Initial enthusiasm about discoveries in our field has evolved into questions about the meaning and future of current scientific approaches. The first set of questions has to do with the validity of information derived from biological and psychosocial sciences in understanding psychopathology. The second set of questions has to do with the value system dictating the sciences of psychopathology and their implications for future discoveries.

A scientific response to these questions would consist of a critique of research methods. This editorial attempts an epistemological analysis rather than a scientific approach. This analysis assumes methodological integrity and thus starts from the point where the scientific critique ends. We advance the position that exclusive reliance on the experimental process produces incomplete knowledge in basic, clinical, or psychosocial fields. For this reason, pragmatic-social processes have been and will continue to exert major influence on our scientific initiatives. We argue that experimentally sound scientific findings have only relative value and that awareness of the influence of nonscientific processes on our scientific endeavors can promote and guide cross-fertilization of the sciences of psychopathology.

Empirical sciences follow a common process. A general theory leads to specific hypotheses, which, after experimentation, lead to explanations of specific events. This process allows bias at all three levels.

There has been a debate on the process leading to theory development. The traditional Baconian view +(1) has been that theory development relies on an inductive process during which a number of negative observations are analyzed before an affirmative conclusion is reached. The opposite view, advocated by Popper +(2), argues that theory development relies mainly on creative intuition best understood through psychosocial, rather than logical, analysis. The principal difference between these approaches has to do with the role of the verification process (experimentation) in the growth of a theory. The first approach includes part of the experimental verification work in the process that leads to theory development. The second argues that a theory can be tested only after it has been advanced.

Regardless of the relative contribution of prior observations, theory development is influenced by nonobservational factors. Most theories are ruled out before they are subjected to experimental testing. The criteria are pragmatic and social rather than experimental and include the ability of a theory to provide explanations for problems posed at the time that it is advanced, as well as intellectual appeal +(3, +4). Treatment efficacy and experimental technology have played a decisive role in theory formulation. Psychological theories were favored during a time that drug treatment had a rather limited role in patient care. The advances of molecular genetics and cellular and neuroimaging technology are currently a major factor in theory selection.

Experimentation has led to disagreement among philosophers of science. The inductivists, starting with Bacon, suggested that a "conclusion on the affirmative instances" comes after a number of negative observations +(1). In contrast, Popper argued that experimentation could never verify a theory. Simply, failure to falsify a theory allows it to survive one more round of testing +(2, +5). In the clinical and biological sciences, in which events are multidetermined, even Popper’s conservative criterion of falsification is hard to apply. A weak association between a predictive (independent) variable and a target (dependent) variable cannot be falsifying +(6). Borderline statistical significance gives probabilistic, but not logical, grounds for rejecting a hypothesis. Therefore, experimentation plays an important yet limited role in the falsification or corroboration of a theory, thus allowing socially determined, pragmatic concerns to influence the survival of a biological or psychosocial theory.

Even if concerns about theory and experimentation are set aside, information derived from the theory-observation interaction does not safely explain clinical or biological events. Logical positivists view explanation as a conclusion derived from observations interpreted in the context of a general law +(7). Philosophers argue that this logical sequence has at least two problems +(8, +9). First, the causation of a phenomenon may not be explained, even when the attempted explanations are derived from a general law and observation. Second, an event may be explained even though it is not derived from a general law. The sciences of psychopathology, even when they rely on an underlying law, e.g., frontostriatal-limbic dysfunction contributes to the chronicity of geriatric depression +(10), explains only part of the variance of abnormal behavior so that explanations derived from an experiment cannot be fully supported or rejected. These and other formal problems suggest that scientific explanations derived from theory, observation, and their interactions have, at best, relative value.

But how do philosophers and scientists cope with the limitations of the scientific process? Scientific realism and empiricism offer two distinct approaches +(4, +11, +12). Scientific realism sees scientific theories and corroborating experiments as approximations of reality. For example, we are justified in believing that the affective and cognitive subdivisions of the anterior cingulate are modulating behaviors related to depression +(13) because this theory offers an explanation of relevant experimental results. Unlike scientific realism, empiricism avoids making a statement concerning the truth of theories about nonobservable entities. Instead, it views theories as statements that are adequate to predict what is observed. In the example we gave, the anterior cingulate properties offer an explanation about some aspects of depression, but this statement is not final. Experiments evaluate the empirical adequacy of theories but cannot assert whether a theory is true. Regardless of their differences, both scientific realists and empiricists view scientific explanations as incomplete and subject to correction.

As there is no safety in theory, experiment, and explanation, the selection of scientific directions is influenced not only by scientific factors but also by pragmatic and social values. Two pragmatic factors seem central to the selection of theories: 1) the potential to function as the intellectual platform of new hypotheses and experiments and 2) the potential to bring to bear the most advanced means of experimentation. At least three sets of social factors influence the direction of biological and the psychosocial sciences of psychopathology: 1) their esthetic appeal, i.e., simplicity and plausibility, has been the essence of Occam’s razor since the 14th century; 2) their cognitive merit, i.e., their explanatory power, predictive content, and usefulness; and 3) their responsiveness to contemporaneous scientific and social concerns. The relative value of each of these factors depends on the specific scientific field of psychopathology, its history, its current state of development, the means of experimentation, and social demands. Pragmatic-social factors not only influence current theory selection but also guide the process of theory replacement when a new theory better satisfies evolving pragmatic-social demands.

Historically, pragmatic and social values have either helped or harmed knowledge development in psychopathology. They have helped when attempts were made to synthesize the sciences of psychopathology into comprehensive models. Engel’s biopsychosocial model +(14) allowed parallel developments of the biological and psychosocial sciences of psychopathology and some degree of integration. The limited experimental contributions of the original biopsychosocial model may be explained by the limited experimental means of that time. In contrast, adherence to singular approaches, e.g., "biologism" and "psychologism," that do not recognize the importance of other fields led to scientific fragmentation, inappropriate for a complex field like psychopathology.

There are at least two scientific reasons to reject the singular assumptions of the biological and psychosocial sciences. First, there is no reliable way to differentiate behavioral "abnormalities" constituting normal responses to an abnormal environment from abnormal responses to a normal environment. Second, sequential interactions between the brain and environment blur the differences between environmental and nonenvironmental etiology. For example, in posttraumatic stress disorder, an abnormal environment causes brain changes that subsequently lead to abnormal responses to a normal environment. Therefore, there is no compelling scientific reason to choose assumptions favoring exclusively biological or psychosocial determinants of psychopathology.

Along with an absence of scientific reasons, no formal grounds exist for reducing clinical biology, psychology, and sociology to more basic sciences +(15, +16). Moreover, scientific growth began to permit the development of integrated models appropriate to the complexity of psychopathology. The conceptualization of "endophenotypes" is such an example. Ideal endophenotypes are trait-related, behavioral, or biological abnormalities associated with vulnerability to a psychiatric disorder +(17). They are heritable, cosegregate within families and illness, and are found in affected and nonaffected family members at a higher rate than in the general population. Sensory-motor gating abnormalities, eye-tracking dysfunction, and working-memory abnormalities are examples of abnormalities that meet some of the endophenotype criteria in schizophrenia. Focusing on an endophenotype, e.g., eye tracking, simplifies genetic analysis because endophenotypes may be determined by fewer genes than complex syndromes such as schizophrenia. Work on endophenotypes brings together clinical phenomenology, cognitive neuroscience, and structural and functional neuroanatomy to work along with genetics. Moreover, it fosters the development of simpler animal models based on the endophenotypic abnormality rather than the whole syndrome. Nonetheless, work on endophenotypes would only be feasible if these scientific fields reach a level permitting meaningful integration.

Awareness that social processes influence the trajectory of the sciences of psychopathology does not weaken their epistemic value. However, it creates the responsibility of identifying scientific priorities without the comfort of exclusive reliance on scientific criteria. Scientists can no longer adhere to singular approaches emphasizing the importance of one type of science at the expense of another. The limitations inherent in theory, experimentation, and explanation do not allow safe confirmation or rejection of scientific insights and make convictions impossible to defend on scientific grounds. Policy makers need to be aware of the relative value of the scientific process while they form a vision about scientific initiatives likely to be most productive. They, too, cannot rely only on scientific findings and must take into consideration social and pragmatic influences.

Two directions seem appropriate. First, pluralism ("let all flowers bloom") in supporting the specific sciences of psychopathology is a rather safe route, as the competitive nature of research funding provides a Darwinian selection of the most productive approaches. Second, integrative approaches need to be cultivated because this research is at an early stage of conceptual and experimental development. Agenda-setting conferences and appropriate announcements and review criteria need to be part of such encouragement. However, there is little justification to support singular approaches that neglect the contribution of other fields. Historically, "biologism" and "psychologism" have done little for our field, and there is even less reason to believe that similar singular approaches will do better during an era of greater sophistication.

Address reprint requests to Dr. Alexopoulos, Cornell Institute of Geriatric Psychiatry, 21 Bloomingdale Rd., White Plains, NY 10605; gsalexop@med.cornell.edu (e-mail).