Gene × Environment interaction: What exactly are we talking about?
A big gene-times-environment statistic does not prove the two factors physically meet—stay skeptical.
01Research in Context
What this study did
Yeung (2018) wrote a theory paper, not an experiment.
The author asked: when we see a big gene-times-environment (GxE) number in a chart, what does that number really mean?
The paper warns readers not to treat a tall bar or a tiny p-value as proof that genes and surroundings physically shake hands inside a child.
What they found
The study found no new data.
Instead it showed that a statistical interaction is just math.
It can hide in the data even when genes and surroundings never touch each other in real life.
How this fits with other research
Doughty et al. (2015) looked at Jamaican kids with autism. They reported that carrying two certain gene types at once almost tripled autism odds.
That finding is exactly the kind of shiny interaction term Yeung (2018) says we should not over-trust. The numbers look exciting, but they may not point to a real biological handshake.
Roll (2005) reviewed earlier autism-gene work and also warned that no single gene story is ready for prime time. Yeung (2018) extends this caution to the newer GxE claims.
Facon et al. (2011), Jarrold et al. (2004), and Flapper et al. (2013) all push the same theme: stop worshipping p-values and start asking harder design questions.
Why it matters
As a BCBA you may read that a gene-times-environment link predicts severe behavior. Before you add that story to a parent meeting, remember Yeung (2018): the math may be real, but the mechanism is still a guess. Use the data to guide further assessment, not to lock in a life-long label.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →When you next see a GxE bar graph in a team meeting, ask: "Does this show a real mechanism or just a math overlap?"
02At a glance
03Original abstract
An ambiguity exists in how psychological scientists use the word "interaction." This word can refer to physical interactions between components that form the mechanisms in complex systems, but it can also refer to statistical interactions revealed by General Linear Statistical Models (e.g., Analyses of Variance). Statistical interactions indicate that the nature of the relationship between two variables depends on a third variable, but the discovery of such interactions does not constitute evidence of physical interactions between components in a system. Studies conducted using traditional behavioral genetics methods sometimes reveal statistical interactions between genes and environments, but the presence or absence of such interactions tell us surprisingly little about actual, physical interactions between genes and their contexts. This is important, because it is only the latter kinds of interactions that cause the development of behavioral phenotypes, including developmental disabilities. Therefore, when behavioral scientists discover (or fail to discover) Genotype × Environment interactions, it is important to exercise care in interpreting their meaning and in assessing the utility of such findings.
Research in developmental disabilities, 2018 · doi:10.1016/j.ridd.2018.04.012