Dr. Lieberman is affiliated with the Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York City, and with the New York State Psychiatric Institute, New York City. Dr. Drake is with the Department of Community and Family Medicine and the Department of Psychiatry, Dartmouth Medical School, Hanover, New Hampshire. Dr. Sederer is medical director of the New York State Office of Mental Health, New York City. Dr. Belger, Dr. Perkins, and Dr. Stroup are with the Department of Psychiatry, University of North Carolina School of Medicine, Chapel Hill. Dr. Keefe is with the Department of Psychiatry and the Department of Behavioral Sciences, Duke University School of Medicine, Durham, North Carolina. Send correspondence to Dr. Lieberman at the New York State Psychiatric Institute, 1051 Riverside Dr., New York, NY 10032 (e-mail: jlieberman@columbia.edu). Preliminary results of this study were presented at the annual meeting of the American Psychiatric Association, May 1—6, 2004, New York City.

Two conceptualizations of serious mental illness—the "broken brain" and "recovery" models—have dominated recent attempts to understand and treat severe and persistent mental disorders, such as schizophrenia (1,2). Studies of the neurobiology of schizophrenia have identified pathological changes and mechanisms in the brain associated with vulnerability and the manifest illness (3). Research reveals that neuropathological brain changes and information-processing deficits in schizophrenia are mild preceding the onset of symptoms but worsen after acute illness and often progress with subsequent clinical episodes. These findings imply irreversible loss of brain volume and deterioration of mental functions. At the same time, a more hopeful view of the course of illness emerges from the recovery movement, which emphasizes developing a meaningful life beyond illness, and this view is supported by long-term outcome studies, which consistently show heterogeneity, with many patients improving and in some cases achieving substantial remission of illness (4,5).

Although these two models may seem contradictory, the Surgeon General's report on mental health (5) and the report of the President's New Freedom Commission on Mental Health (6) attempt to bridge them. Indeed, the report of the New Freedom Commission (6), intended to be a roadmap for the mental health system, states that recovery is the "single most important goal" of people served by the mental health system and that we should "accelerate research to promote recovery." In this article, we suggest a working definition of recovery that may facilitate communication and permit a research-based synthesis of the two models. Following a brief historical overview of models for understanding schizophrenia, we discuss recent research on course and treatment in relation to recovery.

Both neuroscience and recovery models of schizophrenia have deep historical roots. In the original conceptualization of dementia praecox by Kraepelin (7), the illness was seen as progressive and deteriorating. This pessimistic view was reinforced for many years by the limitations of existing treatments (8) and recently by research on brain abnormalities and neurodevelopmental theories of schizophrenia (summarized below). The presumption of a deteriorating course and structural brain abnormalities has sometimes fostered an attitude of therapeutic nihilism (9). The neurodevelopmental theory postulates that etiologic and pathogenic factors occur long before the onset of the illness (probably in gestation), disrupt the course of normal neural development, and produce alterations of specific neural circuits that confer vulnerability and ultimately lead to biological and psychosocial malfunction (10,11,12). According to this view, once the neurodevelopmental diathesis of schizophrenia has been established, the illness runs an inexorable course.

Kraepelin's pessimistic view and circular reasoning (that is, course was used to determine diagnosis) were challenged by Bleuler (13), who observed that patients with apparent dementia praecox constituted a heterogeneous "group of schizophrenias" with highly variable outcomes. He noted that some patients' symptoms remitted substantially or even completely, with return to normal functioning, and that the modal course was one of eventual improvement, which he termed "healing with scarring." Since Bleuler's early observations, the evidence from numerous long-term follow-up studies has supported this heterogeneity, the capacity for symptomatic remission and functional improvement early in the course of the illness, and also, though less completely, improvements in later life (14,15,16). Mental health advocates have embraced the longitudinal research as evidence of the potential for recovery.

Much of the tension between neuroscientists' emphasis on brain pathologies and mental health advocates' emphasis on recovery appears to arise from the lack of definitional clarity. Recovery has been defined in literally dozens of ways (17). Neuroscientists, on the one hand, often assume that recovery connotes the complete absence of disease or a cure. Even current studies of outcome and recovery tend to emphasize returns to normal function (8,18,19,20,21,22,23,24,25). Advocates, on the other hand, often use recovery to describe a process of managing one's mental illness, moving beyond its devastating psychological effects, and pursuing a personally meaningful life in the community (26,27,28,29,30). This latter meaning involves hope, motivation, personal responsibility, the pursuit of individual goals, and participation in community life, but not necessarily the absence of symptoms. Recovery in this more complex sense implies certain types of outcomes and treatments but is also connected to civil rights, stigma, self-help, opportunities, and other community concepts that are much broader than usual definitions of illness and level of functions of the health care system. Davidson and colleagues (31) offered the simple but useful clarification that neuroscientists are describing recovery "from" illness, which implies cure, whereas advocates are describing recovery "in" illness (that is, being "in recovery"), which refers to moving ahead with one's life, often despite persistent symptoms.

Given the protean definitions of recovery and the potential for miscommunication, how can we understand and discuss the links between science and recovery? One possible solution, which we do not endorse, would be to avoid the term recovery entirely as another unhelpful piece of mental health jargon. This path would ignore the power and value of the term for consumers and providers and would fail to respond to the New Freedom Commission's recognition of the importance of recovery and its call for research on this concept. We therefore propose an alternative path of definitional specificity and consistency based on using qualifying terms for recovery. Because the meaning of recovery lacks precision (and thus promotes ambiguity and confusion), we propose using standard qualifiers—for example, "recovery of cognitive functioning" or "recovery of vocational functioning"—to signify improvements in specific areas. The danger remains that some will read recovery as implying complete cure, but we endorse the current emphasis on improvement, which realistically emphasizes states of partial cure. The emphasis on a range of improvements in specific areas should allow us to communicate more clearly regarding the current findings and goals of research. We also acknowledge that civil rights, stigma, housing, vocational opportunities, and other community issues are relevant to the recovery movement and extremely important to those with disabilities but are generally outside the scope of the psychiatric research reviewed here.

Schizophrenia is a complex brain disorder that spans the life course (Figure 1). Although the etiology is uncertain, current evidence indicates that polygenic factors combine with a variety of environmental insults—including intrauterine stressors (for example, malnutrition and infection), birth complications, and postnatal factors (for example, psychoactive drug use)—to influence and perturb neurodevelopment and enhance the vulnerability to develop the heterogeneous disorder that we call schizophrenia (3). Although the disorder has been defined for the past century primarily in terms of psychotic symptoms, schizophrenia more recently has been characterized in terms of brain abnormalities and other symptom dimensions, such as negative symptoms and impairments in neurocognitive functioning in addition to psychosis (9,32).

Neurobiology

The evidence for neurobiological abnormalities underlying schizophrenia is robust. Several susceptibility genes for schizophrenia have been identified and replicated, with evidence rapidly emerging of the functional significance of allelic variation, mutations, and copy number variation (33,34). These genes give rise to the (intermediate) phenotype, or observable aspects, of the illness (35,36). Neuroimaging research has demonstrated decreased gray matter volume in the cerebral cortex, reduced volumes of the hippocampus and thalamus, and enlarged lateral and third ventricles and subarachnoid space (37,38,39). Studies of brain function using neuroimaging, electrophysiology, and cognitive tests have also consistently found evidence of hypofunctioning in specific tasks mediated by the frontal and temporal lobes (40) and of diminished perfusion or activation on measures of positron-emission tomography or functional magnetic resonance imaging (41,42,43,44,45,46). At the same time, studies of sensory gating among patients with schizophrenia show deficiencies in the normal inhibitory mechanisms that help to manage modulation of various sensory inputs (47,48,49). Milder forms of these biological abnormalities appear to be present at the onset of overt psychotic illness and may be present premorbidly, with areas of cognitive dysfunction carrying implications for brain localization, as well as for greater specificity for cognitive remediation interventions (50).

Progressive structural pathology may occur with additional psychotic episodes, at least for a subgroup of patients (51), leading to the current hypothesis that active psychosis can be neurotoxic (9). Some patients clearly demonstrate further deterioration of brain structure and cognitive function over time (52). The natural age-dependent attrition of dopamine activity may also explain why positive symptoms tend to decrease over time and in later life (53).

Importantly, all of these studies show a range of findings from little or no differences through extreme abnormalities relative to persons without the disorder. The large number of genes and types of environmental insults involved in etiology almost certainly underlie the extreme heterogeneity of brain abnormalities and of vulnerabilities, although these relationships are not yet well understood.

Psychotic symptoms

Many people experience psychotic symptoms (for example, in relation to drug use or trauma), but only a small proportion experience severe disturbances that endure for months and thus meet the criteria for schizophrenia (54). Although a substantial proportion of individuals experience a full remission of psychotic symptoms, few fully recover without recurrences. It is unclear if this is due to the recurrent and progressive nature of the illness, to the limitations of our treatments, or to problems with access to and delivery of clinical care. On the other hand, only a minority of patients fail to respond to treatment and remain severely psychotic for many years after their first episode. Although the early phase of illness is characterized by positive symptoms of psychosis, each successive episode of psychosis may bring less resolution of positive symptoms and an increasing prominence of negative symptoms, which often become enduring. Positive symptoms tend to diminish in intensity in the fifth and sixth decades of life (53,55).

Cognitive symptoms

There is increasing recognition that cognitive symptoms (also termed neuropsychological or neurocognitive deficits related to abnormalities in perception and information processing) constitute a more persistent dimension of the illness and one that more directly and permanently limits functioning (32,56,57). Subtle cognitive and academic performance differences between individuals who do not develop schizophrenia and future schizophrenia patients emerge in childhood and precede the onset of formal signs of the illness, but these are not outside the normal range (58,59,60). At the time of the first psychotic episode, most patients appear to sustain a further decrement in cognitive functioning (61). There is little or no evidence for natural improvements in cognitive symptoms over time, and some elderly patients may gradually worsen over time (52).

Psychosocial functioning

Functional behaviors, such as living independently in the community, working competitively, and developing social and intimate relationships, have been studied for at least a century with several consistent findings. First, functional capacity is highly variable among persons with schizophrenia, but it shows some continuity over time (13,62,63). Some people have premorbid functioning in the normal range, whereas others show prolonged poor levels of function or severe deterioration in functioning before the onset of overt illness; these differences were the basis for the theory of good and poor prognosis and the theory of process and nonprocess schizophrenia (64) and more recently for concepts of deficit and nondeficit schizophrenia (56,65).

Second, for persons with schizophrenia, functioning in adult roles is strongly affected by environmental opportunities, supports, stimulation, and stigma. Stigma works its ill effects in ways that are hard to quantify, but it produces a major impact on opportunities for normative life and for self-esteem and morale. During the era of long-term hospitalization, for example, patients with schizophrenia experienced profound atrophy of social, vocational, and basic living skills as a result of prolonged isolation and exposure to neglectful, stultifying conditions (66,67), which as iatrogenic effects may have reinforced false beliefs about intrinsic deterioration. Conversely, affording people opportunities to develop in social and vocational ways engenders optimism and better functioning (68,69). International studies by the World Health Organization show that patients with schizophrenia experience better functional outcomes in agrarian societies or developing economies with greater family and social supports, simpler vocational roles, and less stigma regarding the illness (70).

Third, functional performance in schizophrenia has a highly variable course, with substantial restoration of psychosocial functioning after initial episodes, increasing deterioration following subsequent episodes, and the potential for some degree of functional improvement in later life (69). The evidence that many people with schizophrenia improve in social and vocational functioning over time is counterbalanced by the evidence that many others suffer sustained disability, high rates of institutionalization in nursing homes, and excessive medical morbidity and mortality (71,72,73). In the United States, people with schizophrenia die approximately 25 years earlier than others in the general population, and the majority of older people with schizophrenia are not living independently in the community.

The magnitude of effects has been small to moderate from a host of variables thought to influence behavioral function in schizophrenia, including gender, birth trauma, age at onset, mode of onset, premorbid adjustment, intelligence, severity of symptoms, neurocognitive deficits, and severity of abnormal brain morphology (62,63,74,75,76,77). However, recent research shows that cognitive impairments probably have the strongest associations with functional outcomes (32,78,79,80,81,82) and predict an estimated 40% of the variance (83), far more than any other domain (32).

Quality of life

Quality of life encompasses many of the same concepts as current multifaceted views of recovery (84). Quality of life refers to the nonillness aspects of the person's life and generally includes subjective views of physical health status, functional ability, psychological status and well-being, social interactions, and economic status (85). In mental health research, quality-of-life measures also address satisfaction with life, functioning in daily activities and social roles, personal preferences regarding goals, living conditions, safety, environmental restrictions, finances, and opportunities (84,86,87,88,89). Although quality of life has many of the same problems of nonspecificity as the concept of recovery, it offers the advantages of clear communication and harmony with other areas of medicine faced with the same challenges.

Mental health research shows that patients with schizophrenia report worse quality of life than the general population and than those with physical illnesses; that younger, female, married, and less educated patients report higher quality of life; that length of illness correlates with lower quality of life; and that more symptoms, especially negative and cognitive symptoms, are related to lower quality of life (90). In general, subjective quality of life is related to the discrepancy between circumstances and expectations, which probably explains the inverse relationship with education. Many psychological constructs, such as self-esteem and self-efficacy, are considered to be mediating variables in models of quality of life (89).

Self-agency

The concept of recovery clearly extends beyond quality of life to encompass a process of assuming self-agency by developing hope, taking responsibility for one's life, participating actively in one's treatment, managing one's illness, and defining and pursuing personally meaningful goals. These issues have been described extensively in the recovery literature (26,27,91,92,93,94).

Recovery of biological functions

Several lines of evidence are beginning to show that biological functions in the brain can improve with treatment among adults with schizophrenia. Recent studies of nicotinic agonists have been reported to improve sensory gating (95), and dopamine agonists have been found to enhance frontal cortical blood flow and metabolism (96). In addition, mood stabilizers, antidepressants, and some antipsychotics have been found to promote growth in neural and glial cells (97,98,99,100). Moreover, clinical studies have shown that administration of selective mood stabilizers (for example, lithium) and antipsychotics (for example, clozapine-like second-generation drugs) can attenuate the loss or increase the volume of brain gray matter (101,102,103). Recent research also suggests that at the cellular level medications improve neurochemical function and that learning-based interventions, such as cognitive training, may also improve neural connectivity (104).

Recovery from psychosis

Interventions during the early phase of psychosis aim to eliminate symptoms completely (full remission) and prevent future episodes of psychosis. For the majority of first-episode patients, antipsychotic medications and evidence-based psychosocial treatments can achieve full or substantial remission of positive symptoms (105). Elimination of negative symptoms is much less clear. Combinations of current medications and psychosocial interventions can prevent future episodes of psychosis in most cases, but only if patients continue to receive treatment. Approximately 90% of patients will have at least one psychotic relapse within five years (106), partially because of high attrition rates from treatment, failure to implement evidence-based interventions, and the limitations of available therapeutics.

In general, medications are less effective for patients with chronic schizophrenia than for those with first-episode or early-stage schizophrenia. Combining psychosocial interventions (such as assertive community treatment, family psychoeducation, adherence interventions, social skills training, and relapse prevention techniques) with medications has an additive effect on relapse prevention (107,108,109,110,111). Cognitive-behavioral interventions can also reduce the functional impact of residual positive psychotic symptoms, as well as diminish negative symptoms (109, 112). The combination of optimal psychosocial and pharmacological intervention for management of symptoms has been termed "illness management and recovery" (109).

Recovery of cognitive functions

Cognitive symptoms are rarely, if ever, eliminated when present at the onset of illness and the initial episode of psychosis. Although researchers have just started to target cognitive symptoms in schizophrenia, initial studies suggest that specific cognitive symptoms can be reduced (not eliminated) by second-generation antipsychotic medications (113,114), by cognitive retraining (115,116,117,118), and perhaps by supportive psychotherapy (110).

Improvements in existing cognitive symptoms, although modest, may at some point become sufficient to improve more complex functional behaviors, such as social, educational, and vocational functioning (110,115,119,120,121). Functional magnetic resonance imaging data suggest that cognitive rehabilitation programs may even change the way the brain processes information (116). Improvements in cognition lead to immediate benefits, such as increased learning potential (83), social cognition (122), and functional capacity (123,124,125); longer-term effects on cognitive functions remain to be demonstrated.

Recovery of functional capacity

Psychiatric rehabilitation targets specific behavioral functions, such as independent living, competitive employment, education, family relationships, and social relationships (107,126,127). Effective interventions include skills training, psychoeducation, cognitive-behavioral therapies, cognitive enhancement therapies, and a variety of supports within and around normative community settings (5,6,121,128,129,130). Medications alone do not have a direct impact on behavioral functional performance but may facilitate participation in psychiatric rehabilitation (131).

Research shows that independent living is a realistic goal for the great majority of individuals with schizophrenia. Supported housing and "housing first" approaches have demonstrated that most persons with schizophrenia (over 90%) can live in independent housing, without requiring stepwise progression through a continuum of group homes and without requiring participation in mental health treatment, provided they have supports (132,133). This capacity for independent community living includes individuals who are chronically homeless and mentally ill. For the small minority of patients with schizophrenia who cannot live independently, substance abuse and comorbid medical conditions, rather than mental illness, seem to be the critical barriers (134,135).

Competitive employment and education in routine community settings also appear to be realistic and beneficial goals for most people with schizophrenia. Supported employment, which provides supports for finding jobs that match persons' interests and for maintaining employment, has robust empirical support (128,136,137). Supported employment is also effective for patients with severe disabilities and for those with comorbid conditions (128). Although 70% or more of participants in supported employment can achieve competitive employment, nearly all work part-time, and very few people entirely stop receiving disability payments (128). Limitations in the success of supported employment appear to be due to some combination of cognitive deficits and disincentives in the insurance and benefits systems (138,139). For example, patients who are working full-time can lose their health insurance and not be able to obtain insurance through employment. Supported education has received minimal research attention compared with supported employment, but it follows the same principles, uses similar techniques, and is often combined with supported employment (140). Helping people with schizophrenia return to work or school early in the course of illness, rather than after disability has been established for years, appears to be even more effective (141,142).

The research on improving social relationships is considerably more complicated. Social skills training helps to improve specific social behaviors among patients with schizophrenia, and some transfer of these skills occurs from the training setting to patients' day-to-day lives (143,144). Nevertheless, after more than 40 years of research, social skills training remains controversial because the magnitude of effects on social performance behaviors in routine community settings remains uncertain. Recent efforts to include in vivo practice and to involve natural supports in reinforcing social skills are promising (145,146). Although many other approaches to helping people with schizophrenia achieve their social goals are commonly used and show promise (for example, consumer-run social programs, supported socialization, recovery centers, and clubhouses), they have not been studied rigorously (147).

One area of social performance that has been studied separately is family relationships. Family psychoeducation programs of several types consistently decrease negative emotional expression and conflict within families and thus improve family relationships (148). This has generally been attributed to changing the family unit's behavior rather than the individual patient's social behaviors.

As research has clarified the importance of cognitive deficits in relation to performance behaviors, researchers have shown that cognitive training (121) and environmental adaptation and compensation (149) can improve performance behaviors in a variety of settings. How much these efforts will add to standard supported education, supported employment, and social skills training is as yet unclear.

Recovery of quality of life

In general, treatment improves subjective quality of life among people with schizophrenia through its effects on other outcomes (for example, symptoms and cognition). Interventions that improve symptom control without increasing side effects and those that improve functional behaviors also improve subjective quality of life (90). Quality of life is consistently increased by interventions such as supported employment that lead to better vocational functioning and better incomes (150,151,152).

Recovery of self-agency

As noted above, the recovery concept goes beyond quality of life as an outcome to also emphasize an active process of self-agency: developing hope, taking charge of one's life, and pursuing one's personal goals. Psychiatric rehabilitation has emphasized these issues, including personal goal attainment, for many years (153). Current research includes several effective approaches that improve motivation and self-management of illness, such as education, behavioral tailoring, and cognitive-behavioral interventions (109). In addition, current psychotherapies (154), self-help programs (155), and peer-directed medication management (156) emphasize the process of hope and taking responsibility. As yet, however, controlled research on these approaches has not been conducted (157).

In this first decade of the 21st century, intervention research supports the New Freedom Commission's (6) call for research on the multifaceted concept of recovery. As reviewed above, researchers are making progress in understanding and improving various aspects of brain function, symptom control, cognitive function, psychosocial behaviors, quality of life, and self-agency. In most of these areas, effective interventions exist, and within each area a number of promising interventions are under investigation.

Several findings are clear. People with schizophrenia are tremendously heterogeneous in each domain of recovery, and the various domains of recovery are themselves relatively independent from one another. Current interventions are effective for specific dimensions of the illness and functions, are usually ameliorative rather than curative, and are effective only for a proportion of patients. Hence, we suggest defining recovery in terms of improvements in specific domains rather than globally.

The limitations of current research on recovery are also apparent. First, a consistent and more specific terminology is needed to facilitate communication among advocates, researchers, and policy makers. Second, because our knowledge of the neurobiology, pathophysiology, and heterogeneity of schizophrenia is incomplete, definitive interventions are lacking. In the immediate future, interventions are likely to be restorative for some patients, ameliorative for most, and ineffective for some others. Third, because the dissemination and adoption of effective treatments lag far behind their development (Drake RE, Skinner JS, personal communication, 2008), the great majority of patients with schizophrenia will probably not have access to evidence-based interventions until there is extensive reform of health care and disability policies in the United States.

Neuroscience, clinical and services research, and the recovery movement are addressing the same important issues in regard to schizophrenia, albeit in different ways (5,6). Research might be seen as more relevant and advocacy might be seen as more effective if the two groups could adopt a more precise, uniform terminology and work collaboratively. We have suggested a simple modification of terms to foster collaboration and have illustrated how such language might be useful in clarifying areas and degrees of progress, as well as areas in which we are in need of more intensive research.

Dr. Lieberman has received research grants from, served as a consultant for, served on the advisory board for, or served on the Data and Safety Monitoring Board for AstraZeneca, Bristol-Myers Squibb, Eli Lilly and Company, GlaxoSmithKline, Forest Laboratories, Janssen Pharmaceuticals, Merck, Pfizer, Solvay, and Wyeth. Dr. Lieberman also holds a patent with RepliGen Corporation. Dr. Lieberman does not receive financial compensation or salary support for being a consultant or for being on an advisory board. Dr. Keefe has received research grants from, served as a consultant for, or received speaker's honoraria from Abbott Laboratories, Bristol-Myers Squibb, Cephalon, Dainippon Sumitomo Pharm, Eli Lilly and Company, Gabriel Pharmaceuticals, Johnson & Johnson, Lundbeck, Memory Pharmaceuticals, Merck, Orexigen Therapeutics, Otsuka America Pharmaceutical, Pfizer, Sanofi-Aventis, and XenoPort. Dr. Keefe also receives royalties from the Brief Assessment of Cognition Symbol Coding of the MATRICS Consensus Cognitive Battery. Dr. Perkins has received research grants, consulting fees, or educational fees from AstraZeneca, Bristol-Myers Squibb, Eli Lilly and Company, Forest Laboratories, GlaxoSmithKline, Janssen Pharmaceuticals, Otsuka America Pharmaceutical, Pfizer, and Shire Pharmaceuticals. Dr. Stroup has served as a consultant for or has received speaker's honoraria from AstraZeneca, Janssen Pharmaceuticals, Eli Lilly and Company, Pfizer, and Solvay Pharmaceuticals. The other authors report no competing interests.