1.1 Reproducible research

Board members’ views underscore the major point, as articulated by one of them: “As any neuroscientist knows, no two brains are identical”. However, despite the major impact of neuroscience on psychological research and large-scale funding of neuroimaging research over the previous 25 years, systematic differences between individuals have generally been neglected. The days are now gone when small sample sizes (stemming from cost restrictions) made this neglect almost unavoidable. Yet, the nagging doubt remains: do reproducible studies in personality neuroscience require sample sizes running into the hundreds or thousands (Marek et al., Reference Marek, Tervo-Clemmens, Calabro, Montez, Kay and Dosenbach2022; Spisak, Bingel, & Wager, Reference Spisak, Bingel and Wager2022)? The Marek et al. (Reference Marek, Tervo-Clemmens, Calabro, Montez, Kay and Dosenbach2022) paper, arguing for thousands, garnered considerable attention and comment. The concern of these authors related specifically to reproducible “brain-wide association studies” (BWAS) – defined as “studies of the associations between common inter-individual variability in human brain structure/function and cognition or psychiatric symptomatology” (p. 654). Reproducible BWAS may require large sample sizes, and luckily these are becoming increasingly attainable with resources like the UK Biobank, Human Connectome Project, and the Adolescent Brain Cognitive Development Study. However, such large sample sizes are not invariably required for neuroimaging studies of individual differences. Indeed, the findings of Marek et al. (Reference Marek, Tervo-Clemmens, Calabro, Montez, Kay and Dosenbach2022) support certain study designs that require much smaller sample sizes – a theme taken up by those who responded to their paper.

Encouragingly, there is evidence that smaller sample sizes can achieve robust results, as discussed by Rosenberg and Finn (Reference Rosenberg and Finn2022). They correctly note that Marek et al. do not undermine MRI research as a whole. As they state (p. 835): “some of the most well-replicated findings in human neuroscience come from studies that used carefully designed task paradigms to measure well-characterized cognitive process in a small number of individuals” (e.g., face recognition). Sensibly, Rosenberg and Finn (2002) advise testing the generalizability of models that predict variables from brain features; and the building of more robust brain-based predictive models. (See also Editorial in Nature Neuroscience, 2022.) At this time, though, it would be fair to say that these issues remain subject to debate (DeYoung et al., Reference DeYoung, Sassenberg, Abend, Beaty, Allen and Wacker2022b) – inevitable in a field still low in years and high in ambition.

We are now at a stage where, as a result of the availability of more sophisticated forms of technology, as well as larger sample sizes afforded by unit cost reductions and more investment from funders, neuroscientists are increasingly coming to recognise that they need not – indeed, should not – ignore systematic individual differences in their data. These developments hold crucial implications for such areas as functional connectomics, large-scale neurogenomics, and studies of the molecular basis of gene–environment interactions.

1.2 Published papers

Looking back on the papers published since 2018 supports our contention that personality neuroscience is not only a vibrant but also highly diverse field, with the potential to unify otherwise separate traditions. For example, a special issue, edited by Robert Latzman, Robert Krueger, Colin DeYoung and Giorgia Michelini (Reference Latzman, Krueger, DeYoung and Michelini2021), tackled the important topic of connecting quantitatively-derived personality–psychopathology models and neuroscience. Other papers published since its inception attest to the diversity of the field, covering such areas: narcissism (Jauk & Kanske, Reference Jauk and Kanske2021); machine learning approaches for parsing comorbidity and heterogeneity in antisociality and substance use disorders (Shane & Denomme, 2001); ketamine and neuroticism (McNaughton & Glue, Reference McNaughton and Glue2020); anxiety and mindfulness (Jaiswal, Mugglen, Juan, & Liang, Reference Jaiswal, Muggleton, Juan and Liang2019); openness to experience and dopamine effects on divergent thinking (Käckenmester, Bott, & Wacker, Reference Käckenmester, Bott and Wacker2019); behavioural inhibition system dysfunction and ADHD (Sadeghi et al., Reference Sadeghi, McIntosh, Shadli, Healey, Rostami and McNaughton2019); the neurobiological and environmental aetiology of psychopathy (Frazier, Ferreira, & Gonzales, Reference Frazier, Ferreira and Gonzales2019); curiosity as a fundamental aspect of personality (Zurn & Bassett, Reference Zurn and Bassett2018); virtual personalities neural network models and their neurobiological underpinnings (Read, Brown, Wang, & Miller, Reference Read, Brown, Wang and Miller2018); neuroanatomical correlates of hierarchical personality traits in chimpanzees, and their associations with limbic structures (Latzman, Boysen, & Schapiro, Reference Latzman, Boysen and Schapiro2018); and the opportunities for personality neuroscience of network approaches to understanding individual differences in brain connectivity (Tompson, Falk, Vettel, & Bassett, Reference Tompson, Falk, Vettel and Bassett2018). The capacity to integrate across such diverse areas is a powerful feature of personality neuroscience – indeed, it was one of the reasons the journal was founded.

2.1 Psychopathology/psychiatry

Psychopathology is identified as being particularly important, especially as there is a move towards hierarchical personality models being mapped onto hierarchical psychopathology models – this conceptualisation returns to a much older idea of an essential continuity between personality and psychopathology. As one board member notes: “Of course for me they ARE the same, but neither side tends really to notice the brain in the middle”. Given its inherently multidimensional and hierarchical approach, personality neuroscience has the potential to provide a cogent account of the extensive comorbidities seen in clinical data.

Despite these promising developments, another member remarked that clinical neuroscience still needs to pay much more attention to individual differences. This might seem like an odd assertion, given that clinical research is inherently about individual differences (comparing those who are ill with those who are not). But, relying on diagnostic groups to study and treat mental illness is often limited in its effectiveness because of the heterogeneity within diagnostic categories and the lack of biological grounding of the categories in the first place. As discussed by Latzman et al. (Reference Latzman, Krueger, DeYoung and Michelini2021), clinical neuroscience has much to gain from personality neuroscience, especially in relation to the dimensions of symptomatology, not least as a supplement to a focus on broad diagnostic categories. There is already a movement in this direction, following the NIMH’s RDoC system for studying transdiagnostic neurobehavioral circuits (Cuthbert, Reference Cuthbert2022). This is facilitated by the integration of the Hierarchical Taxonomy of Psychopathology (HiTOP; Kotov et al., Reference Kotov, Krueger, Watson, Cicero, Conway and Wright2021) system, which is an attempt from various mental health disciplines to improve the organisation, description, and measurement of psychopathology.

HiTOP serves to highlight the limitations of traditional nosologies, such as the DSM and ICD, including: arbitrary boundaries between psychopathology and normality; often unclear boundaries between disorders; frequent disorder co-occurrence; heterogeneity within disorders; and diagnostic instability. As noted by Latzman and DeYoung (Reference Latzman and DeYoung2020), an empirically-derived dimensional approach promises to accelerate clinical neuroscience. Developmental neuroscience should also have much to say about typical and atypical trajectories of development, given that these are generally conceptualised as dimensional. However, simply moving to dimensional trait-based ideas will not be enough: both brains and specific psychological presentations are highly individual, and the ‘devil’ in their detail lies in the interface among brain, living conditions, and presentation, as discussed below.

When thinking about how best to enable these advances, disciplinary boundaries continue to present an obstacle to adopting an individual differences perspective. As noted by a board member and endorsed by others: “I work in a psychiatry department, where the role of individual differences is woefully ignored. I sense that disorder-specific research groups fear losing the flag they have planted in their categorical disorder(s) of interest. There would have to be some systematic studies demonstrating the benefit of adding individual differences measures (ones that span normal and clinical range) to clinical neuroscience studies”. A member who works in the commercial world has the same experience: “I work in the industry now and I feel the same. Individual differences also exist in the biological domain; for example, in their expression of neuropsychiatric disorders like Alzheimer’s disease. The clinical trial world is not ready yet to take into account the profound variability that exists across people, even when rather reliable biomarkers to track that variability exists (e.g., PET tracer)”. Below, we see similar professional boundary issues with human and non-human experimental animal studies of personality.

One major problem to be acknowledged and overcome reflects the fact that personality theory and neuroscience have developed along quite distinct epistemological lines. With notable exceptions (e.g., Jeffrey Gray’s seminal work), they have managed to avoid thinking about each other. In particular, trait theories have tended to ignore detailed state control of behaviour. Neural analysis of state control ignores individual differences, usually treating them as “error”. For both, there is the problem of how the “objective data” are chosen, analysed and interpreted. This is not a trivial issue, and nor is it necessarily easy to solve once acknowledged.

Nevertheless, the years to come will surely witness personality traits increasingly finding their way into biologically-informed clinical perspectives, used as tractable indicators of the risk of developing disorders of many types. In consequence, a as-yet unrealised promise points to treatment development and truly personalised precision medicine. This is especially relevant in clinical psychopharmacology where different patients often respond very differently to the same drugs.

2.2 Opportunities and challenges ahead

It is trite to say that there are many challenges ahead. Some readily spring to mind. For example, consider the lexical approach, which is central to current-day personality description and measurement. It still needs to be better integrated with standard neuroscience data. However, any attempt will need to face the ever-present danger of failing to bridge the conceptual gap between the objective (i.e., neuroscience and verifiable self-report) and the subjective (i.e., private mental states). However, as we can assume that the latter is based in neural processes, in principle, this should not be unbridgeable. Relatedly, there is the need to consider a merger between wider issues; for example, from social neuroscience (e.g., person perception neuroscience; Mitchell, Reference Mitchell2009), to explore how the brain responds when people assess self and others according to various traits, much of which is based on the lexical tradition in one form or another.

In the age of the Internet of Things (IoT; Montag, Dagum, Hall, & Elhai, Reference Montag, Elhai and Dagum2021a), we have the opportunity to move beyond self-reports, important as they might remain. The IoT allows us to collect a much broader raft of (in particular, digital) behavioural data at unprecedented size and timescale. Intriguingly, this might even lead to the development of completely new structures of personality description. More generally, there is much potential in exploiting the opportunities afforded by the digital world. These include the investigation of behavioural signatures of personality, as seen, for example, with ‘digital phenotyping’ that applies methods from psychoinformatics that when combined with neuroscientific data allows the emergence of psychoneuroinformatics – this is especially relevant because it is ecologically valid (e.g., call behaviour on smartphones as a putative marker of sociality/extraversion; Montag et al., Reference Montag, Baumeister, Kannen, Sariyska, Meßner and Brand2019). Surprisingly, this is seldom done in the neurosciences at this time, despite the tendency of neuroscience researchers to be early adopters of new technologies (Montag, Elhai, & Dagum, Reference Montag, Elhai and Dagum2021b). In addition to the accessibility of intensive ecological sampling, virtual reality is opening up further intriguing possibilities, especially in the experimental manipulation of environmental/situational variables (e.g., Krupic, Zuro, & Corr, Reference Krupic, Zuro and Corr2021).

Molecular (genetic) studies clearly fall within the realm of personality neuroscience (Montag, Ebstein, Jawinski, & Markett, Reference Montag, Ebstein, Jawinski and Markett2020) and this, too, could benefit from digital phenotyping (Montag, Dagum, Hall, & Elhai, Reference Montag, Dagum, Hall and Elhai2022). For example, whether it is possible to carve out digital biomarkers (Montag, Dagum, Hall, & Elhai, Reference Montag, Dagum, Hall and Elhai2021); and it would be interesting to see if we find patterns of digital footprints (Montag, Elhai, & Dagum, Reference Montag, Elhai and Dagum2021b) giving insights into both the neurobiology and personality of the person.

2.3 Conceptual, statistical, and task considerations

In addition to the many conceptual problems that attend the field, there remain long-standing ones from personality psychology. For example, to reconcile the idiographic with the nomothetic, which is an old problem with little real progress to its name. Related to this problem is the question of whether traits evolve uniquely within each person, as opposed to uniformly across persons: this has implications for the extent to which structural-descriptive models of personality are truly “carved at nature’s joints” and are, therefore, representative of any population. Also, there is the need to consider identity, coherence, regulatory/control processes, as well as the motivational core of personality traits – these present formidable challenges, conceptually and empirically. In this sense, personality is not just about individual differences, although in much research practice, it reduces to them.

Personality neuroscience is at the forefront of developments when it manages to integrate personality and neural approaches and not implicitly (or explicitly) privilege one over the other. This relies on deep theoretical integration – a challenge in itself. This need is seen, for example, with the NIMH approach going awry by privileging the neural over the psychological, at least according to its former director, Thomas Insel (Reference Insel2022). We see this critique in other representative reviews of standard practices in psychiatry (e.g., Nour, Liu, & Dolan, Reference Nour, Liu and Dolan2022; Scull, Reference Schull2021).

There is the ‘effect size challenge’, too – related to how we chose to characterise personality in relation to neuroscience. Many effect sizes for associations between personality traits and other variables (such as biomarkers) may well be tiny, and so we need to power studies appropriately. One solution might be to acknowledge that the personality index needs to be at the right level to maximise the association effect size. It may even be unlikely that the whole trait score is the best level to use because it may not be underpinned by a single latent causal variable. In some cases, a more fine-grained analysis would be more powerful; however, as ever, with any such approach we run the risk of data snooping, so there would need to be protection from false positives resulting from running multiple tests. The greater use of a structural equation modelling approach might help, providing greater power at the higher levels as the latent variable modelled would be the shared variance between the trait indicators and that shared variance might be the part which is shared with a biomarker.

There is also the problem of the nature of the task used in typical neuroscience studies. For much of the time, methods have not been developed with individual differences in mind. Often, we look for individual differences in performance on a task that was basically designed to elicit a typical response and which have become popular precisely because between-subject variability is low (e.g., the stop signal task; Hedge, Powell, & Sumner, Reference Hedge, Powell and Sumner2018). It is much harder to develop approaches de novo that are aimed squarely at having good psychometric properties for individual differences research, although some attempts have been made (e.g., Allen et al., Reference Allen, Johnson, Bruke, McLean Sammon, Wu and Hooley2021, with the emotional signal task). This problem is confounded further by modelling and statistical issues.

A particular lesson to be learned from computational/statistical models may help when analysing behavioural indices, even very familiar ones such as response choice or reaction time measures. As remarked by a board member, the behaviour can be critical to maximizing the chances of finding an association with a personality measure, or between two behavioural measures purportedly measuring the same construct (see Haines et al., Reference Haines, Kvam, Irving, Smith, Beauchaine and Turner2020; Rouder & Haaf, Reference Rouder and Haaf2019; these papers present illustrations using delay discounting, implicit association tests, Stroop, flanker, and cueing tasks).

Relatedly, in personality research, the behavioural measure of interest (such as a difference in performance between two task conditions) is usually captured as a fixed effect. A fixed effect is assumed to have the same value for every person. Despite this, we correlate the individuals’ scores on the behavioural measure with personality measures. Personality researchers routinely do this even though, under a fixed effect model, the variation in the score across individuals is considered to be an error. It would be more appropriate (and, as Haines et al., and Rouder & Haaf show, more powerful) to model the individual differences in the behaviour as a random effect; that is, an effect that varies across individuals. It is ironic that the use of random effects modelling is not more widespread in individual differences research, given that it has become common in cognitive psychology for more than a decade (e.g., Baayen, Davidson, & Bates, Reference Baayen, Davidson and Bates2008) and in neuroimaging for even longer (see Friston, Stephan, Lund, Morcom, & Kiebel, Reference Friston, Stephan, Lund, Morcom and Kiebel2005). The potential value of models including random effects, and their patchy application in personality research, has been noted for some time (e.g., see West, Ryu, Kwok, & Cham, Reference West, Ryu, Kwok and Cham2011, for a review). Personality effects need to be seen as a source of signal, not noise.

Although any such list could get very long, editorial board members identified an additonal number of challenges that deserve mention:

1. 1. Many-to-many mapping of traits to neural variables (see DeYoung et al., Reference DeYoung, Beaty, Genc, Latzman, Passamonti and Wacker2022a).

2. 2. Individual differences in brain structure are obscured by warping to a common template (as is typical in MRI).

3. 3. Variability in neurochemistry, as assessed by molecular imaging (PET).

4. 4. Individual differences in functional localisation relative to anatomical landmarks (even if physical alignment is perfect, the same function can be carried out in different brain regions, especially in the cortex).

5. 5. The need to pay proper attention to psychometrics in both behavioural and neural measurement, and to consider basic issues around reliability.

6. 6. Determining causality – manipulating traits is very difficult, and merely manipulating (state) behaviour in the short-term may not be equivalent to manipulating the trait corresponding to that behaviour.

7. 7. Not abandoning theory as atheoretical approaches like machine learning become more powerful, and alluring – the alternative would be to combine psychological theory building with machine learning, which is feasible (e.g., Elhai & Montag, Reference Elhai and Montag2020).

More generally, there seems to be a major potential for a confluence of neuroimaging and neurogenetics that includes epigenetics, to get deeper into the precise molecular mechanisms by which environmental exposure influences gene expression. The real challenge lies in designing studies where several layers of the brain are included in one study design. Collecting data from molecular genetics, epigenetics, mRNA levels, proteomics, hormone levels, brain scans, together with personality assessments, is a huge challenge – both intellectually and economically. But bridging the different brain layers and reuniting these with personality assessments/personality theory is of prime importance (Montag & Elhai, Reference Montag and Elhai2019). If this is not achieved, we end up with a set of puzzle pieces on the scientific table – and no picture on the box to guide us. The years to come will surely witness the slotting together of these different pieces to present the overall scientific picture.

Collaboration and cutting across specialism boundaries are especially needed when, for example, we attempt to triangulate brain imaging, neurochemistry, and personality data. We see this need, too, when employing other methods. For example, the potential of using Mendelian Randomisation methods to understand causality between personality and related traits (including potential endophenotypes).

Such collaboration is especially important for developing a cumulative, error-checking, progressive research system. This has already been seen in neuroimaging genetics, where the Enhancing NeuroImaging Genetics through Meta Analysis (ENIGMA) consortium of over 1400 researchers from 43 countries have made major discoveries in neuroscience. NATO-type meetings might go some way toward addressing the issues that have evaded solution to now. In addition to academia-to-academia collaborations, more is needed in the form of academia-to-industry collaborations, especially in the neuro-psychiatry field.

3.1 The value of special issues

In advancing the aims of the journal, we believe that special issues are especially valuable, as a source of concentrated science on a specific theme. We have already seen this with the Latzman et al. (Reference Latzman, Krueger, DeYoung and Michelini2021) one on connecting quantitatively-derived personality–psychopathology models and neuroscience. The next planned special issue is on the value of non-human studies to personality neuroscience (from fish through primates), marshalled by Yury Lages and Neil McNaughton (Reference Lages and McNaughton2022). As noted by these guest editors, the conceptual, theoretical, and philosophical tend to almost coalesce into something of a religion when we compare human descriptive studies and experimental non-human studies (McNaughton & Corr, Reference McNaughton, Corr and Corr2008) – “Descended from the apes! My dear, we will hope it is not true. But if it is, let us pray that it may not become generally known.”Footnote ¹ Lurking in many areas of psychological neuroscience are similar conceptual and philosophical problems – typically, not acknowledged, often not even recognised.

As these special issue editors note, it is not just that non-humans cannot use a language to respond. Rather, it is more the case that many researchers seem unwilling to use the same methods to study human and non-human animals, or even the same words for description and explanation. In any event, as the special issue intends to show, non-human animal studies are important for personality neuroscience, and they can even be linked to the major descriptive/structural model of the Big Five (e.g., Latzman et al., Reference Latzman, Boysen and Schapiro2018). In addition, selected rodent strains can provide models of key personality factors (Broadhurst, Reference Broadhurst1975) and trait-linked disorders (Macedo-Souza, Maisonnette, Filgueiras, Landeira-Fernandez, & Krahe, Reference Macedo-Souza, Maisonnette, Filgueiras, Landeira-Fernandez and Krahe2019). Cross-mammalian work is of relevance to better understanding personality and it adds an important evolutionary perspective on why we are the creatures we are – something which has previously been put forward in the context of Panskepp’s Affective Neuroscience Theory (Montag & Panksepp, Reference Montag and Panksepp2017). When thinking more broadly on this issue, there is a clear need for more sensitive behavioural measures and new experimental designs that bridge the gap between non-human and human-animal work (Mobbs et al., Reference Mobbs, Wise, Suthana, Guzman, Kriegeskorte and Leibo2021).

3.2 Cambridge University Press

As this editorial attests, much has already been achieved but much more is needed. None of this would have been possible, however, without Cambridge University Press who have supported the journal from its inception and continue to support it in new ways. As evidence of this support, they have sponsored an annual cycle of special issues on topical themes that have the capacity to move the field forward in very significant ways. Special issues are, indeed, valuable but they take time and commitment on the part of editors – however, we believe the end product is well worth it, especially as they are an effective way to consolidate past research and provide guidance for the future (they also tend to attract considerable attention and citations, all to the good for careers, as well as science!). We welcome suggestions for future issues. To facilitate the process, fees will be waived for those who cannot find the funds from existing research budgets – for regular submissions, Personality Neuroscience continues to operate a policy to ensure that a genuine lack of research funds is never an obstacle to publication.

The journal exists to serve the personality neuroscience community, and for it to achieve its full potential it needs support. We especially encourage younger researchers to submit their best work, especially work that may not be readily appealing to other journals that do not quite appreciate the unique value of combining neuroscience and personality. We can assist researchers to get their papers into proper shape for publication, and while quality must be protected, we prefer not to adopt a ‘judge and jury’ approach to editorial decisions which can, so often, be disheartening to the early career researcher. Our ambition is to make the journal inclusive to all.

We end with an apt quote from DeYoung et al. (Reference DeYoung, Beaty, Genc, Latzman, Passamonti and Wacker2022a, pp. 13–14): “The field of personality neuroscience has a crucial role to play in understanding why people do what they do and how they differ from each other”. The subtitle of their paper inspires the subtitle of this editorial: “An emerging field with bright prospects”.