Introduction
When we discuss developmental psychopathology, when we write about attachment, or the dyad, or Gene×Environment interaction, an elephant in the room is often overlooked. This is the fact that the most powerful predictor of mental illness is socioeconomic deprivation (Lund et al., Reference Lund, Brooke-Sumner, Baingana, Baron, Breuer, Chandra and Saxena2018). This is the starting point for the perspective we would like to bring to this paper. We are not aiming to present a naively political or materialist position, in which everything is reduced to socioeconomic forces. Rather, we argue that this epidemiological reality needs to be more meaningfully integrated into theoretical models of developmental psychopathology. We will set out here an approach that seeks to locate the social environment around the individual as central to their subjective experience, and their vulnerability to psychological disorder. This is, in part, an evolutionary approach, informed by recently emerging thinking around the social and culturally driven nature of human cognitive development and the ways in which humans are primed to learn and communicate culture, fostered by the capacity for mentalizing, which enables the highly intersubjective nature of our capacity for affect regulation and social functioning. We will set this out in the second half of this paper, but first we would like to outline the case for a new social communicative model of psychopathology in which we suggest that the mechanisms through which culture is maintained and transmitted across generations are overlapping and in places isomorphic with the developmental generation of mental disorder. In support of this argument, we will turn to recent research on the structure of psychopathology and findings on a general psychopathology, or “p” factor.
Why do we Need a New Model?
We will begin with discussion of the difficulties that face our current conceptualization of psychopathology, and which, we will propose, support thinking about the construction of a unified model of disorder. We suggest that the fragmentation of mental disorder into hundreds of diagnostic groupings driven by the pragmatism of practice and the application of prescriptive models of treatments has impeded the understanding of the way global influences such as culture affect the development of mental disorder. The Research Domain Criteria (RDoC) movement demonstrated that the understanding of disease mechanisms requires abandoning phenomenologically rooted categories of disorder (Insel et al., Reference Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn and Wang2010). We will review evidence that points to the appropriateness of a single, unified conceptualization of psychopathology which offers a frame for understanding how culture interfaces with the emergence of mental disorder.
Clinical problems: Comorbidity and severity
Perhaps the most acutely felt source of dissatisfaction with diagnostic taxonomies is the high degree of comorbidity which affects the majority of individuals who meet criteria for any one diagnosis (Kessler, Chiu, Demler, Merikangas, & Walters, Reference Kessler, Chiu, Demler, Merikangas and Walters2005). Comorbidity transcends diagnoses, and overlap is observed between most symptoms known to be associated with mental disorder (e.g., Budde et al., Reference Budde, Anderson-Schmidt, Gade, Reich-Erkelenz, Adorjan, Kalman and Heilbronner2018). Comorbidity is ubiquitous among all common mental disorders (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014; Copeland, Shanahan, Erkanli, Costello, & Angold, Reference Copeland, Shanahan, Erkanli, Costello and Angold2013). From the perspective of developmental psychopathology, comorbidity may be considered largely artefactual due to nosology splitting disease entities into subcategories that likely lack validity (Beauchaine & Cicchetti, Reference Beauchaine and Cicchetti2016; Forbes, Tackett, Markon, & Krueger, Reference Forbes, Tackett, Markon and Krueger2016). If disorders were reconceptualized as indicators of latent transdiagnostic spectra, comorbidity would no longer be a problem. If disorders share phenotypic and genotypic variance, comorbidity seems an inappropriate and unhelpful term that could be readily replaced by the phrase “frequent co-occurrence” (Goldberg, Reference Goldberg2015). The term “comorbidity” should be reserved for etiologically distinct disorders such as bipolar disorder and hiatus hernia, as was originally intended (Feinstein, Reference Feinstein1970).
A second, equally problematic area in relation to the diagnosis of mental disorder is the concept of severity. A paper by Zimmerman, Morgan, and Stanton (Reference Zimmerman, Morgan and Stanton2018) demonstrates that the generally used concept of severity has no reliable agreed referent, either across diagnostic categories or, at closer scrutiny, even within disorders. It is used variously to refer to number of symptoms, the intensity of symptoms, the frequency or persistence of symptoms, the impact of symptoms on quality of life or general adaptation, the likelihood of permanent disability or death, and so on. Although the fifth edition of Diagnostic and Statistical Manual of Mental Disorders (DSM-5), in a number of instances, offers specific criteria for quantification or qualification, these are not consistently validated against external criteria or in universal use (American Psychiatric Association, 2013). Zimmerman et al. (Reference Zimmerman, Morgan and Stanton2018) systematically examined the way severity has been conceptualized and researched in relation to depression and personality disorders. In both cases, treatment recommendations are made with reference to treatment severity (National Institute for Health and Care Excellence, 2009; National Institute for Health and Clinical Excellence, 2013). For example, two meta-analyses used a cut-off of 20 or more on the 17-item Hamilton Rating Scale for Depression (HAM-D) (DeRubeis, Gelfand, Tang, & Simons, Reference DeRubeis, Gelfand, Tang and Simons1999; Gibbons, Hur, Brown, Davis, & Mann, Reference Gibbons, Hur, Brown, Davis and Mann2012) using Elkin et al.'s (Reference Elkin, Gibbons, Shea, Sotsky, Watkins, Pilkonis and Hedeker1995) landmark study as justification. A difficulty with this is that Elkin et al. (Reference Elkin, Gibbons, Shea, Sotsky, Watkins, Pilkonis and Hedeker1995) cited no evidence for using that particular cut-off score. Other studies cite different cut-offs for the same instrument (Dunner, Lipschitz, Pitts, & Davies, Reference Dunner, Lipschitz, Pitts and Davies2005; Kirsch et al., Reference Kirsch, Deacon, Huedo-Medina, Scoboria, Moore and Johnson2008; Zimmerman, Martinez, Young, Chelminski, & Dalrymple, Reference Zimmerman, Martinez, Young, Chelminski and Dalrymple2013) but each also uses clinician's global judgement of severity (CGI). Of course, a further issue is that we do not know how clinicians weight various symptoms as they assign CGI ratings alongside a HAM-D assessment. Using different scales, we find quite different metrics for the classification of severity (Zimmerman et al., Reference Zimmerman, Martinez, Friedman, Boerescu, Attiullah and Toba2012). The picture in relation to the severity of personality disorder is even more confusing. Does it refer to co-occurring disorders (Links & Eynan, Reference Links and Eynan2013) or other personality disorder diagnoses (Bateman & Fonagy, Reference Bateman and Fonagy2013) or more severe symptoms such as self-injury (Zanarini et al., Reference Zanarini, Yong, Frankenburg, Hennen, Reich, Marino and Vujanovic2002) or indication of social functioning (Yang, Coid, & Tyrer, Reference Yang, Coid and Tyrer2010)? These and over 200 references in the review of Zimmerman et al. (Reference Zimmerman, Morgan and Stanton2018) point to a profound difficulty in the use of the severity concept as nested within diagnoses. However, apart from severity, how do we judge the general seriousness of psychopathology? Is co-occurrence/comorbidity equivalent to severity? Is persistence/treatment resistance an overlapping construct? One potential answer may be in the psychometric construct of the p factor.
Psychometric evidence
A compelling tranche of evidence for a singular model for psychopathology derives from recent work in factor analysis. A method of capturing underlying covariation amongst a number of variables is the so-called bifactor model, which assumes that all variables have common variance explained by a single factor independent of covariance accounted for by clusters of such variables (spectral level factors, e.g., internalizing and externalizing) (Markon, Reference Markon2019). The p factor is thus a statistical summary of the variance common to symptoms, across diagnoses and spectral clusterings (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014). It was so labelled by Caspi and colleagues to assist in making the conceptual leap from Spearman's intelligence manifold (the g factor). Both the general factor (e.g., p factor) and specific factors (e.g., internalizing, externalizing) contribute to the ways in which symptoms or disorders positively co-occur with each other, but the majority of what we measure using interviews or questionnaires is underpinned by the general factor (Constantinou & Fonagy, Reference Constantinou and Fonagy2019). The bifactor model consistently outperforms other models in describing the covariation in psychopathology data across children (Lahey et al., Reference Lahey, Rathouz, Keenan, Stepp, Loeber and Hipwell2015; Moore et al., Reference Moore, Kaczkurkin, Durham, Jeong, McDowell, Dupont and Lahey2020; Olino et al., Reference Olino, Bufferd, Dougherty, Dyson, Carlson and Klein2018; Sallis et al., Reference Sallis, Szekely, Neumann, Jolicoeur-Martineau, van, Hillegers and Evans2019), adolescents (Castellanos-Ryan et al., Reference Castellanos-Ryan, Briere, O'Leary-Barrett, Banaschewski, Bokde and Bromberg2016; Laceulle, Vollebergh, & Ormel, Reference Laceulle, Vollebergh and Ormel2015; Patalay et al., Reference Patalay, Fonagy, Deighton, Belsky, Vostanis and Wolpert2015; Snyder, Young, & Hankin, Reference Snyder, Young and Hankin2017), and adults (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014; Gluschkoff, Jokela, & Rosenström, Reference Gluschkoff, Jokela and Rosenström2019; Lahey et al., Reference Lahey, Applegate, Hakes, Zald, Hariri and Rathouz2012, Reference Lahey, Zald, Perkins, Villalta-Gil, Werts, Van Hulle and Waldman2018). Some of the above studies indicate that the p factor best accounts for covariation among disorder-level indicators, while others demonstrate the bifactor model's superiority using symptom-level indicators, though the p factor may not represent all covariation between specific spectral factors and common variance remains for specific dimensions to be correlated (Carragher et al., Reference Carragher, Teesson, Sunderland, Newton, Krueger, Conrod and Slade2016; Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014). The p factor appears to capture an underlying propensity for any kind of psychopathology (Caspi & Moffitt, Reference Caspi and Moffitt2018) or an overall index of seriousness or impairment (Smith, Atkinson, Davis, Riley, & Oltmanns, Reference Smith, Atkinson, Davis, Riley and Oltmanns2020), which can be represented as a global statistic that is open to external validation.
A number of methodological issues need to be resolved before the concept of a general psychopathology factor could be widely adopted by developmental researchers and clinicians. First, symptoms are shared across disorders, both within and between spectral level factors; this could generate the appearance but not the reality of a general psychopathology factor. For example, both adolescent depression and antisocial personality disorder are characterized by irritability. However, when these overlapping symptoms are removed, the general model remains unchanged (Lahey et al., Reference Lahey, Zald, Perkins, Villalta-Gil, Werts, Van Hulle and Waldman2018). Some have suggested that common method variance, shared biases, positive or negative halo effects and so on could account for a general factor. Modern psychometrics, however, recognizes that traditional “nuisance” variables (such as confirmatory bias) are better seen as part of what one wishes to measure rather than something to exclude (Podsakoff, MacKenzie, Lee, & Podsakoff, Reference Podsakoff, MacKenzie, Lee and Podsakoff2003). In a comprehensive, multitrait, multimethod analysis, positive covariation among a range of disorders was shown to exist across ratings from different informants (Lahey, Krueger, Rathouz, Waldman, & Zald, Reference Lahey, Krueger, Rathouz, Waldman and Zald2017), meaning that it is not simply biased responding from a given informant that drives the positive manifold. It is more difficult to rebut criticisms that the bifactor model may be repeatedly favored over competing models such as the traditional correlated factor model because it is better at accommodating noise in the data (i.e., the problem of “overfitting” [Greene et al., Reference Greene, Eaton, Li, Forbes, Krueger, Markon and Kotov2019; Murray & Johnson, Reference Murray and Johnson2013; Reise, Kim, Mansolf, & Widaman, Reference Reise, Kim, Mansolf and Widaman2016]) or is statistically more complex than its counterparts (Bonifay & Cai, Reference Bonifay and Cai2017), meaning fit indices favor the bifactor model for statistical rather than substantive reasons. Others have argued that the bifactor model simply redistributes the covariance among observed indicators rather than providing a new model of sorts (Vine et al., Reference Vine, Byrd, Mohr, Scott, Beeney and Stepp2020; Watts, Poore, & Waldman, Reference Watts, Poore and Waldman2019). Some have even called into question whether a general factor is necessary or sufficient to explain the positive manifold among symptom items (van Bork, Epskamp, Rhemtulla, Borsboom, & van der Maas, Reference van Bork, Epskamp, Rhemtulla, Borsboom and van der Maas2017). In order for the bifactor model to be meaningfully applied to research and practice, there have been calls to move beyond model fit indices and evaluate the general and specific dimensions against external criteria to test theoretical predictions (Bornovalova, Choate, Fatimah, Petersen, & Wiernik, Reference Bornovalova, Choate, Fatimah, Petersen and Wiernik2020; Sellbom & Tellegen, Reference Sellbom and Tellegen2019).
If the p factor signals a general predisposition to mental disorder, as we wish to establish here, then it should be associated with independent measures of risk factors for psychopathology. Higher p factor scores are associated with early adverse experiences and a family history of psychopathology (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014; Deutz et al., Reference Deutz, Geeraerts, Belsky, Dekovic, van Baar, Prinzie and Patalay2020; Hyland et al., Reference Hyland, Murphy, Shevlin, Bentall, Karatzias, Ho and McElroy2020; Lahey et al., Reference Lahey, Applegate, Hakes, Zald, Hariri and Rathouz2012; Martel et al., Reference Martel, Pan, Hoffmann, Gadelha, do Rosario, Mari and Salum2017; Schaefer et al., Reference Schaefer, Moffitt, Arseneault, Danese, Fisher, Houts and Caspi2018). Of particular relevance to the current paper, the p factor is also associated with markers of socioeconomic disadvantage, whether they are assessed cross-sectionally (Belsky et al., Reference Belsky, Caspi, Arseneault, Corcoran, Domingue, Harris and Odgers2019; Lahey et al., Reference Lahey, Applegate, Hakes, Zald, Hariri and Rathouz2012; Patalay et al., Reference Patalay, Fonagy, Deighton, Belsky, Vostanis and Wolpert2015), prospectively (Blanco et al., Reference Blanco, Wall, Hoertel, Krueger, Liu, Grant and Olfson2021; Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014; Schäfer et al., Reference Schäfer, McLaughlin, Manfro, Pan, Rohde, Miguel and Salum2020), or longitudinally (Snyder, Young, & Hankin, Reference Snyder, Young and Hankin2019; Wade, Fox, Zeanah, & Nelson, Reference Wade, Fox, Zeanah and Nelson2018). Furthermore, heightened p factor scores in adolescence are associated with experiences of racial discrimination (Liu, Mustanski, Dick, Bolland, & Kertes, Reference Liu, Mustanski, Dick, Bolland and Kertes2017), and higher p scores in adulthood are independently predicted by experiences of childhood and adolescent victimization, after accounting for genetic liabilities and pre-existing symptomatology (Schaefer et al., Reference Schaefer, Moffitt, Arseneault, Danese, Fisher, Houts and Caspi2018). This growing body of research demonstrates that experiences of social exclusion and disadvantage in the family, peer and neighborhood environments contribute to one's general susceptibility to mental ill-health, calling for a comprehensive developmental psychopathology account of putative mechanisms.
Genetic evidence
Genetic evidence from both behavior genetics and molecular biology is also consistent with the general psychopathology argument. Behavior genetic studies on families and twins tend to show that genetic risk is not specific to particular disorders. Rather it is largely a transdiagnostic vulnerability. However, the literature also offers evidence for genetic associations at some specific spectral level factors. Several studies report factor analyses of genetic correlations based on twin samples. In these studies, a variable number of syndromal categories are identified and the configuration of spectral factors differs, but evidence consistently points to almost all the genetic covariance being explained by a general genetic factor (Lahey, Van Hulle, Singh, Waldman, & Rathouz, Reference Lahey, Van Hulle, Singh, Waldman and Rathouz2011; Pettersson, Larsson, & Lichtenstein, Reference Pettersson, Larsson and Lichtenstein2016; Selzam, Coleman, Caspi, Moffitt, & Plomin, Reference Selzam, Coleman, Caspi, Moffitt and Plomin2018; Waldman, Poore, Luningham, & Yang, Reference Waldman, Poore, Luningham and Yang2020). The notion of a general vulnerability to psychopathology, regardless of diagnosis, is supported by family studies that show a general rising of risk of mental disorder with increased relatedness given a diagnosis of severe mental disorder in a proband (Song et al., Reference Song, Bergen, Kuja-Halkola, Larsson, Landen and Lichtenstein2015). There is little evidence for diagnosis-specific genetic influence (Wolf et al., Reference Wolf, Miller, Krueger, Lyons, Tsuang and Koenen2010). In terms of specific spectral levels, externalizing (e.g., aggression, conduct problems) has a high genetic influence (Waldman, Poore, van Hulle, Rathouz, & Lahey, Reference Waldman, Poore, van Hulle, Rathouz and Lahey2016). In contrast, nonshared environmental influences may be specific to diagnoses (Waldman et al., Reference Waldman, Poore, van Hulle, Rathouz and Lahey2016).
Molecular genetic studies confirm that phenotypic similarity can be accounted for by concordance at the molecular level (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013; Lee et al., Reference Lee, Ripke, Neale, Faraone, Purcell and Perlis2013). A genome-wide association study of 25 brain disorders and 17 mental disorders from over 1 million participants reveals that mental disorders shared common variance risks (Brainstorm Consortium et al., Reference Anttila, Bulik-Sullivan, Finucane, Walters, Bras and Murray2018). This was less evident for brain disorders, normally considered neurological, which appear more distinct from each other. The findings confirm the importance of common genetic variation as a risk factor for mental disorders. Genetic evidence suggests that heritability will be increased at the higher order within a hierarchical structure and lowest at first-order dimensions at the levels of symptoms (Budde et al., Reference Budde, Anderson-Schmidt, Gade, Reich-Erkelenz, Adorjan, Kalman and Heilbronner2018). Single nucleotide polymorphism studies confirm that roughly 50% of the variance in heritability estimates is explained by a general psychopathology factor, with lower estimates associated with specific internalizing or externalizing factors (Harden et al., Reference Harden, Engelhardt, Mann, Patterson, Grotzinger, Savicki and Tucker-Drob2020; Neumann et al., Reference Neumann, Pappa, Lahey, Verhulst, Medina-Gomez, Jaddoe and Tiemeier2016; Rosenström et al., Reference Rosenström, Gjerde, Krueger, Aggen, Czajkowski, Gillespie and Ystrom2019). This is consistent with the assumption that the same gene may have a role to play in the range of disorders (Lee et al., Reference Lee, Ripke, Neale, Faraone, Purcell and Perlis2013; Smoller, Reference Smoller2013). This certainly seems to be evident from studying mental health risks associated with a single gene such as 22q11.2, whose deletion appears to create mental health risks well captured by a bifactor model (Niarchou et al., Reference Niarchou, Moore, Tang, Calkins, McDonald-McGuinn, Zackai and Gur2017).
Neuroscientific evidence
Neurobiological causal models that link to the assumption of a singular model of mental disorder most commonly invoke dysfunction of the prefrontal cortex (PFC) (Macdonald, Goines, Novacek, & Walker, Reference Macdonald, Goines, Novacek and Walker2016, p. 1148). There is evidence from meta-analytic studies of reduced gray matter volume in the PFC and limbic regions of individuals with mental disorder (Wise et al., Reference Wise, Radua, Via, Cardoner, Abe, Adams and Arnone2017). In a bifactor analysis of over 2,500 11-year-olds, the p factor predicted reduced gray matter volume in the dorsal, orbito-frontal and ventro-lateral regions in the PFC (Snyder, Hankin, Sandman, Head, & Davis, Reference Snyder, Hankin, Sandman, Head and Davis2017). In a sample of over 1,000 16-year-olds, Kaczkurkin et al. (Reference Kaczkurkin, Moore, Calkins, Ciric, Detre, Elliott and Satterthwaite2018) found that the p factor predicted blood perfusion abnormalities in regions of the anterior cingulate cortex, a frontal region involved in effortful control. These findings are of particular developmental interest as these were relatively young samples and therefore an explanation couched in terms of the shared accumulated consequences of mental disorders or their treatment is less plausible.
Other studies have linked the p factor to disturbed connectivity in fronto-temporal (Alnæs et al., Reference Alnæs, Kaufmann, Doan, Córdova-Palomera, Wang, Bettella and Westlye2017) and fronto-parietal networks (Elliott, Romer, Knodt, & Hariri, Reference Elliott, Romer, Knodt and Hariri2018), delayed maturation of the default mode network (which includes medial-prefrontal, cingulate, and parietal regions), and structural alterations in fronto-thalamo-cerebellar networks (Moberget et al., Reference Moberget, Alnaes, Kaufmann, Doan, Cordova-Palomera, Norbom and Westlye2019; Romer et al., Reference Romer, Knodt, Houts, Brigidi, Moffitt, Caspi and Hariri2018, Reference Romer, Knodt, Sison, Ireland, Houts, Ramrakha and Hariri2019). Any neurobehavioral mechanism(s) mediating the association between neural function and psychopathology would need to be broad enough to account for the involvement of multiple networks involved in coordinating and controlling thought and behavior (Romer et al., Reference Romer, Elliott, Knodt, Sison, Ireland, Houts and Hariri2021).
One candidate is executive function, which encompasses the higher order processes for directing and regulating one's thoughts, emotions and behaviors according to one's goals and context (Diamond, Reference Diamond2013). Neuroimaging studies of executive functioning point to the same core problem across many disorders. A meta-analysis of almost 300 studies of executive functioning in 5,500 patients and 5,700 controls reported common impairments associated with the neuro-circuitry during cognitive control paradigms across diagnostic groups (McTeague et al., Reference McTeague, Huemer, Carreon, Jiang, Eickhoff and Etkin2017). This entailed the multiple demand network (Crittenden, Mitchell, & Duncan, Reference Crittenden, Mitchell and Duncan2016): the left PFC, the interior insula, the right ventro-lateral PFC, the right intra-parietal sulcus, and the mid-cingulate presupplementary motor area. To note, this comprehensive analysis also revealed anomalous activation associated with executive function (EF) in the anterior dorsal and anterior cingulate cluster, including the insula which, in previous large-scale meta-analyses, has been shown to be prone to gray matter loss across mental disorders irrespective of diagnoses (Goodkind et al., Reference Goodkind, Eickhoff, Oathes, Jiang, Chang, Jones-Hagata and Etkin2015). In a study involving 1,600 participants (Shanmugan et al., Reference Shanmugan, Wolf, Calkins, Moore, Ruparel, Hopson and Satterthwaite2016) who were evaluated using a structured clinical interview and who performed a working memory task with functional magnetic resonance imaging (fMRI), bifactor analysis of the 112-item interview revealed a general factor and four spectral level factors. Specific spectral-level factors (externalizing, internalizing, and thought disorders) did not relate to EF, but the p factor predicted both performance and executive system activation in the frontal pole, the anterior cingulate cortex, anterior insula, thalamus, and precuneus. There were also significant regional diminutions associated with the specific spectral-level factors (e.g., behavioral symptoms were associated with hypo-activation of the fronto-parietal cortex and cerebellum; psychosis-spectrum symptoms with hypo-activation of the left dorsolateral PFC). The results confirm that anomalous brain function areas concerned with the executive system are associated with transdiagnostic psychopathology but once these are identified, specific associations with particular groupings of disorders are likely to emerge.
In line with this argument, EF deficits have been demonstrated for almost all diagnoses, including the most serious psychotic disorders (Sellers, Wells, & Morrison, Reference Sellers, Wells and Morrison2018), as well as pre-psychotic conditions (Dickson et al., Reference Dickson, Cullen, Jones, Reichenberg, Roberts, Hodgins and Laurens2018) and bipolar disorder (Lima, Peckham, & Johnson, Reference Lima, Peckham and Johnson2018). EF deficit has been associated with autism (Demetriou et al., Reference Demetriou, Lampit, Quintana, Naismith, Song, Pye and Guastella2018), unipolar depression (Knight, Air, & Baune, Reference Knight, Air and Baune2018) including suicidal tendencies (Ho, Hsu, Lu, Gossop, & Chen, Reference Ho, Hsu, Lu, Gossop and Chen2018), attention-deficit/hyperactivity disorder (ADHD) (Kofler et al., Reference Kofler, Sarver, Harmon, Moltisanti, Aduen, Soto and Ferretti2018), conduct problems (Delfin, Andine, Hofvander, Billstedt, & Wallinius, Reference Delfin, Andine, Hofvander, Billstedt and Wallinius2018), anxiety (Unterrainer et al., Reference Unterrainer, Domschke, Rahm, Wiltink, Schulz, Pfeiffer and Beutel2018), binge eating (Kittel, Schmidt, & Hilbert, Reference Kittel, Schmidt and Hilbert2017), antisocial personality disorder (Dolan, Reference Dolan2012), and borderline personality disorder (Ernst et al., Reference Ernst, Mohr, Schott, Rickmeyer, Fischmann, Leuzinger-Bohleber and Grabhorn2018). Studies that test more than one clinical group commonly find EF deficits in all (Nigg et al., Reference Nigg, Jester, Stavro, Ip, Puttler and Zucker2017; Sedgwick et al., Reference Sedgwick, Young, Baumeister, Greer, Das and Kumari2017) and executive dysfunction correlates positively with both internalizing and externalizing dimensional scores (Snyder, Miyake, & Hankin, Reference Snyder, Miyake and Hankin2015).
How can we understand these consistent associations? The link may be to some general underlying dysfunction that all mental disorders share, such as the inability to control attention and direct it to goal-relevant information (Beauchaine & Thayer, Reference Beauchaine and Thayer2015; Cirino et al., Reference Cirino, Ahmed, Miciak, Taylor, Gerst and Barnes2018; Friedman & Miyake, Reference Friedman and Miyake2017). It is possible that the shared dysfunction is restricted to emotional challenges such as the capacity to shift attention away from threatening stimuli (Drabick, Ollendick, & Bubier, Reference Drabick, Ollendick and Bubier2010) or the control of behavioral impulses (McGrath et al., Reference McGrath, Braaten, Doty, Willoughby, Wilson, O'Donnell and Doyle2016). There is certainly a growing body of evidence that links EF deficits with the p factor (Bloemen et al., Reference Bloemen, Oldehinkel, Laceulle, Ormel, Rommelse and Hartman2018; Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel and Moffitt2014; Castellanos-Ryan et al., Reference Castellanos-Ryan, Briere, O'Leary-Barrett, Banaschewski, Bokde and Bromberg2016; Harden et al., Reference Harden, Engelhardt, Mann, Patterson, Grotzinger, Savicki and Tucker-Drob2020; Hatoum, Rhee, Corley, Hewitt, & Friedman, Reference Hatoum, Rhee, Corley, Hewitt and Friedman2018; Martel et al., Reference Martel, Pan, Hoffmann, Gadelha, do Rosario, Mari and Salum2017; Shields, Reardon, Brandes, & Tackett, Reference Shields, Reardon, Brandes and Tackett2019; Snyder, Friedman, & Hankin, Reference Snyder, Friedman and Hankin2019; White et al., Reference White, Moore, Calkins, Wolf, Satterthwaite, Leibenluft and Gur2017), and preliminary evidence showing that poorer EF mediates the link between childhood neglect (as indexed by institutionalization) and later p factor scores in adolescence (Wade, Fox, Zeanah, & Nelson, Reference Wade, Fox, Zeanah and Nelson2019). There is a limitation to this literature, however, in that demonstrations of EF deficit in different clinical groups often use different measures of EF pertaining to different EF domains (Hatoum et al., Reference Hatoum, Rhee, Corley, Hewitt and Friedman2018).
Taken together, structural neuro-imaging and behavioral evidence points to a specific, plausible common cause for mental disorder as a category. The implication here is not so much whether EF anomalies do or do not offer an adequate account of general psychopathology. The point of interest is that the variance shared by symptoms and syndromes can be meaningfully predicted by specific behavioral observations, which in turn connect with likely neuro-developmental anomalies. EF deficits may not be the cause of psychopathology but may be part of a general model of mental disorder that we are seeking to identify – such deficits may generate vulnerability, creating the potential for mental disorder in response to specific experiences/stressors.
In previous publications we have advanced the view that developmental limitations to mentalizing may be a shared by a wide range of mental disorders (Fonagy & Luyten, Reference Fonagy and Luyten2009; Luyten, Campbell, Allison, & Fonagy, Reference Luyten, Campbell, Allison and Fonagy2020). Can in fact the “p” factor itself be related to impairments in mentalizing (i.e., the capacity to understand ourselves and others in terms of inner mental states) and its consequences? Here we would like to consider the idea that in order to experience a sense of purposeful connection to their broader social community, the individual needs to feel that s/he is recognized as an agent and that s/he is being mentalized by their social system (Fonagy et al., Reference Fonagy, Twemlow, Vernberg, Nelson, Dill, Little and Sargent2009; Twemlow, Fonagy, & Sacco, Reference Twemlow, Fonagy, Sacco, Midgley and Vrouva2013). Mentalizing theory has always maintained that when our immediate social context, conceptualized dyadically, does not provide this kind of experience, we are vulnerable to the development of psychopathology (Fonagy & Luyten, Reference Fonagy and Luyten2009; Luyten et al., Reference Luyten, Campbell, Allison and Fonagy2020). In addition, we will explore the proposition that psychopathology and social alienation and inequality are linked in the individual's experience of a broader social context that fails to communicate the subjective richness of individual minds.
The Development of Culture and the Role of Culture in Human Development
In this section we will set out the context to our thinking about the evolution of social-cognitive and culture-building capacities that has inspired our social-communicative approach to developmental psychopathology. The notion that the roots of psychopathology and culture are shared has a distinguished history, not least in the ideas advanced by Sigmund Freud in Civilisation and its Discontents (Freud, Reference Freud and Strachey1930). Bronfenbrenner's ecological approach where culture is often considered as part of the exo- and macrosystem which has distal, contextual, or situational influence, has been particularly influential (Bronfenbrenner, Reference Bronfenbrenner1979). More recently, Causadias (Reference Causadias2013) pointed out that even when culture is considered in developmental contexts it is contemplated in terms of environmental influences on development, or in nondevelopmental terms, as an attribute of individuals in particular populations. In line with the aims of this paper, Causadias (Reference Causadias2013) proposes to conceptualize culture in developmental terms and examine these at the level of both the individual and social group. Causidias maps out four scientific concerns (Causadias, Reference Causadias2013; Causadias & Cicchetti, Reference Causadias and Cicchetti2018): (a) to explore at the level of an individual's development how may continuities and discontinuities in cultural processes impact on that person's trajectories of adaptation and maladaptation; (b) to consider how culture shapes development in general at a social (situational or contextual) and at an individual level (immediate and personal); (c) to investigate the interaction of culture with genetic, neurological, and temperamental influences; (d) to develop the measurement of cultural influence through direct assessment of culture, exploring surface- and deep-level diversity and social and individual-level processes, opening the way to improved cross-cultural translation of interventions and culturally valid assessment. Our approach fits into the second scientific concern described in this framework by proposing an evolutionarily informed developmental approach to the relationship between developmental experiences of social, epistemically valid communications, and an individual's capacity to navigate and participate in the complex form of social functioning that constitutes cultural activity.
We will begin by explaining what we mean by “culture”, and we will then consider thinking about the origins of human social cognition, most notably described by Tomasello (Reference Tomasello2014). Our working definition of culture is derived from Sperber, who encapsulated it in the phrase “culture is the precipitate of cognition and communication in a human population” (Sperber, Reference Sperber, Fraser and Gaskell1990, p. 42). Gergely and Csibra have described the “demand characteristics of cognitively opaque cultural forms” (Gergely & Csibra, Reference Gergely, Csibra, Enfield and Levinson2006, p. 8) as a central feature of human culture. From our standpoint this idea is also critical. Artefacts, whether in the form of abstract ideas or physical technology, that have the quality of opacity – that is, their purpose, function, rationale, or method of usage is not obvious or easily worked out without explanation from an instructor – require communication, that is, some form of education or teaching. To make the efficient learning of complex/opaque things, “humans evolve specialized cognitive resources that form a dedicated interpersonal system of mutual design in which one is predisposed to ‘teach’ and to ‘learn’ new and relevant cultural information to (and from) conspecifics” (Gergely & Csibra, Reference Gergely and Csibra2005, p. 472). This is the process of natural pedagogy, which we will describe in more detail below with reference to developmental psychopathology. According to Gergely and Csibra, culture is not only the outcome of communication, it is communication – but of a particular type and purpose, deriving from natural pedagogy (Gergely & Csibra, Reference Gergely, Csibra, Enfield and Levinson2006).
Culture, as understood here, is both the process and outcome of the communication of opaque knowledge. Culture therefore can be defined as a dynamic system of explicit and implicit rules, involving attitudes, values, beliefs, traditions, customs, norms, and behaviors – a system of knowledge – shared by a relatively large group of people and passed on from generation to generation. Culture is a system that enables the accumulation and transmission of a body of shared techniques and practices to optimize people's interactions with the world around them. The primatologist, linguist, and developmentalist Tomasello described the process of knowledge accumulation as “the ratchet effect” (Tomasello, Carpenter, Call, Behne, & Moll, Reference Tomasello, Carpenter, Call, Behne and Moll2005) – adopting new elements, which are superior to and enable us to dispense with prior solutions. This process allows the accumulation of knowledge within a community to be transmitted through interpersonal learning. Critically, social learning in this way is dependent on levels of “thinking together” – joint attention, intellectual collaboration, empathic entanglements in relation to understandings and misunderstandings – that involves interpersonal engagement and functioning at quite a sophisticated level (O'Madagain & Tomasello, Reference O'Madagain and Tomasello2019; Tomasello, Reference Tomasello2020).
Thinking together requires robust and, at moments highly reflective, mentalizing. Perhaps the most persuasive understanding about how mentalizing develops is an account that centers on collaboration as an essential species-specific attribute (Tomasello, Reference Tomasello2018). Human sociality is explained by the remarkable capacity we have to share the mental states of others. Recent research on mentalizing across species has shown surprisingly high levels of mental state understanding not just in infants, but more surprisingly in other nonhuman primates (Sandel, MacLean, & Hare, Reference Sandel, MacLean and Hare2011) and other mammals such as domestic dogs (Call, Brauer, Kaminski, & Tomasello, Reference Call, Brauer, Kaminski and Tomasello2003) and goats (Tomasello, Call, & Kaminski, Reference Tomasello, Call and Kaminski2006). This has prompted some researchers to look more closely at what is unambiguously unique about human mentalizing. It seems that great apes can imagine and track the mental states of others; they are able to track what the other sees and even anticipate fairly accurately how this information will affect the animal's behavior. They appear to have an understanding of mental states such as seeing, intending, and knowing (Karg, Schmelz, Call, & Tomasello, Reference Karg, Schmelz, Call and Tomasello2015).
Around 9 months of age, human infants, clearly capable of tracking the mental states of others (Kovacs, Teglas, & Endress, Reference Kovacs, Teglas and Endress2010), acquire a crucial additional competence: joint attention. This is the experience when the infant and the infant's partner understand themselves to be attending to the same thing at the same time but to be doing so from different perspectives (Tomasello, Reference Tomasello2018). Tomasello (O'Madagain & Tomasello, Reference O'Madagain and Tomasello2019; Tomasello, Reference Tomasello2019) points out that the key difference between humans and other species is this capacity to coordinate perspectives. We would suggest, along with Tomasello, that this joining of minds, in which a common object is identified at the same time as recognizing different perspectives on it, is the crucial ingredient of mentalizing in humans. Tomasello describes this as a “dual level structure” of shared intentionality (Tomasello, Reference Tomasello2020), because it encompasses both a shared focus and individual, separate perspectives upon the same thing (Tomasello, Reference Tomasello2016). Developmentally, as appreciation of objective reality develops, the infant and then child becomes able to triangulate the self (subjective) view, with the view of the other (the parent) concerning the same aspect of reality, and with both their own and their parent's views of the actual physical reality (Davidson, Reference Davidson2001). Tomasello argues that the move from epistemic tracking to shared intentionality may have been driven by the organizing function of social collaboration (Tomasello, Reference Tomasello2020). Epistemic tracking alone serves social competition well: simply knowing what the competitor wants is sufficient, and there is no great selective advantage to coordinating the competitor's mental state with the contents of one's subjectivity or with external objective reality. Cooperation, however, is immeasurably advanced by being able to compare and coordinate different perspectives on the same situation (Colle et al., Reference Colle, Dimaggio, Carcione, Giuseppe, Semerari and Chiavarino2020).
Collaboration is supported by a special set of mental processes reserved for shared cognition – described as “we-mode”, or relational mentalizing. A line of thinking has emerged depicting social cognition in the relational, or “we”, stance (Higgins, Reference Higgins2020). It is suggested that each individual who intends to accomplish some outcome together with another is requited to adopt a “first person plural perspective”, the “we-mode” (Gallotti & Frith, Reference Gallotti and Frith2013, p. 160). The we-mode may be organized around cognitive and neural structures that are intrinsic to our individual make-up and are the product of a distinct developmental and evolutionary history (Tomasello, Reference Tomasello2019). Building on joint intentionality, the joint agent emerges where mental states are aligned to achieve a common goal, grounded in respect born of each having a role in the collaborative activity (Tomasello, Reference Tomasello2016). The we-mode assumes a mutual recognition of the subjectivity and humanity of the other: “recognition of the other persons as agents or persons just as real as oneself…the recognition of an inescapable fact that characterizes the human condition” (Tomasello, Reference Tomasello2016, p. 56).
Acting jointly and collaboratively requires a distinct set of processes. The we-mode is jointly held and shared by other(s), and can be distinguished from the I-mode where the individual's beliefs, feelings, and wishes are self-contained and in an ontologically separate category. In the I-mode, applying the intentional stance to others serves the self in that others are used instrumentally as social obstacles or social instruments (Tomasello, Reference Tomasello2019). From a systemic psychotherapeutic clinical perspective we have discussed a similar set of processes under the heading of relational mentalizing (Asen & Fonagy, Reference Asen and Fonagy2012, Reference Asen and Fonagy2017; Bateman & Fonagy, Reference Bateman and Fonagy2016), referring to the shared thinking and feeling within a social system, a dyad, a family, or other social group.
The we-mode, or relational mentalizing, concerns intentional states that are assumed by individuals in the system to be joint or shared by everyone. It is assumed that in the “we-mode” (Gallotti & Frith, Reference Gallotti and Frith2013), the social context (the mere presence of others) improves a person's potential for mentalizing by broadening awareness of the options available for action and generating new solutions for action. This involves co-representing the other's viewpoint, a precondition for acting jointly. When people (families or any other collection of individuals) decide to be and act together, to join forces, there is a sense in which no member of the group can be assumed to be doing it “on their own” or can be appropriately considered as thinking or feeling in isolation from others in that “psychological collective”. This sense of “we-ness” of shared minds has an irreducibility which means that it must be addressed separately from individual mentalizing of self and others as joint actions are experienced in a qualitatively different way and involve shared or “we-intentions”. This is relational mentalizing, which relies on underlying mutually accepted yet often implicit conceptual and situational presuppositions and does not necessarily involve agreement making (Tuomela, Reference Tuomela2005).
This readiness to act together, to share an action and jointly execute a plan was probably the capacity that has enabled us to conquer, if that is an appropriate term in the light of the dubious consequences of our undoubted success, all other species on the planet. The feeling associated with this joint intentionality, the feeling of We-ness, may be underpinned by and certainly generates the potential for social collaboration. This sharing of minds in an irreducibly collective mode of cognition has been recognized by many including developmentalists (e.g., Tronick, Reference Tronick2008), by psychoanalysts of most classical schools (e.g., Winnicott, Reference Winnicott and Winnicott1956) and increasingly by neuroscientists (e.g., Gallotti & Frith, Reference Gallotti and Frith2013).
Psychopathology as a By-product of our Capacity for Culture
How does the we-mode relate to psychopathology? This thinking begins with Csibra and Gergely's theory of natural pedagogy (Csibra & Gergely, Reference Csibra and Gergely2009), which posits a human-specific, cue-driven social-cognitive adaptation of mutual design dedicated to ensuring the most effective and efficient transfer of culturally relevant knowledge. They argue that human communication is the evolutionary product of the requirement to transmit cognitively opaque cultural knowledge: knowledge that is robust to interference, is generalizable, and becomes experienced as shared in the sense that it immediately generates an expectation that others belonging to one's social group also possess this knowledge.
Csibra and Gergely take the concept of “ostensive cues” (Csibra & Gergely, Reference Csibra and Gergely2009; Csibra & Gergely, Reference Csibra and Gergely2011) – discussed originally by Bertrand Russell (Reference Russell1940/1967), but extensively used by Sperber and Wilson (Reference Sperber and Wilson1995) – to mean that certain signals are employed by an agent and prepare the addressee for the intent of the agent to communicate. Examples of ostensive cues are eye contact, eyebrow raising, contingent reactivity, and infant-direct speech (“motherese”). Ostensive cues may also serve the function of counteracting natural “epistemic vigilance” – the self-protective suspicion towards potentially damaging, deceptive, or inaccurate information (Sperber et al., Reference Sperber, Clement, Heintz, Mascaro, Mercier, Origgi and Wilson2010). Ostensive cues generate a particular attentional state, which we link to the we-mode discussed above, where epistemic vigilance is momentarily suspended and the addressee feels that the subsequent communication contains information specifically relevant to them which should be remembered and encoded as knowledge relevant generally to social situations involving the self with others (Butler, Gibbs, & Levush, Reference Butler, Gibbs and Levush2020; Gergely, Reference Gergely, Metcalfe and Terrace2013). Such information can be about an object, or about the other's views and attitudes about the object or the beliefs communicated by the other about the self that can be regarded as generalizable and relevant across situations (Egyed, Király, & Gergely, Reference Egyed, Király and Gergely2013; Futo, Teglas, Csibra, & Gergely, Reference Futo, Teglas, Csibra and Gergely2010). The information can be laid down and used as part of procedural and semantic memory, not uniquely or primarily episodic memory (Kovacs, Teglas, Gergely, & Csibra, Reference Kovacs, Teglas, Gergely and Csibra2017).
We have found the notion of ostensive cues helpful as it links directly to our understanding of the we-mode of social cognition. To learn about reality, we have to turn to others. The human world, outside and inside, is so complex that it is impossible to discover it all by ourselves. We also learn about ourselves from others. We build and constantly unconsciously update representations of ourselves through our social interactions imbued with learning opportunities (Gergely & Jacob, Reference Gergely, Jacob, Benson, Xu and Kushnir2012; Gopnik & Wellman, Reference Gopnik and Wellman2012). The knowledge we acquire about who we are does not emerge from within but is absorbed from others, kept current, updated in response to changing social contexts, enabling social adaptation crucial to human survival. Throughout development, the key to keeping the learning channel open is the experience of self-recognition that ideally precedes genuine learning from the object. That experience is based on detecting how one is seen. By its very nature, this assessment of how one's mind is being seen by another must be highly contingent on the feedback one is receiving from the other. This allows for highly responsive and adaptive social learning about the particular nature of the social environment, and as such it generates the particular flexibility of human social cognition. However, we suggest that it also creates a potential vulnerability in terms of psychopathology.
The theory of natural pedagogy underlines the highly interpersonal nature of the process by means of which epistemic trust – trust in communicated information – is generated in infancy, and this has significant implications for the field of developmental psychopathology. Associative learning assessed early in development appears predictive of later social functioning (Reeb-Sutherland, Levitt, & Fox, Reference Reeb-Sutherland, Levitt and Fox2012) and learning strategies emerging early in infancy may come to significantly influence behavior in complex social situations (Hammock & Levitt, Reference Hammock and Levitt2006). Caregiving experiences affect how someone extracts and processes information from the environment, and learning capacity is influenced by an individual's caregiving history. This has been classically shown for macaques (Capitanio, Reference Capitanio1985; Mason & Capitanio, Reference Mason and Capitanio1988) and also for human infants (Bigelow & DeCoste, Reference Bigelow and DeCoste2003). Less contingently responsive caregiving is associated with slower rates of associative learning in infants (Papousek & Papousek, Reference Papousek and Papousek1975). We suggest that if the caregivers around the infant are not reliably responsive, not benign, and/or not able to recognize what is meaningful and relevant to the infant's self, this can undermine the capacity to learn through the underdevelopment of epistemic trust. There is evidence that the quality of the relationship of a child to a communicator determines in large measure the extent to which the child will acquire information from that communicator and generalize this (Corriveau et al., Reference Corriveau, Harris, Meins, Fernyhough, Arnott, Elliott and de Rosnay2009; Lane & Harris, Reference Lane and Harris2015; Mascaro & Sperber, Reference Mascaro and Sperber2009; Shafto, Eaves, Navarro, & Perfors, Reference Shafto, Eaves, Navarro and Perfors2012). This emphasis on the significance of reputation within interpersonal processes is in keeping with recent work exploring the inherently socially driven nature of higher order cognitive processes (Rudrauf, Reference Rudrauf2014).
Our hypothetical model for social learning moderated by epistemic trust would be assumed to be contingent on the establishment of the we-mode. The prototypical path by which the we-mode influences the learning situation might be: (a) the learner's imagined sense of self (their personal narrative) (b) is imagined by the instructor, establishing a prospect for the we-mode; (c) this image is perceived, reinforcing the potential we-mode, and (d) compared with the learner's personal narrative; (e) in the case of a match, the co-representation has been created and the we-mode removes the I-mode's protection from change, and the channel for rapid, efficient knowledge transfer is opened. Relational mentalizing is thus key to establishing epistemic trust. The communicator needs to be able to mentalize the addressee well enough for the addressee to feel accurately mentalized. This imaginative experience (we-mode) unlocks the barrier epistemic of vigilance. Throughout development the key to keeping the learning channel open is generating experiences of recognition that enables genuine learning from the communicator. That experience is based on detecting how one is seen – what we term the epistemic match. Putting it at its pragmatic simplest: if I feel that I am understood, I will be disposed to learn from the person who understood me, who I feel is a trustworthy potential collaborator. This will include learning about myself but also learning about others and about the world I live in.
Our developmental, interpersonal view on the stimulation of epistemic trust in the context of early relationships creates a new role for the attachment relationship. We propose that human social-cognitive developmental processes have hijacked the attachment relationship to use it as one, though not the only, major mediator of epistemic trust. One of the great social advantages bestowed by secure attachment, we suggest, is that it promotes the individual's capacity for social learning because it facilitates a general capacity for epistemic trust (Luyten et al., Reference Luyten, Campbell, Allison and Fonagy2020) and may account for some of the benefit which secure attachment brings in terms of mental health benefit (Groh, Fearon, van IJzendoorn, Bakermans-Kranenburg, & Roisman, Reference Groh, Fearon, van IJzendoorn, Bakermans-Kranenburg and Roisman2017). Research evidence supports the developmental nature of trust and the social factors that influence it (Corriveau et al., Reference Corriveau, Harris, Meins, Fernyhough, Arnott, Elliott and de Rosnay2009; Eaves & Shafto, Reference Eaves and Shafto2017; Markson & Luo, Reference Markson and Luo2020; Tong, Wang, & Danovitch, Reference Tong, Wang and Danovitch2020).
The attachment relationship constitutes a powerful source of social information about the environment and the extent to which a strong orientation towards other people's mental states may be an appropriate strategy (Fonagy & Campbell, Reference Fonagy and Campbell2017a). The great gift of a secure attachment relationship when considered in relation to groups not just individuals (Hrdy, Reference Hrdy, Carter, Ahnert, Grossmann, Hrdy, Lamb, Porges and Sachser2006, Reference Hrdy2009, Reference Hrdy2013; Mesman et al., Reference Mesman, van IJzendoorn, Behrens, Carbonell, Carcamo, Cohen-Paraira and Zreik2016b; Mesman et al., Reference Mesman, Minter, Angnged, Cisse, Salali and Migliano2018; Mesman, Minter, & Angnged, Reference Mesman, Minter and Angnged2016a) is that it enables the child to orient themselves to opportunities for cultural learning from their environment (Fonagy, Luyten, Allison, & Campbell, Reference Fonagy, Luyten, Allison and Campbell2017a, Reference Fonagy, Luyten, Allison and Campbell2017b). This idea is embodied in Tomasello's (Reference Tomasello2016) proposal of collective intentionality, a form of shared intentionality where the child comes to coordinate their actions with an increasingly large group that ultimately becomes our culture. Ultimately, this enables us to communicate with, learn from, and collaborate with non-kin who are around us because we share with them a sense of identity obtained through common practices, beliefs, attitudes, and a sense of belonging or identity.
This is not to negate the importance of dyadic approaches to attachment. Caregiver sensitivity is likely to teach the child to be sensitive to ostensive cues that trigger the joint intentionality of the we-mode and may be a part of the explanation of the educational advantages which secure attachment brings (Belsky & Fearon, Reference Belsky and Fearon2002; Stievenart, Roskam, Meunier, & Van de Moortele, Reference Stievenart, Roskam, Meunier and Van de Moortele2011; van IJzendoorn, Dijkstra, & Bus, Reference van IJzendoorn, Dijkstra and Bus1995). We would argue that sensitivity serves to instill “basic trust” that a caregiver can function with synchronous behavior to the infant and reliably establish shared intentionality by a variety of means involving creating (for the child) a perceptible representation of their self-narrative which generates a co-representation of sufficient clarity to engender the epistemic trust necessary to explore, establish joint intentionality, and learn from the caregiver and from others, and re-establish the safe haven of the we-mode when returning in need of comfort or support. Of course, the we-mode ensures that the infant learns about themselves (their emotions, their thinking and their identity) from the safe instruction of the caregiver. Thus sensitivity fostering secure attachment also fosters the capacity for social learning and adaptations across cultures. Recently, Mesman (Reference Mesman2020) and colleagues demonstrated cross-cultural relevance of sensitivity showing that it could be reliably coded on the basis on interactions in seven different cultures.
Dyadic sensitivity generating a predisposition to epistemic trust is unlikely to be an exclusive strategy by which this social orientation can be achieved. Many non-Western communities have simultaneous multiple caregiving without clear place-bound or time-bound task division (Hrdy, Reference Hrdy2009). Anthropological studies consistently describe the wide prevalence of collectively distributed alloparenting – infants receiving person-directed care from a range of caregivers – engendering multiple natural attachments (Hrdy, Reference Hrdy, Meehan and Crittenden2016; Marlowe, Reference Marlowe, Lamb and Hewlett2005; Meehan & Hawks, Reference Meehan, Hawks, Otto and Keller2015). Looking for a predefined characterization such as sensitivity of dyadic interactions in cultural context is known as the etic approach to studying behavior and is contrasted with the emic approach that takes each culture as its own starting point (Harris, Reference Harris1976). Taking the emic approach, sensitivity has a different timbre in non-Western cultures. It is not that environments do not differ in terms of the amount of security they provide, but this is a function of the social network which is foundational to the child's development (Meehan, Helfrecht, & Malcom, Reference Meehan, Helfrecht, Malcom, Meehan and Crittenden2016). Even when there is proximal caregiving for the infant from a carer, the child is metaphorically “facing outward”, their priority is to learn to see the world as others see it. Caregivers orient children literally outward by placing them in the same direction in which they are facing (Ochs & Izquierdo, Reference Ochs and Izquierdo2009); they lead them in activity and children follow their lead (Keller, Kartner, Borke, Yovsi, & Kleis, Reference Keller, Kartner, Borke, Yovsi and Kleis2005). Thus the carer's priority is not to demonstrate knowledge about and teach the infant about the self but rather to support them to take the perspective of others. Non-Western adult–child play and talk tend to be less common and when they do take place, they usually do not follow the typical interactional “script” of Western lifestyle communities. Face-to-face serve-and-respond interaction may be a somewhat anomalous form of ostensive cueing in human evolutionary experience (Keller & Bard, Reference Keller and Bard2017; Keller & Chaudhary, Reference Keller, Chaudhary, Keller and Bard2017). There are many ways that a child can have their agency recognized that do not follow the Western model of parental interaction. Care may be less overtly psychological or even emotion focused and far more based on physical contact (Morelli et al., Reference Morelli, Chaudhary, Gottlieb, Keller, Murray, Quinn, Vicedo, Keller and Bard2017). The caregiver speaks to the infant not about them but rather to instruct, guide, and direct them to appreciate understanding the actions of others. The infant is a “social apprentice” who only learns secondarily about self by identifying the ways their experience is similar to those of others in the community. While this care might not appear to fall in with some of the one-on-one interactive sensitive caregiving, and what is critical for security and trust may partially be culture specific (Keller & Chaudhary, Reference Keller, Chaudhary, Keller and Bard2017), what it does provide is a powerful piece of interpersonal communication about the capacity of the environment to accommodate and be oriented to one's presence (Fonagy & Campbell, Reference Fonagy and Campbell2017b). The nature of that communication creates moments of recognition and a sense of joint intentionality albeit these are experienced in the context of the social network and privilege behaviors that are most likely to ensure security in that community.
One-to-one sensitive responsiveness from a primary caregiver is one route to establishing we-mode and epistemic trust, but there are other routes. The we-mode can be as readily created around a joint intention to explore others, jointly seeing to it to see how others feel. Epistemic trust might in many cultures be affectively isomorphic with attachment (particularly in normative Westernized cultures), but epistemic trust and attachment may also be two separate developmental processes. Dyadic sensitivity may be a somewhat anomalous form of ostensive cueing of the we-mode in human evolutionary experience (Keller et al., Reference Keller, Bard, Morelli, Chaudhary, Vicedo, Rosabal-Coto and Gottlieb2018). There are many ways that a child can have their interest and agency recognized that do not follow Ainsworth's model of infant–parent interaction. For example, the young child's experiences of having their physical needs met, discomforts resolved, and their desire to explore tolerated, may not involve any conventional sensitive interaction. It may not always involve the same caregiver, can engage a broader social group with an investment in the child, it may not involve conversation, or even eye contact, there may be little overt acknowledgement or much of a pause while pursuing other tasks – but such interactions can still generate a powerful experience of having one's “knowledge” (what I perceive myself to need) affirmed and shown to be aligned with what the social environment can provide and tolerate (the experience of having those needs met) (Keller et al., Reference Keller, Bard, Morelli, Chaudhary, Vicedo, Rosabal-Coto and Gottlieb2018; Keller & Chaudhary, Reference Keller, Chaudhary, Keller and Bard2017). Commentary on psychological states (mind-mindedness) is one way caregivers may align intentional stances, supporting early motor independence may be another. It is engagement in collaboration and shared intentionality that underpins the we-mode and it is epistemic trust that ensures that social learning can take place as the I-mode is replaced by the we-mode.
In focusing on social learning we see vulnerable or sensitive developmental periods stretching beyond the early years into later childhood and adolescence. Critically, these are the times when children and young people, through school and ever-widening exposure to the outside world, have their perceptions of the world as safe and reliable, or dangerous and unpredictable, reinforced. Whilst early attachment increases the chances of a positive outcome (Simpson, Collins, Tran, & Haydon, Reference Simpson, Collins, Tran and Haydon2007), peer, not parental, support and acceptance turns out to be the best predictor of resilience in adolescence (van Harmelen et al., Reference van Harmelen, Kievit, Ioannidis, Neufeld, Jones, Bullmore and Goodyer2017). Yet the mechanisms we identify in the context of studying early development may be of great help in understanding later influences. Shared intentionality will be or will fail to be established with a range of agents. Further, epistemic trust can be understood, on a heuristic level at least, as a generalized trust in one's social community: shared intentionality may be experienced by the individual at a group as well as on an individual level, and is related to the expectation that the social environment will protect, care for, and help realize aims and ambitions.
It is clear that the absence of epistemic trust would deeply disadvantage an individual in many social contexts. The loss of this key process for the efficient acquisition of self-knowledge as part of cultural knowledge has significant implications for social functioning. It is here that we draw the connection between our culture-making capacity and vulnerability to what we typically understand as psychopathology. The individual may become limited in their ability to update their understanding of potentially rapidly changing social situations and would appear inflexible or even rigid in the face of social change. Why would an individual fail to experience epistemic trust even in situations where trust was warranted – that is, where their personal narrative was appreciated? There are two obvious reasons. First, adversity and deprivation, when tantamount to trauma, can generate chronic mistrust by inhibiting imagination in relation to mental content, creating an overarching avoidance of thinking about either one's own or other people's mental states, leaving the individual deeply vulnerable in most social situations (Ensink et al., Reference Ensink, Normandin, Target, Fonagy, Sabourin and Berthelot2015; Levy, Goldstein, & Feldman, Reference Levy, Goldstein and Feldman2019; Macintosh, Reference Macintosh2013; Taubner & Curth, Reference Taubner and Curth2013). Even in the absence of such a pervasive failure of imagination, a compromised capacity to mentalize – that is, to make sense of both other people's behavior in terms of underlying mental states – may lead the traumatized individual to be biased in their perception of social reality (Cicchetti & Curtis, Reference Cicchetti and Curtis2005; Germine, Dunn, McLaughlin, & Smoller, Reference Germine, Dunn, McLaughlin and Smoller2015; Kay & Green, Reference Kay and Green2016) and misrepresent how others represent them, leading them to feel persistently misunderstood and to experience an intense and consistent sense of injustice. Second, the long-term outcome of epistemic isolation secondary to the failure of imagination we describe here may create problems for individuals who have distorted personal narratives that generate inaccurate views of the self, so that even an accurate perception of one's personal narrative by others is not experienced as a match, and a painful experience of interpersonal alienation persists. For example, an individual whose personal narrative entails a sense of failure or badness will not paradoxically benefit from a supportive, positive reflection of their self-narrative, and they are left with a sense of not having been understood. Conversely, in yet other instances, deprivation and trauma may generate inappropriate trust. We understand such excessive epistemic credulity as triggered by a hyperactive or unmoored social imagination generating a personal narrative that is too diffuse to provide an accurate sense of differential awareness of others’ capacity to perceive oneself. Excessive credulity results as all personal narratives feel as if they “fit” sufficiently for trust to be generated, making the person vulnerable to exploitation. Of course, limited imagination may cause profound misperceptions of the other's representations of one's personal narrative, and an illusory fit is created where none in reality exists. There may be many other possibilities, but they share the quality of disrupting the individual's capacity to experience the we-mode based on the triangulation of self, other, and external reality.
If epidemiological figures are to be relied on, only one in five people will go through life without experiencing a diagnosable mental health condition (Schaefer et al., Reference Schaefer, Caspi, Belsky, Harrington, Houts, Horwood and Moffitt2017): perhaps it is mental health rather than mental illness that is something of a myth (Szasz, Reference Szasz1960). Looking at such prevalence figures from the perspective of natural selection, it is clear that whatever the neural systems are that underpin mental disorder, they must have other functions that are critical for survival. A defining feature of mental disorder is the experience of “wild imagination”. Human imagination is essential for mentalizing, and hence for the transmission of culture. We need to be able to imagine others’ perspectives to collaborate effectively in a social world, which seems to be a small price to pay for the errors – or wildness – of imagination that can ensue following adversity or trauma.
Carver, Johnson, & Timpano, (Reference Carver, Johnson and Timpano2017) have helpfully drawn attention to the contrast between reflective and reflexive mode of cognition suggested by Epstein (Reference Epstein1994) to delineate a dichotomy between a basic, responsive to emotion mode (reflexive) and deliberative mental function (reflective). A claimed shared characteristic of the reflexive mode is relative spontaneity, simplicity, and responsiveness to affect (Metcalfe & Mischel, Reference Metcalfe and Mischel1999; Strack & Deutsch, Reference Strack and Deutsch2004). There is inherent uncertainty to social communication because of the ultimate opaqueness of other minds. Thus a reflexive mode of functioning may be better adapted to the task of establishing joint intentionality especially in human environments dominated by uncertainty. Yet if emotions are responded to spontaneously, without constraint, the individual may be at risk or vulnerable to mental disorder (Carver et al., Reference Carver, Johnson and Timpano2017). The idea that different modes of cognition might be triggered by the environment, and how this might interact with an individual's stance on social learning will be explored further in the final section of this paper to consider the risk factors for psychopathology arising from socioeconomic circumstances.
Reintegrating the Social Environment with the Key Questions of Psychopathology
How does the social learning/epistemic trust model help us understand the profound influence of the social environment – in particular economic deprivation or exposure to discrimination, social isolation, or a hostile community – on child development and the emergence of mental health problems? We assume that such social environments can signal to the child that they are not facing a reliable, safe environment in which it is adaptive to be trusting and open to social learning (Luyten et al., Reference Luyten, Campbell, Allison and Fonagy2020). There is now good evidence supporting these assumptions.
The Five Year Forward View for Mental Health recognized that children from low-income families are at three times higher risk of mental health problems than those from the highest income families (NHS England, 2014), and the recent survey of the prevalence of diagnosable mental disorder in children and young people in England found that mental disorders occurred more commonly in children living in low-income households (NHS Digital, 2018). In this survey, other factors that increased the likelihood of diagnosable disorder included having parents who showed signs of common mental health problems and living in families with less healthy family functioning. Maternal depression and couple conflict are also associated with economic pressure (Kavanaugh, Neppl, & Melby, Reference Kavanaugh, Neppl and Melby2018) and the impact of these factors on young people's mental health has been conceptualized in terms of the Family Stress Model: economic hardship puts economic pressure on families, leading to parental psychological distress which results in both interparental relationship problems and disrupted parenting, ultimately leading to adjustment problems in children and young people (Masarik & Conger, Reference Masarik and Conger2017). Help with these difficulties is likely to be less available to lower income families and lower socioeconomic status (SES) has been shown to be associated with poorer care for mental health problems generally (Jokela, Batty, Vahtera, Elovainio, & Kivimaki, Reference Jokela, Batty, Vahtera, Elovainio and Kivimaki2013).
Other suggested mediators of the impact of low SES on mental health outcomes for children and young people include stressful life events, literacy difficulties, and the experience of living in a disadvantaged neighborhood (Piotrowska, Stride, Maughan, & Rowe, Reference Piotrowska, Stride, Maughan and Rowe2019). It can be difficult to disentangle family and neighborhood level effects of disadvantage, although it has been suggested that as children's autonomy and independence from their caregivers increase, they are likely to be more exposed to potential direct neighborhood effects on development, which have been hypothesized to include increased likelihood of interaction with antisocial peers, lack of access to social facilities, tensions between majority and minority residents, relative lack of mechanisms of social control, and low levels of cohesion, support, and collective efficacy (Ingoldsby et al., Reference Ingoldsby, Shaw, Winslow, Schonberg, Gilliom and Criss2006) as well as awareness of income differentials (Wilkinson & Pickett, Reference Wilkinson and Pickett2009).
However, it is important not to make simplistic assumptions about the quality of relationships implied by socioeconomic disadvantage. It has also been shown that high-SES individuals feel more powerful and manifest cognitive and behavioral tendencies with a greater focus on self (Inesi, Botti, Dubois, Rucker, & Galinsky, Reference Inesi, Botti, Dubois, Rucker and Galinsky2011). By contrast, individuals with low-SES feel relatively less powerful and manifest tendencies with a greater focus on others and the community they are in. In other words high-SES individuals have a focus on the I-mode driven by personal agency, while low-power, low-SES individuals are more likely to have a communal focus and prioritize we-processes (Rucker & Galinsky, Reference Rucker and Galinsky2017). Individuals who are less socioeconomically privileged tend to behave in more community and socially oriented ways in interpersonal trust experiments and tend to spend more money proportionately on social activities and charity than more affluent individuals (Dubois, Rucker, & Galinsky, Reference Dubois, Rucker and Galinsky2015). Less affluent individuals are more dependent on their community; wealthier and more socially protected individuals have a stronger perception of their self-agency and do not need to be so community focused (Overbeck & Droutman, Reference Overbeck and Droutman2013). As a result, individuals functioning in a lower SES environment are also likely to be more sensitive to their social environment and its reliability and how benign or supportive it may be (Gruenfeld, Inesi, Magee, & Galinsky, Reference Gruenfeld, Inesi, Magee and Galinsky2008; Liu et al., Reference Liu, Mustanski, Dick, Bolland and Kertes2017). This greater orientation towards the community places individuals who lack power at greater risk when they encounter hostile social contexts and explicit adversity. Thus, whether the impact of deprivation is felt via the reduction of social trust or if the lack of individual power of members in the high deprivation social groups places them in positions of greater vulnerability when social networks break down, the prediction would be for a linear relationship between socioeconomic disadvantage and poor mental health outcomes.
When children live in such social contexts, high-risk life strategies (Del Giudice, Reference Del Giudice2016), as generated by imaginative solutions that may bear little on objective reality, could be more advantageous than solutions arrived at through more exhaustive logical inference. Along with others, we have suggested that the quality of family relationships in an attachment context in early life may provide the child with information about the safety or predictability of their environment (Chisholm, Reference Chisholm1999). Creative solutions inevitably entail high risk and are most likely to be adaptive in the context of unpredictable environments (Frankenhuis & Del Giudice, Reference Frankenhuis and Del Giudice2012). This is simply a reflection of how natural selection shapes developmental strategies that in turn give rise to phenotypes adapted to a local ecology (Panchanathan, Frankenhuis, & Barrett, Reference Panchanathan, Frankenhuis and Barrett2010). A possible general model for psychopathology is provided by the notion of a developmental mismatch (Gluckman, Low, Buklijas, Hanson, & Beedle, Reference Gluckman, Low, Buklijas, Hanson and Beedle2011). A change of environment might create problems of a developmental mismatch. Thus, the prioritization of reflexive cognition indicated by hostile early environmental conditions might actually generate maladaptive responses under more favorable conditions.
Why is there more mental disorder in low-SES communities? It is not at all improbable that if dependence on the community is greater in less powerful individuals, then losing that support through the breakup or fragmentations of the social fabric will impact more on an low than high-SES individuals. Failure of epistemic trust will affect low-SES people more than high-SES people because low-SES individuals are more dependent on their community. If resilience is maintaining We-ness, joint intentionality, the openness to social learning, and epistemic trust embedded into the links with the community, then the individual who is more dependent on those links because of their relative lack of power will be more likely to succumb to mental disorder than the individual whose position of power enables them to act as independent agents.
A further characteristic of uncertain and adverse environments is that they trigger reflexive thinking, and with it spontaneity and imagination both to find imaginative solutions and to optimize the chance of establishing joint intentionality with individuals who could be a source of support through joint action. In adversity, imaginative solutions that may bear little on objective reality could be more advantageous than solutions arrived at through more exhaustive logical inference (Frankenhuis & Del Giudice, Reference Frankenhuis and Del Giudice2012). This has been suggested as the evolutionary basis of the risky strategies commonly generated by adolescents characteristic of young people under environmental pressure (Del Giudice, Reference Del Giudice2016). Congruent with this, our work has also demonstrated that maladaptive solutions at this developmental period are characterized by an unconstrained excess of mentalizing that Carla Sharp and colleagues have termed hypermentalizing (Sharp & Vanwoerden, Reference Sharp and Vanwoerden2015). Thus, the prioritization of reflexive cognition indicated by hostile early environmental conditions might generate adaptive responses under favorable conditions, but could also increase the risk of profoundly maladaptive developments. If calling forth imagination is a double-edged sword, bringing a risk of imagination unconstrained by objectivity, perhaps it is unsurprising that children are most at risk of mental disorder when they grow up in adverse environments in the absence of social support and the potential of binding unconstrained imagination within an agreed social reality. We speculate that the epistemic trust that comes with the experience of being mentalized may also act to regulate imaginative, mentalizing activity: the we-mode, involving the triangulation of different perspectives and external reality, functionally moors these social cognitive processes. Perhaps, the benefit of social support may rest in the assurance that if a young person feels they have found joined intentionality with concerned others, they do not need to expend excessive efforts to understand themselves, and the experience of trust can provide a secure platform that enables them to turn outwards to others in their social world.
Funding Statement
PF is in part supported by the NIHR Applied Research Collaboration (ARC) North Thames at Barts Health NHS Trust. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
Conflicts of Interests
PF and PL are involved in the development, training, and dissemination of mentalization-based treatments. While PF derives no personal gain from these activities, he is chief executive of Anna Freud National Centre for Children and Families, a charity that financially benefits from trainings in this modality. Apart from these, the authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.
