A human’s ability to interpret the cries of a baby in pain isn’t innate but learned from experience


Before young children learn to speak, crying is their only means of vocal communication. But do adults know when a baby is in pain as opposed to being mildly uncomfortable?

A new study reported in Current Biology on August 8, 2022 finds that the answer to this question is that it depends.

“We found that the ability to detect pain in cries—that is, to identify a pain cry from a mere discomfort cry—is modulated by experience of caring for babies,” says Nicolas Mathevon, University of Saint-Etienne, France.

“Current parents of young babies can identify a baby’s pain cries even if they have never heard this baby before, whereas inexperienced individuals are typically unable to do so.”

The findings show that humans’ ability to interpret babies’ cries isn’t innate but learned from experience. Parenting young babies shapes our ability to decode the information conveyed by babies’ communication signals.

Mathevon and his University of Saint-Etienne colleagues including David Reby and Roland Peyron made this discovery as part of a broader research program investigating how information is encoded in babies’ cries and how human listeners extract this information. In the new study, they wanted to find out how prior caregiving experience with babies shaped the ability to identify when they were in pain.

They recruited people with different amounts of experience caring for babies, ranging from people with no experience at all to current parents of young children. They also included people with occasional experience babysitting and non-parents with more extensive professional experience in caregiving.

Next, they gave everyone in the study a short training phase in which they heard eight discomfort cries from one baby over a couple of days. Next, their ability to decode the cries as discomfort or pain was put to the test.

And it turned out that experience was everything. People with little to no experience couldn’t tell the difference between cries any better than chance. Those with a small amount of experience performed slightly better.

Current parents and professionals did better than chance. But parents of younger babies were the clear winners. They were able to identify the crying contexts of babies even when they’d never heard the cries of that youngster before. Parents of older kids and those with professional experience didn’t do well with unfamiliar cries.

“Only parents of younger babies were also able to identify the crying contexts of an unknown baby they had never heard before,” says first author of the study Siloe Corvin.

“Professional pediatric caregivers are less successful at extending this ability to unknown babies,” says study co-author Camille Fauchon. “This was surprising at first, but it is consistent with the idea that experienced listeners may develop a resistance that decreases their sensitivity to acoustic cues of pain.”

The findings show that babies’ cries contain important information that’s encoded in their acoustic structure. While adults are attuned to that information, our ability to decode it and identify when a baby is in pain gets better with exposure and experience.

The researchers hope that learning more about how babies communicate pain may help parents learn how to recognize and respond to it even better. They’re now conducting neuroimaging studies to further explore how experience and parenthood shape brain activity when babies cry.

All infants cry to motivate their caregivers to respond to their needs.1

As a result, caregivers tend to interpret a baby crying as a signal of distress or need. Infants follow a predictable cry curve with a peak in intensity at around 6–8 weeks, and persistence after 3 months may be considered pathologic.2 The ability to distinguish pathological cries in infants using acoustic feature extraction and classification algorithms is validated in the literature; 27 prior studies were able to discriminate pathological infant cries (Down’s syndrome, brain damage, Cri du Chat) with an average accuracy rate of 96.9%.3

Acoustic analyses of an infant’s cry could be instrumental in the home setting. Despite caregivers’ best intentions, interpretation of infant cries can be difficult. The perceptions of the listener can be influenced by their sleep habits, mental state, their own physiologic response to the cry, and other sociodemographic factors.4,5 Machine learning could offer an objective assessment of the acoustic features of infant cries to translate their behavioral states.6 This would contribute significantly to infant care by distinguishing if an infant was experiencing pain or if they were responding to another behavioral state (i.e., hunger or being fussy).

It is not only in the home environment that machine learning could aid in infant care. Clinical care and especially hospital settings focus on mitigation of infant pain. Historically, it was believed that infants were incapable of feeling pain.7 However, recent research into the developmental physiology of nociception indicates that the opposite is true. Untreated pain in neonates can leave a lasting neurophysiological footprint associated with decreased brain8,9 and body growth,10 altered neural connections and organization,11,12 poorer cognitive and motor function,13 impaired visual–motor integration, and poorer executive functioning skills.14,15

To assess pain, providers rely upon rating scales such as the Neonatal Infant Pain Scale,16 premature infant pain profile,17 Face, Legs, Activity, Cry, and Consolability scale,18 and Crying, Oxygenation, vital signs, facial Expression, and Sleeplessness scale,19 among others.

Most estimates of inter-rater reliability of infant scales are high16,20,21 with some studies showing poor agreement across these scales in measurements,22,23 suggesting that both clinical factors and the choice of scale may strongly influence the magnitude and the reliability of these pain measurements.

In addition to measurement of pain using subjective infant pain scales, smaller-sample studies have found that infants in pain cry differently from infants who are not experiencing pain—with algorithms showing between 74% and 90% accuracy, as discussed further in the Supplementary Material.

These small-sample algorithms were not portable by nature; this leaves room for a universally applicable machine learning program to help home caregivers and medical providers accurately assess infant cry and determine when the infant is experiencing pain vs. another behavioral state. On the basis of finding a quantitative measure of infant cries, we created a free phone app, ChatterBabyTM, as an accessible and portable algorithm deployment to predict whether a baby’s cry was due to one of the three behavioral states: pain, hunger, or fussiness.

The algorithms were then applied to infant cries where parents reported their infants as having colic. This process simulates an initial clinical visit where the parent has complaints of colic and a workup for conditions like reflux esophagitis or infantile migraine may be initiated and diagnosed. We hypothesized that colic cries would be acoustically similar to pain cries, a finding that would explain and validate caregiver distress regarding caring for an infant with colic.

reference link : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7033040/

Original Research: Open access.
Adults learn to identify pain in babies’ cries” by Nicolas Mathevon et al. Current Biology


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.