What We Don’t Know About Temperament and Sleep Training

Research has clearly shown that temperament plays a direct and forceful role in children’s sleep. Children with a “difficult” temperament (hard to soothe, prone to crying) have a much higher degree of sleep problems. Traits like intensity and reactivity, adaptability and distractibility, low and high sensory thresholds, as well as some positive traits like sociability, extroversion, and engagement, all directly and negatively impact sleep.

In a large parent survey on temperament and sleep, I also found that the temperament traits above predicted every aspect of challenging sleep: poor sleepy signals, difficulty falling asleep independently, significant parental bedtime and middle-of-the-night effort, and frequent wakings (Gordon, 2020).

The sleep training status quo is all about behavior

Existing mainstream advice on sleep, however, focuses exclusively on parental behavior as the primary cause of sleep problems. Such advice (and the research it’s based on) is rooted in behaviorism—a theory from the 1920s that suggests that all behavior is the result of experience. Inborn traits, like temperament, are not considered as relevant. In fact, sleep books suggest that a difficult temperament is caused by poor sleep. Wakefulness is a behavior that is reinforced by the environment (parents) and will persist unless the reinforcement is removed. This process is called extinction and the removal of reinforcement can be total (pure extinction, cold turkey) or incremental (graduated extinction, Ferber, pop-ins). It’s the most researched and most endorsed intervention for infant sleep.

Experts submit that extinction is straightforward, quick, effective, and without negative side effect. For many families, this is exactly the case—three or four nights of very limited crying before sleep and—ta da!—lots of sleep.

For parents of intense, sensitive children, it’s a different situation

For families of intense, persistent children, however, the crying is not limited or quick. Data from my parent survey found that parents who had attempted extinction-based methods reported significantly more crying than they expected or were told to expect by the books they read. In my work with parents, those with alert, sensitive, intense infants often report hours of crying across a week or more with no improvement or reduction in distress.

Further, crying in these cases can be intense. These children can rapidly become hysterical and often do not or cannot calm down on their own. Extinction-based methods only allow parents to briefly check in but not pick up or soothe the infant, potentially leaving these children in a dysregulated state for longer periods than the books indicate is expected or possible. The context of both sleep and trying to work on sleep is fundamentally different for these parents and these children.

Temperament is not a factor in sleep training research

Given the important role temperament potentially plays in sleep and extinction-based sleep training, it’s surprising that temperament is practically absent from research or parenting books. In a literature review on evidence-based behavioral sleep interventions (Gordon, 2022), of 58 studies included, only five assessed temperament at all and often only at the end of the intervention. We know virtually nothing about how these more reactive children experience extinction-based interventions or whether they are even effective for this group. The review further found that across studies, extinction did not work for between 16% and 50% of participants. In real-world samples, the figure is between 40% and 60% (Loutzenhiser et al., 2011). Because temperament isn’t assessed in extinction studies, we don't know what percentage of these are the more reactive children.

“No negative side effects”

Research and advice clearly suggest that extinction has no negative side effects. However, it almost never indicates how much crying happened—how much crying was “without effect.” In the only study that reported the amount of crying (France & Blampied, 2005), one child cried a total of 130 minutes over three weeks (and was only crying for five minutes by day five), while another cried a total of 900 minutes and cried 110 minutes on day 21. These are two very different experiences of the same intervention and may have had very different outcomes for those children.

The conclusion of this study was that the intervention was successful and no negative side effects were found. However, were these conclusions equally true for both of those children? While it’s likely that the child who never cried more than 15 minutes at any one time acclimated to the intervention quite quickly, the second child clearly did not.

What’s missing in sleep training research is an appreciation of differential susceptibility—meaning that an event that might be tolerated reasonably well by one child can have a marked and negative impact on another because of differences in individual neurological and sensory systems (Boyce, 2016). Unfortunately, this question has simply not been investigated. We don’t know if there are impacts of different amounts of crying, for certain infants, at certain ages that might have effects that we’d want to know about.

Sleep Essential Reads

When Sleep Banking May Be a Viable Strategy

It's Not Just What You Eat, but When You Eat That Matters

Consequences for parents of intense, reactive children

Parents are the ones experiencing the consequences of the one-size-fits-all approach to sleep training. They are led to expect outcomes that are possible for “easier” children, but unrealistic for their child. What are parents to do when they are on hour two of hysterics on night number six? Sleep books don’t talk about a “Plan B” but instead imply that parents may just need to be firmer or try harder. None of the research or books account for the level of difficulty or distress these parents are experiencing. Parents of temperamentally intense children end up blaming themselves and then just soldiering through without an alternative.

It really shouldn’t be this way. Research needs to start asking some different questions.