In a detailed and compelling synthesis, Burt questions the validity of genomics to gain knowledge about social science issues. Burt's cautionary note is a much-needed reminder of the limitations of sociogenomics, a field that has seen a fair amount of hype in the last few years. Here, we wish to comment on a central argument Burt makes – that because they are already well-measured behaviorally, constructs like academic achievement or cognitive aptitudes have little to benefit from the tools of sociogenomics. In our view, this argument potentially disregards the serious challenges psychologists face in measuring these constructs, no matter how well-defined they may seem behaviorally.
Psychological constructs are not pure, assumption-free operationalizations of the underlying traits or abilities they are meant to represent. Rather, the way constructs are validated and refined over time follows a process whereby convergent validity dictates the empirical instantiations – in the form of tests, tasks, or measures – that are hypothesized to probe the same constructs, and those that in contrast are understood to tap different ones. As a result, constructs are heavily influenced by initial operationalizations, in a process that is biased toward convergence at the risk of failing to explore valid – or sometimes better – alternatives (Moreau & Wiebels, Reference Moreau and Wiebels2022).
In this context, we should be cautious about uncritically ascribing validity to psychological constructs on the basis of psychometric convergence or divergence with one another. No matter how objective they might seem, behaviorally assessed constructs remain subjective and far from assumption-free. It does not follow that genomics is necessarily the answer to help refine our understanding of psychological constructs, but we should refrain from thinking that the measurement of constructs in the behavioral sciences is as good as it can be, or that only improvements in psychometric properties will lead to better, more valid assessments. Sociogenomics may or may not be the solution, but epistemic validity remains a challenge in the field.
Not all fields to which genomic tools are applied suffer equally from this limitation. For example, this bias is arguably less problematic when genomics is applied to medicine, where it has led to major advances in our understanding of cancer, heritable disorders, or infectious disease outbreaks (McCarthy, McLeod, & Ginsburg, Reference McCarthy, McLeod and Ginsburg2013). Success in the clinical domain remains highly heterogeneous, however, with most significant advances having been achieved for conditions within which constructs of interest (e.g., diagnosis) are well defined and fairly objective, often because they are based on the presence or absence of biological features. In contexts where diagnosis is more subjective and constructs of interest less well defined – for example with psychiatric disorders diagnosed primarily from the presence of behaviors or related symptoms – genomic-based advances have been less prominent, for the same reasons they have been of somewhat limited benefit in psychology thus far.
So what could sociogenomics contribute to our understanding of aptitudes and achievement that current behavioral measures do not? The potential is wide-ranging and multifaceted, but one application that stands out is with respect to behavioral interventions designed to improve cognitive performance or abilities (Madole & Harden, Reference Madole and Harden2023). Recent attempts to improve cognitive abilities have suffered from major setbacks, with strong initial claims failing to stand up to scrutiny (Moreau, Reference Moreau2022; Moreau, Macnamara, & Hambrick, Reference Moreau, Macnamara and Hambrick2019; Sala & Gobet, Reference Sala and Gobet2019). One of the main issues that has been identified is the lack of mechanistic understanding for the behavioral dynamics elicited by interventions, especially given the important heterogeneity in individual responses (Moreau, Reference Moreau2021). Some individuals show promising improvements post-interventions, whereas others do not appear to benefit at all, and current models provide little insight into the determinants of individual differences (Moreau, Reference Moreau2022). Together with efforts to improve and refine measurement in the context of interventions (Moreau & Wiebels, Reference Moreau and Wiebels2021), the field of genomics has the potential to shed light on the complex interactions at play to determine – and eventually predict – individual responses in a personalized manner. To be successful, such behavioral interventions are likely to require precision regimens, whereby individual characteristics – potentially including genomic information – are leveraged to establish responder profiles and thereby determine the optimal blend for a particular person at a particular time.
Despite the potential for sociogenomics in this space, tangible progress remains dependent on addressing current limitations in the use of polygenic scores, especially issues such as confounding and stratification. Although these limitations might be alleviated in the context of interventions because of the controlled nature of these designs, they generally remain issues that the field of sociogenomics will need to grapple with. In addition, when genetic data are incorporated into intervention designs and individual response predictions, researchers should explicitly specify in what ways they can lead to qualitative improvements, and the potential downsides. Polygenic scores remain probabilistic, and as such include wide individual differences in the target trait at all levels (Plomin, DeFries, Knopik, & Neiderhiser, Reference Plomin, DeFries, Knopik and Neiderhiser2016); fair and accurate assessments of what sociogenomics can and cannot contribute at this time are to the benefit of all.
Finally, efforts to incorporate sociogenomics within behavioral interventions should not divert from attempts to address the structural scarcity and inequality inherent to systems and institutions. In particular, the notion that success is primarily driven by aptitudes or abilities that can be targeted by interventions has been shown to be problematic or even dangerous in some instances (Moreau, Reference Moreau2022; Nathan, Reference Nathan2017). When unchallenged, this view can prevent the implementation of institutional reforms that are known to effectively reduce systemic inequalities (Furnham, Reference Furnham2003). Addressing inequalities is an endeavor that often requires collective action on multiple fronts, and gaining a better understanding of individual differences via genomics is, albeit promising, only one of the many facets that can be leveraged to make progress in this direction.
In a detailed and compelling synthesis, Burt questions the validity of genomics to gain knowledge about social science issues. Burt's cautionary note is a much-needed reminder of the limitations of sociogenomics, a field that has seen a fair amount of hype in the last few years. Here, we wish to comment on a central argument Burt makes – that because they are already well-measured behaviorally, constructs like academic achievement or cognitive aptitudes have little to benefit from the tools of sociogenomics. In our view, this argument potentially disregards the serious challenges psychologists face in measuring these constructs, no matter how well-defined they may seem behaviorally.
Psychological constructs are not pure, assumption-free operationalizations of the underlying traits or abilities they are meant to represent. Rather, the way constructs are validated and refined over time follows a process whereby convergent validity dictates the empirical instantiations – in the form of tests, tasks, or measures – that are hypothesized to probe the same constructs, and those that in contrast are understood to tap different ones. As a result, constructs are heavily influenced by initial operationalizations, in a process that is biased toward convergence at the risk of failing to explore valid – or sometimes better – alternatives (Moreau & Wiebels, Reference Moreau and Wiebels2022).
In this context, we should be cautious about uncritically ascribing validity to psychological constructs on the basis of psychometric convergence or divergence with one another. No matter how objective they might seem, behaviorally assessed constructs remain subjective and far from assumption-free. It does not follow that genomics is necessarily the answer to help refine our understanding of psychological constructs, but we should refrain from thinking that the measurement of constructs in the behavioral sciences is as good as it can be, or that only improvements in psychometric properties will lead to better, more valid assessments. Sociogenomics may or may not be the solution, but epistemic validity remains a challenge in the field.
Not all fields to which genomic tools are applied suffer equally from this limitation. For example, this bias is arguably less problematic when genomics is applied to medicine, where it has led to major advances in our understanding of cancer, heritable disorders, or infectious disease outbreaks (McCarthy, McLeod, & Ginsburg, Reference McCarthy, McLeod and Ginsburg2013). Success in the clinical domain remains highly heterogeneous, however, with most significant advances having been achieved for conditions within which constructs of interest (e.g., diagnosis) are well defined and fairly objective, often because they are based on the presence or absence of biological features. In contexts where diagnosis is more subjective and constructs of interest less well defined – for example with psychiatric disorders diagnosed primarily from the presence of behaviors or related symptoms – genomic-based advances have been less prominent, for the same reasons they have been of somewhat limited benefit in psychology thus far.
So what could sociogenomics contribute to our understanding of aptitudes and achievement that current behavioral measures do not? The potential is wide-ranging and multifaceted, but one application that stands out is with respect to behavioral interventions designed to improve cognitive performance or abilities (Madole & Harden, Reference Madole and Harden2023). Recent attempts to improve cognitive abilities have suffered from major setbacks, with strong initial claims failing to stand up to scrutiny (Moreau, Reference Moreau2022; Moreau, Macnamara, & Hambrick, Reference Moreau, Macnamara and Hambrick2019; Sala & Gobet, Reference Sala and Gobet2019). One of the main issues that has been identified is the lack of mechanistic understanding for the behavioral dynamics elicited by interventions, especially given the important heterogeneity in individual responses (Moreau, Reference Moreau2021). Some individuals show promising improvements post-interventions, whereas others do not appear to benefit at all, and current models provide little insight into the determinants of individual differences (Moreau, Reference Moreau2022). Together with efforts to improve and refine measurement in the context of interventions (Moreau & Wiebels, Reference Moreau and Wiebels2021), the field of genomics has the potential to shed light on the complex interactions at play to determine – and eventually predict – individual responses in a personalized manner. To be successful, such behavioral interventions are likely to require precision regimens, whereby individual characteristics – potentially including genomic information – are leveraged to establish responder profiles and thereby determine the optimal blend for a particular person at a particular time.
Despite the potential for sociogenomics in this space, tangible progress remains dependent on addressing current limitations in the use of polygenic scores, especially issues such as confounding and stratification. Although these limitations might be alleviated in the context of interventions because of the controlled nature of these designs, they generally remain issues that the field of sociogenomics will need to grapple with. In addition, when genetic data are incorporated into intervention designs and individual response predictions, researchers should explicitly specify in what ways they can lead to qualitative improvements, and the potential downsides. Polygenic scores remain probabilistic, and as such include wide individual differences in the target trait at all levels (Plomin, DeFries, Knopik, & Neiderhiser, Reference Plomin, DeFries, Knopik and Neiderhiser2016); fair and accurate assessments of what sociogenomics can and cannot contribute at this time are to the benefit of all.
Finally, efforts to incorporate sociogenomics within behavioral interventions should not divert from attempts to address the structural scarcity and inequality inherent to systems and institutions. In particular, the notion that success is primarily driven by aptitudes or abilities that can be targeted by interventions has been shown to be problematic or even dangerous in some instances (Moreau, Reference Moreau2022; Nathan, Reference Nathan2017). When unchallenged, this view can prevent the implementation of institutional reforms that are known to effectively reduce systemic inequalities (Furnham, Reference Furnham2003). Addressing inequalities is an endeavor that often requires collective action on multiple fronts, and gaining a better understanding of individual differences via genomics is, albeit promising, only one of the many facets that can be leveraged to make progress in this direction.
Financial support
D.M. and K.W. are supported by a Marsden grant from the Royal Society of New Zealand and a University of Auckland Early Career Research Excellence Award.
Competing interest
None.