Emotional Capacity: regulating, relating and adapting.

A project review begins with a delivery problem and turns emotional within minutes. A client says the team has wasted three months. The programme lead feels a rush of anger and immediately wants to defend the work. Across the table, one colleague becomes unusually quiet while another starts answering every criticism before it has finished. The factual questions still matter, yet the emotional conditions are already changing what people notice, how they interpret one another and what each response is likely to produce next.

The programme lead has several kinds of work to do before the meeting can recover. She needs to recognise her own reaction early enough to avoid being driven by it, understand why the accusation has landed so strongly and decide how much of that anger should influence what she says. She also needs to notice that the client's frustration may be directed at uncertainty about the launch rather than at her personally, read the silence of her colleague without treating it as proof of agreement or fear, and adjust the timing, tone and content of her response as the conversation develops.

We define Emotional Capacity as the capacity to understand and manage emotions in oneself and respond effectively to emotions in others. Within the Unified Skills Map, we organise it through six interrelated dimensions: Awareness of one's emotions, Understanding one's emotions, Regulating one's emotions, Awareness of others' emotions, Understanding others' emotions, and Responding to others' emotions. Research across affective science, emotion regulation, social cognition, empathic accuracy and interpersonal emotion regulation provides substantial support for the functions represented by these dimensions, although their academic foundations are uneven. Emotion regulation has a mature direct literature, while the evidence for responding to others as a distinct underlying capacity is more distributed across neighbouring research traditions (Park, Zhan and Naragon-Gainey, 2022; Niven et al., 2024; Nozaki and Gross, 2025).

The six dimensions describe a dynamic system rather than a psychological sequence that people complete in order. Recognising an emotion can alter its intensity or meaning, and a new interpretation can change what subsequently needs regulating. When someone reduces their own defensiveness, attentional resources may become available for another person's reaction; any response then changes the interaction and produces further emotional information. These loops make Emotional Capacity relevant to thinking and action without turning it into a general explanation for social success.

The surrounding academic vocabulary can obscure this territory because emotional intelligence, empathy, emotion regulation, emotional stability and social skill overlap selected parts of it while referring to different constructs. Ability models of emotional intelligence are a close neighbour because they examine the processing of emotional information, whereas trait and mixed approaches often include self-perceptions, personality characteristics or broad social competencies (Mayer, Caruso and Salovey, 2016). Empathy is equally heterogeneous: sharing another person's feeling, taking their perspective and accurately inferring what they feel can occur together, but the research treats them as distinguishable processes (Arioli et al., 2021; Salazar Kämpf et al., 2023). We therefore use those literatures to illuminate specific emotional functions rather than treating any one established label as equivalent to our complete construct.

Emotion becomes useful when it is represented accurately enough

The programme lead's anger arrives before she has formulated an explanation for it. Her body changes, attention narrows towards the accusation, and a response begins to feel urgent. Awareness of one's emotions concerns detecting that change and representing the emerging state with enough differentiation to work with it.

Research on intrapersonal emotion knowledge complicates the familiar idea that people either know how they feel or do not. Park, Zhan and Naragon-Gainey's (2022) meta-analysis found that constructs such as emotional clarity, awareness, affect labelling and differentiation overlap incompletely. In particular, emotion differentiation was not significantly associated with emotional clarity in the way a single unified awareness faculty would predict. Attention to emotion also differs from the quality of the representation produced. A person can monitor distress repeatedly while continuing to experience it as an undifferentiated sense that something is wrong.

Useful awareness can therefore include several levels of resolution. Someone may first detect a shift in valence or arousal, later recognise its intensity, and eventually distinguish anxiety from frustration or shame. Mixed states add another complication because relief and disappointment can coexist, as can anger and fear. The emotional label is one part of the representation rather than a complete account of the experience.

Interoceptive information contributes to that representation because changes in heart rate, breathing, muscle tension or visceral sensation can become part of how emotion is experienced. The measurement literature gives little support to treating physiology as a direct decoder of discrete emotional states. Meta-analytic work finds relationships between interoceptive accuracy and perceived emotional intensity, while also documenting substantial methodological problems in common interoception measures (Coll et al., 2021; Parrinello et al., 2022). Bodily information is therefore evidence available to emotional awareness, with interpretation still required.

The malleability of emotional representation is supported by experimental work. Vedernikova, Kuppens and Erbas (2021) increased participants' emotion knowledge and found improved differentiation of negative emotions relative to a control condition, with the group difference still present at follow-up. Positive emotion differentiation did not show the same specific pattern. The result is useful because it demonstrates change in one component of awareness while keeping the claim narrow enough to match what the intervention actually tested.

Greater awareness can also become unproductive when attention turns into repetitive self-focus without improved discrimination. Rumination can keep emotion highly available while making the person's interpretation increasingly rigid. For development and assessment, the relevant question is therefore how timely, differentiated and calibrated the representation becomes, rather than how often someone reports thinking about their feelings.

Understanding one's emotions begins once the person tries to explain what the state reflects and how it is influencing cognition or behaviour. The programme lead may recognise anger accurately and still misunderstand its source. The reaction could be tied to perceived unfairness, fear of reputational damage, accumulated exhaustion, or a belief that the client has ignored earlier warnings, with each interpretation creating different implications for what follows.

Appraisal traditions and affect-and-judgment research provide a strong basis for this dimension because emotions are connected to how situations are evaluated relative to goals, expectations and perceived control. Experimental and meta-analytic research also shows that emotions with the same valence can influence judgment differently. Bartholomeyczik, Gusenbauer and Treffers (2022) found emotion-specific patterns in decisions under risk and uncertainty, while Marini's (2023) meta-analysis found a small causal effect of fear towards greater risk aversion without comparable reliable mean effects for several other emotions. A positive-versus-negative classification loses much of the information that matters.

Understanding remains fallible because people have privileged access to the experience of an emotion without equivalent access to every cause that produced it. Gangemi, Dahò and Mancini (2021) describe the danger through emotional reasoning, where the presence of a feeling becomes evidence used to infer external reality. The programme lead may correctly observe that she feels threatened and still make an unsupported inference that the client intends to humiliate her. Stronger emotional understanding keeps the internal signal available while leaving its causal interpretation open to revision.

This is one point where Analytical Capacity can work alongside Emotional Capacity. Emotional Capacity concerns the representation of the state, its plausible triggers and its influence on attention or action. Analytical processing can compare explanations and examine whether the external evidence supports the interpretation. The two capacities can operate on the same episode without needing a perfectly clean division between individual cognitive operations.

Regulation is a problem of fit

Once an emotional state is sufficiently represented, Regulating one's emotions concerns influencing its trajectory, expression or consequences in a way that fits the situation and the person's goals. Emotion-regulation research gives this dimension the strongest direct scientific foundation within our six-part architecture.

Gross's process and extended-process models describe regulation as involving judgments about whether regulation is needed, strategy selection, implementation and monitoring (Gross, 2015). Regulation can act through the situation itself, attention, interpretation or response. The intended outcome can vary because a person may seek to reduce an emotion, maintain it, increase it or change how it is expressed.

The programme lead's anger therefore creates several possible regulatory choices. She could reinterpret the client's statement, pause before answering, ask for the specific concern behind it, reduce exposure to an escalating exchange by moving part of the discussion offline, or allow enough of the anger to remain that she addresses an unfair accusation directly. None of these strategies can be evaluated independently of what the meeting requires.

Research on regulatory flexibility makes this contextual requirement explicit. Bonanno and Burton (2013) described flexibility through sensitivity to context, access to a repertoire of possible responses and responsiveness to feedback. Matthews et al. (2021), reviewing 219 studies on emotion-regulation choice, identified affective, cognitive, motivational, individual and sociocultural determinants of strategy selection. Intensity, goals, available options and features of the event all influence what people choose. Specker, Sheppes and Nickerson (2024) provide more direct contemporary evidence that the effectiveness of regulatory selection depends on how strategy choices fit emotional demands.

These findings make frequency-based prescriptions weak indicators of capacity. Frequent reappraisal can be useful in many contexts, while a poorly timed reinterpretation can become a way of avoiding an issue that requires action. Expressive suppression has average costs, including a meta-analytic association with lower positive affect, yet a brief decision to contain expression can serve a legitimate immediate goal when visible escalation would damage the interaction (Fernandes and Tone, 2021). Regulation quality becomes visible through fit, monitoring and adjustment rather than allegiance to a universally preferred strategy.

Emotional labour illustrates why the same outward behaviour can arise from different internal processes. Surface acting changes displayed expression to meet a role requirement, while other forms of regulation may attempt to change the underlying experience. Workplace research finds overlap between emotion regulation and emotional labour while preserving their empirical distinction (Joffe, Kangas and Peters, 2025). A customer-service employee smiling despite anger has demonstrated control of expression; that observation alone tells us little about whether the emotion was understood, whether the strategy is sustainable or whether another response would have served the situation better.

Pressure raises the demands on regulation without producing one uniform cognitive effect. Meta-analytic and systematic work on acute stress finds heterogeneous effects that depend on timing, task characteristics and individual response (McManus et al., 2022). Emotional Capacity under pressure therefore concerns maintaining enough flexibility to preserve useful choice as affective load increases. Emotional numbness would remove information along with disruption, while uncontrolled reactivity would allow the state to dominate action.

Other people's emotions are visible only through imperfect signals

The meeting changes again when the client stops challenging the programme lead and turns towards the silent analyst. His gaze drops, his answers shorten and his posture becomes less open. These cues may signal anxiety, embarrassment, frustration, fatigue or a deliberate attempt to avoid entering the dispute. Awareness of others' emotions begins with detecting probable affective change while resisting the temptation to treat one cue as a transparent reading of another person's internal state.

Emotion-perception research has moved decisively away from a simple facial fingerprint account. Barrett et al. (2019) reviewed a large body of evidence and concluded that facial movements vary substantially across people and contexts, limiting confident one-to-one inference from a facial configuration to a discrete emotion. Durán and Fernández-Dols's (2021) meta-analysis similarly found low average co-occurrence between experienced emotions and the facial configurations predicted by classic basic-emotion accounts.

Facial movement therefore provides probabilistic information alongside voice, body movement, conversation history and situational context. Irwantoro et al. (2023) showed experimentally that temporal context can improve facial-expression classification and that the effect is especially relevant for people who perform poorly when expressions are isolated. Other work demonstrates that channels such as voice and touch carry different advantages depending on the emotional content being perceived (Oya and Tanaka, 2023).

Culture adds another layer because display rules, conceptual categories and habitual attention shape both expression and interpretation. Laukka and Elfenbein's (2021) meta-analysis found above-chance recognition of vocal emotion across cultures together with an in-group advantage. Li et al. (2025), reviewing 105 studies, identify cultural influences operating through attentional and social-cognitive mechanisms in facial emotion perception. Emotional awareness of others therefore depends partly on Contextual Knowledge about the setting in which cues acquire meaning.

A manager who interprets quietness as disengagement may be correct in one team and seriously mistaken in another, because the same visible restraint can reflect deference, disagreement, discomfort, concentration or a cultural norm around emotional display. Strong Emotional Capacity keeps that uncertainty visible while integrating additional information and calibrating confidence in the emerging interpretation.

That calibrated detection gives us a useful boundary between Awareness of others' emotions and Understanding others' emotions. Awareness concerns the probable state suggested by available signals. Understanding concerns what that state may mean in this person's situation: what concern has been activated, which expectation has changed, what pressure they may be experiencing and how the relationship or context shapes the emotion.

Understanding another person requires inference without mind reading

The research traditions surrounding empathy can easily blur this distinction because empathy itself refers to several different phenomena. Emotional resonance sits inside affective empathy, whereas cognitive empathy and perspective taking concern representations of another person's point of view. Empathic concern adds a motivational element, while mentalizing also includes inference about beliefs and intentions. Meta-analytic evidence supports partial relationships among these processes while preserving substantial differences between them (Arioli et al., 2021; Salazar Kämpf et al., 2023; Yavuz et al., 2024).

For Understanding others' emotions, empathic accuracy offers one of the strongest operational anchors because it asks how closely an observer's inference corresponds with the target's reported experience. He and Côté's (2023) preregistered research is particularly useful because demonstrated empathic accuracy and favourable self-beliefs about empathy showed different relationships with adjustment. People who believe they understand others well are therefore measuring something different from people who demonstrate accuracy in a performance task.

Recognition and understanding also separate empirically because identifying that the analyst is probably anxious is a narrower achievement than understanding what the anxiety concerns. He could be worried about being blamed, uncertain about the technical facts, anxious about contradicting a senior colleague or dealing with a completely unrelated pressure. Each explanation can fit the same broad emotional label.

Good interpersonal understanding consequently operates through hypotheses that remain revisable. Gehlbach and Mu (2023) describe perspective taking as a process shaped by motivation, bias and context rather than an automatic transfer into another person's mind. Research on hypermentalizing provides a useful warning from the opposite direction: increased attribution of mental states can become inaccurate when the observer generates elaborate interpretations without adequate evidence (McLaren et al., 2022). The number or confidence of interpersonal inferences is therefore a poor proxy for their quality.

Contextual Knowledge becomes indispensable because culture, relationship history, role expectations and institutional stakes constrain what an emotional signal is likely to mean. A team member who knows that a particular client becomes terse when worried about executive scrutiny has information that a generic emotion-recognition test cannot supply. The contextual knowledge narrows plausible interpretations, while Emotional Capacity concerns how the person notices, tests and updates emotionally relevant hypotheses within that context.

This separation also protects the construct from the assumption that empathy is inherently benevolent. Someone can infer another person's fears accurately and use that knowledge manipulatively. Affective empathy can produce personal distress that interferes with response quality, while cognitive understanding can remain strong. Emotional Capacity therefore requires accurate enough processing and effective calibration without assuming that understanding automatically creates prosocial motivation.

Responding is emotional calibration inside a larger action

The client eventually explains that the anger is coming from an executive launch commitment made before several dependencies were understood, changing the emotional problem the programme lead is trying to manage. She now has a better hypothesis about what is driving the interaction, while the appropriate response still has to be chosen. Responding to others' emotions concerns calibrating behaviour in light of the other person's probable emotional state, the goals of the interaction and the feedback generated by each response.

Interpersonal emotion regulation provides the strongest research base for this function. People attempt to influence one another's emotional states through attention, reframing, changes to the situation, support and modulation of expression. Niven et al. (2024) review individual differences in success across these processes and emphasise that the literature on who regulates others effectively remains less mature than the literature establishing that interpersonal regulation occurs. Nozaki and Gross (2025) integrate supportive communication with interpersonal emotion-regulation models and describe identification, goal setting, strategy selection, implementation and monitoring within social interaction.

The monitoring component matters because another person's response is new evidence. If the programme lead acknowledges the client's pressure and the client becomes more agitated, the initial interpretation or chosen strategy may have been wrong. She may need to ask a different question, change timing or stop trying to regulate the emotion at all, because interpersonal calibration develops through this feedback loop.

Immediate comfort is only one possible criterion of effectiveness. A useful response to anger may involve setting a boundary, while someone overwhelmed by bad news may need silence before explanation; difficult feedback can remain difficult even when the emotional calibration around it is strong. Interpersonal emotion-regulation research distinguishes prosocial and instrumental motives, and the target's immediate emotional improvement is only one possible outcome among task progress, trust, conflict reduction and longer-run relational consequences (Niven, Henkel and Hanratty, 2019; Vowels and Carnelley, 2022).

This dimension has the most consequential boundary with Practical Skills. Emotional Capacity contributes the calibration that changes timing, tone, degree of directness, attentional focus or response strategy because of the emotional conditions of the interaction. Conducting an entire negotiation, feedback conversation, customer interaction or conflict-resolution process requires a wider configuration of language, procedural knowledge and practical capability. An emotionally well-calibrated choice can support the action without becoming equivalent to the complete action.

The project review makes the distinction concrete: recognising that the client is overloaded by uncertainty and choosing to slow the discussion expresses emotional calibration. Explaining the revised delivery plan clearly requires communication skill and technical knowledge. Negotiating a new launch commitment draws on additional Practical Skills. Emotional Capacity changes how those capabilities are deployed as the interaction evolves.

Chavira Trujillo, Gallego Tomás and López-Pérez (2022) provide direct empirical evidence that cognitive and affective empathy relate differently to interpersonal emotion-regulation directions and strategies. The study relies on correlational self-report and therefore cannot establish a separate D6 capacity, but it supports the broader proposition that understanding and response are connected without being identical. The emerging IER literature gives us enough evidence to defend the function while keeping its scientific status more qualified than intrapersonal emotion regulation.

Emotional Capacity works through feedback loops

The six dimensions become easier to understand once their apparent symmetry is relaxed. Self-directed emotional functioning moves through awareness, interpretation and regulation, while interpersonal functioning places particular weight on detecting others' states, interpreting them under uncertainty and calibrating a response. The architecture separates useful functions without claiming two perfectly matched triples or six independent latent abilities.

A more faithful description follows the movement of emotional information through detection, interpretation, intrapersonal regulation and interpersonal calibration, with several loops operating at once. Labelling or reappraisal can change awareness, while regulation can alter attention and make another person's cues easier to notice. The other person's response can reveal that an interpretation was inaccurate and simultaneously change the observer's own emotional state. Social interaction creates a coupled system in which each participant becomes part of the other's emotional context (Reeck, Ames and Ochsner, 2016; Nozaki and Gross, 2025).

Substantial dissociation can occur across the architecture: a person may identify their anger accurately and still regulate it poorly. Someone who remains composed under pressure can still misread everyone else in the room. High affective empathy can coexist with weak empathic accuracy, and good understanding of another person's concern can coexist with poor behavioural execution when communication skill is weak. These differences support keeping the six dimensions conceptually visible even though their boundaries remain permeable in actual performance.

Capacity is only one determinant of what becomes observable. Fatigue, incentives, power, conflicting goals and available options can prevent a person from using processes they possess. A manager may read a colleague's distress accurately and choose to ignore it because an organisational incentive rewards speed. Another may know several regulation strategies but have no practical way to leave a high-pressure situation. Assessment therefore has to distinguish weak capacity from situations in which capacity was unavailable, unnecessary or deliberately unused.

Emotional meaning depends on context and knowledge

Contextual Knowledge shapes almost every part of Emotional Capacity. A direct challenge from a senior executive, a colleague or a close friend carries different implications even when the words are identical. Silence can signal respect, uncertainty, resistance or disengagement depending on the culture, relationship and task. Emotional restraint may reflect role expectations rather than low emotional intensity.

The in-group advantage found in vocal emotion recognition and the cultural mechanisms reviewed by Li et al. (2025) make this dependency measurable, indicating that emotional interpretation is partly learned through exposure to local display rules and social expectations. Matsumoto and Wilson's (2022) review of a half-century of culture-and-emotion research similarly shows why context-free claims about expression and regulation are difficult to sustain.

Within the Unified Skills Map, this creates a productive division of labour. Knowledge Domains provide conceptual and contextual structures that help a person interpret what an emotional signal could mean in this setting. Emotional Capacity concerns the processes through which those signals are noticed, represented, regulated and used to adjust interaction. Strong contextual knowledge can narrow the set of plausible interpretations, while emotional sensitivity remains necessary to detect that the relevant change has occurred.

Professional expertise creates a similar relationship because clinicians, negotiators or teachers can learn domain-specific patterns that make certain emotional dynamics more intelligible, while those learned expectations still remain fallible. Expertise can also create expectations strong enough to bias perception, so emotionally informed professional judgment still benefits from uncertainty and revision.

Development requires practice across changing emotional situations

The evidence supports deliberate development of constituent emotional processes, although it becomes progressively weaker as the claim moves from immediate learning towards durable transfer across unpredictable situations.

Emotion knowledge can be changed experimentally, as Vedernikova, Kuppens and Erbas (2021) demonstrate for negative emotion differentiation. Emotion-recognition training also produces gains on targeted tasks, and Roheger et al.'s (2022) systematic review and meta-analysis of social-cognition training found strong effects for theory-of-mind training in several age groups while identifying transfer to other processes as a major unresolved problem.

A substantial empathy-training literature provides another source of intervention evidence. Ngo et al. (2022) synthesised 36 randomized studies involving 1,616 mental-health practitioners and trainees and found a medium-to-large average effect, with instruction, observation and rehearsal among useful components. Such findings establish that relevant behaviours and processes can improve through structured intervention, while the heterogeneity of empathy measures limits claims about what exactly changed in every study.

Workplace evidence reaches closer to the setting in which Emotional Capacity eventually matters. Mehler et al. (2024) reviewed 50 studies of emotional-competency training and found moderate improvements, including evidence of persistence beyond three months. High heterogeneity and generally limited study quality make broad causal claims about a unified workplace capacity premature. The pattern still supports intentional development when training targets identifiable processes rather than relying on motivational language about becoming emotionally intelligent.

A useful development sequence therefore moves from conceptual knowledge into repeated practice under variation. People can learn a richer emotion vocabulary, practise distinguishing similar states, compare regulatory options, work through dynamic interpersonal scenarios and receive feedback on the consequences of their choices. Role-play and observation can expose timing and calibration, while reflection after real interactions can identify patterns that were invisible in the moment. Variation matters because memorising one preferred response creates a script that may fail as soon as emotional conditions change.

Transfer remains the harder standard because improvement on a recognition task demonstrates task improvement, while greater reported confidence demonstrates a change in self-perception. Durable Emotional Capacity requires evidence that learning survives changes in people, power relations, cultural context, intensity and practical demands. The current literature is much stronger at demonstrating change in specific components than at demonstrating that a single higher-order capacity has been strengthened everywhere it may later be required.

Assessment needs evidence from several kinds of performance

A single questionnaire is poorly matched to a capacity whose components include private awareness, dynamic regulation, perception of others and adaptive response. Self-report can measure beliefs about emotional tendencies or perceived difficulty, yet a person who reports high empathy may still perform weakly on empathic-accuracy tasks. He and Côté's (2023) findings provide a particularly clear warning because demonstrated empathic accuracy and favourable self-views were associated with different outcomes.

Performance tasks offer stronger evidence for some dimensions while introducing their own limitations. Static facial-expression tasks can assess selected recognition processes but strip away context. Emotion-regulation scenarios can reveal strategy knowledge without showing what the person would do when highly activated. Informants can observe behaviour while remaining unable to see the person's internal interpretation. Physiological measures register aspects of arousal without supplying the meaning of an emotion.

Ecological momentary assessment is valuable because it samples experience close to the situations in which it occurs. Wrzus and Neubauer's (2023) meta-analysis of 477 EMA studies documents both the feasibility of repeated real-world assessment and the design considerations required to obtain usable data. EMA can help reveal whether someone differentiates emotions across daily situations, which strategies they use and how those choices relate to subsequent states, although the repeated reporting itself can influence awareness.

A credible assessment of Emotional Capacity therefore needs triangulation across multiple methods and situations. Own-emotion awareness can be examined through repeated emotional representations and differentiation. Understanding can be tested by asking people to propose and revise plausible explanations for their reactions. Regulation requires situations that vary in intensity, controllability and goals so that fit and switching can be observed. Other-emotion awareness benefits from dynamic, multimodal and culturally varied stimuli, while understanding others requires target-validated or context-rich inference tasks with confidence estimates.

Responding to others requires the most interactive design because the target's reaction is part of the evidence. A branching simulation can present an emotionally charged interaction, record the initial response, reveal how the other person reacts and then observe whether the participant recalibrates. The complete quality of the conversation should be scored separately from the emotional calibration inside it, because language skill, technical knowledge and other Practical Skills contribute to the final performance.

Assessment becomes stronger when the same underlying emotional demands appear across different contexts. Repeated evidence allows us to ask whether the person possesses reusable processes or has learned the conventions of one task. It also gives uncertainty a legitimate role: calibrated recognition that another person's state is ambiguous can represent stronger Emotional Capacity than a confident forced-choice answer unsupported by context.

AI can improve emotional output without revealing the same underlying process

AI-mediated interaction makes the distinction between output and human capacity unusually visible. Large language models can produce text that readers judge highly empathic even though the process generating that text differs materially from human emotional experience and interpersonal history.

Sorin et al. (2024) reviewed twelve early studies of large language models and empathy and found that systems could produce emotionally supportive responses and perform strongly on selected cognitive-empathy-like tasks. Most studies evaluated outputs through subjective ratings, with substantial limitations in prompts, comparison standards and operational definitions. High-rated language therefore establishes something about the response a system generated under the test, while leaving the human chain of emotional awareness, lived affect, regulation and relational interpretation outside the measurement.

Human-AI collaboration shows a related distinction in Sharma et al.'s (2023) test of an AI feedback tool in a real peer-support setting, where assistance helped human supporters produce more empathic text. The intervention augmented observable communication, while any subsequent improvement in independent detection, understanding or emotional calibration would require a different design with unaided follow-up and transfer measures.

Source beliefs also change how emotional language is experienced. Rubin et al. (2025), across nine studies involving 6,282 participants, found that identical or comparable responses were evaluated differently depending on whether people believed they came from a human or AI source. Wenger, Cameron and Inzlicht (2026) extend this emerging literature with four studies showing an AI empathy choice paradox: participants preferred to receive empathy from humans, while AI-generated responses were often rated as higher in quality and more effective at making them feel heard when chosen. The result reinforces the separation between linguistic performance, perceived relational source and the broader human experience of receiving empathy.

AI can therefore participate in Emotional Capacity development or substitute for parts of its expression depending on the workflow. A person may use a model to rehearse alternative interpretations, compare possible responses or receive feedback that later changes independent behaviour. The same person may instead outsource the interpretation of a colleague's emotion and accept a plausible machine-generated explanation with unjustified confidence. Facial-expression research already shows why automated claims about another person's internal state require caution, because even human observers do not have direct access to discrete emotion from isolated cues.

Longitudinal evidence remains too limited to determine whether routine AI assistance will strengthen, leave unchanged or displace human emotional processes over time. Current evidence supports a narrower assessment distinction: unaided human Emotional Capacity, AI-augmented emotional performance and performance after assistance should be observed separately whenever development is the question.

Emotional Capacity keeps action responsive to what people are actually experiencing

The project review can recover because several processes change together. The programme lead recognises her anger before it controls the exchange, revises her explanation of why the client's criticism matters, and selects a response that keeps the substantive problem discussable. She notices the analyst's withdrawal without pretending to know its cause, tests her interpretation through the conversation and adjusts the meeting once the client's underlying pressure becomes clearer.

No single act demonstrates the whole capacity because apparent calm can result from suppression, a warm sentence can be scripted, and accurate recognition of frustration can still be followed by a response that escalates the conflict. Emotional Capacity becomes visible across the quality of detection, interpretation, regulation and interpersonal calibration, with contextual knowledge and other capabilities determining how those processes eventually appear in action.

Our six dimensions provide a functional architecture for that work. Awareness of one's emotions and Understanding one's emotions make the internal state sufficiently legible to influence deliberate action. Regulating one's emotions keeps emotional trajectories responsive to goals, context and feedback. Awareness of others' emotions and Understanding others' emotions extend the same discipline into inherently uncertain interpersonal territory, where signals require multimodal and contextual interpretation. Responding to others' emotions concerns the calibration that follows, while the full communication or interpersonal performance remains part of a broader capability system.

The scientific evidence is strongest when this architecture is treated as recursive and context-sensitive. Emotion carries information that can improve attention, motivation and interaction, yet that information requires interpretation and can be wrong about the external world. Regulation works through flexibility rather than universal strategy rankings. Other people remain partially hidden from us even when their expressions seem obvious. Effective response grows from continuing adjustment as the interaction produces new evidence.

Emotional Capacity therefore supports adaptation because changing situations often change emotional demands alongside technical or cognitive demands. A new role may alter status relationships, failure may create frustration or threat, and feedback can provoke defensiveness at the same moment that uncertainty generates anxiety or collaboration exposes people to emotions they did not create and cannot control. Remaining capable under those conditions requires emotional information to stay usable enough for thinking and action to continue changing with the situation.