Expressed Emotion

Abstract

Expressed emotion (EE) is a measure of the emotional climate in a patient's home, indexed by the attitudes relatives voice about the patient during a standardized interview. As a construct in social psychology, it captures five components — critical comments, hostility, emotional over-involvement, warmth, and positive remarks — of which the first three define a household as high-EE. Its importance rests on one durable finding: patients who return to a high-EE home relapse at roughly twice the rate of those in a low-EE home, a relationship holding across schizophrenia, mood disorders, and eating disorders. This article covers what EE is, its component structure, how it is measured by the Camberwell Family Interview and Five-Minute Speech Sample, its predictive validity, and the cognitive model of caregiving that explains it.

Keywords: expressed emotion, criticism, emotional over-involvement, relapse, caregiving

Few constructs in clinical psychology have travelled as far from their origin as expressed emotion. It began as an attempt to explain a puzzling epidemiological observation — that discharged psychiatric patients who returned to live with relatives fared worse than those sent to hostels — and became one of the most robust psychosocial predictors of relapse in the whole of psychiatry (Brown, Birley, & Wing, 1972). EE is not a measure of how much emotion a family feels, nor a judgement of blame; it is a specific, reliably rated index of the critical, hostile, or over-involved attitudes a relative expresses when speaking about a patient, and it forecasts the course of illness with a consistency that has survived five decades of replication (Butzlaff & Hooley, 1998).

What Expressed Emotion Is

Expressed emotion is defined operationally by the attitudes a relative voices about a patient during a semi-structured interview, and theoretically by what those attitudes predict about the course of the patient's illness. The construct grew from a specific empirical surprise. In the late 1950s George Brown noticed that men discharged from psychiatric hospitals to live with wives or parents relapsed more often than those who went to live in lodgings or with siblings — the closer the emotional tie, paradoxically, the worse the outcome (Brown et al., 1972). The explanation was not the relationship itself but its emotional tenor: some households were marked by criticism and over-involvement, and it was those, not close relationships in general, that predicted return of symptoms.

EE is therefore best understood as a property of the relative's expressed attitude, assessed at a single point — usually around the time of an acute admission — and used to predict relapse over the following months. It is measured by trained raters who listen to how a relative talks about the patient, counting critical remarks, judging the presence of hostility, and scoring emotional over-involvement. The rating is remarkably reliable between trained coders and, once established, is a stable characteristic of the household over the medium term (Vaughn & Leff, 1976b).

The construct must be distinguished from adjacent ones. It is not emotional regulation, which concerns how an individual manages their own affective states; it is not affect or the patient's own emotions, but the relative's expressed stance toward the patient; and it is not a diagnosis or a judgement of a family's worth. High EE is common, understandable, and often a response to the strain of caregiving rather than a cause of illness in the first place — a point the cognitive model below makes central.

The Components of Expressed Emotion

EE is not a single dimension but a profile of five rated components, three of which carry the predictive weight. Critical comments are statements of dislike or disapproval of the patient's behaviour, counted as a frequency across the interview; a comment qualifies by its content or by a critical tone of voice. Hostility is a more generalized, global rejection of the patient as a person — criticism that has spread from specific behaviours to the whole individual. Emotional over-involvement (EOI) is exaggerated, self-sacrificing, or over-protective behaviour by the relative, including dramatic accounts of distress, constant reassurance-seeking, and an inability to lead a separate life (Vaughn & Leff, 1976b).

The remaining two components — warmth and positive remarks — capture the constructive side of the relationship, but they do not enter the standard high/low classification and have weaker and less consistent associations with outcome. A household is rated high-EE if the relative makes six or more critical comments, shows any hostility, or scores at the top of the emotional-over-involvement scale; falling below all three thresholds classifies the household as low-EE (Vaughn & Leff, 1976a). This cutoff, derived empirically from the distribution that best predicted relapse, is what turns five continuous ratings into the single dichotomous predictor used throughout the literature.

The first demonstration lets a reader set the three key ratings and see how the standard rule classifies a household.

Criticism and hostility tend to co-occur and are thought to share an underlying dimension of rejection, while emotional over-involvement is partly separable — a relative can be over-involved without being critical, and the two components sometimes respond differently to intervention (Hooley, 2007).

Measuring Expressed Emotion

The reference standard for measuring EE is the Camberwell Family Interview (CFI), a semi-structured interview with a relative conducted around the time of the patient's admission. Originally lasting several hours, the CFI elicits an account of the patient's symptoms, the household's routines, and the relationship, while trained raters count critical comments and score hostility, over-involvement, warmth, and positive remarks from an audio recording (Vaughn & Leff, 1976b). Its psychometric properties are excellent, but its length and the specialist training it demands make it impractical for routine clinical use, and this cost is the single greatest obstacle to using EE outside research.

That obstacle motivated a briefer instrument. The Five-Minute Speech Sample (FMSS) asks the relative to talk without interruption for five minutes about the patient and their relationship; raters then classify the sample as high- or low-EE from the opening statement, the quality of the relationship described, and the frequency of critical or over-involved remarks (Magaña et al., 1986). The FMSS is far quicker and needs less training, and it agrees substantially with the CFI, though it tends to identify fewer high-EE households — its brevity trades a little sensitivity for a large gain in feasibility.

A related but distinct measure is perceived criticism — a single question asking the patient how critical their relative is of them, rated on a ten-point scale. Despite its simplicity it predicts relapse, sometimes as well as the interview-based measures, suggesting that the patient's perception of the family climate carries independent predictive weight (Hooley & Teasdale, 1989).

The second demonstration contrasts the relapse rates associated with high and low EE across several disorders.

Predictive Validity and Generality

The reason EE became a central construct is its predictive power. Across studies, patients with schizophrenia discharged to a high-EE household relapse over the following nine to twelve months at roughly two to three times the rate of those in a low-EE household — commonly around 50–65% versus 20–35% (Bebbington & Kuipers, 1994). The aggregate analysis by Bebbington and Kuipers, pooling data from many studies across several countries, established that the association was not an artefact of any one sample or culture, and that its size was clinically meaningful. A comprehensive overview of the schizophrenia literature reaches the same conclusion, and situates EE within the broader set of psychosocial factors shaping the disorder's course (Amaresha & Venkatasubramanian, 2012).

Note. The pathway runs from a Camberwell Family Interview rating at admission to relapse over nine-month follow-up. High-EE households carry roughly 2.4 times the relapse risk of low-EE households (Bebbington & Kuipers, 1994).

The definitive statement of EE's generality came from a meta-analysis of 27 studies, which confirmed that high EE reliably predicts relapse in schizophrenia and, crucially, that the effect is even stronger for mood disorders and eating disorders than for the psychotic illness in which it was discovered (Butzlaff & Hooley, 1998). This was a pivotal result: it moved EE from a schizophrenia-specific finding to a trans-diagnostic marker of the family environment's influence on the course of illness. Subsequent work has extended the finding to depression, bipolar disorder, eating disorders, and physical health conditions, wherever a patient's outcome depends on an ongoing relationship with a caregiver (Wearden et al., 2000; Rienecke, 2018).

Two cautions temper the picture. First, EE is a correlational predictor measured at one time point; a high-EE attitude may partly reflect the burden of caring for a more severely ill patient rather than acting purely as a cause of relapse. Second, the direction of influence is bidirectional — patient symptoms provoke critical responses, which in turn worsen symptoms — so EE is best read as one node in a reciprocal system, not a one-way cause.

The Cognitive Model of Caregiving

Why do some relatives become critical or over-involved while others, facing the same illness, do not? The leading answer is a cognitive model in which the relative's appraisals of the illness drive their emotional response (Kuipers, Onwumere, & Bebbington, 2010). A relative who attributes a patient's symptoms to factors within the patient's control — laziness, wilful difficulty — tends toward criticism and hostility; one who attributes the same symptoms to an uncontrollable illness tends toward acceptance, and one who over-identifies with the patient's suffering tends toward emotional over-involvement. Criticism, on this view, is not a personality flaw but the downstream product of a particular, and correctable, way of understanding the patient's behaviour.

The model is explicitly a system: the relative's appraisal shapes their expressed attitude, the attitude affects the patient's stress and symptoms, and the patient's behaviour feeds back to confirm or revise the relative's appraisal. High-EE and low-EE households settle into self-sustaining cycles — a critical relative and a symptomatic patient reinforce each other, while an accepting relative and a stabilizing patient do the same in the benign direction. Because the entry point is the relative's cognition, the model predicts that changing how relatives understand the illness should change the emotional climate, and this is the rationale for family intervention.

The third demonstration renders this appraisal loop and lets a reader see how a relative's attribution shifts the household toward a high- or low-EE cycle.

The cognitive model also reframes EE ethically. Because criticism and over-involvement are understandable responses to genuine strain and to a natural but mistaken reading of symptoms, the construct points toward supporting relatives rather than blaming them — a stance that has shaped a generation of psychosocial interventions (Kuipers et al., 2010).

Worked Example

Consider a rater scoring a Camberwell Family Interview with the mother of a young man recovering from a first episode of schizophrenia. Across the interview the rater counts the mother's critical comments — statements such as “he never lifts a finger” and “I'm sick of the mess” — and tallies seven of them. She shows no generalized hostility (her criticism stays tied to specific behaviours rather than rejecting him as a person), and her emotional-over-involvement score is moderate, a 2 on the 0–5 scale.

Applying the standard rule, the household is classified high-EE on the strength of the critical-comment count alone: seven comments meets the threshold of six or more, and any one of the three criteria — criticism ≥ 6, any hostility, or EOI ≥ 3 (a top-of-scale rating) — is sufficient. The moderate EOI and absent hostility do not lower the classification, because the rule is disjunctive, not additive.

The clinical implication follows from the predictive data. Pooled estimates put nine-month relapse at roughly 50% for patients returning to a high-EE home versus about 21% for a low-EE home (Bebbington & Kuipers, 1994). Expressed as a relative risk, 0.50 ÷ 0.21 ≈ 2.4 — the high-EE household carries roughly two-and-a-half times the relapse risk. That single number is what makes this mother a priority for family intervention: reducing her critical comments below the threshold is a concrete, measurable target with a documented effect on outcome. The first demonstration lets a reader vary the three ratings and confirm how a single component can flip the classification.

Discussion

Expressed emotion is one of the strongest and most replicated psychosocial predictors in psychiatry, and its trans-diagnostic reach — from schizophrenia to depression to eating disorders — makes it a rare example of a family-level variable with generality across very different illnesses (Butzlaff & Hooley, 1998). What remains genuinely debated is mechanism. Whether high EE acts primarily as a stressor that overwhelms a vulnerable patient's coping, as a signal of a more severe underlying illness, or as one half of a reciprocal cycle in which patient and relative shape each other, is not settled by the predictive data alone, because each account is compatible with the same correlation.

The construct also carries a cautionary history. Early presentations risked implying that families caused relapse, an inference that stigmatized relatives already under strain and that the evidence does not support. The modern, cognitively grounded reading is the corrective: EE reflects appraisals and burden as much as it drives outcome, criticism is an intelligible response to a difficult situation, and the appropriate clinical response is to support and educate relatives, not to blame them (Kuipers et al., 2010; Cherry et al., 2017). Family interventions built on this principle, which lower EE by changing how relatives understand the illness, reduce relapse and are recommended in treatment guidelines for schizophrenia. The founding demonstration was a controlled trial in which lowering relatives' EE — by reducing critical comments or the amount of face-to-face contact — cut the nine-month relapse rate from roughly half to under a tenth relative to routine care (Leff et al., 1982).

Current Directions

Three strands define the field's active front. The first is cross-cultural: a meta-analytic review of caregiving and EE in psychosis across cultures found that the association between EE and outcome generalizes internationally, but that the base rate of high EE and the relative weight of criticism versus over-involvement vary systematically with cultural norms about family obligation and emotional expression (O'Driscoll et al., 2019). This work cautions against applying a single Western cutoff uncritically and motivates culturally calibrated measurement.

The second strand extends EE beyond psychosis and beyond the patient. Updated reviews document robust EE effects in eating disorders, where high parental criticism predicts poorer treatment response, sharpening EE into a target for family-based treatment (Rienecke, 2018). A parallel line examines the caregiver's own wellbeing — the guilt, shame, and burden that accompany high EE — reframing the relative not merely as a predictor of the patient's course but as a person with needs of their own; a systematic review found that guilt and shame in carers are closely tied to critical and over-involved attitudes (Cherry et al., 2017). The third strand is mechanistic and interventional: refining the cognitive model of caregiving into carer-focused interventions that target appraisals directly, and testing whether lowering EE mediates their effect on relapse (Kuipers et al., 2010).

Common Misconceptions

High expressed emotion means a family caused the patient's illness.

EE predicts the course of an already-diagnosed illness — the risk of relapse — not its origin, and the relationship is reciprocal: patient symptoms provoke criticism as much as criticism worsens symptoms. High EE is often an understandable response to the strain of caregiving, and the modern cognitive model treats it as a target for support rather than a mark of blame (Kuipers et al., 2010).

Expressed emotion measures how much emotion a family shows.

Despite its name, EE is a specific rated index of critical, hostile, and over-involved attitudes voiced about the patient, not a gauge of emotional intensity in general. A warm, expressive family can be low-EE, and an emotionally restrained one can be high-EE if its remarks are critical (Vaughn & Leff, 1976b).

Expressed emotion matters only in schizophrenia.

Although EE was discovered in schizophrenia, meta-analysis shows the relapse effect is actually stronger in mood disorders and eating disorders, and the construct predicts outcome across a wide range of conditions in which a patient depends on a caregiving relationship (Butzlaff & Hooley, 1998).

Glossary

Camberwell Family Interview (CFI).

The semi-structured reference interview from which expressed emotion is rated, conducted with a relative around the time of the patient's admission.

Critical comments.

Statements of dislike or disapproval of the patient's behaviour, counted as a frequency; six or more classifies a household as high-EE.

Emotional over-involvement (EOI).

Exaggerated, self-sacrificing, or over-protective behaviour by a relative toward the patient; one of the three components that define high EE.

Emotional regulation.

The processes by which a person influences their own emotional states; distinct from EE, which rates a relative's expressed attitude toward another.

Expressed emotion (EE).

A rated measure of the critical, hostile, and over-involved attitudes a relative voices about a patient, used to predict relapse.

Family intervention.

A psychosocial treatment that educates and supports relatives to lower expressed emotion, with demonstrated reductions in patient relapse.

Five-Minute Speech Sample (FMSS).

A brief measure of EE in which a relative speaks uninterrupted for five minutes about the patient, then is classified high- or low-EE.

High-EE household.

A home in which the relative makes six or more critical comments, shows any hostility, or scores at the top of the emotional-over-involvement scale.

Hostility.

Generalized rejection of the patient as a person, rather than criticism of specific behaviours; one of the three high-EE components.

Low-EE household.

A home falling below all three thresholds for criticism, hostility, and over-involvement; associated with markedly lower relapse.

Perceived criticism.

A single-item measure asking patients how critical their relative is of them; predicts relapse despite its brevity.

Positive remarks.

Statements of praise or approval of the patient; a component rated in the CFI but not part of the high/low classification.

Relapse.

The return or marked worsening of symptoms after a period of improvement; the primary outcome EE predicts.

Warmth.

Sympathy, affection, and concern expressed toward the patient; a constructive component rated but not counted in the high-EE cutoff.

Key Researchers

George W. Brown (b. 1930). Medical sociologist at Royal Holloway, University of London; introduced the expressed-emotion concept through the Camberwell Family Interview and the founding studies of relapse in schizophrenia. Wikipedia - Wikidata

Jill M. Hooley (living). Professor of psychology at Harvard University; conducted the meta-analysis establishing EE's trans-diagnostic predictive validity and developed the perceived-criticism paradigm. Faculty Page - ORCID - Google Scholar - Wikipedia

Elizabeth Kuipers (living). Professor of clinical psychology at the Institute of Psychiatry, Psychology & Neuroscience, King's College London; developed the cognitive model of caregiving in psychosis and the aggregate analysis of EE's predictive utility. Faculty Page - ORCID - Google Scholar - Wikipedia

Julian P. Leff (1938–2021). Psychiatrist at the Institute of Psychiatry, King's College London; co-developed the EE rating scales and pioneered the family-intervention trials that lowered EE and reduced relapse. Faculty Page - Wikidata

David J. Miklowitz (b. 1957). Professor of psychiatry at the University of California, Los Angeles; developed family-focused therapy for bipolar disorder and helped create the Five-Minute Speech Sample. Faculty Page - ORCID - Google Scholar - Wikidata

Juliana Onwumere (living). Clinical psychologist at the Institute of Psychiatry, Psychology & Neuroscience, King's College London; studies caregiving processes in psychosis and carer-focused interventions. Faculty Page - ORCID

Frequently Asked Questions

What is expressed emotion in psychology?
Expressed emotion is a measure of the emotional climate in a patient's home, rated from the critical, hostile, and over-involved attitudes a relative voices about the patient during a standardized interview, and used to predict relapse (Brown et al., 1972).

What are the components of expressed emotion?
EE has five rated components: critical comments, hostility, emotional over-involvement, warmth, and positive remarks, of which the first three define a household as high-EE (Vaughn & Leff, 1976b).

What makes a household high-EE?
A household is classified high-EE if the relative makes six or more critical comments, shows any hostility, or scores at the top of the emotional-over-involvement scale; any one criterion is sufficient (Vaughn & Leff, 1976a).

How is expressed emotion measured?
The reference standard is the Camberwell Family Interview, with the shorter Five-Minute Speech Sample used when time and training are limited; a single perceived-criticism question offers a brief patient-report alternative (Magaña et al., 1986).

Does expressed emotion predict relapse?
Yes; patients returning to a high-EE home relapse at roughly two to three times the rate of those in a low-EE home, an effect confirmed by meta-analysis across several disorders (Butzlaff & Hooley, 1998).

Does expressed emotion matter outside schizophrenia?
Yes; the relapse effect is actually stronger for mood disorders and eating disorders than for schizophrenia, and EE predicts outcome across many conditions involving a caregiving relationship (Butzlaff & Hooley, 1998; Rienecke, 2018).

Does high expressed emotion mean the family is to blame?
No; EE predicts the course of an existing illness, not its cause, and high EE is often an understandable response to caregiving strain, which is why the cognitive model treats it as a target for support rather than blame (Kuipers et al., 2010).

Can expressed emotion be changed?
Yes; family interventions that educate relatives and reshape how they understand the illness lower EE and reduce relapse, and are recommended in treatment guidelines for schizophrenia (Kuipers et al., 2010).

References