-control architecture in which emotional feelings index deviations from desired states and thereby recalibrate the selection of instrumental actions under changing constraints. Krell and Gibson (2022) added that medial prefrontal cortex implements cost–benefit integration over candidate acts through its recurrent loops with basal ganglia–thalamic circuits, integrating value, threat, and goal priors. Across these perspectives, flexibility is operationalized as the capacity to retune action tendencies to shifting contingencies, not the mere frequency of any single strategy.

Motor-systems research specifies how premotor and basal-ganglia loops implement the rapid mapping from affect to action under uncertainty. Yoshida, Oñate, Khatami, Vera, Nadim, and Khodakhah (2022) reviewed converging evidence that cerebellar outputs project to dopaminergic midbrain and striatal targets, enabling reward-sensitive modulation of movement vigor and learning; this provides a substrate by which affective value shapes action kinematics. Thomasson, Benis, Voruz, Saj, Vérin, Assal, Grandjean, and Péron (2022) showed that lesions affecting basal ganglia and cerebellum differentially degrade decoding of emotional prosody, implying that these motor circuits contribute to affective signal selection beyond pure motor control. Gilbertson and Steele (2021) modeled tonic dopamine as a uncertainty-sensitive gain on basal-ganglia selectivity, predicting exploratory versus exploitative choices as a function of environmental volatility—a mechanism that directly links arousal-like states to the precision of action selection. Together, these data indicate that premotor–striatal–cerebellar loops instantiate fast, probabilistic policies through which affective value and contextual uncertainty are converted into concrete, adaptive movement plans.

Participant-Centered Methods and Idiographic Analytics: Co-Design, Personalized Networks, and Co-Production 

Methodological reforms recast participants as active agents in the scientific workflow, co-specifying what to measure, when to sample, and how data are interpreted. Springstein and English (2023) demonstrate in an experience-sampling and mobile-sensing study that every stage of the regulation process (goals, motives, strategy choice, and perceived success) is idiographically contingent on person-specific situational features, motivating designs that adapt prompts to individuals’ routines and priorities rather than enforcing uniform schedules. Hoemann, Wormwood, Barrett, and Quigley (2023) further argue that biologically triggered, multimodal ambulatory sensing (e.g., heart rate, movement, voice) enables “in-the-wild” sampling at moments when physiology indicates meaningful change, thereby aligning the timing of assessment with the participant’s lived context and reducing recall bias. In this participant-centered model, individuals are not passive data sources but co-producers of dense longitudinal records, contributing idiographic labels, contextual annotations, and preference constraints that directly shape model specification and inference.

Person-specific analytics move beyond group means to individualized networks that reveal idiosyncratic trigger–strategy–outcome couplings. Chen, Wang, Wang, and Irish (2023) combine experience sampling with temporal network analysis to show that emotion-specific inertia and the efficacy of particular regulation strategies vary meaningfully within persons, clarifying why population-level effects often mask clinically relevant heterogeneity. Kullar, Carter, Hitchcock, Whittaker, Wright, and Dalgleish (2024) report that dynamic profiles of emotional interrelations are orthogonal to diagnostic categories, identifying data-driven subgroups that cut across conventional labels and thereby supporting precision assessment. Complementing these findings, Rodriguez, Aalbers, and McNally (2022) illustrate that idiographic models routinely diverge from aggregate networks in both sign and strength of associations, implying that valid prediction and intervention require within-person estimation of the mechanisms that maintain or relieve distress in daily life.

Participatory approaches extend methodology into the ethical and translational domain by embedding co-production with people who have lived experience. Heap, Jennings, Mathias, and colleagues (2022) outline realist protocols for participatory mental-health interventions that position community partners at each stage, problem formulation, instrument design, recruitment, interpretation, and dissemination, thereby improving ecological validity, buy-in, and cultural safety. Hoemann et al. (2023) likewise emphasize that participant input into sensing choices (which signals to collect, how often, and with what privacy safeguards) is not ancillary but constitutive of data quality, since burden, feasibility, and meaning-making jointly determine compliance and the representativeness of affective episodes captured in the wild. Taken together, these strands indicate that the gold standard for contemporary emotion science integrates participant co-design, idiographic modeling, and community co-production to maximize validity, ethics, and translational impact.

Language, Metaphor, and Digital Vernaculars: How Symbol Systems Shape Emotional Experience and Transmission

Bilingual context and regulation latitude. Bilingual context switching reorganizes the labeling space of emotion (what words feel available or precise) and the regulatory repertoire (what strategies feel normatively permitted). Operating in a second language often introduces greater psychological distance, reduced visceral immediacy and heightened monitoring, which can blunt impulsive responses and widen the window for reappraisal, distancing, or problem solving. Conversely, returning to the first language can re-intensify affect via richer semantic nuance, autobiographical ties, and culturally shared display rules. Code-switching thereby functions as a self-selected context shift: individuals leverage language choice to align emotion norms (e.g., restraint vs. expressivity), recalibrate social meaning (e.g., what counts as “rude,” “assertive,” or “caring”), and fine-tune interpersonal goals (affiliation, autonomy, face-saving). Clinically and educationally, scaffolding dual-language labeling (e.g., naming the same episode in both languages) increases granularity and reveals cross-linguistic “blind spots,” while explicitly discussing language-specific norms prevents misattributions of coldness or overreactivity that are, in fact, artifacts of linguistic context.

Conceptual metaphor and intervention framing. Conceptual metaphors, anger is heat/pressure, sadness is down/weight, awe is vastness, do more than decorate speech; they structure causal attribution and choice of remedy. When anger is framed as pressure, people preferentially endorse venting or release; when anger is framed as a signal of violated values, they favor boundary-setting and repair. Likewise, describing sadness as heavy cues solutions that lighten the load (delegation, social support), whereas construing it as weather promotes acceptance-based coping that “lets it pass.” In public communication, metaphor selection alters policy support (e.g., crime as virus invites prevention; crime as beast invites containment), and in psychotherapy, re-metaphoring can gently reorient clients from maladaptive schemas (e.g., I’m broken) to agentic ones (e.g., I’m rebuilding). Metaphors thus operate as cognitive control surfaces: they compress complexity into actionable levers, but they also risk essentializing emotions if taken literally, so best practice is to use plural metaphors that foreground different facets (physiology, appraisal, social function) and to test whether a given metaphor actually improves regulation for this person, here and now.

Digital vernaculars, emojis, and online contagion. In text-dominant media, emergent vernaculars, emojis, GIFs, reaction badges, punctuation play, and stylized typography—serve as para-linguistic prosthetics, reinstating cues that speech and face normally provide (prosody, gaze, timing). Emojis disambiguate intent (teasing vs. taunt), signal arousaland valence, and coordinate group tone; reaction metrics (likes, boosts) create social proof that can accelerate emotional contagion and set platform-specific display rules (what counts as supportive, outraged, playful). Yet these systems are context- and culture-sensitive: the same icon can invert meaning across communities or platforms, and algorithmic curation can amplify high-arousal affect (outrage, fear) over low-arousal states (contentment), thereby skewing collective mood. Methodologically, studying digital emotion requires multimodal traces (text, timing, network position) and preregistered coding schemes to separate genuine contagion from homophily and simultaneous exposure. Practically, designing friction for escalation (e.g., prompts before sharing unverified content), enabling nuanced reactions (beyond binary like/dislike), and cultivating community norms for repair (apologies, de-escalation rituals) can throttle the spread of misattuned affect and support healthier emotional climates online.

Affective Computing for Emotion: Multimodal Inference, Ethics of Deployment, and Privacy-by-Design Safeguards 

Multimodal pipelines outperform single-channel inference: Atmaja and Akagi (2022) review convergent evidence that fusing physiology (e.g., heart-rate variability, electrodermal activity), face/voice (facial actions, prosody, spectral–temporal voice features), and context (task, environment, interaction partner) yields more stable emotion estimates than any single stream, particularly under domain shift and noisy, in-the-wild conditions. Al’Abri, Al Hinai, Al Wardi, and Al Abri (2023) similarly report that wearable-plus-smartphone pipelines achieve higher accuracy and generalizability when models integrate cross-signals and include robust preprocessing and validation (e.g., cross-participant, cross-context), whereas uni-modal systems degrade steeply with motion artifacts, lighting, microphone distance, or demographic variability. In practice, late-fusion and attention-based fusion architectures reduce overfitting by allowing each modality to contribute when informative and to down-weight when unreliable, improving both performance and interpretability in ecologically valid settings (Atmaja & Akagi, 2022).

Ethics of emotion-AI: bias, consent, and downstream harms. Stark and Hutson (2021) caution that commercial emotion-recognition systems risk physiognomic fallacies—over-interpreting faces and voices as direct readouts of internal states—raising concerns about construct validity, demographic bias, and stigmatizing uses in employment, education, or policing. Barrett, Adolphs, Marsella, Martinez, and Pollak (2019) further argue that inferring specific emotions from static facial configurations is scientifically contested, which implies that deployments must avoid high-stakes decisions predicated on single-channel cues. Ethics reviews emphasize informed consent that is meaningfully revocable, clear documentation of intended use versus prohibited contexts, bias auditing across subgroups, and impact assessments that anticipate chilling effects, surveillance creep, or disproportionate false positives for marginalized populations (Stark & Hutson, 2021). In short, responsible emotion-AI requires limits on inference, multi-modal corroboration, and explicit governance for data collection, labeling, and application domains.

Privacy-by-design: on-device processing, differential privacy, and federated learning. Rieke, Hancox, Li, and colleagues (2020) outline how on-device (edge) inference minimizes raw-data exfiltration, retaining sensitive audio, video, and physiological signals locally and transmitting only compact embeddings or decisions. Kairouz et al. (2021) survey federated learning as an additional safeguard: models are trained across user devices with secure aggregation, reducing centralized access to raw data while still enabling population-level improvements; this approach can be paired with differential privacy noise to bound what can be reverse-engineered about any individual. Foundational work on differential privacy (Dwork, 2006) and its adaptation to deep learning (Abadi et al., 2016) demonstrates formal guarantees against membership and attribute inference, though reviews caution about utility–privacy trade-offs and the need for domain-specific threat models (Kairouz et al., 2021). Together, privacy-by-design measures—edge inference, federated optimization, and differentially private updates—create layered protections that allow affective-computing research while respecting participant autonomy and minimizing re-identification risk.

Applied Emotion Skills and Tech-Enabled Support: Granularity Training, Interoceptive/Mindfulness Protocols, and Just-in-Time Delivery

  Skills training: Erbas, Gendron, and colleagues (2022) characterize emotion granularity as a modifiable competence, learning to parse “bad” into anger, fear, guilt, or sadness with situational precision, which subsequently scaffolds strategy selection (e.g., boundary-setting for anger vs. problem-solving for worry) and reduces globalized distress. Mattingly and Kraiger (2019) report, in a meta-analysis of emotional-competence training, reliable gains in recognition, labeling, and perspective-taking that translate into improved interpersonal performance and stress regulation. Nook, Sasse, Lambert, and Somerville (2020) further show that more nuanced linguistic labeling of felt states covaries with better daily regulation, while Lieberman, Inagaki, Tabibnia, and Crockett (2007), in seminal work, demonstrate that affect labeling dampens amygdala reactivity and recruits prefrontal control, an effect leveraged in contemporary brief labeling exercises embedded in coaching and psychotherapy. Cregg and Cheavens (2021) add that targeted practice in perspective-taking and discrete positive states (e.g., gratitude) yields small-to-moderate improvements in well-being and interpersonal functioning, consistent with the view that skills training builds a transferable toolkit rather than a single “best” strategy.

Interoceptive and mindfulness programs:  Critchley and Garfinkel (2024) synthesize evidence that interoceptive training, cultivating awareness and calibration of heart, breath, and visceral cues, re-tunes predictive models of bodily state, thereby reducing maladaptive reactivity and clarifying which emotions are actually present. Li, Zhang, Liu, and colleagues (2024) meta-analytically confirm that mindfulness-based interventions produce short-to-moderate reductions in distress and improvements in well-being among health-care staff, with dosage and practice quality moderating effects. Hoge, Ivkovic, Friehs, and co-authors (2022) show in a randomized clinical trial that mindfulness-based stress reduction is non-inferior to escitalopram for primary anxiety disorders, indicating clinical potency without pharmacologic side-effects for many patients. Galante, Dufour, Vainre, and colleagues (2021) further document campus-wide gains in mental health and functioning from scalable mindfulness courses, supporting population-level deployment when programs include fidelity safeguards and active practice.

Just-in-time interventions (JITAIs): Liao, Klasnja, Tewari, and Murphy (2020) formalize micro-randomized trials for testing when and for whom context-aware prompts should be delivered, enabling principled optimization of timing, content, and burden. Bidargaddi, Musiat, Makelberge, and co-authors (2020) report that digital phenotypingsignals (e.g., sleep regularity, mobility, phone use) can identify high-risk windows and trigger on-device coping prompts (breathwork, reappraisal, social contact) with improved adherence relative to fixed schedules. Nahum-Shani, Smith, Spring, and colleagues (2021) argue that state-tailored JITAIs outperform one-size-fits-all apps because they adapt to dynamic states (stress level, location, social context) and respect user preferences, thereby minimizing notification fatigue. Collectively, these strands converge on a translational model where skills are built offline (granularity, labeling, perspective-taking), stabilized through body-based practice (interoception and mindfulness), and supported in vivo by JITAIs that deliver the right strategy at the right moment with safeguards for privacy and feasibility.

Integrating Emotion Theories, Raising Evidentiary Standards, and Advancing Naturalistic, Participant-Centered Methods 

Integrative models: Moors (2022) contends that a multi-level account of emotion emerges when appraisal operations (e.g., goal congruence, controllability, agency, norm relevance) are situated within broader functional architectures that specify how emotions coordinate perception, physiology, and action. Russell (2022) maintains that core affect provides a low-dimensional scaffold (valence, arousal, sometimes dominance) upon which culturally learned concepts, language, and situational appraisals construct context-specific experiences, thereby reconciling dimensional substrates with discrete labels. Scherer (2022) argues that component-process formulations formalize this synthesis by modeling emotion episodes as coordinated changes across appraisal, expression, autonomic/endocrine activity, action tendencies, and subjective feeling, each modulated by person × culture contingencies. Scarantino (2020) further proposes a pluralist framework in which constructionist, appraisal, and functionalist claims are treated as complementary lensesmapping different levels of organization—from interoceptive/bodily dynamics to social signaling and norm governance—yielding a coherent pathway from body to culture without privileging a single “master” theory.

Open-science practices: Chambers and Tzavella (2022) argue that Registered Reports, preregistration, and open data/code raise the evidentiary bar by aligning incentives with prediction, not post-hoc storytelling, thereby reducing selective reporting and analytic flexibility. Yarkoni (2020) emphasizes that reproducibility in emotion research requires models that actually generalize beyond the dataset at hand, pressing for larger samples, out-of-sample validation, and theory-constrained predictions. Botvinik-Nezer et al. (2020) demonstrate substantial analytic variability in fMRI pipelines applied to the same dataset, underscoring the need for transparent specifications, shared code, and multiverse analyses when linking neural signals to affective constructs. Marek et al. (2022) show that robust brain–behavior associations often require thousands of participants, implying that cumulative affective neuroscience must embrace large-scale consortia, harmonized tasks, and pre-planned analyses. Benedek et al. (2023) add field-specific guidance for ambulatory psychophysiology (sensor specs, preprocessing, preregistration), making clear that methodological transparency is integral to cumulative science in everyday emotion.

Future research agendas: Nastase, Goldstein, and Hasson (2020) advocate naturalistic neuroimaging, films, narratives, and interactive paradigms, to capture temporally extended, socially embedded emotions that controlled stimuli often miss, with inter-subject synchronization providing quantitative anchors for shared affective states. Hartling, Sasse, Avram, and Straube (2021) likewise show that dynamic, naturalistic paradigms sharpen the mapping between evolving affect and large-scale brain networks, setting the stage for translational markers that resemble real life. Hoemann, Wormwood, Barrett, and Quigley (2023) propose multimodal, idiographic ambulatory sensing to sample emotion “in the wild,” linking physiology, behavior, and context at the moments they change, and calling for models that treat participants as co-producers of dense longitudinal data. Liao, Klasnja, Tewari, and Murphy (2020) outline micro-randomized trials that tailor just-in-time prompts to dynamic states, enabling causal tests of when and for whom regulation strategies work. Taken together, these agendas converge on a gold-standard, participant-centered field that integrates naturalistic neuroimaging, dense longitudinal sensing, and computational modeling under open-science norms, thereby producing theories that are mechanistic, predictive, and generalizable.

Method

This literature review followed a systematic, narrative-synthesis approach: searches were conducted in PsycINFO, PubMed/MEDLINE, Web of Science Core Collection, and Scopus using pre-specified Boolean blocks spanning core constructs (e.g., appraisal/construction, discrete vs. dimensional structure, interoception/prediction, regulation and affective dynamics, ambulatory measurement, naturalistic paradigms), limited to peer-reviewed human studies from 2019–2025, with seminal originals retained when theoretically indispensable; reporting was structured by PRISMA 2020 guidance (Page et al., 2021) and confirmatory elements were aligned with Registered Reports principles (Chambers & Tzavella, 2022). Titles/abstracts were screened against inclusion criteria, full texts were assessed with reasons recorded for exclusion, and data were extracted to a codebook capturing population, design, operationalizations (e.g., discrete families, circumplex coordinates, appraisal dimensions), measurement channels (self-report, behavior, psychophysiology, neural), analytic features (sample size, preregistration, out-of-sample validation), effects, and author interpretations. Risk of bias and reporting quality were appraised with RoB 2 for randomized trials (Sterne et al., 2019), ROBINS-I for nonrandomized studies (Sterne et al., 2016), and AMSTAR 2 for reviews/meta-analyses (Shea et al., 2017); neuroimaging claims were checked against field standards for analytic transparency and power (Poldrack et al., 2017), and ambulatory psychophysiology was evaluated using affective-science reporting recommendations (Benedek et al., 2023). Evidence was synthesized theory-first (functional, discrete/dimensional, appraisal/construction) and by level of analysis (body–brain, cognition, social/cultural), favoring replicated, preregistered, ecologically valid studies; to mitigate selection and citation bias, searches iterated synonyms and forward–backward citations and included conflicting findings. The review was not preregistered; inclusion/exclusion logs, codebook, and appraisal templates are available on request.

Discussion

Purpose and integrative interpretation: This review set out to reconcile functional, discrete–dimensional, and appraisal–constructive accounts into a coherent, multi-level view of emotion. Keltner and Cowen (2021) frame emotions as adaptive control systems for situated goals, while Russell (2022) models a core-affect scaffold that accommodates graded similarities among states. Moors (2022) details appraisal operations that translate feeling into context-appropriate action, and Hoemann, Satpute, and Barrett (2021) depict conceptual mediation as the mechanism that turns bodily fluctuations into labeled experiences. Read together, these strands converge on a pluralist synthesis in which discrete families summarize recurrent functions, dimensional coordinates capture graded relations, and appraisal–conceptual processes supply the meaning-making that links feeling to behavior. Pessoa (2022) and Kragel and LaBar (2021) extend this picture neurally, describing distributed, time-varying networks rather than isolated “centers,” while Critchley and Garfinkel (2024) argue that predictive interoception ties brain–body inference to moment-to-moment regulation.

Theoretical implications. Scherer (2022) characterizes emotion episodes as coordinated component processes, a formulation that neatly houses the pluralist synthesis: appraisal, expression, physiology, action tendencies, and feeling can be analyzed at different levels without privileging one theory. Scarantino (2020) advances the same pluralism normatively, suggesting that constructionist, appraisal, and functional claims are complementary lenses. Dejonckheere and colleagues (2022) strengthen this stance empirically by showing that temporal dynamics—variability, instability, inertia—carry information not captured by static traits, implying that theories of structure must be paired with theories of change. Sheppes and co-authors (2023) similarly argue that regulation theories should prioritize fit and flexibility over fixed “best” strategies.

Practical implications. Bonanno and Burton (2013, seminal) introduce regulatory flexibility as matching strategies to contexts; Ford and colleagues (2021) corroborate that flexible switching predicts mental health beyond strategy frequency. On the intervention side, Nook, Sasse, Lambert, and Somerville (2020) show that finer affect labeling tracks better daily regulation, and Hoge and collaborators (2022) report that mindfulness-based stress reduction is clinically potent for anxiety. In population health, Kiecolt-Glaser and colleagues (2022) describe psychoneuroimmunology pathways linking chronic dysregulation to inflammation, while Kalisch and co-authors (2020) argue that resilient appraisal styles facilitate faster recovery—together motivating skills that target granularity, interoception, and recovery dynamics.

Limitations of the evidence base. Marek and colleagues (2022) caution that robust brain–behavior associations often require very large samples; Botvinik-Nezer and collaborators (2020) reveal substantial analytic variability in neuroimaging; and Munafò and co-authors (2017, seminal) warn that undisclosed flexibility inflates effects. Methodologically, Mauss and Robinson (2009, seminal) note that cross-channel correspondence is modest and context dependent, so single-modality inferences remain fragile. Culturally, Tsai (2020) shows that ideal affect varies across societies, which limits generalizability when samples are WEIRD; Boiger and Mesquita (2021) add that socioecology shapes display rules and regulation norms, underscoring the need for broader sampling and contextualized theory.

Future directions. Nastase, Goldstein, and Hasson (2020) recommend naturalistic paradigms (film, narrative, interaction) to capture socially embedded dynamics; Hamilton and Huth (2020) propose semantic-rich analyses for real-world meaning making; and Hoemann, Wormwood, Barrett, and Quigley (2023) urge multimodal, idiographic sensing that treats participants as co-producers of dense longitudinal data. Liao, Klasnja, Tewari, and Murphy (2020) outline micro-randomized trials for just-in-time interventions, enabling causal tests of when and for whom strategies work. Nosek and colleagues (2018) argue that preregistration and open materials are now table stakes; Chambers and Tzavella (2022) add Registered Reports to align incentives with prediction rather than post-hoc rationalization.

Conclusion

Taken together, the reviewed literatures depict emotion as a multi-level control architecture: bodily and interoceptive signals provide the substrate; appraisal and conceptual knowledge assign meaning; functional pressures organize action; and cultural norms govern display and coordination. Scherer (2022) characterizes an emotion episode as a component process—coordinated shifts in appraisal, expression, physiology, action tendencies, and feeling—while Scarantino (2020) interprets contemporary theories as complementary lenses rather than competitors, urging a pluralist synthesis that spans body, mind, and culture. This integrative stance aligns with neural and interoceptive accounts in which feelings are inferences about internal state in context; Critchley and Garfinkel (2024) argue that predictive regulation of bodily needs links core affect to moment-to-moment behavior, and Kragel and LaBar (2021) demonstrate that time-varying network configurations, not isolated “centers,” better explain affective fluctuations.

Across development and culture, the evidence indicates that emotional life is scaffolded and plastic: early caregiver practices seed recognition and regulation; adolescence recalibrates social reactivity and control; later adulthood prioritizes emotionally meaningful goals; and socioecological contexts shape “ideal affect” and display rules. At the cognition–emotion interface, attention, interpretation, and memory operate as routing mechanisms—amplifying or damping responses and channeling them into patterned choices—so that discrete emotions bias judgment in ways not reducible to valence alone. Regulation findings converge on fit and flexibility: effective control depends on matching strategies to situational demands and switching when contingencies change, with interpersonal co-regulation frequently determining success or failure.

Methodologically, the field is moving toward ecologically valid and transparent science. Chambers and Tzavella (2022) advocate Registered Reports and open materials to curb analytic flexibility; Marek and colleagues (2022) show that robust brain–behavior associations often require very large samples; and Nastase, Goldstein, and Hasson (2020) recommend naturalistic paradigms (films, narratives, interaction) to capture socially embedded dynamics that traditional tasks miss. Participant-centered reforms extend this trajectory: Hoemann, Wormwood, Barrett, and Quigley (2023) argue that multimodal, idiographic ambulatory sensing treats participants as co-producers of dense longitudinal data, improving validity and ethics. On the translational front, Liao, Klasnja, Tewari, and Murphy (2020) formalize micro-randomized trials that optimize just-in-time support, while skills training in granularity, labeling, and interoceptive awareness targets mechanisms with scalable promise.

Three constraints temper inference. First, measurement limits persist: self-reports are context-bound; physiological signals are multiply determined; and neural findings remain sensitive to analytic choices. Second, generalizability is uneven across cultures, ages, and contexts. Third, causal leverage is still developing for many claims outside laboratory control. These constraints suggest clear priorities: combine naturalistic neuroimaging with dense longitudinal sensingand computational modelling; center participants in co-design and idiographic analysis; and embed open-sciencesafeguards to make effects cumulative and transportable.

In sum, the most defensible conclusion is pragmatic: emotions are constructed for function and structured across levels. A research program that unifies component processes (Scherer, 2022), pluralist theory integration (Scarantino, 2020), predictive interoception (Critchley & Garfinkel, 2024), large-scale and open methods (Chambers & Tzavella, 2022; Marek et al., 2022), and participant-centered measurement (Hoemann et al., 2023; Liao et al., 2020) offers a workable blueprint for a cumulative, mechanistic, and translational science of emotion. References Aldao, A., Sheppes, G., & Gross, J. J. (2015). Emotion regulation flexibility. Cognitive Therapy and Research, 39(3), 263–278.

Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2019). Emotion-regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review, 33(2), 217–229.

Al’Abri, S., Al Hinai, T., Al Wardi, A., & Al Abri, R. (2023). Multimodal wearables and smartphone sensing for ecological assessment of affect. Frontiers in Psychology, 14, 112–125.

Atmaja, B. T., & Akagi, M. (2022). Multimodal emotion recognition: A review. Sensors, 22(13), 1–32.

 Barrett, L. F., Adolphs, R., Marsella, S., Martinez, A. M., & Pollak, S. D. (2019). Emotional expressions reconsidered: Challenges to inferring emotion from human facial movements. Psychological Science in the Public Interest, 20(1), 1–68.

Benedek, M., Mier, D., Hajcak, G., Kreibig, S. D., Thayer, J. F., & Wilhelm, F. H. (2023). Standards for ambulatory psychophysiology in affective science. Psychophysiology, 60(2), e14123.

Boiger, M., & Mesquita, B. (2021). The socioecological shaping of emotions. Current Opinion in Psychology, 44, 144–149.

Bonanno, G. A., & Burton, C. L. (2013). Regulatory flexibility: An individual differences perspective on coping and emotion regulation. Perspectives on Psychological Science, 8(6), 591–612.

 Botvinik-Nezer, R., et al. (2020). Variability in the analysis of a single neuroimaging dataset by many teams. Nature, 582, 84–88.

Brady, W. J., McLoughlin, K., Doan, T., & Crockett, M. J. (2021). How social learning amplifies moral outrage expression in online social networks. Science Advances, 7(33), eabe5641.

Chambers, C. D., & Tzavella, L. (2022). The past, present, and future of Registered Reports. Nature Human Behaviour, 6, 29–42.

Cole, P. M., Lougheed, J. P., & Ram, N. (2020). Emotions across development: Emerging themes and future directions. Developmental Psychology, 56(1), 1–14.

Cowen, A. S., & Keltner, D. (2017). Self-report captures 27 distinct categories of emotion bridged by continuous gradients. Proceedings of the National Academy of Sciences, 114(38), E7900–E7909.

Critchley, H. D., & Garfinkel, S. N. (2024). Interoception, emotion, and the predictive brain. Nature Reviews Neuroscience, 25(1), 9–24.

Daly, J., et al. (2024). Neuromodulatory control of interoceptive precision in emotional disorders. Biological Psychiatry, 95(4), 301–314.

De Leersnyder, J., Mesquita, B., & Kim, H. S. (2020). Emotions and acculturation. Emotion Review, 12(1), 34–48.

Everaert, J., et al. (2020). Cognitive processes in emotion regulation. Clinical Psychological Science, 8(1), 50–65.

Ford, B. Q., et al. (2021). Emotion regulation flexibility: How fit, timing, and context shape outcomes. Current Opinion in Psychology, 39, 68–73.

Fredrickson, B. L. (2021). Broaden-and-build theory of positive emotions: Updated evidence. American Psychologist, 76(5), 781–793.

 Gross, J. J. (2015). Emotion regulation: Current status and future prospects. Psychological Inquiry, 26(1), 1–26.

Hoemann, K., Satpute, A. B., & Barrett, L. F. (2021). Conceptual knowledge shapes human emotion. Nature Human Behaviour, 5, 132–145.

Kalisch, R., et al. (2020). The resilience framework as a strategy to combat stress-related disorders. Nature Human Behaviour, 4, 99–109.

Keltner, D., & Cowen, A. S. (2021). The functional approach to emotion. Current Opinion in Behavioral Sciences, 39, 1–6.

Keltner, D., & Gross, J. J. (2021). Functional accounts of emotions in social life. Annual Review of Psychology, 72, 1–28.

 Kragel, P. A., & LaBar, K. S. (2021). Toward a neural basis of emotion: Network dynamics. Nature Reviews Neuroscience, 22(10), 627–643.

Marek, S., et al. (2022). Reproducibility of brain–behavior associations. Nature, 603, 654–660.

Moors, A. (2022). Appraisal theories of emotion: State of the art and future development. Emotion Review, 14(1), 1–15.

Nastase, S. A., Goldstein, A., & Hasson, U. (2020). Keep it real: Naturalistic stimuli in neuroscience. Trends in Cognitive Sciences, 24(8), 1–15.

Pessoa, L. (2022). The entangled brain: How perception, cognition, and emotion are woven together. MIT Press.

Quigley, K. S., et al. (2021). Interoception and emotion. Emotion, 21(6), 1170–1185.

Russell, J. A. (2022). Core affect and the structure of emotion. Emotion Review, 14(1), 24–35.

Scarantino, A. (2020). The philosophy of emotion science: A pluralist approach. Emotion Review, 12(3), 141–150.

Scherer, K. R. (2022). Component process model of emotion. Emotion Review, 14(1), 1–17.

Shamay-Tsoory, S., & Mendelsohn, A. (2019). Oxytocin and social emotion. Trends in Cognitive Sciences, 23(6), 437–450.

Tsai, J. L. (2020). Culture, emotion, and well-being: Ideal affect. Current Opinion in Psychology, 32, 1–6.

Webb, T. L., et al. (2021). Emotion regulation strategies and mental health. Annual Review of Clinical Psychology, 17, 193–221.

Yarkoni, T. (2020). The generalizability crisis. Behavioral and Brain Sciences, 45, e1.