Circumplex Model of Affect: Structure, Uses, and Limits
Author: Angela Bogdanova · Published: October 13, 2026 · Editorial Policy
Research cluster: Emotion & Affect Science · Written by Artificial Sapiens
The circumplex model of affect is a two-dimensional map of feeling states. Its horizontal dimension is valence, ranging from unpleasant to pleasant; its vertical dimension is arousal, ranging from low activation or sleepiness to high activation or alertness. An excited, happy state and a calm, contented state can therefore share pleasant valence while occupying different positions on the arousal dimension. The model is particularly useful for organizing affect ratings and testing how affective terms relate to one another. It does not, by itself, identify the cause, meaning, or complete psychological structure of a particular emotion. Russell (1980) established the influential modern formulation.
The scientific value of the circumplex lies in its economical description of affective similarities and differences. Its limitations are equally important: two ratings cannot necessarily distinguish anger from fear, capture mixed feelings, or justify inferring inner experience from a person's face, voice, heart rate, or algorithmic score. Research also disputes whether affect consistently forms a mathematically ideal circle. Kuppens et al. (2013) and Stanisławski et al. (2021) provide especially informative tests of its assumptions.
What Is the Circumplex Model of Affect?
A circumplex is a spatial arrangement in which related variables or concepts occur in a circular order, with nearby positions representing relatively similar meanings or patterns and opposite positions representing contrasting patterns. In emotion research, the circle usually organizes affective terms or reported states according to pleasantness and activation. The word 'circumplex' refers to more than the visual appearance of a wheel: in formal research it implies testable restrictions on the pattern of associations among variables.
James A. Russell's 1980 article, A Circumplex Model of Affect, brought together ratings of affect-related words, perceived similarities, dimensional ratings, and participants' reports of their current states. Russell reported evidence for a systematic circular ordering. The argument was that terms such as pleasure, excitement, arousal, distress, displeasure, sleepiness, and relaxation occupy related positions rather than constituting independent, interchangeable emotional dimensions. The original article is a model of the structure of affective concepts and reports, not a demonstration that exactly two dedicated neural circuits produce every emotion. Original research.
In later work, Russell developed the concept of core affect: a person's momentary, consciously accessible sense of feeling pleasant or unpleasant and more or less activated. The circumplex can represent core affect; core affect is the proposed psychological state, while the circumplex is the representational and measurement framework. The broader claim that core affect is central to emotional episodes belongs to Russell's psychological construction account and remains theory-dependent. Russell (2003).
The Two Axes: Valence and Arousal
Valence: the pleasant–unpleasant dimension
Valence describes the hedonic quality of an experience: how pleasant, agreeable, disagreeable, or aversive it feels. At the pleasant end are states such as contentment and enjoyment; at the unpleasant end may be worry, misery, irritation, or discomfort. A person might also describe a state as approximately neutral. Researchers often use the word 'positive' for pleasant valence and 'negative' for unpleasant valence, but these are descriptions of how an experience feels, not judgments that it is healthy, unhealthy, morally good, or pathological.
Valence is not a direct measure of whether something helps a person. Fear may feel deeply unpleasant and still guide protective action. Grief may be painful and carry profound relational meaning. Conversely, a pleasant sensation can occur in circumstances that have harmful consequences. The affective coordinates describe a quality of current experience, while adaptive function and long-term outcome require separate investigation.
Arousal: the activation dimension
Arousal refers to the felt or measured level of activation, alertness, or energization. Toward the high-arousal end lie states commonly described as excited, alarmed, agitated, or highly alert. Toward the low-arousal end lie relaxed, sluggish, sleepy, or quiet states. Arousal in this model is a broad psychological dimension and should not be equated mechanically with any one physiological signal or with a person's willingness to act.
Arousal also differs from emotional intensity. Someone may report an intensely sad, immobilized state that has low overt activation, whereas another person may feel mildly excited yet highly energized. Strength of experience, bodily activation, motivational direction, and behavioral expression overlap in some circumstances but are conceptually and empirically distinguishable. The circumplex maps activation; it cannot substitute for asking about intensity, bodily sensation, or the perceived significance of an event. Posner, Russell, and Peterson (2005) discuss dimensional affect and its proposed psychobiological interpretations.
How the dimensions combine
The two axes create four illustrative regions. The upper-right region is relatively pleasant and highly activated; the upper-left is unpleasant and highly activated; the lower-left is unpleasant and less activated; and the lower-right is pleasant and less activated. The center represents an approximately neutral or moderate location on both dimensions. This quadrantal summary is a teaching aid: a person's report may fall anywhere in the continuous space, and words assigned to a quadrant have approximate, context-sensitive positions.
How to Read the Affect Circumplex
Pleasant, high-arousal experiences
Excitement, enthusiasm, exhilaration, and some forms of joy are commonly used as examples of pleasant states with substantial activation. Imagine receiving unexpectedly good news just before stepping on stage. The experience may be unmistakably pleasant and highly energized. Its precise interpretation as pride, anticipation, relief, or joy requires information beyond those two coordinates.
Unpleasant, high-arousal experiences
Alarm, tension, anger, and fear are often mapped as unpleasant and activated. Imagine hearing a loud unexplained crash. A high-arousal, unpleasant rating might be appropriate, but the score cannot establish whether the individual experienced fear, frustration, startle, anger, or some combination. Object, appraisal, bodily context, and personal history matter.
Unpleasant, low-arousal experiences
Dejection, weariness, and some forms of sadness may occupy an unpleasant and relatively low-activation region. A person can feel drained or subdued after disappointment. The diagram makes this distinguishable from agitated distress without claiming that sadness always involves low arousal. Emotional states can change quickly, and some sad experiences are restless or highly activated.
Pleasant, low-arousal experiences
Serenity, relaxation, peacefulness, and contentment are familiar examples of pleasant states with low activation. Someone enjoying a quiet afternoon may be as positively valenced as an excited person at a celebration, yet their bodily and subjective activation differ. This distinction is central to research comparing calm happiness with energized enthusiasm.
Russell's original ordering placed pleasure, excitement, arousal, distress, displeasure, depression, sleepiness, and relaxation at successive illustrative directions around the circle. The word 'depression' in this historical sequence refers to a low-activated unpleasant affect label, not a diagnostic determination about major depressive disorder. The model's angular positions are hypotheses or observed estimates; they are not universal coordinates assigned once and for all to every person's emotions. Russell (1980).
What Do Distance, Direction, and Opposite Positions Mean?
In a circumplex representation, the angle of a term or rating indicates the mixture of valence and arousal it expresses. Two states with nearby angles tend to be conceptually similar or empirically correlated when the model holds. States located roughly opposite one another can show opposing patterns. Angle, position, and distance must be interpreted using the actual estimation and scaling method; an attractive circle drawn in an infographic does not establish that its exact spacing is supported by data.
The center of a two-dimensional rating grid and the center of a formal statistical circumplex are not always interpreted identically. In a self-report grid, the center typically records moderate or neutral valence and arousal. In correlational models, the geometry describes relations among variables that have been centered, standardized, or otherwise transformed. A word can have a strong loading or high communality without the participant experiencing an unusually intense emotion. Mixing individual scores, group averages, and factor loadings is a common source of misleading claims.
A perfect circumplex imposes stronger requirements than merely finding two dimensions. Equal angular spacing and equal distances from the center are often idealized features; weaker circular or quasi-circumplex structures can retain systematic ordering while violating those restrictions. Remmington, Fabrigar, and Visser (2000) emphasized formal model testing rather than relying on suggestive plots. Stanisławski and colleagues (2021) found patterns closer to elliptical than perfectly circular structures in their analyses.
Where the Model Came From
Dimensional approaches to affect predate 1980, but Russell's paper offered an unusually explicit account of how commonly used affect words may form a circular order. The original studies combined multiple techniques, including similarity judgments, ratings on proposed bipolar dimensions, and participants' self-reports. That combination matters because people can organize emotion concepts in one way while reporting current lived states in another; converging results across tasks provide stronger support than any single diagram.
Later investigations elaborated the structure with more finely sampled positions. Yik, Russell, and Steiger (2011) proposed a 12-Point Affect Circumplex and developed self-report scales in several formats. Their four correlational studies examined current and remembered affect, providing a more detailed map than a four-quadrant teaching wheel. Such work strengthens the circumplex as a measurement program, while leaving open the separate question of whether its geometry captures the complete organization of emotions.
What Does the Evidence Actually Show?
Evidence for an organized dimensional structure
The basic observation that affective reports have important pleasantness and activation properties has considerable empirical support. Russell's original studies found an interpretable circular arrangement of descriptors. A 1989 cross-cultural investigation by Russell, Lewicka, and Niit obtained broadly similar circular ordering in several language samples and expression-judgment tasks. Replication across different words, samples, and procedures makes the organization a serious scientific proposal rather than a purely illustrative convention. Russell, Lewicka, and Niit (1989).
In a reanalysis of 47 correlation matrices taken from published datasets, Remmington, Fabrigar, and Visser (2000) reported generally acceptable fit for a model designed to test circumplex relations and often observed strong negative associations between opposing states. This finding supports meaningful structure, but the authors also highlighted weaknesses in some earlier statistical evaluations. Statistical adequacy depends on the exact model, its constraints, its indicators, and the population being studied.
Valence and arousal are not related in one immutable way
The relationship between activation and pleasantness is a particularly revealing problem. In a critical review plus models fitted to eight datasets, Kuppens, Tuerlinckx, Russell, and Barrett (2013) found an average weak V-shaped pattern: experiences toward either extreme of valence tended, on average, to be somewhat more arousing than moderately valenced experiences. At the same time, relationships varied greatly across individuals and situations. The finding challenges a fixed universal formula connecting the two dimensions.
This distinction also explains why calling the axes 'independent' needs care. Orthogonal axes on a drawing mean that the coordinates describe different directions. It does not follow that the two psychological properties are empirically uncorrelated in every person, dataset, emotional context, or choice of items. Nor does statistical separability show that the brain contains two wholly isolated systems. Geometric convenience, observed correlations, and neurophysiological mechanism are three different claims.
The circle is a testable approximation
A 2021 systematic comparison of several proposed affect circumplexes evaluated equal angles, equal communalities, and weaker circular models in a sample of 863 people. Stanisławski, Cieciuch, and Strus (2021) found support for circular ordering but not for a full ideal circumplex in the tested models; patterns were more elliptical, with arousal particularly implicated in deviations. This result does not erase the two-dimensional information. It shows that an exactly symmetrical wheel is a stronger hypothesis than the available measurements uniformly support.
More recently, Nowakowska and Świderska (2026) studied a Polish adaptation of a Circumplex Affect Questionnaire in a daily-diary design with 305 adults. Such instrument-specific research illustrates continuing efforts to validate dimensional measures in particular languages and settings. A successful scale adaptation is evidence about that scale and population; it is not proof that affect has an invariant neural architecture everywhere.
Evidence for richer emotional structure
Valence and arousal do not exhaust the information people give when they name emotions. In a large study of reactions to 2,185 short video clips, Cowen and Keltner (2017) found reports that could be organized into 27 varieties of emotional experience linked by graded transitions. Their results suggest that category labels and richer semantic structures can capture distinctions lost by a simple two-axis summary. These findings arise from the particular stimuli, response formats, and statistical methods used; they do not prove that the number of universal emotions is precisely 27.
A scientifically productive interpretation is that different maps describe different aspects of emotional experience. A low-dimensional representation compresses broad hedonic and activation properties; a richer category space may preserve distinctions such as admiration, envy, awe, and amusement. Each representation gains some tractability and gives up some specificity. The appropriate choice follows the research question rather than a claim that one coordinate system renders every alternative obsolete.
Circumplex Affect, Core Affect, Mood, and Emotion Are Related but Distinct
In the terminology used here, affect is a broad family of affective qualities and states, especially their pleasantness and activation. Core affect is Russell's more specific theoretical construct describing a person's current, accessible sense of pleasure–displeasure and activation. Mood usually describes a more diffuse affective background that may persist without a sharply identified object. An emotion episode may involve appraisal, object or event relevance, bodily change, action preparation, expression, remembered meaning, and a consciously identified category such as fear or gratitude. Researchers disagree on the exact definitions and component relations, so careful writing attributes stronger claims to their framework. Russell (2003).
For example, a person may report unpleasant valence and high activation before an examination. The circumplex locates that current tone. Whether it is apprehension about failure, anger at an unfair procedure, excitement mixed with worry, or physical arousal from caffeine depends on additional information. A report of prolonged low mood likewise describes a time course and context that no single position on an affect wheel can establish.
How Scientists Measure Affect Within This Framework
Affect Grid and direct dimensional ratings
The Affect Grid asks a participant to select a location on a two-dimensional pleasure–arousal space. Russell, Weiss, and Mendelsohn (1989) introduced it as a rapid single-item method and reported evidence of reliability and convergent and discriminant validity across their studies. Its chief advantage is low response burden: a single mark yields a rough coordinate. Its principal trade-off is limited detail about distinct feelings, ambiguity of the scale for some respondents, and restricted opportunities to assess internal consistency within a single observation.
Researchers also administer separate valence and arousal rating scales, emotion adjective inventories, repeated smartphone surveys, or assessments after experimental stimuli. The choice of time frame is consequential: a rating of 'right now' is a different measurement from 'in the past week' or 'how I usually feel.' The context of prompting, response scale anchors, participant language, and ability to distinguish activation from intensity all affect validity.
PANAS and the rotated circumplex
The widely used Positive and Negative Affect Schedule, or PANAS, was designed to measure two sets of affect descriptors, traditionally labeled Positive Affect and Negative Affect. Its factor structure should not be confused with a direct pleasure–displeasure axis. In a rotated representation, high positive activation combines pleasantness with substantial activation, and high negative activation combines unpleasantness with substantial activation. Thus 'positive affect' in this measurement tradition is not synonymous with all pleasant experiences, and 'negative affect' is not synonymous with all unpleasant experiences. Watson, Clark, and Tellegen (1988).
LaRowe, Bohlen, and Williams (2024) reviewed how researchers describe the rotated circumplex and found frequent confusion between activation factors and hedonic valence. Their analysis of 140 sampled empirical articles identified the mistaken interpretation in more than half of the articles they classified as using that model. The practical lesson is to name the measured construct and its item content precisely instead of treating the words 'positive' and 'negative' as a universal measurement key.
Multidimensional scaling and circumplex tests
Multidimensional scaling places concepts in a space according to similarity or dissimilarity judgments. Factor analysis and structural equation models estimate dimensions and relations among reported items. A circumplex-specific test additionally asks whether the relative angles, correlations, and shared variance follow the model's restrictions. Graphs based on averaged adjective ratings, within-person momentary reports, and semantic comparisons may look similar but answer different questions. Remmington et al. (2000) and Yik et al. (2011) offer examples of the latter, more formal research tradition.
Good measurement practice specifies whose affect is being measured, at what moment, on which scale, under which conditions, and for what inferential goal. Test–retest reliability, sensitivity to change, cross-language measurement invariance, and evidence that people interpret the two axes as intended require empirical checking. A precise-looking numeric coordinate does not make an imprecise self-report into a physiologically exact reading.
What the Circumplex Model Is Useful For
Comparing broad patterns across time
Repeated valence and arousal ratings can summarize whether a participant's reported affect shifts toward comfort, distress, calmness, or excitement across a day, experiment, or activity. Plotting trajectories can reveal patterns worth investigating and may reduce respondent burden compared with repeatedly administering long lists of emotion categories. The inference remains descriptive unless a study design supports causal conclusions. Repeated ratings can also be affected by how often people are prompted and how the prompts interfere with experience.
Designing studies and choosing stimuli
Researchers often aim to select experiences or stimuli varying in pleasantness and activation—for example, calming music, exciting images, or distressing scenes. The two-dimensional design helps compare materials without assuming that each belongs to exactly one universal emotion category. Individual ratings should still be gathered: a video that is highly arousing on average may be boring, frightening, or joyful to different viewers.
Exercise, everyday activities, and affective change
Exercise psychology offers one setting where immediate pleasantness and activation are relevant but distinct outcomes. Ekkekakis and Petruzzello (2002)00028-0) argued that affect circumplex measures can capture broad responses without favoring any one emotion category. The model helps researchers describe how an activity feels at a given moment; it does not independently establish that the activity produces lasting improvements in mood or health.
Clinical research and mental-health communication
Dimensional ratings can help characterize experiences appearing in many psychological and medical contexts, such as an unpleasant energized state or a pleasant subdued one. They may support research on patterns of affect, but no coordinate constitutes a diagnosis, symptom threshold, or treatment indication. Depression, anxiety disorders, mania, trauma-related symptoms, and other clinical conditions require criteria, history, duration, impairment, and contextual assessment beyond an affect map. Posner et al. (2005) offered a theoretical application to psychopathology, which should be read as a research proposal rather than a diagnostic rule.
Human–computer interaction and machine-learning research
Designers and researchers may use the valence–arousal space as a labeling scheme for participant reports in interactive technologies or emotion-related datasets. That is a practical coding decision. A computer model's prediction of these labels is not direct access to the person's subjective experience. Facial movement, vocal characteristics, and physiological signals vary with social context, culture, individual physiology, and other causes. An important scientific review found that facial movements cannot be treated as a transparent, one-to-one code for internal emotions. Barrett and colleagues (2019).
Major Limitations of the Circumplex Model
Two coordinates compress distinct emotional meanings
Fear and anger can both be unpleasant and highly activating, yet differ in appraisals, perceived threats, motivational tendencies, and social meaning. Pride and excitement may occupy nearby regions while reflecting different social and self-evaluative processes. The circumplex intentionally summarizes broad dimensions. It cannot substitute for measures designed to capture specific emotion categories, social meanings, appraisal patterns, or the person's own description.
The geometry varies by method, person, and context
Many presentations show a clean, symmetric circle with fixed named emotions around its perimeter. Empirical results support important dimensional ordering but reveal departures from ideal spacing and person-specific relations between arousal and valence. Scale wording, available response options, time horizon, cultural meanings, and statistical transformations can change the observed shape. The evidence from Kuppens et al. (2013) and Stanisławski et al. (2021) warrants treating the circle as a useful model subject to tests.
Cultural variation matters
Finding an approximately circular ordering in multiple languages is valuable evidence for shared features of affect concepts; it does not imply that every term translates one-to-one or that every culture values the same pleasant state. The cultural salience of low- versus high-activation pleasantness, the situations in which people label an emotion, and their response styles may vary. Cross-cultural use requires suitable translation, local validation, and attention to within-culture diversity. Russell, Lewicka, and Niit (1989).
Mixed emotions and dynamic emotional episodes
A single point reports one compressed coordinate at a chosen time. A person may simultaneously feel gratitude and sadness, or excitement and dread. An averaged valence score can fall near neutral despite two strong experiences of opposite hedonic qualities. Multiple-category questions, time-sensitive repeated assessment, and open-ended descriptions can retain information that is lost in one coordinate pair. A moving point over time also cannot show all overlapping feelings without extra measurement channels.
Physiological activation is not uniquely interpretable
Heart rate, skin conductance, movement, facial activity, and other observable measures are influenced by exercise, medication, physical conditions, temperature, attention, and multiple emotional states. Strong physiological activation may coexist with very different appraisals and subjective reports. The circumplex supplies dimensions for comparison; it does not prove that a particular biosignal reveals a specific emotion. Barrett et al. (2019) offer broader reasons to avoid inferring inner states mechanically from outward behavior.
Research measures are not interchangeable
Affect Grid coordinates, PANAS activation factors, intensity ratings, semantic similarity judgments, and categorical emotion reports target overlapping but distinct constructs. The same person can yield different patterns when the question asks how they feel now, how they have felt this month, or which emotion best describes a particular event. A rigorous study states which interpretation its instrument supports, uses an appropriate comparison measure, and avoids treating every result as evidence for a single universally fixed circumplex.
Circumplex Model Versus Other Models of Emotion
Circumplex versus basic emotion theories
Basic emotion approaches investigate whether certain emotion categories have distinctive evolutionary, expressive, physiological, or functional features. The circumplex instead describes broad affective similarities using continuous dimensions. They can be compared empirically—for example, whether valence–arousal ratings account for all reliable variation in self-reported anger, fear, disgust, or joy. Different theories offer different predictions, and findings vary by measurement domain. Dimensional location cannot by itself decide the evolutionary or biological status of an emotion category.
Circumplex versus constructionist accounts
The circumplex is frequently associated with psychological construction perspectives because valence and arousal describe broadly shared affective ingredients. A constructionist account goes further by offering hypotheses about how concepts, prediction, bodily signals, learning, and situations shape emotion episodes. Russell's theory of core affect is one influential account in this family; accepting a two-axis affect map does not logically require accepting every mechanism proposed in a constructionist theory. Russell (2003).
Circumplex versus appraisal approaches
Appraisal approaches examine how a person evaluates the relevance of events to goals, control, certainty, responsibility, or other dimensions. These appraisals may help distinguish emotional episodes that share similar affect coordinates. A circumplex rating can say that fear and anger are both unpleasant and activated while appraisal questions examine what the person thinks is happening and how they might respond. These frameworks can address complementary levels of the same event.
Circumplex versus Plutchik's emotion wheel
The circumplex of affect and Plutchik's wheel are commonly confused because both may appear circular. Russell's model organizes affect along valence and arousal, using spatial proximity to represent measured or proposed relations among feeling states. Plutchik's wheel represents a different theoretical taxonomy of emotion families, combinations, and intensity. A circular graphic is therefore insufficient to identify the underlying psychological theory; its axes, constructs, and testable assumptions must be specified.
How to Use the Model Carefully in Practice
A careful use begins with a clearly defined question. If a person wants to describe their current state, ask separately, 'How pleasant or unpleasant does this feel?' and 'How activated or sleepy do I feel?' Treat the answers as approximate self-reports. Then, if meaningful, ask what is happening, what name the person gives the experience, whether more than one emotion is present, and what has changed over time. The added information restores the context intentionally omitted by the compact map.
For a research study, specify the time frame, population, response scale, instructions, statistical model, and planned interpretation before collecting data. Test the properties of the chosen instrument and inspect whether the hypothesized circular structure fits those data. When reporting results, distinguish a participant's state from a group-level map and a predicted label from a confirmed subjective report. This approach preserves the circumplex model's practical advantages without granting it explanatory powers it has not earned.
The central conclusion is straightforward: valence and arousal give researchers and readers a powerful first description of affect. Their combination can reveal whether an experience is pleasant or unpleasant, calming or energizing, and how reports change across circumstances. Understanding which emotion is present, why it arose, how it is expressed, and what it means requires additional evidence and a richer account of human experience.
Frequently Asked Questions
Who developed the circumplex model of affect?
James A. Russell published the influential modern formulation in 1980 in the Journal of Personality and Social Psychology. Earlier dimensional work helped shape the field, and later researchers expanded, revised, and tested circumplex structures. Russell (1980).
What are the two main dimensions?
Valence (pleasant to unpleasant) and arousal (high to low activation). Both dimensions are continuous and can be reported separately. They describe qualities of affective experience rather than provide a list of diagnostic states.
What are the four quadrants of the affect circumplex?
Pleasant/high activation, unpleasant/high activation, unpleasant/low activation, and pleasant/low activation. Examples often include excitement, distress, dejection, and calm contentment, respectively. Actual emotion labels and individual ratings need not occupy exactly these positions.
Does the circumplex prove that emotions are only two-dimensional?
No. It demonstrates the usefulness of summarizing important affective variation along two broad dimensions and proposes a testable geometry. Other research identifies additional emotional distinctions and category structure beyond two dimensions. Cowen and Keltner (2017).
Are valence and arousal independent?
They are conceptually separable axes. Their observed statistical association depends on the person, situation, and measurement design. An average weak V-shaped relation and substantial individual differences were reported in a comprehensive analysis. Kuppens et al. (2013).
Is the circumplex model the same as an emotion wheel?
A circumplex is often drawn as a wheel, but the scientific model includes specific dimensional and correlational claims. Other emotion wheels, including Plutchik's, arise from different theories and cannot be treated as interchangeable visualizations.
What is the difference between the circumplex and core affect?
The circumplex is a map or proposed structure of affective relations. Core affect is a proposed moment-to-moment psychological state characterized by pleasantness and activation. Russell's later theory relates them but distinguishes the representation from the experience being represented. Russell (2003).
What is the Affect Grid used for?
It is a fast single-item method for indicating a position on the pleasure–arousal space. It can be useful when repeated or brief measurement matters, but it does not identify all specific emotion categories or explain causes. Russell, Weiss, and Mendelsohn (1989).
Are positive and negative affect the same as valence?
Not necessarily. Positive and negative affect factors in PANAS-style research frequently emphasize activation and correspond to rotated directions in a circumplex. Pleasantness and unpleasantness are opposite poles of the valence dimension, whereas the traditional PANAS factors are different measurement constructs. LaRowe et al. (2024).
Can a circumplex identify depression, anxiety, or bipolar disorder?
No. Temporary affect ratings are not diagnostic instruments for those disorders. Clinical assessment considers symptom patterns, duration, impairment, medical context, and validated diagnostic criteria. The wheel may be used descriptively in some research settings.
Can AI determine a person's emotions from valence and arousal predictions?
An AI model can estimate labels or patterns in data that are associated with affect reports. Such an estimate is a statistical prediction, not direct knowledge of a person's conscious experience. Contextual information and human self-report remain essential to responsible interpretation. Barrett et al. (2019).
Is the empirical affect circumplex always a perfect circle?
No. Some datasets approximate circular patterns, while others fit an ellipse or depart from ideal angular spacing and communalities. Testing model fit matters more than drawing a visually tidy wheel. Stanisławski et al. (2021).
For additional scientific context, see Core Affect: Valence, Arousal, and Subjective Experience.
Related Articles
For adjacent English-language topics, see Sleep and Emotional Regulation: Effects on Mood and Stress for one applied relationship involving affective states; Anhedonia in Depression: Loss of Interest, Pleasure, and Motivation for the clinical distinction between hedonic experience and depression-related symptoms; and Bipolar Disorder vs Mood Swings: When Changes Become Clinical Episodes for the boundary between ordinary mood variation and clinical episodes.
References
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. https://doi.org/10.1177/1529100619832930
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. https://doi.org/10.1073/pnas.1702247114
Ekkekakis, P., & Petruzzello, S. J. (2002). Analysis of the affect measurement conundrum in exercise psychology: IV. A conceptual case for the affect circumplex. Psychology of Sport and Exercise, 3(1), 35–63. https://doi.org/10.1016/S1469-0292(01)00028-000028-0)
Kuppens, P., Tuerlinckx, F., Russell, J. A., & Barrett, L. F. (2013). The relation between valence and arousal in subjective experience. Psychological Bulletin, 139(4), 917–940. https://doi.org/10.1037/a0030811
LaRowe, L. R., Bohlen, L. C., & Williams, D. M. (2024). There is no happiness in positive affect: The pervasive misunderstanding of the rotated circumplex model. Frontiers in Psychology, 15, 1301428. https://doi.org/10.3389/fpsyg.2024.1301428
Nowakowska, I., & Świderska, J. (2026). Polish adaptation of the Circumplex Affect Questionnaire (CAQ). Journal of Psychopathology and Behavioral Assessment, 48, Article 29. https://doi.org/10.1007/s10862-026-10292-6
Posner, J., Russell, J. A., & Peterson, B. S. (2005). The circumplex model of affect: An integrative approach to affective neuroscience, cognitive development, and psychopathology. Development and Psychopathology, 17(3), 715–734. https://doi.org/10.1017/S0954579405050340
Remmington, N. A., Fabrigar, L. R., & Visser, P. S. (2000). Reexamining the circumplex model of affect. Journal of Personality and Social Psychology, 79(2), 286–300. https://pubmed.ncbi.nlm.nih.gov/10948981/
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161–1178. https://doi.org/10.1037/h0077714
Russell, J. A. (2003). Core affect and the psychological construction of emotion. Psychological Review, 110(1), 145–172. https://doi.org/10.1037/0033-295X.110.1.145
Russell, J. A., Lewicka, M., & Niit, T. (1989). A cross-cultural study of a circumplex model of affect. Journal of Personality and Social Psychology, 57(5), 848–856. https://doi.org/10.1037/0022-3514.57.5.848
Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect Grid: A single-item scale of pleasure and arousal. Journal of Personality and Social Psychology, 57(3), 493–502. https://doi.org/10.1037/0022-3514.57.3.493
Stanisławski, K., Cieciuch, J., & Strus, W. (2021). Ellipse rather than a circumplex: A systematic test of various circumplexes of emotions. Personality and Individual Differences, 181, 111052. https://doi.org/10.1016/j.paid.2021.111052
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. https://pubmed.ncbi.nlm.nih.gov/3397865/
Yik, M., Russell, J. A., & Steiger, J. H. (2011). A 12-Point Circumplex Structure of Core Affect. Emotion, 11(4), 705–731. https://doi.org/10.1037/a0023980