Source: Unknown

Source: Unknown

A couple years ago a preliminary study came out looking at the cortisol levels of infants undergoing extinction sleep training[1].  You see, for ages people had tried to claim that infants didn’t experience distress during this type of extinction sleep training.  They were simply “protesting” and were quite fine being left to cry on their own.  The researchers found that, on average, infants (all babies were under 12 months, but ranged in age from 4 to 10 months) experienced high levels of cortisol during sleep training, but more importantly, they continued to experience high levels of physiological distress days later when they weren’t crying.  In short, the infants had stopped crying themselves to sleep but remained distressed.  More importantly and tellingly, the infants and their mothers had lost what is called “synchrony”, that is, the physiological link between mother and infant, the link that allows mothers to help soothe their babies when upset, the link that is associated with attachment status.  This latter part has sadly been largely ignored even though that was actually the primary take-home message from the research.

Those who are against extinction sleep training methods have used this study to argue that we have preliminary evidence that such training results (or rather, can result) in high levels of cortisol during and even after the period of sleep training and can have negative effects on the parent-child relationship.  (It’s worth noting the study had many limitations including high variability between participants, which is to be expected given the high variability in cortisol amongst people more generally, and the fact that that infants were distressed before bedtime as well.  I’ve given a more in-depth look at my take on the study here after some very intelligent questions were raised on the issue; however, these limitations are why I call it a “preliminary” study – they couldn’t know what to expect and thus were limited to some initial exploration research with, hopefully, more to follow.)

We all know about the degree to which prolonged and extreme levels of cortisol negatively affect the developing brain.  Extreme stress during infancy is a horrible, horrible thing for a child to endure.  The question becomes, however, how stressful is extinction sleep training?  Although few would argue that it constitutes the type of neglect seen in the more extreme clinical cases, it’s also hard to see it as anything but a form of neglect – even if done “out of love”.  With the aforementioned research coming out, the argument seems to have shifted from the idea that extinction sleep training does no harm to the acknowledgement that extinction sleep training does result in high levels of cortisol, as [1] has demonstrated, but only temporarily.

The argument thus becomes… Babies may experience spikes in cortisol during extinction sleep training but as humans experience stressful situations regularly – and the associated rises in cortisol – then this is no different from what humans experience throughout life.  Indeed, some have argued that the experience of these situations actually benefits us as we need the occasional boost of cortisol now and again as abnormally low levels of cortisol are also problematic.

It sounds lovely, doesn’t it?  I mean, really, our babies are experiencing something that we all experience with no harm (after all, we’re not talking chronic stress here) and is something that they need to adapt to in our world.  A little bit of cortisol never hurt anyone, did it?

There’s a problem with this philosophy, however, and it needs to be discussed: The period of hyporesponsivity.

What is the Period of Hyporesponsivity?

Basic_HPA_AxisThe period of hyporesponsivity was first detected in rodent development and has since been found to occur in human infants as well; in short, it involves a temporal period in development in which it is difficult to elicit a cortical response in an otherwise securely-attached child (insecure attachment is associated with cortisol increases to various situations)[2][3].  For human babies, this temporal period starts at about two months of age and develops throughout the first year of life with infants showing a strong hyporesponsive response by one year (often by 4 months), and a currently-unknown end-date though it looks to be around the 3 year mark.  Specifically, the usual markers of mild-moderate stressors (in the case of research with humans it is inoculations which are actually rather painful) will continue to elicit a behaviour response of crying, but there will be no associated cortisol spike.  Thus, there is a behavioural-physiological mismatch that results from this period of hyporesponsivity.

Importantly, one behaviour that does elicit a cortisol response is separation from a caregiver as it is us caregivers who help buffer these negative emotional situations[4].  Specifically, in one study, securely-attached 9-month-old infants did not show a rise in cortisol for approximately 30 minutes when separated from their mother, but were left with a babysitter so long as the babysitter was responsive to their distress[5].  However, if the babysitter was non-responsive there was a significant increase in cortisol.  Why does this happen?  Because hyporesponsivity is obtained by social buffering – our infants need us to buffer this cortical response[2-4].  In one study with non-human primates, the mere presence of the mother during a stressful event (like capture) buffered the elevations in cortisol associated with the stressful event[6].  (Though don’t believe you can just sit there stone-faced and be of comfort, responsiveness involves reacting to and helping your child in some way.  See the effects of the still-face paradigm if you believe sitting and not responding will actually help.)

You may still be asking, “What’s the big deal?”  I mean, if you subscribe to the idea that infants’ experience of cortisol is just like adults’, why should this matter?  After all, isn’t a bit of cortisol good for them (as it is us)?  Or even, why should we care about temporary rises in cortisol if they aren’t so extreme as to fall into the severe neglect situation?

The Evolutionary Dilemma

This is the where we hit the evolutionary dilemma: For human infants (and infants of other species) to have a period of hyporesponsivity, there is likely an important developmental reason.  Although we don’t know exactly what it is, there is good reason to believe is has to do with our neurological development.  Let me take a step back and explain in more detail.

We have to start with the knowledge that extreme stress and prolonged bouts of high cortisol negatively influence neural development.  We see problems in the development of white matter in cases of extremely severe neglect[7], negative effects on the hippocampus and amygdala (brain structures utilized in memory and emotion)[8], and even brain shrinkage from a lack of love[9], but even in lesser conditions we can see what is termed a “stress-reactive profile” in which the child develops a heighted response to stress[4].  Evolutionarily, this makes sense.  An infant raised in an environment in which he or she is not safe has to be acutely aware of the stress around them.  It may mean the difference between life and death down the line.  Extinction sleep training, however, is different because there really is no threat to the child, although the child doesn’t actually know this.

We also know that the effects of stress start very early and even moderate levels can influence development.  For example, mothers who experience higher levels of cortisol during pregnancy have children with different behavioural patterns (i.e., more behavioural problems) than those who have lower levels of cortisol during pregnancy[10].  Amazingly, this isn’t related to a mother’s own identification of stress, but rather the physiological levels, regardless of coping skills.  Again, evolutionarily this makes sense as the mother’s environment is providing valuable information for the infant.  Namely, the infant needs to know what kind of world he or she is coming into and maternal stress is one valuable piece of the puzzle.

Now, as with any evolutionary trait, there are two possibilities: (1) The effect is a byproduct that does no harm, but also has no benefit, or (2) The effect is beneficial to our development.  If the third option – a negative effect on humans – were to hold true, we should have evolved out of it by now as that’s how evolution works.  Now we have to question if the period of hyporesponsivity is a benign effect or a positive one.

My hypothesis is that it is a positive one, or rather, it is the expected, normal outcome and anything that deviates is actually providing critical information for the developing brain.  First, looking at the extreme examples of neglect and abuse, we see the role of cortisol on neural development and it is certainly not benign.  Furthermore, as more research examines the effects of cortisol on development, we see more and more “moderately negative” or even “normal” childhood or family events (in today’s society, not necessarily long-term) resulting in impeded development and attachment[11][12].  For example, one recent study looking at the effects of parental conflict on child cognitive development[11] found a negative relationship between the degree of parental conflict witnessed at age 8 and the ability to regulate emotions and attention at age 9 and was also associated with slower development of fluid cognitive abilities.

Second, the period of hyporesponsivity occurs during peak neurological development (i.e., the first three years of life) – especially for areas of the brain associated with emotional and social processing[13] (cognitive development continues to accelerate well into late adolescence).  Specifically, the areas of the brain associated with early stress are the amygdala[8], limbic brain structures[8], and white matter[7]. (For a great review of brain development in the first three years and the associated structures, see here.)  Given the aforementioned link between severe, chronic stress and later outcomes, is it any surprise the brain tries to protect itself?  Of course there is a difference between temporary stress and chronic maltreatment, but the combination of what is developing during this timeframe and the evolutionary mechanism of socially-buffered hyporesponsivity suggests (to me) the main goal of parenting is to attempt to avoid these stressors during this time-frame.

Third, the behaviour or circumstances that elicit cortisol spikes in infancy (during this period of hyporesponsivity) are moderate-severe pain, abuse, neglect, or abandonment of a caregiver with no responsive substitute.  Indeed, as previously mentioned, if a caregiver (or substitute, for up to 30 minutes) is present and responsive, a child can be in mild-moderate pain or crying hysterically, but there is a disconnect between the behaviour (crying) and physiology (no cortisol response)[3].  It’s why I always tell parents that although it may seem futile to comfort a crying baby who seems inconsolable, you are actually providing a very valuable service.  As neuroscientists Megan Gunnar and Bonny Donzella aptly wrote in [2]:

Tracy Maddy“The functional effect of sensitive, responsive, attentive caregiving is that it allows children to express and experience distress, communicate those emotions to caregivers in ways that can elicit help, without stimulating increases in glucocorticoids.” (p. 207)

You may be wondering why anything can elicit this cortical response.  I mean, if it’s so bad for the developing brain, why not develop to experience no stress in early infancy?  First, clearly if this were beneficial overall, we likely would have evolved for it, so there is a reason it doesn’t actually help us.  One potential explanation would be that cortisol can provide very important information as an infant does not develop outside of an environmental context.  An infant needs to take in the most important environmental information in order to survive (or give up, as has been the case with extreme neglect and abuse) – certainly, experiencing stressors without the comfort of a responsive carer are critical pieces of the environmental puzzle.  The problem, as pointed out above, is that these cues and the subsequent changes to development are only adaptive if they take place within a context which makes them adaptive.  For the most part, infants who experience these cortisol spikes in extinction sleep training are not in an environment that would typically necessitate the encoding of such information.  By engaging in extinction sleep training, parents are providing their infants with information they may not wish to provide, such as the knowledge that the world is not as safe as we want our children to believe or that they are, in essence, alone during difficult (and scary) periods (kind of like setting up the belief of a dog-eat-dog world).

Finally, one of the foundations of my hypothesis is that the effects of cortisol on the developing brain are not discrete, but rather continuous.  That is, I don’t believe there is some magic amount of neglect or stress that needs to be experienced before we see any effect, but rather every amount of stress adds incrementally to change the neural development from that of expected neural development to something different.  Of course we cannot say that “different” is “bad” because, as discussed above, it’s more likely a matter of providing the infant with critical information about their environment.  It can, however, become “bad”, and the stage at which that happens is likely different for each child based on a host of factors including temperament and overall responsiveness, amongst others.


If we consider extinction sleep training within the framework of the period of hyporesponsivity, we are left with some very important, and potentially disturbing, ideas.  First and foremost is the awareness that extinction sleep training – which is the removal of responsiveness and presence – is exactly the thing that elicits the type of cortisol spike that is nearly impossible to obtain in infants and younger toddlers.  Although most of us believe that crying = stress and not crying ≠ stress, it’s not nearly so simple.  For close to three years, there can be a behaviour-physiological mismatch, with infants continuing to express themselves through crying, but rarely showing the associated cortisol response when parents are responsive.  This is essential for parents to understand because it can remove the mental barriers to being responsive when it’s most difficult; by knowing that comfort is associated with a lack of cortisol flooding the system, it can become much easier for a parent to respond and assist their infant, even if it doesn’t end the bout of crying in the moment.

Second, one concern most parents have about infant sleep is the fear that a lack of consolidated sleep in infancy is associated with rising cortisol levels.  Although there is strong evidence for “poor sleep” in adulthood being associated with a rise in cortisol levels[14] (although they then drop to lower-than-normal levels with prolonged sleep deprivation), the definition of “poor sleep” in infancy and toddlerhood is drastically different and the effects on mothers whose hormones are different to begin with is still not fully understood.  Indeed, longitudinal studies find that the types of “sleep problems” parents complain about (and sleep trainers or “experts” tell you to worry about) are actually very normal for children and are not associated with long-term sleep problems[15][16] (even if parents complain about them).  There will be some infants who present with signs of sleep deprivation or other behaviours of concern, but they are almost always due to third variable problems such as feeding issues or other health concerns (see more here).  As such, these parents would be advised to search for the root cause of their child’s sleep problem instead of just treating the symptom.

Third, if we accept that there is an evolutionary reason for this period and for the lack of responsiveness to be key in eliciting said cortisol spikes, we need to look at alternatives for families who are in need of a change.  If we rid ourselves of extinction sleep training, what are we left with?  Actually quite a lot.  I never want to be seen as saying that parents have to stick with a situation that doesn’t work for the entire family, but I also will never support “solutions” that put the brunt of the solution on the youngest and most vulnerable member of the family.  Especially when we have good cause to question the methods being used.  There are gentle alternatives that you can look into (see here for a pretty comprehensive list) and even though they won’t get your child sleeping 12 hours straight at three months (no, that’s not a realistic goal if your baby isn’t doing this normally, even though there will be some babies for whom this is their natural sleep stage; these babies are the rarity, not the norm), they will get you enough sleep.  Furthermore, none of them will give you results overnight – they take time – but given that extinction methods often take much longer and fail at a rather dramatic rate when done at home[17], you probably are saving time in the long-run as you won’t need to do these methods over and over again.

In short, we should not dismiss the influence of cortisol during extinction sleep training even though more research is certainly needed to better understand the totality of effects, the interactions with other variables like temperament and maternal attitudes (early evidence suggests certain children are especially prone to problems associated with ‘less than optimal care’ in this regard; see [2]), and the length of time in which these effects are seen.  Arguably even more importantly, we need more parents to be aware of the period of hyporesponsivity and the behavioural-physiological mismatch their infant may display in order to make them aware of their paramount role in calming and soothing a child that may seem inconsolable (and if you need some help dealing with those crying times, you can read up on some ideas here).  When parents feel like they are helping their child – even if no outward sign is changing – it can ease feelings of parental helplessness.  That is something that I hope everyone can agree on.



[1] Middlemiss W, Granger DA, Goldberg WA, Nathans L.  Asynchrony of mother-infant hypothalamic-pituitary-adrenal axis activity following extinction of infant crying responses induced during the transition to sleep.  Early Human Development 2012; 88: 227-32.

[2] Gunnar MR, Donzella B.  Social regulation of the cortisol levels in early human development.  Psychoneuroendocrinology 2002; 27: 199-220.

[3] Gunnar M, Quevedo K.  The neurobiology of stress and development.  Annual Reviews of Psychology 2007; 58: 145-73.

[4] Gunnar MR. Social regulation of stress in early childhood. In K. McCartney & D. Phillips (Eds.),Blackwell Handbook of Early Childhood Development (pp. 106-125). Malden: Blackwell Publishing, 2006.

[5] Gunnar MR, Larson M, Hertsgaard L, Harris M, Brodersen L.  The stressfulness of separation among 9-month-old infants: effects of social context variables and infant temperament.  Child Development 1992; 63: 290-303.

[6] Bayart F, Hayashi KT, Faull KF, Barchas JD, Levine S.  Influence of maternal proximity on behavioral and psychological responses to separation in infant rhesus monkeys (Macaca mulatta). Behavioral Neuroscience 1990; 104: 98-107.

[7] Bick J, Zhu T, Stamoulis C, Fox NA, Zeanah C, Nelson CA.  Effect of early institutionalization and foster care on long-term white matter development.  JAMA Pediatrics 2015; DOI: 10.1001/jamapediatrics.2014.3212

[8] Hanson JL, Nacewicz BM, Sutterer MJ, Cayo AA, Schaefer SM, et al. Behavior problems after early life stress: contributions of the hippocampus and amygdala.  Biological Psychiatry 2014; DOI: 10.1016/j.biopsych.2014.04.020


[10] Robinson M, Mattes E, Oddy W, Pennell C, van Eekelen A, Mclean N, et al. Prenatal stress and risk of behavioral morbidity from age 2 to 14 years: the influence of the number, type, and timing of stressful life events. Development and Psychopathology 2011; 23: 155-168.

[11] Hinnant JB, El-Shiekh M, Keiley M, Buckhalt JA.  Marital conflict, allostatic load, and the development of children’s fluid cognitive performance.  Child Development 2013; 84: 2003-14.


[13] Petchel P, Pizzagalli DA.  Effects of early life stress on cognitive and affective function: an integrated review of human literature.  Psychopharmacology 2011; 214: 55-70.

[14] Leproult R, Copinschi G, Buxton O, Van Cauter E.  Sleep loss results in an elevation of cortisol levels the next evening.  Sleep 1997; 20: 865-70.

[15] Weinraub M, Bender RH, Friedman SL, Susman EJ, Knoke B, et al.  Patterns of developmental change in infants’ nighttime sleep awakenings from 6 through 36 months of age.  Developmental Psychology 2012; 48: 1511-28.

[16] Hysing M, Harvey AG, Torgersen L, Ystrom E, Reichborn-Kjennerud T, Siversten B.  Trajectories and predictors of nocturnal awakenings and sleep duration in infants.  Journal of Developmental and Behavioural Pediatrics 2014; 35: 309-16.

[17] Loutzenhiser L, Hoffman J, Beatch J.  Parental perceptions of the effectiveness of graduated extinction in reducing infant night-wakings.  Journal of Reproductive and Infant Psychology 2014; DOI: 10.1080/02646838.2014.910864[/fusion_builder_column][/fusion_builder_row][/fusion_builder_container]