Emotional Responses (Intensity/Valence) Fulfill RDoC Criteria but Remain Excluded From Diagnostic Classifications of Mood Disorders

ABSTRACT

Objectives

Categorical classifications of depression, such as DSM‐5 and ICD, are limited by high clinical heterogeneity, arbitrary diagnostic thresholds, and weak links to underlying pathophysiology. In line with the Research Domain Criteria (RDoC) framework, this article proposes a dimensional and mechanistic approach centered on emotional responses as core transdiagnostic features of mood disorders. Emotional responses are conceptualized along two fundamental dimensions: intensity (arousal) and hedonic valence.

Methods

This perspective integrates evidence from clinical, psychophysiological, neuroimaging, and preclinical studies.

Results

Depressive episodes may be characterized by either emotional hyporeactivity (emotional blunting and reduced responsiveness) or emotional hyperreactivity (heightened distress, anxiety, irritability, and emotional lability), the latter being closely associated with mixed features, particularly in bipolar depression. In parallel, depression is marked by maladaptive valence attribution, encompassing anhedonia related to reward‐system dysfunction and a pervasive negativity bias. Preclinical findings implicate dysregulated basolateral amygdala circuits in valence attribution and demonstrate the reversibility of negativity bias with antidepressant interventions.

Conclusions

Integrating emotional intensity and valence within an RDoC‐informed framework provides a biologically grounded account of depressive heterogeneity, refines the characterization of mixed states, and offers clinically relevant guidance for treatment selection. This perspective supports the development of personalized therapeutic strategies and the systematic inclusion of emotional dimensions in clinical research.

1. Dimensional Versus Categorical Classifications of Depression

1.1. The Research Domain Criteria (RDoC) Framework

The Research Domain Criteria (RDoC) initiative, launched by the National Institute of Mental Health (NIMH) in 2009, was designed to overcome the limitations of categorical psychiatric classifications such as the Diagnostic and Statistical Manual of Mental Disorders (DSM) and the International Classification of Diseases for Mortality and Morbidity Statistics (ICD), which rely on heterogeneous symptom clusters and are unrelated to pathophysiology [1]. Instead, RDoC relies on a dimensional approach, focusing on transdiagnostic symptoms—such as cognitive deficits, anhedonia, or impulsivity—that cut across traditional diagnoses [2]. Each dimension can be examined across seven levels of analysis: genes, molecules, cells, neural circuits, physiology, behavior, and self‐report. Moreover, they are organized into five broad functional domains: positive valence systems, negative valence systems, cognitive systems, social processes, and arousal/regulatory systems [2, 3]. By linking symptoms to dysfunctional neural mechanisms and identifying objective biomarkers, RDoC aims to advance the research in precision psychiatry, fostering diagnostic refinement and the development of targeted therapeutic interventions [2, 3].

1.2. Major Depressive Episodes in the DSM‐5: Limitations of Categorical Approaches

The categorical definition of major depressive episodes (MDE) in the DSM‐5 [4] has been widely criticized for its arbitrary thresholds, its inability to account for subsyndromal but disabling forms, and its high intradiagnostic heterogeneity, whereby two patients may meet criteria without sharing any symptoms [5, 6]. Recent analyses of DSM‐5 diagnostic criteria further underline this issue, showing substantial symptom non‐specificity and questioning the meaningfulness of studying MDD as a unitary construct [7]. This heterogeneity suggests that distinct underlying physiopathological mechanisms may be grouped under the same label, raising questions about diagnosis and treatment specificity [8]. Moreover, the categorical system inadequately addresses the frequent comorbidities of depression, such as anxiety or substance use, which constitutes an additional limitation for therapeutic decision‐making [5].

Bipolar depression poses greater clinical challenges compared to unipolar depression because it is commonly associated with mixed features [9, 10], and the treatment choice is more complex [11]. The CANMAT‐ISBD [12] guidelines offer, in association with a mood stabilizer, a wide range of therapeutic options, sometimes with markedly different mechanisms of action (antidepressants, antipsychotics, anticonvulsants, dopaminergic agonists, etc.). However, in the absence of predictive biomarkers, no well‐established strategy currently guides treatment selection [12, 13].

To sum up, both unipolar and bipolar depression are heterogeneous clinical entities. Their physiopathology remains to be elucidated, and current treatments are often ineffective in achieving stabilization. These limitations highlight the need for dimensional and mechanistic approaches to improve phenotypic characterization and treatment personalization. To address these gaps, we propose a new framework in which emotional responses are critical dimensions for describing depression and guiding treatment selection.

1.3. The Relevance of Studying Emotional Responses as Dimensions

Emotions are defined as brief, adaptive responses to salient stimuli, distinct from diffuse, enduring mood states. They rapidly engage neural circuits such as the amygdala, hypothalamus, and brainstem, triggering measurable physiological reactions—including autonomic changes, hormonal activation, and facial expressions—that prepare the organism for action [14, 15, 16]. Emotions are themselves characterized by two fundamental dimensions: the intensity of the response (arousal) and the hedonic valence attributed to stimuli, ranging from pleasant to unpleasant [17]. Describing emotion within this two‐axis framework intensity and valence, enabling continuous representation of affective states, interindividual variability, and objective measurement [17, 18]. Even though different theoretical frameworks exist to conceptualize emotions [19], this model is particularly relevant, as we will see, for studying emotional disturbances in mood disorders.

2. The Intensity of Emotional Responses in Depression: Reduced or Heightened Responses?

2.1. Clinical Assessment: Emotional Intensity Response—Hypo Versus Hyperreactivity

A meta‐analysis has shown that MDE are characterized by reduced emotional reactivity to both positively and negatively valenced stimuli, with a larger reduction for positive stimuli than for negative stimuli [20]. These results were comparable with self‐reported experience, expressive behavior, and peripheral physiology. However, in bipolar depression, we have shown that the intensity of emotional response allows for the distinction between two types of depression [21, 22, 23, 24, 25, 26]. One form is characterized by emotional hypo‐reactivity, in which patients report blunted emotions and indifference toward their environment, as seen in MDD. As Kraepelin described in Manic‐Depressive Insanity and Paranoia, in such cases, one may deliver either good or bad news without eliciting any reaction from the patients [27]. Conversely, some depressive episodes are characterized by emotional hyperreactivity, in which patients report being overwhelmed by highly distressing emotions such as profound psychic pain, severe anxiety, and marked irritability. Being highly sensitive to their environment, their mood therefore appears extremely labile. These opposite alterations are ignored in current classification systems. However, this phenomenon is not confined to bipolar disorders even if emotional hyporeactivity is far more common in MDD [20, 28].

In bipolar disorders, beyond acute episodes, the same phenomenon occurs in subsyndromal symptoms. We have shown that emotional intensity response (hypo vs. hyperreactivity) allows us to distinguish subgroups of partially remitted patients with divergent functional outcomes, inflammatory profiles, and cardiovascular risk [29, 30, 31]. Such findings underscore the value of evaluating emotional response intensity as a central dimensional marker to refine clinical characterization.

2.2. Correlates of Altered Emotional Intensity Response

Alterations in the intensity of emotional responses in depression are evident across autonomic, neural, and behavioral systems. Psychophysiological studies and meta‐analytic work indicate attenuated reactivity to emotional stimuli—particularly for positive material—across skin conductance, startle modulation, facial expressivity, and self‐reported experience [20, 28, 32, 33, 34]. Heart‐rate variability is reliably reduced in major depression, reflecting diminished vagal control and reduced flexibility of arousal regulation [35, 36]. EEG studies show similarly blunted modulation of P3/LPP amplitudes by emotional intensity, indicating reduced sustained engagement with affective stimuli [37, 38, 39]. Behavioral and ERP studies using graded intensity‐recognition tasks also show reduced accuracy and abnormal ERP across intensity levels, suggesting a specific impairment in encoding how strongly an emotion is expressed rather than its valence per se [40].

Importantly, biological modulators appear to differentiate hypo‐ versus hyper‐reactive profiles: in remitted bipolar patients, higher high‐sensitivity CRP is associated with emotional hyperreactivity and poorer functioning, indicating an inflammatory contribution to heightened emotional intensity in a clinically meaningful subgroup [29, 41].

These findings indicate that disturbances in emotional intensity in mood disorders reflect biologically grounded alterations in arousal, and that distinct hypo‐ and hyper‐reactive profiles may characterize different depressive subtypes and residual symptom patterns. Unfortunately, as this feature is not recognized in existing classification systems, it has received little attention in empirical research.

3. Perturbations of Valence Attribution

3.1. Anhedonia as an Impairment of the Reward System

Anhedonia, defined as a reduced capacity to experience pleasure, is a core diagnostic criterion of MDE. Anhedonia is considered one of the key consequences of reward system dysfunctions, which have been extensively studied in both humans and animals [42, 43, 44, 45]. Reward processing refers to the set of psychological and neurobiological mechanisms involved in detecting, valuing, anticipating, and consuming rewarding stimuli, as well as the learning processes that link actions to outcomes. Classically, it is decomposed into three dissociable components: (1) reward anticipation or “wanting”, reflecting motivational drive; (2) reward consummation or “liking,” corresponding to hedonic responses during the reward receipt; and (3) reward learning, which relies on prediction errors to update future expectations.

Depression is associated with convergent impairments across all three major components of reward processing. On the motivational side, patients show reduced willingness to exert effort for rewards and heightened effort‐cost sensitivity, together with diminished anticipatory pleasure and attenuated nucleus accumbens (NAc) activity during reward anticipation [46, 47, 48, 49]. Hedonic consummation is likewise frequently blunted, with depressed individuals reporting lower in‐the‐moment pleasure and exhibiting reduced striatal and pallidal responses during reward receipt [49, 50, 51, 52]. Reward learning is also disrupted, as evidenced by reduced sensitivity to positive outcomes, impaired reinforcement learning, and attenuated frontostriatal prediction‐error signals [43, 45, 53].

Together, these alterations in wanting, liking, and learning reflect pervasive distortions in how reward value is represented and integrated, providing a mechanistic basis for anhedonia in depression.

3.2. Beyond Anhedonia: A Global Negativity Bias

In addition to diminished sensitivity to pleasurable experiences, depression is characterized by a global valence attribution bias associating augmented sensitivity to aversive stimuli, which are perceived as even more unpleasant [54, 55]. This negativity bias reflects a global shift toward more negative appraisals of emotional stimuli and operates across multiple levels of information processing—including perception, attention, and memory. In depressed patients, numerous studies have shown that negative stimuli are perceived as disproportionately aversive. For example, patients tend to interpret neutral or ambiguous facial expressions as negative and show heightened sensitivity to sad or angry faces [55, 56, 57]. Beyond facial perception, negative interpretations of ambiguous situations and selective encoding or retrieval of negative information are frequently observed, indicating a pervasive shift in valence attribution [58, 59, 60]. This negativity bias likely contributes to the frequent co‐occurrence of anxiety symptoms during depressive episodes by amplifying the perceived relevance of threat signals in daily life.

This negativity bias contrasts sharply with the positive bias observed in mania, where patients tend to overvalue rewarding stimuli and downplay negative signals [61, 62]. As described by Henry and Berkovitch [63], such positive valence bias combined with elevated emotional reactivity fuels pleasure‐seeking behaviors and risk‐taking. Taken together, the asymmetry in a global valence bias in depression and mania illustrates a core dynamic of bipolar disorder, with significant clinical implications.

3.3. The Basolateral Amygdala (BLA) and Valence Attribution: Insights From Preclinical Research

Recent animal studies have significantly advanced our understanding of the mechanisms involved in the attribution of valence to emotional stimuli. The basolateral amygdala (BLA) plays a central role in valence attribution, with distinct neuronal populations encoding positive and negative valence. Projection‐defined circuits, along with spatial and genetic characteristics, support this segregation. For instance, BLA neurons projecting to the nucleus accumbens (NAc) encode both positive and negative valence depending on their nature and specific neuronal target, whereas those projecting to the central amygdala (CeA) convey aversive valence [64, 65, 66, 67, 68, 69, 70, 71, 72]. Associative learning studies further demonstrate that conditioned stimuli acquire valence‐specific neural representations in the BLA that resemble their unconditioned counterparts, and these representations can dynamically remap during reversal learning [73].

We have found a global negativity bias in a mouse model of depression, characterized by a tendency to over‐attribute negative valence to emotional stimuli (positive and negative). This bias was reversed by chronic antidepressant treatment, and such reversal was associated with the remission of depressive‐like behaviors [74]. Mechanistically, the bias reflected an imbalance in BLA activity, with hyperactivation of neurons projecting to the CeA and reduced activity of neurons projecting to the NAc. Importantly, selective activation of the BLA‐NAc pathway was sufficient to normalize both the negativity bias and the depressive phenotype, highlighting this circuit as a causal substrate of depressive episodes and a potential therapeutic target.

Taken together, these findings underline the BLA's central role in valence attribution and highlight how its dysregulation may contribute to the pathophysiology of depression, thereby providing promising targets for novel therapeutic interventions.

Given the positivity bias observed in patients with manic states, we aimed to investigate this bias in a classical animal model of mania. Unfortunately, although the model exhibits locomotor agitation, we did not observe any positivity bias, casting doubt on the model's validity [75].

3.4. Emotional Bias and Neuroimaging Studies in Humans

A systematic review of neuroimaging studies in major depressive disorder (MDD) identified consistent abnormalities in facial emotion processing. Patients showed a mood‐congruent bias, linked to hyperactivation to negative stimuli and hypoactivation to positive stimuli, particularly in the amygdala, insula, parahippocampal gyrus, fusiform face area, and putamen. Despite these converging findings, inconsistencies remain—especially in prefrontal regions—likely reflecting heterogeneity in experimental paradigms and patient samples [57]. These facial recognition biases, along with alterations in amygdala activation, are likewise evident under conditions of subliminal stimulus presentation [76].

Antidepressant treatments modulate emotional biases and restore activation within specific neural circuits. Thus, the response of the pregenual anterior cingulate cortex (pgACC), which is closely connected to the amygdala, is higher before treatment than after antidepressant therapy. The treatment‐related difference is attributable to a reduced activation to sad faces and an enhanced response to happy faces [77].

4. Emotional Response in Clinical Practice

4.1. Emotional Response Patterns as a Framework for Defining Episodes With Mixed Features

A dimensional approach based on emotional responses might allow a more precise definition of subgroups of depression. Current definitions of depression with mixed features remain far from satisfactory, as they rely on criteria that do not accurately reflect clinical reality. The DSM‐5 defines this specifier as at least 3 of the following symptoms associated with the criteria of a depressive episode: elevated or expansive mood, inflated self‐esteem or grandiosity, unusual talkativeness or pressure to keep talking, flight of ideas or racing thoughts, increased energy or goal‐directed activity with potential painful consequences, and decreased need for sleep. However, the most common symptoms are not represented, and specific symptoms—such as increased self‐esteem or expansive euphoria—are rarely observed [78, 79, 80].

Among bipolar patients, the prevalence of mixed features is higher than in MDD [81, 82]. In patients with MDD, mixed features are estimated to occur in approximately 15% of cases [83]. However, there is considerable variability across prevalence studies [82]. Perugi et al. [84] reported prevalence rates of mixed features in MDD ranging from 7.5% using DSM‐5 criteria to 29.1% using research criteria that included agitation.

Nevertheless, it is essential to clearly distinguish these mixed depressions from other forms of depression, as they carry a higher risk of suicidal behavior and are more frequently associated with substance use [81]. Mixed features also pertain to the hypo/manic pole and therefore encompass a large number of patients, and individuals with mixed presentations manifest more frequent and severe episodes, spend less time in the euthymic state, and exhibit poorer response to treatment [78, 79, 85]. Furthermore, as will be demonstrated in the following sections, this may necessitate distinct therapeutic strategies.

Because it is crucial to characterize mood states with mixed features better, we have proposed a new model based on emotional responses. In this model, depressive episodes and elevated phases (hypomanic or manic episodes) with mixed characteristics are associated with an overall increase in emotional response intensity. Valence determines the polarity of the mood state [63].

Thus, according to this model, depressions with mixed features are characterized by an increase in emotional intensity associated with a negative bias (pleasant stimuli are perceived as less pleasant—though some degree of reactivity remains—while aversive stimuli are experienced as more unpleasant). Depending on the context and external stimuli, this accounts for the pronounced emotional lability of these patients and the heightened expressiveness of their affective responses.

The intensity of the emotional response makes it possible to distinguish these conditions from other forms of depression characterized by a decrease in emotional intensity (blunted affect). In such cases, the first affective tone to disappear is the capacity to experience pleasure, and progressively, a global narrowing of the entire affective sphere may occur, with reduced reactivity to both pleasant and aversive stimuli, eventually leading to a general indifference to emotional stimuli (emotional blunting).

4.2. Emotional Response as a Clinical Marker for Treatment Choice

Proposing a new perspective on mood disorders—no longer centered on mood itself (sad or euphoric), but rather on emotional responses—offers a better understanding of therapeutic outcomes and may help guide treatment choices. Despite the existence of depression with mixed features within MDD, there are currently no specific recommendations for its management in clinical guidelines, and the recommended treatments are antidepressant medications [86].

Treatment selection in bipolar depression is more complex. Clinicians are frequently faced with the following scenario: a patient receiving lithium or an anticonvulsant who relapses into a depressive episode. According to current guidelines [12, 87], the clinician may either add an antipsychotic or an antidepressant; however, until recently, no specific recommendations were provided to support this decision.

In 2021, the Canadian Network for Mood and Anxiety Treatments (CANMAT) proposed recommendations for the management of bipolar patients with mixed presentations [88]. However, these recommendations are based on a few studies using DSM‐5 criteria, and proxy criteria are generally used. No agents have sufficient evidence for first‐line recommendation, and two antipsychotics—Asenapine and Cariprazine—are recommended as second‐line treatment with a 2 level of evidence. In 2025, a systematic review examined the efficacy of pharmacological treatments for depression with mixed features in patients with MDD or Bipolar Depression [89]. The criteria used to define mixed features were: an MDE with two or more hypomanic symptoms, or an MDE with a Young Mania Rating Scale (YMRS) score ≥ 4. The need to include a lenient criterion, such as a score on a mania rating scale, underscores the difficulty of identifying studies that adequately address this crucial question. Only one study specifically addressed treatment response in MDE with mixed features and demonstrated the efficacy of Lurasidone [90], whereas most studies used post hoc analyses. Second‐generation antipsychotics were the only class of drugs studied, except one of them, which assessed a combination of Olanzapine and Fluoxetine that demonstrated good efficacy.

By contrast, in depression with mixed features, antidepressant treatment is associated with an increased risk of worsening manic symptoms within a few months [91]. Therefore, a more accurate identification of bipolar depressive episodes at risk of antidepressant‐induced switching should provide a stronger rationale for their use. It is also important to acknowledge that antidepressants continue to be frequently prescribed in bipolar depression and retain therapeutic value in well‐selected cases.

Considering emotional responses may provide helpful guidance, the option of prescribing an antipsychotic appears preferable when patients display highly intense emotional responses (emotional hyperreactivity), which is frequently associated with marked internal tension and a subjective experience of a “restless mind”—a triad highly characteristic of depression with mixed features [9, 92]. This observation seems consistent with the pharmacological profile of antipsychotics, which are known to attenuate emotional responsiveness—a feature that frequently constitutes a source of dissatisfaction or complaint among patients.

Conversely, when depression manifests as global inhibition accompanied by emotional hyporeactivity and a negativity bias, an antidepressant seems to be the most appropriate option. Indeed, antidepressants have been shown to correct such biases, as demonstrated by Harmer and colleagues [93, 94]. By avoiding the prescription of antidepressants in patients with depression presenting mixed features, the risk of switching to a manic episode can be reduced [13].

The strength of our model lies in its ability to provide a comprehensive framework and a simple clinical assessment to guide the choice between antipsychotic and antidepressant treatments in bipolar depression. It also helps explain why many bipolar depressive episodes do not respond to antipsychotic treatment, owing to a worsening of overall inhibition (Figure 1).

FIGURE 1.

Schematic representation of different depressive subtypes according to the intensity and valence of emotional responses.

5. Emotional Responses and RDoC

Kraepelin initially conceptualized mixed states within a dimensional framework encompassing three domains: mood, psychomotor activity, and ideation (flow or speed of thought). Each of these domains may shift toward excitation or inhibition, yielding a range of intermediate states between mania and depression (Kraepelin, as summarized in modern reviews, e.g., [95], and further explored in later studies: [80, 96]). While thought and activity may vary along a continuum of increase or decrease, mood traditionally represents a qualitative rather than a quantitative variable.

The incorporation of the notion of response intensity enables all three dimensions to be assessed on a quantitative scale, thereby allowing for a more refined and operational characterization of states exhibiting mixed features, as demonstrated with the Mathys scale, which we developed to assess mood states using this dimensional approach [22, 23, 25, 26]. A dimensional framework, as proposed by the RDoC initiative, is well‐suited to capturing these patterns, as it organizes psychopathology in terms of fundamental constructs—particularly positive and negative valence systems and arousal/regulatory systems—measured across multiple units of analysis [97, 98]. From this perspective, mood states can be conceptualized as a specific configuration of emotional intensity and valence attribution, an approach consistent with multidimensional models of mood pathology [99].

Such a framework also clarifies why mixed‐like emotional dysregulation is frequently observed across bipolar disorder, borderline personality disorder, and attention‐deficit/hyperactivity disorder. All these disorders are strongly associated with chronic mood lability and low grade of emotional hyperreactivity [100, 101]. It also appears necessary to investigate the intensity of emotional responses during depressive episodes across different stages of life. In particular, the higher prevalence of emotional hyperreactivity in adolescents may help explain antidepressant‐induced activation syndromes [102]. Furthermore, this perspective may enhance our understanding of treatment effects across these conditions [99].

An intriguing question pertains to the mechanism of action of rapid‐acting antidepressants, such as ketamine and esketamine. It remains to be determined whether these agents primarily act by modifying the intensity of emotional experiences or by shifting their affective valence.

6. Conclusion

Taken together, evidence from clinical, physiological, and preclinical studies demonstrates that emotional response—defined by its intensity and hedonic valence—is a central feature of mood disorders. Altered responses in intensity emotional responses (hyper/hypo), and emotional biases, are linked to distinct functional outcomes, neurobiological mechanisms, and therapeutic challenges. These findings highlight the limits of purely categorical diagnostic systems and argue for integrating emotional responses as a transdiagnostic construct.

Within the RDoC framework, emotional response meets the criteria for a valid dimension: it can be reliably assessed across multiple levels of analysis—from subjective reports to physiology, electrophysiology, and neuroimaging—and it reflects clinically meaningful dimensions across disorders, accounting for their heterogeneity. Recognizing emotional processing as a core dimension may therefore facilitate the development of personalized pharmacological and psychosocial interventions, as treatments can be tailored to specific emotional profiles, as recently emphasized by Henry and Berkovitch [63].

It is essential to incorporate these dimensions into clinical trials.

In parallel, we are currently testing an application that could help decode emotional intensity and emotional biases in routine clinical practice during depressive episodes.

Finally, if negativity bias correction is essential for recovery from depression, novel antidepressant compounds must be tested in animals to determine whether they alleviate this bias [103].

Author Contributions

Chantal Henry: Conceptualization, writing – original draft, writing – review and editing. Chlöé Müh: Writing – review and editing, visualization. Claire‐Hélène de Badts: Writing – review and editing. Mariana Alonso: Writing – review and editing. Lucie Berkovitch: Writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.