Zero-sum science I: Paradigms and paradigm shifts

The term paradigm shift was introduced to the physical sciences (astronomy, chemistry, earth sciences, physics) by Kuhn (Reference Kuhn1962), and quicky taken up by social scientists in economics, psychology, sociology, and beyond (Coats, Reference Coats1969; Reese & Overton, Reference Reese and Overton1972). Kuhn’s original definition is captured in the American Psychological Association Dictionary of Psychology, which defines a paradigm shift as, “…substantial and fairly rapid change in the pattern of ideas and assumptions defining the nature of a science and determining the methods and procedures used” (American Psychological Association, 2018). This definition applies best to the physical sciences, where single paradigms that dominate entire fields sometimes give way to new paradigms with greater explanatory reach. Displacement of Newtonian physics by Einsteinian physics is the prototypic example. Confirming Einstein’s theory required new experimental methods conducted at previously uninterrogable levels of analysis, changing physics and other fields forever (Holton, Reference Holton1969). Other paradigm shifts in the history of the physical sciences include transitions to genetic inheritance (Mendel) and evolution by natural selection (Darwin) (see Goldstein, Reference Goldstein2012).

Kuhn (Reference Kuhn1962) acknowledged a major difference between the physical and social sciences, with the latter guided by no single paradigm. Kuhn attributed this to social science being immature, or preparadigmatic. The contemporary and more apt interpretation is that psychological science cannot and will never converge on one paradigm given the complexity of human behavior and its transacting determinants across levels of analysis spanning genes to cultures (Beauchaine & Constantino, Reference Beauchaine and Constantino2017; Cicchetti & Dawson, Reference Cicchetti and Dawson2002; Sanbonmatsu et al., Reference Sanbonmatsu, Cooley and Butner2021; Weimer & Palmero, Reference Weimer and Palmero1973). Paradigms in the social sciences are multiple and sometimes obscure (Polsby, Reference Polsby2003).

The nature of this paradigmatic heterogeneity is twofold. First, given the complexity of human behavior, all social sciences, to a lesser or greater extent, comprise various subdisciplines with proponents who operate from different theoretical perspectives, carry different epistemological assumptions, and use widely varying methods to test scientific propositions, often at different levels of analysis (Beauchaine & Cicchetti, Reference Beauchaine and Cicchetti2019; Strauman, Reference Strauman2001; Varpio & MacLeod, Reference Varpio and MacLeod2020). No single paradigm can capture such divergence. As a result, few if any paradigm shifts affect entire social science disciplines, and some paradigms cross disciplines (e.g., De Vincenzo et al., Reference De Vincenzo, Stocco and Modugno2023). Second, paradigm shifts are often slower than Kuhn (Reference Kuhn1962) described. Some competing paradigms coexist in equilibrium for many years before one ascends to prominence (Beauchaine & Haines, Reference Beauchaine, Haines, Beauchaine and Crowell2020). In rare cases, competing paradigms coexist in perpetuity, unknowable future events notwithstanding.

Most paradigm shifts follow periods of intradisciplinary debate. In contemporary psychology, several subdisciplinary themes rise to this level or hold potential to do so. These concern the nature of psychological constructs (De Boeck et al., Reference De Boeck, Pek, Walton, Wegener, Turner, Andersen, Beauchaine, Lecavalier, Myung and Petty2023), the value of inductive (top-down, theory-driven) versus deductive (bottom-up, data-driven) research in clinical science (Achenbach, Reference Achenbach2020), phylogenetic versus ontogenetic origins of human emotion (Beauchaine & Haines, Reference Beauchaine, Haines, Beauchaine and Crowell2020; Barrett et al., Reference Barrett, Lindquist, Bliss-Moreau, Duncan, Gendron, Mize and Brennan2007; Panksepp, Reference Panksepp2007), methodologic approaches to longitudinal data analysis (Haines et al., Reference Haines, Kvam, Irving, Smith, Beauchaine, Pitt, Ahn and Turner2024), whether certain subdisciplines should maintain their basic science portfolios or accept populist calls for strictly applied research (Berkman & Wilson, Reference Berkman and Wilson2021; Mede & Schäfer, Reference Mede and Schäfer2020), and whether psychology and its subdisciplines should retain their logical positivist identities or abandon/replace them to in response to increasingly common prescriptive and proscriptive ideological dicta (Holman & Wilholt, Reference Holman and Wilholt2022; Silander et al., Reference Silander, Geczy, Marks and Mather2020).

Examples of inter- as opposed to intradisciplinary paradigm shifts include replacement of clinical judgement with actuarial prediction and empirically derived classification—a change that originated in clinical psychology and was later adopted by psychiatry (Achenbach & Edelbrock, Reference Achenbach and Edelbrock1978; Dawes et al., Reference Dawes, Faust and Meehl1989); and adoption of the biomedical model of mental illness—a change that originated in psychiatry and was later adopted by clinical psychology (Deacon, Reference Deacon2013). Both paradigm shifts were contentious, and the latter was partly pragmatic; many clinical scientists acceded to biomedical research agendas to maintain and extend their federal funding (NIMH, 2024).

Whether field-defining, subdisciplinary, or cross-disciplinary, the predominant approach to science pits paradigmatic and subparadigmatic hypotheses against one another in a zero-sum game. The prevailing tradition in psychology and other social sciences when writing papers and grants is to juxtapose competing theories and devise hypotheses and experiments to adjudicate a winner. As I discuss below, zero-sum thinking hamstrings progress in the social sciences by relegating useful scientific knowledge from the “losing” paradigm to the annals of history. In clinical science, this can harm clients. I provide three examples below, then discuss how continued pursuit of Dante Cicchetti’s transdisciplinary vision affords an enduring, positive-sum solution.

Zero-sum science II: Discarding useful paradigms

Paradigm shifts do not guarantee scientific progress (Kuhn, Reference Kuhn1962). Even when progress is realized, scientific knowledge, useful principles, and effective therapeutic techniques from “defeated” paradigms are lost in zero-sum science. The plight of behaviorism provides an example. Although behaviorism has persisted in clinical psychology beyond its lifespan more broadly, use of behavior modification principles has declined sharply in recent years under longstanding criticisms from cognitive psychology and neuroscience, and, more recently, humanism (Guercio, Reference Guercio2020)Footnote ⁵ . Ardent humanists reject, among other empirically supported treatments, response-cost contingencies in the classroom and applied behavior analysis for treating autism (Anderson, Reference Anderson2023; Elkins, Reference Elkins2009; Jung & Smith, Reference Jung and Smith2018; Moss, Reference Moss2018). Both are viewed as unethical because they diminish self-determinism (see Footnote 4; Gnaulati, Reference Gnaulati2022; Smith, Reference Smith1978). In this instance, rejection of the behavioral paradigm is based on ideological, not practical or empirical grounds—operant reinforcement principles are no less effective in the 21st Century than they were in the 20th Century.Footnote ⁶

Regardless of where one falls on these paradigmatic debates, abandoning behavioral principles is likely setting clinical science back in its overriding objective to reduce human suffering (Gee et al., Reference Gee, DeYoung, McLaughlin, Tillman, Barch, Forbes, Krueger, Strauman, Weierich and Shackman2022). Several recently developed school- and classroom-based interventions that largely ignore and sometimes decry behavioral principles are proving iatrogenic (Foulkes & Stringaris, Reference Foulkes and Stringaris2023). Programs that prevent student exposure to anxiety-eliciting and fear-eliciting events, for example, often increase anxiety over time, eroding instead of bolstering resilience (e.g., Jones et al., Reference Jones, Bellet and McNally2020; Limber & Kowalski, Reference Limber and Kowalski2020; Travers, Reference Travers2017; Ttofi & Farrington, Reference Ttofi and Farrington2011). Competent behaviorists would likely have averted such outcomes by spotting reinforcement contingencies that amplify anxiety and reinforce self-narratives of vulnerability (Bellet et al., Reference Bellet, Jones and McNally2018, Reference Bellet, Jones, Meyersburg, Brenneman, Morehead and McNally2020).

In some cases, arguments set forth to “defeat” rival paradigms oversimplify core tenets and mischaracterize opponentsʼ perspectives. Skinner, for example, is often described as a naïve reductionist who renounced genetics, neuroscience, and all endogenous mechanisms of behavior (DeBell & Harless, Reference DeBell, Harless and Skinner1992). Those who have read Skinner, however, know he espoused no such position (Zilio, Reference Zilio2016). Instead, he demanded that explanatory mechanisms of behavior be observable and measurable (Skinner, Reference Skinner1963), a perspective consistent with logical positivist philosophy of scienceFootnote ⁷ and neuroscientific data of the day (Moore, Reference Moore1985; Olds & Milner, Reference Olds and Milner1954). Although Skinner overestimated the promise of behaviorism in complex human environments (Gary, Reference Gary1973), he was not a reductionist, and operant techniques proved invaluable, rivaling and sometimes exceeding the effectiveness of potentially addictive medications for psychiatric disorders (Connors et al., Reference Conners, Epstein, March, Angold, Wells, Klaric, Swanson, Arnold, Abikoff, Elliott, Greenhill, Hechtman, Hinshaw, Hoza, Jensen, Kraemer, Newcorn, Pelham, Severe and Wigal2001; Wardle, Reference Wardle1990).

When Skinner (Reference Skinner1963) wrote his now-famous article on operant conditioning in the American Psychologist, psychophysiological methods were being used and had been used for some time to evaluate endogenous responses to reward and punishment, and endogenous correlates of operant conditioning (e.g., Schmidt, Reference Schmidt1941). Skinner himself appealed to physiological data to understand mechanisms of operant reinforcement (Lambert et al., Reference Lambert, Skinner and Forbes1933). Neural correlates of human behavior simply could not be interrogated until the mid-1990s, when functional neuroimaging became available. In the age of neuroscience, however, the value of which cannot be overstated, Skinner’s contributions to understanding human behavior and reducing human suffering are often further diminished (Guercio, Reference Guercio2020; Wakschlag et al., Reference Wakschlag, Perlman, Blair, Leibenluft, Briggs-Gowan and Pine2017). Operant reinforcement principles—the bedrock of a generation of effective interventions for ADHD, oppositionality, anxiety disorders, and developmental disabilities—may be fading into historical obscurity. Behaviorism lost the zero-sum game (Braat et al., Reference Braat, Engelen, van Gemert and Verhaegh2020).

In other cases, paradigms with overriding empirical support are supplanted based on oversimplified and mistaken rationales contrasting them with fully integrable “rival” paradigms (Loeb, 2018). Wakschlag et al. (Reference Wakschlag, Perlman, Blair, Leibenluft, Briggs-Gowan and Pine2017), for example, built their case for neurodevelopmental mechanisms of disruptive behavior by eschewing coercion theory (an operant framework)Footnote ⁸ , reasoning that some children who are parented competently show aggression. Rejecting coercion theory by this rationale ignores a robust literature demonstrating its validity, based on two logical fallacies that are surprisingly common in science. First, it presents a false dilemma (asrgumentum falsum dilemma; see Tomić, Reference Tomić2013). Both neural and environmental mechanisms of disruptive behavior are well-established (Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Beauchaine et al., Reference Beauchaine, Zisner and Sauder2017), with their interactions often accounting for more variance than their main effects combined (cf. Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008).

A corollary of false dilemma is overgeneralization (fallacy of extension; see Peters et al., Reference Peters, Krauss and Braganza2022). In social science, exceptions to theories are rarely ipso facto evidence of invalidity—human behavior is far too complex for this level of certainty. With such reasoning, Wakschlag et al. (Reference Wakschlag, Perlman, Blair, Leibenluft, Briggs-Gowan and Pine2017) hold coercion theory to an impossible standard that no psychological or neuroscientific theory, including their own, can possibly meet—accounting for every affected person’s disruptive behavior. The bottom line: Wakschlag et al. (Reference Wakschlag, Perlman, Blair, Leibenluft, Briggs-Gowan and Pine2017) and Patterson et al. (Reference Patterson, DeBaryshe and Ramsey1989) are both correct. Bridging their findings, however, requires multidisciplinary grand theory, a point I elaborate in later sections.

More broadly, by implicitly prioritizing main-effects at single levels of analysis (here neuroscientific) and using those effects to mistakenly “disprove” and reject any or all others—including those with well-replicated empirical support – we obscure multi-determining mechanisms of psychopathology, fail to detect important transactions among mechanisms across levels of analysis, and hamstring our predictive models. The proper question to ask is how to integrate such findings. At present, oversimplified neuroscientific explanations of disruptive behavior have ascended, with coercion theory (behavioral) and extended environment theories (social) fading from prominence.Footnote ⁹ As important as neurodevelopmental mechanisms are (Beauchaine et al., Reference Beauchaine, Neuhaus, Brenner and Gatzke-Kopp2008), this edges the field toward biological reductionism. Interventions derived from or informed by coercion theory have benefitted thousands of children across the U.S. and Europe for two generations, with medium to large effects that in some cases endure and even strengthen into adulthood (e.g., Beauchaine et al., Reference Beauchaine, Webster-Stratton and Reid2005; Dodge et al., Reference Dodge, Bierman, Coie, Greenberg, Lochman, McMahon and Pinderhughes2015; Samimy et al., Reference Samimy, Schettini, Fernhoff, Webster-Stratton, Beauchaine and Storch2022; Webster-Stratton & McCoy, Reference Webster-Stratton and McCoy2015). Moreover, biological vulnerabilities predict children’s responses to and are changed by these interventions, with direct implications for precision care (Bell et al., Reference Bell, Shader, Webster-Stratton, Reid and Beauchaine2018).

Identifying interactive mechanisms across levels of analysis is critical to building more effective interventions (Beauchaine et al., Reference Beauchaine, Hinshaw and Bridge2019a). Zero-sum game science—even when multidisciplinary—discourages transactional model building across levels of analysis, constraining our understanding of psychopathology and its development, and obscuring equifinal pathways to what appear to be single disorders. Cumulative, positive-sum science that integrates new findings into theoretically informed, multiple-levels-of-analysis models is needed (Beauchaine & Constantino, Reference Beauchaine and Constantino2017; Cicchetti & Dawson, Reference Cicchetti and Dawson2002).Footnote ¹⁰

In the social sciences, there are many examples of prioritizing single levels of analysis while ignoring known effects at other levels of analysis. Some of these extend beyond clinical science to the field more broadly. In such cases, information can be lost, understanding can be eroded, and our capacity to predict future outcomes, identify those at highest need, and redistribute limited resources can be compromised. This is especially likely when higher-order mechanisms in hierarchical data (e.g., school effects) are tested while ignoring nested effects at lower levels of analysis (e.g., children within classrooms). When this occurs, variance in outcomes accounted for at lower levels of the hierarchy (children, classrooms) is subsumed into the higher-order effect tested (schools). This yields overestimates—sometimes by wide margins—of the higher-order effect (Wampold & Serlin, Reference Wampold and Serlin2000). In this case, a researcher would overestimate school effects, possibly delivering expensive interventions at the school level when delivery to selected classrooms or individual children is indicated. This principle is illustrated in Figure 1.

Figure 1. A nested hierarchy of influences on child development and behavior (e.g., depression), with arbitrary percentages of variance accounted for at each level of nesting. To simplify, I assume (1) only individual-level and environmental contributors to depression, (2) error-free measurement, and (3) non-overlapping and non-interacting effects. Variance attributable to each level of nesting can be ascertained with data collected at each level using multilevel (hierarchical) modeling. When all such data are included in the model but only institutional environment is tested, the effect size for institutional environment is overestimated as the sum of b→g because variance at all lower levels of the hierarchy is subsumed in the effect. Variance attributable to national culture (a) cannot be quantified without data from multiple nations. Collecting such data should be prioritized in years to come.

The field’s recent and long-overdue paradigm shift toward prioritizing physical and mental health disparities provides an example (Carter & Mazzoni, Reference Carter and Mazzoni2021, Knerr & Fullerton, Reference Knerr and Fullerton2012; Wallace, 2013). Disparities research enjoys a rich if not representative history in social and community psychology, and is now a major focus of clinical psychology (Carbado et al., Reference Carbado, Crenshaw, Mays and Tomlinson2013, Major, Reference Major and Zanna1994), Foundational contributions to the field, which extend back four decades (Ulbrich et al., Reference Ulbrich, Warheit and Zimmerman1989; Vega & Rumbaut, Reference Vega and Rumbaut1991), identify predictors of disparities at intrinsic (e.g., temperament, attachment, social cognition, coping style), family (e.g., housing stress, single-parenthood, coercion) and other extrinsic levels of analysis (e.g., discrimination, institutional racism, neighborhood cohesion, culture) (Case et al., Reference Case, Eagle, Yao and Proeschold-Bell2018; Dankwa-Mullan et al., Reference Dankwa-Mullan, Rhee, Stoff, Pohlhaus, Sy, Stinson and Ruffin2010; Evans et al., Reference Evans, Barer and Marmor1994; Keppel et al., Reference Keppel, Pamuk, Lynch, Carter-Pokras, Kim, Mays, Pearcy, Schoenbach and Weissman2005). As the field matured across decades, this work included increasingly sophisticated multilevel measurement and analysis strategies to disentangle overlapping individual, family, institutional, geographic, and other effects (e.g., Balsam et al., Reference Balsam, Beauchaine, Mickey and Rothblum2005; Evans et al., Reference Evans, Williams, Onnela and Subramanian2018). More recently, similar multilevel strategies have been used effectively in research on mental health outcomes among children and adolescents with intersecting identities (e.g., Hahn et al., Reference Hahn, Chisolm and Beauchaine2024; Kern et al., Reference Kern, Duinhof, Walsh, Cosma, Moreno-Maldonado, Molcho and Stevens2020).

Assumptions undergirding multilevel analysis include (1) multidisciplinary approaches are needed to identify and disentangle all essential contributors to mental health disparities (Fleming et al., Reference Fleming, Perkins, Easa, Conde, Baker, Southerland, Dottin, Benabe, Ofili, Bond, McClure, Sayre, Beanan and Norris2008), and, (2) both intrinsic and extrinsic factors must be accounted for to understand synergistic effects (Intrinsic Vulnerability × Extrinsic Risk interactions) that render some individuals in environments of discrimination and racism far more likely to have existing mental health problems, to develop future mental health problems, and to experience early mortality. Only by understanding multilevel Person × Environment interactions can we effectively target these individuals with potentially life-saving preventive interventions and treatments (Holmes et al., Reference Holmes, Lehman, Hade, Ferketich, Gehlert, Rauscher, Abrams and Bird2008). As we work toward the desired ideal of eliminating discrimination, racism, and stigma, delivering resources to the most vulnerable in such contexts minimizes current suffering.

Several authors have argued against multidisciplinary and multilevel analysis of physical and mental health disparities on moral grounds (e.g., Knerr & Fullerton, Reference Knerr and Fullerton2012; Wallace, 2013). These calls are based on two concerns. First, limited progress has been made toward closing disparities or altering deteriorating mental health among racial minorities or other underserved sectors of the U.S. population. Second, focusing on and documenting individual-, family-, and community-level contributors to mental health problems among marginalized groups either is or will be construed as blaming the victim.

Concerns over limited progress are legitimate and deserve our attention and research focus. Although stagnated progress in treating mental illness is a general problem for the field (Bommersbach et al., Reference Bommersbach, McKean, Olfson and Rhee2023; Brennan, Reference Brennan2022), stakes are higher for racialized groups and other marginalized members of the population who already show higher rates of morbidity and mortality than others. Toward addressing and acknowledging the pervasive and persistent nature of mental health disparities, the National Institute of Mental Health recently published their Strategic Framework for Addressing Youth Mental Health Disparities (NIMH, 2023), which codifies and sets funding priorities for closing knowledge gaps in environmental, biological, cognitive, emotional, and behavioral mechanisms of mental health disparities—i.e., at multiple levels of analysis.

Better understanding of common, unique, and diverging mechanisms of risk within and across U.S. subpopulations (e.g., sex, race, sexual orientation, gender identity, socioeconomic status, etc.) is essential to simultaneously and effectively target relevant group- and individual-level mechanisms across levels of analysis. True transdisciplinarity research—in contrast to multiple but insular lines of research contributed by separate disciplines—is a historically recent development with uncharted potential for bringing interventions to children and families who need them most. With positive-sum science, important work on structural racism, discrimination, and stigma need not be deprioritized (e.g., Acker et al., Reference Acker, Aghaee, Mujahid, Deardorff and Kubo2023; Hatzenbuehler et al., Reference Hatzenbuehler, McLaughlin, Keyes and Hasin2010).

In our own work, we are combining information from multiple institutions (local healthcare systems, departments of health, Census Bureau, vaccine uptake) and levels of analysis (medical conditions, prenatal visits, parental ACE exposures) and using machine learning to identify underserved families with expecting mothers in their first trimester who are least likely to attend child well-visits and receive other forms of pre- and postnatal care (Beauchaine & Fox, Reference Beauchaine and Fox2023). We then send community mental health workers directly to mothers’ places of living, including homeless shelters, to provide those services. Some participating mothers attend brief focus groups to improve program efficacy moving forward (see Carter & Mazzoni, Reference Carter and Mazzoni2021). Only by collecting data across all available levels of analysis can we effectively redirect critical services in an existing system of care to those members of underserved communities who need them now. Focusing exclusively on institutions precludes the granular understanding needed to accomplish this objective. Every sliver of data available is needed to maximize prediction accuracy. In this case, the zero-sum framework compromises our understanding of disparities and erodes our capacity to predict current and future utilization of care. Individuals within their communities are adversely affected.

Interim summary

Paradigmatic science is often a zero-sum game. Paradigms, theories, hypotheses, and measurement models are pitted against one another, with proponents on each side seeking “wins.” In a zero-sum context, such wins limit scientific progress by relegating useful findings, principles, and interventions from the “loser” to the annals of history. Many such wins are dubious given (a) rejection of established theories based on ideological over empirical grounds; (b) oversimplification of rival paradigms and mischaracterizations of their proponents’ perspectives; (c) logically suspect arguments based on false choices and overgeneralizations; and (d) the practice of testing superordinate effects in nested data structures while ignoring subordinate effects. The common throughline in these items is prioritization of one’s preferred paradigm, epistemological assumptions, and methods as means of knowing. This neednʼt be intentional and likely isnʼt much of the time. Confirmation bias is well-documented in social science and is not a sign of malice (e.g., Hergovich et al., Reference Hergovich, Schott and Burger2010). At the same time, it curtails scientific advances and is amplified in the zero-sum game context many scholars are socialized in. My assertion here, which I devote the remainder of this paper to, is that the current zeitgeist of multidisciplinary science, although a crowning achievement of developmental psychopathology to date, is necessary but insufficient for overcoming zero-sum thinking and the constraints it imposes on transformative social science. Positive-sum science requires a step forward to paradigmatic and epistemological pluralism.

Developmental psychopathology: From multidisciplinarity to epistemological pluralism

Acknowledging the daunting complexity of psychopathology is historically recent (Cohen, Reference Cohen2016; Hall et al., Reference Hall, Moore and Ritchie2016). Determinants and correlates of human behavior, including psychopathology, are orders of magnitude more complex than understood two decades ago, when, for example, psychiatric geneticists still assumed oligogenic vulnerability to psychopathology. Since then, cross-disciplinary research has implicated thousands of vulnerability genes (Ripke et al., Reference Ripke, OʼDushlaine, Chambert, Moran, Kähler, Akterin, Bergen, Collins, Crowley, Fromer, Kim, Lee, Magnusson, Sanchez, Stahl, Williams, Wray, Xia and Bettella2013); identified overlapping genetic risk to multiple disorders (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013); specified complex genetic, epigenetic, hormonal, and environmental determinants of brain function (Beauchaine & Constantino, Reference Beauchaine and Constantino2017); and demonstrated nonlinear interactions among neural systems in affecting human behavior (Beauchaine & Hinshaw, Reference Beauchaine and Hinshaw2020; Haines et al., Reference Haines, Beauchaine, Galdo, Rogers, Hahn, Pitt, Myung, Turner and Ahn2020). Many more etiological inputs could be described. In this context, valuing multidisciplinary research is unsurprising, but it hasnʼt always been this way.

Among Dante Cicchetti’s sure to be enduring contributions to developmental psychopathology—and there will be many—is the overarching, integrative inclusiveness he brought to the field. Cicchetti saw the benefits of engaging with and bridging to a wide range of disciplines with varying epistemological assumptions, research foci, and preferred analytic methods. This theretofore unseen inclusiveness is evident in his earliest writing in this journal: the first four pages of the first issue of Development and Psychopathology, where his first editorial appeared (Cicchetti, Reference Cicchetti1989). Cicchetti stated, “…developmental psychopathology is a comprehensive approach which strives to integrate elements derived from the fields of developmental psychology, clinical psychology and psychiatry, epidemiology, sociology, and both the physiological sciences and neurosciences” (pp. 1–2; see also Cicchetti, Reference Cicchetti, Rolf, Masten, Cicchetti, Nuechterlein and Weintraub1990). Since then, Cicchetti has co-edited, with carefully chosen colleagues across various disciplines, over 60 special issues in which experts from diverse fields and perspectives contributed. Over the 35-year period beginning in 1989, interdisciplinary research emerged where it had not been before, elucidating ontogenic and transactional influences on psychopathology across levels of analysis spanning genes to cultures (Beauchaine et al., Reference Beauchaine, Constantino and Hayden2018; Masten, Reference Masten2006). This work specified multicausal pathways to assumedly single disorders, described divergent outcomes from common etiological starting points, and identified biological and behavioral mechanisms through which interventions exert their effects (Beauchaine & Slep, Reference Beauchaine and Slep2018; Beauchaine et al., Reference Beauchaine, Zisner, Hayden, Ollendick, White and White2019b; Bell et al., Reference Bell, Shader, Webster-Stratton, Reid and Beauchaine2018).

Cicchetti facilitated—more than any other scholar—progression of developmental psychopathology from unidisciplinary to multidisciplinary to interdisciplinary to transdisciplinary. It is difficult to overstate the value of this sweeping contribution and the extent to which disciplinary boundaries were diffused (Beauchaine et al., Reference Beauchaine, Constantino and Hayden2018). A challenge for the next generation is to finish this work. Despite overwhelming progress, further interdisciplinary integration is essential for constructing the grand theories needed for groundbreaking insights into psychopathology and its development, and for leveraging those insights into more effective prevention and intervention programs. Such progress depends on (1) reformulating zero-sum science; (2) breaking down lingering disciplinary hierarchies; and (3) confronting confirmation bias whereby scientists reject other’s out-of-discipline and out-of-paradigm scholarship using higher standards of evidence than they use to reject within-discipline and within-paradigm science. Addressing these points is essential for constructing integrative grand theories without persistently and repetitively “pulling the curtains down” on one another.

Epistemological pluralism places a broad range of paradigmatic assumptions, scientific perspectives, and methodological practices on equal footing, eliminating disciplinary hierarchies in transdisciplinary research teams and in science more broadly (Healy, Reference Healy2003; Miller et al., Reference Miller, Baird, Littlefield, Kofinas, Chapin and Redman2008). Major assumptions of epistemological pluralism are that no single paradigm or perspective can capture all aspects of complex phenomena, and no discipline or paradigm holds a priori predominance in interpreting events, rejecting paradigms, or constraining research agendas (Wegerhoff et al., Reference Wegerhoff, Ward and Dixon2022). Epistemological pluralists recognize the complexity of natural phenomena, juxtapose that complexity with the limits of knowledge for single human agents, acknowledge the overwhelming level of detail needed to understand even local problems and events, and how that level of detail obscures and biases one’s interpretations of distal problems and events even within one’s field (Potochnik, Reference Potochnik2017). According to this perspective, understanding is maximized by taking a transdisciplinary but “modular” approach to studying complex phenomena, combining multiple models built by “local” experts into a grand theoretical account that exceeds the sum of its parts (Ruphy, Reference Ruphy2016). I provide an example below.

Epistemological pluralism encourages humility in science, respect for paradigmatic assumptions in other disciplines, and trust in others’ training and competencies. Few neuroscientists with expertise in designing tasks to elicit, test, and construct theories about neural reactivity to appetitive, aversive, and emotion-eliciting stimuli (neural substrates of operant conditioning) in the constrained environment of an MRI scanner can fully appreciate or make optimally informed scientific judgements about work by a behaviorist with expertise in operant conditioning, functional analysis of behavior change, and designing response-cost contingencies for different populations at different ages in different settings. Neither can many so-described behaviorists fully appreciate or make optimally informed scientific judgements about advanced work in neuroscience. By themselves, these are complex roles, and both neuroscientists and behaviorists, depending on the scope of the project, are viewed as essential contributors to a transdisciplinary research team. Their “modular” expertise provides for fruitful cross-pollination of ideas, with potential for scientific innovation. Although the transition from single principal investigators to multiple principal investigators on NIH grants is facilitative in this regard, it is a nominal move if not implemented in practice.

With epistemological pluralism and positive-sum science, paradigmatic heterogeneity is a strength, not a limitation. This provides opportunities for growth and construction of overarching theoretical models of psychopathology that accommodate all levels of analysis simultaneously. It discourages scientists from making common errors in logic, including (a) rejecting grand theories based on refutation of local conjectures, and (b) rejecting grand theories that do not account for every person who exhibits a specific behavior or disorder. In both cases, we throw the baby out with the bathwater. Paradigmatic and epistemologic humility and acknowledgement of the complexity of psychopathology obviate these impulses.

I illustrate these principles in Figure 2, using our neurodevelopmental model of antisocial behavior, which describes how biological vulnerability (heritable trait impulsivity) is shaped by environmental risk (e.g., family coercion, deviant peer affiliations, early substance exposure, discrimination and racism) and protection (e.g., trusted adults, neighborhood cohesion) across development (Beauchaine et al., Reference Beauchaine, Zisner and Sauder2017; Beauchaine, Reference Beauchaine, Lejuez and Gratz2020; Reference Beauchaine, Hinshaw and Bridge2019a; McDonough-Caplan & Beauchaine, Reference McDonough-Caplan, Beauchaine and Martel2018). Although I cannot review all effects here, each has well-replicated local support. I acknowledge this is one common pathway to antisociality, not the only pathway, that specific vulnerabilities and cumulative risk exposures extend beyond the examples provided, and that such exposures vary in type and timing within individuals across development. Although not represented in Figure 2 given space limitations, these caveats are readily accommodated.

Figure 2. Depiction of developmental psychopathology (DP) as a positive-sum metaparadigm, using the example of externalizing behavior development from preschool to adulthood for trait impulsive boys (Ahmad & Hinshaw, Reference Ahmad and Hinshaw2017; Beauchaine et al., Reference Beauchaine, Zisner and Sauder2017; Bell et al., Reference Bell, Fristad, Youngstrom, Arnold, Beauchaine, Findling, Birmaher and Horwitz2021; see Beauchaine, Reference Beauchaine, Lejuez and Gratz2020 for a parallel model for girls). Bolded text within this figure caption indicates core DP principles. Predisposing vulnerability to externalizing progression is conferred by heritable trait impulsivity, a primary source of hyperactive-impulsive and combined subtypes of ADHD. Vulnerability is multifactorial, and is instantiated across multiple levels of analysis including genetic, neurohormonal, neural, emotional, temperamental, and behavioral (other levels of analysis, such as psychophysiological, are omitted). Progression (continuity) is potentiated by multiple environmental risk mediators (bottom), which change across development depending on context to “pull” affected individuals along the externalizing spectrum. The list of environmental risk mediators is incomplete, and in the real world, risk mediators overlap. Trait impulsivity confers risk at all ages, whereas emotion dysregulation takes on increasing importance in later childhood and adulthood via deficient prefrontal cortex neuromaturation. Discontinuity, which is not depicted, can occur in contexts of resilience and protection. Arbitrarily chosen examples of discipline-specific paradigms (colored circles and boxes) dictate local research practices, assumptions, and methods, with no “mandate” of epistemological unity. Neither the inclusive DP metaparadigm, which can be applied to internalizing progression and other disorders (e.g., Hankin et al., Reference Hankin, Snyder, Gulley, Schweizer, Bijttebier, Nelis, Toh and Vasey2016), nor the specific theoretical model of externalizing progression, depend on any single local hypothesis, and hypotheses at different levels of analysis are not pitted against one another, as is common in the current zero-sum game context. All relations (arrows) are supported by existing literature (see Beauchaine & Constantino, Reference Beauchaine and Constantino2017; Beauchaine et al., Reference Beauchaine, Zisner and Sauder2017).

Certain other predisposing vulnerabilities, such as in utero substance exposure and head injury, are not readily accommodated, and require different models given different core mechanisms.Footnote ¹¹ I also recognize that heritable temperamental impulsivity does not determine adult antisocial behavior, and that in contexts devoid of risk, progression past ADHD (the diagnostic manifestation of functionally impairing impulsivity) is unlikely. I refer interested readers to selected lab publications for further details (e.g., Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Beauchaine et al., Reference Beauchaine, Zisner and Sauder2017; McDonough-Caplan & Beauchaine, Reference McDonough-Caplan, Beauchaine and Martel2018; Rutter et al., Reference Rutter, Birch and Chess1964).

Developmental psychopathology as a metaparadigm

Just over a decade ago, Micheal Rutter (Reference Rutter2013) considered the question, in this journal, of whether developmental psychopathology initiated a paradigm shift in clinical science. After cataloging discoveries within the field related to several child-onset disorders, Rutter listed defining achievements, including identification of environmental risk mediators, expanding work on gene-environment interaction and correlation, demonstrating effects of parental mental illness on offspring, specifying Vulnerability × Stress interactions, modeling ethnicity effects, and mapping continuities and discontinuities in behavior. Rutter concluded that developmental psychopathology indeed initiated a paradigm shift. He went on to explain (p. 1210):

I have always found Rutter’s conclusion and explanation to be somewhat at odds. The success of developmental psychopathology is undeniable. It is surely an impressive organizing system with broad reach that redirected research emphases of many scholars. Continuity, discontinuity, and mediation were sharpened as areas of focus by developmental psychopathologists, but these, as Rutter alludes to, preexisted in sociology (continuity and discontinuity; e.g., Robins, Reference Robins1966) and social psychology (mediation; Baron & Kenny, Reference Baron and Kenny1986). And if developmental psychopathology was a paradigm shift, what was it shifting from? Almost all of the disciplines it touched, along with their defining scientific paradigms, remain intact (e.g., developmental psychology, epidemiology, clinical psychology, education, psychiatry, etc.). Even in clinical psychology, where many adherents reside, most graduate programs are split along child and adult lines, just as they have always been. Major breakthroughs at biological levels of analysis, such as genome sequencing, epigenetic divergence across development, and neurobiological substrates of human behavior, came from the usual suspects and were later adopted and extended by developmental psychopathologists (Fowles, Reference Fowlesin press; Fraga et al., Reference Fraga, Ballestar, Paz, Ropero, Setien, Ballestar, Heine-Suñer, Cigudosa, Urioste, Benitez, Boix-Chornet, Sanchez-Aguilera, Ling, Carlsson, Poulsen, Vaag, Stephan, Spector, Wu and Esteller2005; Heather & Chain, Reference Heather and Chain2016). Research on Gene × Environment interaction originated in the adult psychopathology literature (Gottesman, Reference Gottesman1963), and advanced techniques for modeling nested longitudinal data originated in education and criminology (e.g., Nagin & Land, Reference Nagin and Land1993; Rogosa et al., Reference Rogosa, Brandt and Zimowski1982). Multiple-levels-of-analysis research rightfully ascended (Cicchetti & Dawson, Reference Cicchetti and Dawson2002), but this too was organizational, with no new levels added. Although resilience research (e.g., Masten & Cicchetti, Reference Masten, Cicchetti and Cicchetti2016) and work on the neurobiology of stress responding (e.g., Loman & Gunnar, Reference Loman and Gunnar2010) may be considered exceptions, they too have firm roots in literatures that emerged before developmental psychopathology’s inception (Gunnar, Reference Gunnar2021; Luthar, Reference Luthar, Cicchetti and Cohen2006).

Rutter’s (Reference Rutter2013) conclusion seems to follow expectations of paradigm shifts in the physical sciences – rapid changes in assumptions that affect the methods and procedures used across entire fields (American Psychological Association, 2018). Yet paradigm shifts in the physical sciences are elegant in their simplicity and unity. A round earth, a heliocentric solar system, Mendelian inheritance, evolution by natural selection, and special relativity can all be distilled to a single idea, expressed in one or two sentences, and connoted by a single epistemology. They are generally unavoidable by members of the scientific community of physicists, chemists, etc. even by those who might advocate an alternative.

This is not the nature of paradigms or paradigm shifts in the social sciences, where disciplinary and subdisciplinary paradigms and associated epistemologies abound, and where no single paradigm constrains scientific activities of all or even most members of a scientific community (De Vincenzo et al., Reference De Vincenzo, Stocco and Modugno2023; Polsby, Reference Polsby2003). Subdisciplinary paradigms tend to be localized to one or two levels of analysis, and may only be vaguely familiar to colleagues in different areas of single departments. This is the very situation where epistemological pluralism is most needed. Developmental psychopathology does not bring about unity in the sense of a classic paradigm in the physical sciences. Its collective principles serve an essential organizing function for disciplinary and subdisciplinary paradigms that otherwise would not mix. It advances social science by imposing epistemological and paradigmatic plurality. It is a metaparadigm. Paradigmatic unification in the social sciences is an unrealistic and impractical ideal. No single account can possibly capture all essential aspects of such complex phenomena (Ritzer, Reference Ritzer1975; Wegerhoff et al., Reference Wegerhoff, Ward and Dixon2022).

Construed as a metaparadigm, developmental psychopathology holds enormous untapped potential to move the field from zero-sum thinking and scholarship to positive-sum science and epistemological pluralism. This integrative vision, which furthers Dante Cicchetti’s legacy of interdisciplinarity, requires broad commitment among scientists to avoid zero-sum scholarship in which important ideas, useful theories and principles, and effective interventions are rejected based on confirmation bias, errors in logic, and ideology. Although talk of transdisciplinarity is now pedestrian and transdisciplinary research teams abound, our approach to scholarship remains largely zero-sum. Causes are surely many, and likely include (a) our disciplinary history of and dogged adherence to null-hypothesis significance testing; (b) highly specialized journals that, wittingly or unwittingly, demand content fully consistent with local paradigmatic and epistemologic views and assumptions; (c) a highly competitive funding climate that prioritizes innovation over integration; (d) historical and entrenched disciplinary hierarchies; and (e) limited education about and continued effectiveness of confirmation bias, oversimplification, false dilemmas, overgeneralization, and ideological appeals in arguing against and rejecting viable theories, ideas, and interventions.

Moving from zero-sum to positive-sum science is not an all-or-nothing endeavor. Much progress is already behind us, thanks to efforts by Dante Cicchetti and other leaders in the field who established its broad interdisciplinary reach and translational research agenda. It is difficult to imagine where the field would be without the metaparadigmatic, now 35-year influence of Development and Psychopathology. As Cicchetti (Reference Cicchetti1989), the only editor to date said in his first editorial, “…the success of this journal, as well as the advancement of developmental psychopathology, is dependent upon our commitment to realizing the potential of the field. I invite you to become an active participant in this process.” (p. 3). It is time for the next generation to heed the call.