This is the research companion to Connection Is Not an Emotion. It Is a Physiology.. Read the main post first.
The post argues that connection is at least three things at once -- a feeling, each adult’s own neurochemistry, and the coordination between the two adults -- and that the developmental literature only saw the fuller picture once it stopped measuring mothers alone. These five studies elucidate that argument. Two of them go to the mechanism: a father’s brain reorganizes around the caregiving he actually does, and the chemistry of forming a new bond runs on more than oxytocin. The other three follow the mechanism downstream -- the father-child bond predicts a young child’s behavior about as strongly as the mother-child bond; a father’s early presence is still legible in his grown son’s stress biology decades later; and the two parents’ mental health rises and falls together, which makes the couple, not the individual, the right unit of concern. Some anchor the claim. Some extend it.
When the Father-Attachment Numbers Finally Came In, They Looked Like the Mother Numbers
Deneault, A.-A., Bakermans-Kranenburg, M. J., Groh, A. M., Fearon, P. R. M., & Madigan, S. (2021). Child-father attachment in early childhood and behavior problems: A meta-analysis. New Directions for Child and Adolescent Development, 2021(180), 43-66.
DOI: 10.1002/cad.20434
What they found: This was the first meta-analysis dedicated specifically to child-father attachment and behavior problems. Across 15 samples (N = 1,304 dyads), insecure child-father attachment in early childhood was associated with more externalizing behavior, a moderate effect (r = .18, d = .37). Across 12 samples (N = 1,073), it was associated with more internalizing behavior, a small effect (r = .09, d = .17). The headline was the comparison: when the authors placed the father effects next to the established mother-attachment meta-analyses, the confidence intervals overlapped -- father d = .37 versus mother d = .31 for externalizing; father d = .17 versus mother d = .15 to .19 for internalizing. Statistically, the two relationships predicted child behavior at indistinguishable strength.
Why this matters for you: For decades the assumption baked into parenting science was that the father-child bond was a fainter copy of the mother-child one. The data say otherwise: an insecure attachment to dad carries about the same weight for how a young child’s behavior goes as an insecure attachment to mom. If you are a father, this is the quantitative version of the post’s point -- your bond is not decorative or secondary, it is load-bearing. It also means that programs and pediatric conversations that include only mothers are leaving half the attachment network unexamined.
What it doesn’t answer: This is correlational, so it shows the relationship matters, not that fixing it guarantees a different outcome. Every sample was Western, middle-class, and overwhelmingly heterosexual two-parent, so it does not speak to other family structures. And “comparable effect sizes” does not mean “identical mechanism” -- how an insecure father bond confers risk may run through different channels than the mother bond.
When Fathers Did the Primary Caregiving, Their Brains Looked Like Mothers’
Abraham, E., Hendler, T., Shapira-Lichter, I., Kanat-Maymon, Y., Zagoory-Sharon, O., & Feldman, R. (2014). Father’s brain is sensitive to childcare experiences. Proceedings of the National Academy of Sciences, 111(27), 9792-9797.
DOI: 10.1073/pnas.1402569111
What they found: This study scanned the brains of 89 first-time parents in three groups -- primary-caregiving mothers, secondary-caregiving fathers, and primary-caregiving gay fathers raising an infant with no mother involved -- using fMRI alongside oxytocin levels and coded video of each parent with their baby. Parenting ran on two brain networks: an emotional-processing network (the amygdala and other oxytocin-rich subcortical structures tied to vigilance and reward) and a socio-cognitive network (the superior temporal sulcus and prefrontal regions tied to reading the infant’s signals). The primary-caregiving mothers showed stronger activation in the emotional network; the secondary-caregiving fathers showed stronger activation in the socio-cognitive one. The primary-caregiving gay fathers engaged both at once -- amygdala activation like the mothers, socio-cognitive activation like the fathers -- plus functional connectivity linking the two. And across all the fathers, the more hands-on time a father spent in direct childcare, the stronger that connectivity ran (r = .33). A standard masculinity-femininity inventory found no difference between gay and straight fathers, so the pattern tracked the caregiving role, not sexual orientation. Oxytocin levels did not differ across the three groups.
Why this matters for you: This is the de-gendering claim from the post, located in the brain. The parental brain is not a fixed maternal circuit that fathers run a fainter copy of. It is built by the caregiving a parent actually does. A father doing the hands-on work activates the same emotional machinery long assumed to belong to mothers, and the more of that work he does, the more tightly the two systems wire together. The lever is doing the care, not being born to it -- which makes “either parent can serve this function” a statement about biology, not only about fairness.
What it doesn’t answer: The primary-caregiving group here was gay fathers, so the mirror-image question -- whether two mothers’ brains reorganize the same way -- is still open. The link between hands-on time and brain connectivity is correlational: it may be that more attuned fathers also choose to do more of the care, rather than the care alone doing the wiring. And this is brain and hormone, not a child outcome. It shows the parent’s brain adapts; it does not show what that adaptation produces in the child.
Oxytocin Gets the Headlines. It Wasn’t Working Alone.
Ulmer-Yaniv, A., Avitsur, R., Kanat-Maymon, Y., Schneiderman, I., Zagoory-Sharon, O., & Feldman, R. (2016). Affiliation, reward, and immune biomarkers coalesce to support social synchrony during periods of bond formation in humans. Brain, Behavior, and Immunity, 56, 130-139.
DOI: 10.1016/j.bbi.2016.02.017
What they found: Researchers drew blood from 189 adults in three groups -- new parents in the first four to six months with a firstborn, new lovers in the first months of a romance, and unattached singles -- and measured three signals at once: oxytocin (the affiliation system), beta-endorphin (the body’s own reward opioid), and interleukin-6 (an immune and stress marker). They also micro-coded the moment-to-moment behavioral synchrony in each pair. Oxytocin rose during both parental and romantic bonding and was highest in the new lovers. Beta-endorphin and interleukin-6 were highest in the new parents and lowest in the singles. During bond formation the three systems became more tightly coupled, and in a path model the effect of beta-endorphin and interleukin-6 on synchrony ran through oxytocin -- the oxytocin system was the hub the other signals worked through.
Why this matters for you: Bonding is not a single-hormone event. Forming a new attachment, whether to a baby or to a partner, brings a small crew of chemical systems online together, and they pull in the same direction. Beta-endorphin, the body’s own feel-good opioid, was elevated in new parents -- part of why the early months can feel quietly rewarding even when they are exhausting. For the post’s claim that connection is a physiology and not only a feeling, this is the receipt: the warm pull toward a new baby is the affiliation, reward, and stress systems coming online at once, with oxytocin at the center.
What it doesn’t answer: This study looked at new parents and new lovers in general. It was not about fathers specifically, and it was not about play. It did not test the idea that fathers bond through rough-and-tumble play by way of a beta-endorphin release. That step is a synthesis by evolutionary anthropologist Anna Machin, who reads this bonding literature alongside the father-play research to argue that beta-endorphin is the chemical that carries the long-term father-child bond (The Life of Dad). It is the same synthesis the post’s beta-endorphin claim rests on, and we treat it the same way in both places: the best explanation available, not the final word. The study itself is also correlational and cross-sectional -- a single snapshot, not a before-and-after -- and the sample was healthy, educated, and mostly in established relationships.
A Father’s Presence at Age 10 Showed Up in His Son’s Cortisol at 37
Choi, J., Kim, H. K., Capaldi, D. M., & Snodgrass, J. J. (2021). Long-term effects of father involvement in childhood on their son’s physiological stress regulation system in adulthood. Developmental Psychobiology, 63, e22152.
DOI: 10.1002/dev.22152
What they found: Using the Oregon Youth Study, which followed at-risk boys from high-crime neighborhoods starting in 1984, researchers measured how much fathers shared activities with their sons at ages 9 to 11, then collected the sons’ diurnal (across-the-day) cortisol nearly three decades later, around age 37. The quantity of shared father-son activities predicted a healthier cortisol rhythm in adulthood -- higher morning levels and a steeper daily decline, the hallmarks of a well-regulated stress system (intercept beta = .22; slope beta = -.26). This held after controlling for childhood and adult income and for the men’s current relationship with their fathers. Notably, the observer-rated quality of the father involvement was not a significant predictor; in this sample, showing up and doing things together was what registered.
Why this matters for you: This is the deepest extension of the post’s “connection is a physiology” thesis: the same lesson Aytuglu found in children’s blood markers, but stretched across an entire lifespan and located in the father-child relationship. The machinery that governs how this man handles stress at 37 was partly tuned by whether his dad showed up to play games, talk, and take trips when he was ten. The quantity-over-quality result is also quietly freeing: for a father worried about doing it “right,” the bigger variable here was simply being there, consistently.
What it doesn’t answer: This is sons only, roughly 90 percent European American, and an at-risk sample, so it cannot speak to daughters or to lower-risk families. There was no cortisol measured between childhood and 37, so the path the effect took is unknown. And the quality “null” almost certainly reflects a brief 1980s lab task that failed to capture everyday relational quality, not real proof that quality is irrelevant.
One in Ten New Fathers Is Depressed -- and It Tracks With the Mother
Paulson, J. F., & Bazemore, S. D. (2010). Prenatal and postpartum depression in fathers and its association with maternal depression: A meta-analysis. JAMA, 303(19), 1961-1969.
DOI: 10.1001/jama.2010.605
What they found: Pooling 43 observational studies across 16 countries (28,004 participants), this meta-analysis found that roughly 10.4 percent of fathers experienced depression somewhere between the first trimester and the baby’s first birthday, and that the rate peaked at 25.6 percent in the 3-to-6-month postpartum window. Paternal and maternal depression were moderately and significantly correlated (r = .308): when one parent was depressed, the other’s risk was meaningfully elevated. The authors’ clinical conclusion was that depression in one parent should prompt clinical attention to the other, and that screening should be framed at the couple level rather than treating each parent as an independent risk unit.
Why this matters for you: This is the partnership angle the post flagged -- that the coordination between the two parenting adults registers separately in the child’s world -- viewed through mental health. The two adults are not independent emotional weather systems; they move together. For a family, that reframes a depressed parent as a couple-level signal, not a private problem, and it makes a strong case that fathers deserve perinatal screening that the system overwhelmingly still aims only at mothers. The 3-to-6-month peak also lines up with the post’s point that the father’s bond often consolidates in exactly that window.
What it doesn’t answer: The maternal-paternal correlation is associational -- it shows the two co-occur, not which one drives the other (shared stressors, assortative mating, and partner effects are all live possibilities). Heterogeneity across studies was very high, so the pooled rates carry wide uncertainty. And the samples skewed toward middle-income, two-parent, health-service-recruited families, so rates among separated, low-resource, or non-cohabiting fathers may differ.
Coming Up
Our next post, “The Architecture of Choice,” explores why self-regulation around screens and food is far less about a child’s willpower and far more about how the surrounding environment is designed.