A University of Michigan study finds that romantic partners who spend a weekend caring for a lifelike infant simulator together experience a notable drop in testosterone when they repeat the caregiving task on Monday. The research, conducted by doctoral student Rachel E. Brandon and psychology professor Robin S. Edelstein, suggests that hands-on practice and pre-existing attitudes toward parenting shape how couples biologically respond to shared childcare.
Testosterone, produced in both men and women, fluctuates with social context and is linked to trade-offs between mating and nurturing behaviours. While higher levels are often tied to competition and status seeking, lower levels tend to align with bonding and cooperative caregiving. Previous studies have shown such hormonal shifts occur in real or simulated parenting scenarios, including short-term declines after soothing a crying baby. The new study extends this line of inquiry to couples without children who care for an infant together.
The researchers recruited 30 childless couples, 60 individuals aged 20 to 40, who had lived together for at least a year and planned to start a family within five years. About 13 per cent of the couples were same-gender. Participants completed online surveys about their comfort around babies and their anticipated parenting abilities before visiting the lab.
The team used the RealCare Baby infant simulator, which looks and behaves like a real newborn and cries at intervals while requiring feeding, burping, rocking and changing. Internal sensors track how caregivers attend to the doll, while wristbands record each partner’s level of participation. The design allowed the researchers to control the infant’s behaviour and observe responses to a novel caregiving experience.
Couples attended a Friday lab session to learn how to care for the doll and then spent 10 minutes caring for it together in a room as it cried and demanded care. Saliva samples were taken just before and 10 minutes after the task to capture short-term testosterone changes, which typically reflect hormonal responses about 20 minutes after an event.
After the Friday session, the doll accompanied the couple home for a weekend of joint caregiving, with half the participants experiencing disrupted sleep (the doll woke three to six times a night) and the other half a quiet weekend. On Monday, participants returned to the lab for the same 10-minute joint care task and another saliva sample. In total, 28 couples completed both lab sessions.
Across the first lab visit on Friday, testosterone did not change significantly: men showed an average increase of 0.60 per cent and women a decline of 3.29 per cent, with the overall average not differing from zero. By Monday, the pattern reversed: testosterone fell on average by 9.09 per cent in men and 11.53 per cent in women, yielding an overall average drop of about 10 per cent, compared with roughly a 1 per cent decline on Friday.
Researchers found the Monday declines occurred across both sexes and were independent of whether the weekend involved sleep disruption. The pattern aligns with earlier work by Edelstein and colleagues, which showed short-term drops in testosterone after soothing a real infant during a stressful task, though the prior research involved actual parents rather than childless couples using a simulator.
To probe why the Monday drop occurred, the team examined weekend data from the doll’s computer and participants’ reported experiences. They found that couples generally performed well at home, with an average effectiveness score around 81 out of 100 and about 26 caregiving check-ins per person. Yet neither objective performance nor self-reported stress or weekend enjoyment explained the Monday hormonal shift. Interestingly, participants whose testosterone rose during the Friday session tended to describe the weekend as less stressful, a finding the authors suggest could reflect a sense of challenge rather than relief.
Brandon and Edelstein emphasised that most laboratory studies of hormonal responses do not track changes across repeated sessions, and they were surprised by stronger hormone declines on Monday than on Friday. “This difference might suggest that something about the novelty of the task, especially among people who do not yet have children, contributed to variability in hormone changes in the first session,” they said. They also noted that sleep disruption did not appear to affect the hormonal outcomes, and that neither how well couples cared for the doll nor each partner’s contribution to caregiving predicted the Monday decline.
Attitude toward parenting and comfort with infants appeared to modulate the pattern. The researchers reported that individuals who entered the study more comfortable around babies or who placed a higher personal value on becoming a parent tended to show a drop in testosterone during both sessions. Conversely, those less comfortable with babies showed no Friday decline but did experience a substantial drop by Monday. A similar interaction emerged with confidence about future parenting, though it did not reach statistical significance in all analyses. The authors cautioned that the moderation patterns weakened when raw hormone levels were analysed rather than percent changes.
“How people responded the first time depended on how comfortable they felt around babies and how much they valued parenting, but those who started out less comfortable caught up after just one weekend,” the authors explained. “Our findings also suggest that the ‘new parents’ in our study adapted to caregiving relatively quickly, as indicated by their hormone responses in the second session.”
The researchers also stressed that hormonal shifts should be viewed in the context of natural biological variation and common misconceptions about what lower testosterone signifies. “People’s testosterone levels differ from one another much more than any one person’s levels change over time,” they noted. “So the declines in testosterone that we observe in our studies are relatively small in magnitude and not likely detected by individuals themselves, but they may nevertheless have real-world effects on behavior.”
They added that while men continued to have higher overall testosterone levels, the percentage changes were similar across sexes, and “low testosterone” should not automatically be regarded as problematic, particularly in caregiving contexts.
Limitations include the relatively small, predominantly white, middle-class sample and the fact that about half of the female participants used hormonal contraceptives. The analyses were exploratory and not preregistered, and the researchers did not determine in advance how large a sample would be needed to detect effects. They cautioned that the study used a simulated infant rather than a real one, so applicability to actual parenting remains to be tested in future work.
As a caveat, there was no non-intervention comparison group that carried out the two lab tasks without weekend practice, making it difficult to attribute the Monday drop solely to weekend caregiving. The researchers acknowledge this and suggest future studies compare home caregiving with brief lab interactions to isolate the effect of hands-on practice from mere familiarity. They also propose measuring participants’ momentary sense of capability and exploring larger, more diverse samples, and perhaps tracking additional hormones such as cortisol to examine endocrine synchrony within couples during caregiving.
“We’re also interested in the extent to which couple members’ hormones may be correlated, or synchronized, during different kinds of interactions,” Brandon and Edelstein said. “Are partners’ testosterone levels more likely to rise and fall together during caregiving interactions, for instance, compared to laboratory stressors? How might other hormones, such as cortisol, change during caregiving, and are couple members’ cortisol responses synchronized?” Cortisol is a hormone involved in the body’s stress response.
The study, titled “Testosterone Changes in Couples in Response to a Caregiving Task with an Infant Simulator,” was authored by Rachel E. Brandon, Annika From, Miranda Reynaga, Helen Devine, Amie M. Gordon, and Robin S. Edelstein.
