Testosterone, cortisol, and the neural response underlying paternal protection in the prenatal period
Winona Mulder (10997865)
Psychobiology, University of Amsterdam Martine Verhees
Annemieke Witte 19-06-2020
Abstract
Parental caregiving is a crucial element in the survival and development of young children. Certain aspects of parental behaviour and quality of care have been associated with steroidal hormones such as testosterone and cortisol. Parental protection, a facet of parental behaviour, is important for child outcomes, higher survival rates and a better quality of life. Testosterone levels decrease when men become fathers. Focussing on expectant fathers, this study will investigate the association of the neural response underlying paternal protection with basal levels of testosterone and cortisol in the prenatal period. Salivary testosterone and cortisol samples were collected. While in magnetic resonance imaging machine, expectant fathers (N=19) watched videos showing an infant in danger and matched neutral videos, while instructed to imagine that the infant was their own child. An effect was found for infant-threatening vs neutral situations in the amygdala (region of interest). However, no correlations were found for testosterone or cortisol levels and amygdala reactivity.
Keywords: Paternal protection, testosterone, cortisol, amygdala reactivity, fMRI, father
Introduction
Young children are dependent on carethey receive from others. Care refers to the behaviours and practices of caregivers to provide the food, health care, stimulation and emotional
support necessary for children’s healthy survival, growth and development. The quality of care can have a powerful effect on a child’s cognitive and social-emotional development (Richter, Griesel and Manegold, 2004). Parental care consists of behavioural aspects such as sensitivity and responsiveness. These aspects are defined together as “the ability to
accurately perceive and interpret the infants’ attachment signals, and to respond to them promptly and adequately” (van IJzendoorn, Juffer and Duyvesteyn, 1995). The majority of studies on parental caregiving are actually researching parental sensitivity, measured for example by encoding behaviour of expectant parents interacting with an inconsolable infant doll, or of parents interacting with their own infant (Bos et al., 2018). In this paper, another facet of parental care will be discussed: parental protection.
Although parental protection has been proposed to be important for child outcomes, granting higher survival rates and a better quality of life, even in modern societies
(Alyousefi-Van Dijk et al., 2019), studies on parental protection are scarce (Bakermans-Kranenburg and van IJzendoorn, 2017). Some behaviours that have been proposed to relate to protection are prenatal pathogen avoidance in the form of morning sickness, using the tend-and-befriend strategy during interactions with conspecifics, putting a child’s needs before your own and protecting a child from life’s difficulties (Taylor et al., 2000; Hancock,
Lawrence and Zubrick, 2014; Bakermans-Kranenburg and van IJzendoorn, 2017). These few studies have mostly involved mothers, even though there is evidence that child deaths were two times higher in conservative societies when the father was absent (Hurtado and Hill, 1992). Although these effects may be smaller in modern society, they emphasize the
possibility that fathers also develop the neurobiological tools and instinctive actions to protect their offspring (Rilling and Mascaro, 2017). Furthermore, these neurobiological tools could influence and relate to traditionally male protective behaviours, e.g. providing (money for) food and shelter, and using the more aggressive tend-and-defend strategy during interactions with conspecifics (Taylor et al., 2000; Bakermans-Kranenburg and van IJzendoorn, 2017).
Parenting behaviour in humans has regularly been associated with hormones such as vasopressin (AVP), testosterone (T) and cortisol (CORT) (Mills-Koonce et al., 2009; Cohen-Bendahan et al., 2015; Edelstein et al., 2017). The primary function of AVP in the body is to regulate extracellular fluid volume. However, similar to T, AVP is also associated with male reproduction and male-typical social behaviours such as aggression and territoriality
(Donaldson and Young, 2008). In addition, AVP is suggested to be strongly related to paternal parenting behaviour, more strongly than to maternal parenting (Alyousefi-Van Dijk et al., 2019). When AVP was administered through a nasal spray, fathers-to-be spent more time watching baby-related avatars, compared to control avatars, however endogenous AVP concentrations were not related to caregiving interest in expectant fathers (Cohen-Bendahan et al., 2015). AVP dependent paternal brain activations and hormonal responses have been suggested to underlie fathers’ ability to interpret others’ intentions in order to accurately defend offspring (Thompson et al., 2006; Atzil et al., 2012). By stimulating selective threat perception, AVP may facilitate protective behaviour in human fathers. This suggests that
AVP might be a relevant endocrine factor associated with protective parenting. However, other hormones, such as T and CORT, may be relevant as well.
Testosterone is the primary male sex hormone and anabolic steroid. T levels affect everything from the reproductive system to muscle mass and bone density in men. This hormone plays a role in certain male-typical behaviours (Birger et al., 2003; Mehta et al., 2017). Levels of T decrease in men when they become fathers and are negatively associated with quality of caregiving in the prenatal and postnatal periods (Gettler et al., 2011;
Weisman, Zagoory-Sharon and Feldman, 2014; Bos et al., 2018). Similarly, men with lower T levels are more sympathetic and more responsive to infant cries than men with higher T levels (Fleming et al., 2002). While inhibiting direct forms of caregiving such as feeding, cleaning and holding, higher levels of T could actually promote indirect forms of caregiving such as provisioning and aggressive defence of child when threatened or in danger (Van Anders, Tolman and Volling, 2012; Rilling and Mascaro, 2017).
Cortisol is a steroid hormone that regulates a wide range of processes throughout the body and has a very important role in helping the body respond to stress. Maternal salivary CORT was negatively associated with maternal sensitivity observed during parent-child interactions in the first two postpartum years (Finegood et al., 2016). However, equivalent changes in CORT across both sexes are associated with very different behavioural
consequences for men and women (Sherman et al., 2017). Bos et al. found a negative
association between CORT and the quality of care in prenatal and postnatal periods, but only in fathers, not in mothers. A paternal responsiveness study found that the CORT levels decreased more during a 30 min free play with their child for fathers who had spent more time alone with their toddler that day (Storey et al., 2011). In all, existing research suggests that T and CORT are associated with parenting. However, to my knowledge, these hormones have not yet been researched in relation to protective parenting. Focusing on expectant fathers, the current study will investigate association of the neural response underlying paternal protection (PP) with basal levels of T and CORT in the prenatal period.
Higher basal CORT levels are related to stronger negative functional connectivity with specific regions in the medial prefrontal cortex (mPFC), even under relatively stress-free circumstances (Veer et al., 2012). This suggests CORT-mediated regulation of the amygdala by the mPFC. In addition, research into the feedback and feed-forward mechanisms of glucocorticoids showed that the amygdala, being a stress-excitatory region, may in fact be activated by CORT (Herman et al., 2012). Higher endogenous T levels are related to
increased brain activation in the amygdala during encoding of neutral pictures (Ackermann et al., 2012). This supports the idea that CORT and/or T levels may relate to amygdala
activation under neutral, stress-free circumstances. That is where the current study will differ from previous research. Prior studies have not investigated relations between hormone levels and the neural processing of situations where protective behaviour was necessary. However, a recent study, using data from the same pilot study as the current study, explored the neural basis for paternal responses to threat to infants, where participants watched videos showing an infant in danger, while instructed to imagine that the infant was their own or someone else’s (Van’t Veer et al., 2019). This study found evidence that infant-threatening situations, compared to neutral situations, elicited more activation in the amygdala of expectant fathers. This strengthens the idea that fathers allotted affective salience to situations where their child
is in danger and suggests that the amygdala may be implicated in the identification of threats, after which the father can take action to protect their child (Van’t Veer et al., 2019). In consequence of these findings, the bilateral amygdalae will be the region of interest (ROI) for the present study.
To summarize, the present study will investigate the association of expectant fathers’ amygdala response to a necessity of protective behaviour with basal levels of T and CORT in the prenatal period. We will simulate the necessity of protective behaviour with an assortment of infant-threatening videos and neutral videos matching each situation (the same videos used by Van’t Veer et al.). Firstly, in line with prior research, we expect the infant-threatening videos to induce a higher activation in the amygdala, compared to the neutral videos.
Secondly, we expect participants with higher basal T and higher basal CORT concentrations, to possess more amygdala activity during the infant-threatening videos, compared to the neutral non-threatening videos.
Method
Participants
Participants were recruited through midwives and online ads. Pre-selection criteria included living together with their pregnant partner, fluency in Dutch and good health, without (using medication for) psychiatric, neuroendocrine or neurological disorders. A screening was done to exclude participants with claustrophobia, metal parts or shards in the body, excessive smoking and/or alcohol use, recreational drug use within 6 months prior to participating and use of any other interfering medications.
Twenty-five first-time expectant fathers participated in the study. Due to technical issues, data of four participants was unusable. The current study included a randomly selected subset of 19 expectant fathers (Mage= 31, SDage= 0.94, range= 24-40 years) in the final analysis. The
mean foetal age of the child was 27.8 weeks (SD= 1.2, range= 20.4-36.1 weeks). All participants provided written informed consent for participation and the use and storage of data.
Procedure
Participants were told to abstain from alcohol and excessive physical activity during the 24h before the start of the session and from caffeine on the day of the session. After receiving general information about the study, participants collected at least 1.5ml saliva by salivating down a straw into a test tube. The sample was immediately stored in a refrigerator at -20 degrees Celsius until laboratory assessment. Participants were instructed to self-administer a nasal spray containing a placebo (Chlor-butanol solution) (Witte et al., 2019). Subjects were informed about what to expect during the MRI. A brief training of the task was done on a laptop to familiarize participants with the task and the infant pictures that would be used to aid the imagination of the child’s identity (own vs unknown), in this study only the task blocks with ‘own’ infant pictures were used.
After saliva was collected for the second time (this sample was not used in the current study), participants were positioned in the scanner. Earplugs and headphones were given to reduce scanner noise and for communication between scans. Their head was fixated in the coil using small cushions, and they were asked to lay as still as possible. Functional
MR-images were acquired during the performance of a few different tasks, of which only the ‘Threat to infant’ task was used in this study (Van’t Veer et al., 2019). Total scanning time was approximately 1 hour.
After the MRI, participants were guided through several behavioural tasks, and collected saliva for the third time (both the behavioural tasks and this third saliva sample were not used in the current study). Lastly, an application was installed on their smartphone to ask questions about involvement with their (un)born child during the months after the session. The total length of the experimental procedure was approximately 3 h. Time of day of the experimental session varied between 0900 h and 2200 h. Subjects received financial rewards of €30 per visit and €10 if the participant and their partner completed 80% of the online questionnaires at home (via the app).
This study was part of a larger study where, in order to investigate the role of
hormones in paternal behaviour, participants self-administered AVP intranasally during two other laboratory visits; a second pre-natal visit and one post-natal visit (Thijssen et al., 2018). The full study procedure was approved by the ethics committee of the Leiden University Medical Centre (LUMC).
Measures
fMRI paradigm. To evoke the neural process of perceiving an infant in need of protection,
participants watched threatening and neutral videos (threat factor) while imagining that either their own, or someone else’s infant (familial factor) was shown in the videos. To aid the imagination of their own (unborn) infant in the videos, a morphed picture was added to the start of the video task. In order to create a likeness of the father’s own (unborn) child, a morphed image was created from a photograph of the participant (25%) combined with an average baby face (75%), using the program Fantamorph (Van’t Veer et al., 2019). As stated earlier, only the threat factor (threat vs. non-threat) to the participants own infant (morphed picture) was used in this study (see Figure 1). This resulted in 2 conditions, each repeated 4 times, making 8 blocks overall. Each block consisted of three consecutive 6 s videos.
Participants were randomly assigned to one of the pre-programmed block orders. Before a given block, the morphed picture of their own infant was shown accompanied by the instructions to imagine that this was the infant in the following videos. After 8 s (and a brief stimulus interval of 250 ms), the instruction screen progressed to a block of three threatening (or neutral) videos that each lasted 6 s, resulting in a block length of 18 s (see Figure 1).
Videos were semi randomly selected out of a pool of six threatening videos (e.g. hot tea is accidentally dropped on a baby, an adult loses grip of a baby stroller that rolls off a bridge and crashes into a cyclist and a baby accidentally falls off a changing table while being changed) and six matched neutral videos (e.g. tea is placed on a table next to the baby, an adult on top of the bridge safely puts baby stroller on the brakes and a baby lies on the
changing table while being changed). These videos contrast situations where protective action by the parent is called for with those where it is not. The videos were filmed by a professional production team, using a lifelike baby doll. Depiction of the doll’s face or that of any actors was minimized to ease the task for participants of imagining their own infant in the videos. After the first four blocks, participants were asked to report on a visually presented scale their feelings of arousal [‘How calm or tense do you feel right now?’, calm (0)-tense (100)],
valence [‘How positive or negative do you feel right now?’, negative (0)-positive (100)], and how well they succeeded in imagining their own infant in the preceding block of videos [very poorly (0)-very well (100)] (see Figure 1).
Figure 1. fMRI paradigm designed to asses neural processing of own infant in threatening vs neutral situations. All stimuli were placed on a black background and were presented via a beamer that projected onto a screen placed in the back of the MRI bore, visible to participants via a mirror on the head coil (adapted from Van’t Veer et al., 2019).
Salivary hormone samples. T and CORT concentrations were measured via a saliva sample
collected with a passive drool collection tube. The baseline saliva sample was taken at the start of the laboratory session, before self-administering the (placebo) nasal spray (Witte et al., 2019). T and CORT samples were analysed by Dresden LabService GmbH. After
thawing, samples were centrifuged for 10 min at 4000 rpm and concentrations of salivary free CORT and T were measured using high performance Liquid Chromatography-tandem Mass Spectrometry in combination with on-line solid phase extraction (Gao, Stalder and
Kirschbaum, 2015).
Saliva samples were collected at different times of the day (as participants did not all visit LUMC in the morning or evening). Therefore, both the CORT and T data was checked for a covariance of time. According to Dabbs (1990), male mean salivary T concentrations (ng/dL) drop about 50% from morning (07:00) to evening (22:00), with the largest drops early in the day. A different study claimed serum T levels were 20-25% lower at 16:00 than at 08:00, in men 30-40 years old (Brambilla et al., 2009). A calculation applied to the data from Dabbs (1990) showed a 34% drop in salivary T concentration between 08:00 and 16:00. This is still a drop of about 10 percent points more than Brambilla claimed. Interestingly, both studies claimed that T concentrations dropped more during the first half of the day, compared to the second half, suggesting a non-linear relation between T levels and collection time (Dabbs, 1990; Brambilla et al., 2009). CORT levels vary throughout the day but are also generally higher in the morning after waking up and then fall throughout the day, this is called a diurnal rhythm (Nicolson, 2004; Ljubijankić et al., 2008; Sjörs, Ljung and Jonsdottir,
2014). In response to stress, extra CORT is released to help the body respond properly. Figure 3 depicts the general diurnal rhythm for salivary CORT (Sjörs, Ljung and Jonsdottir, 2014).
Figure 3. Diurnal salivary cortisol in normal healthy participants (male and female, ages 25-50) (adapted from Sjörs, Ljung and Jonsdottir, 2014).
Based on the research discussed above, there was an expectation of a non-linear negative association between time of saliva collection and the concentrations of both T and CORT. To check for evidence of non-linear covariance, a residuals plot was analysed. However, the sample sizes in this study were too small to justify higher order curve fitting. Thus, a Pearson correlation analysis was conducted. Before analysis, collection time (e.g. 11:00) was transformed into the difference between collection time and a baseline time (08:00) in minutes (e.g. 180 m). This new variable was the time variable used to check for a covariance of time with hormone measurements. Both the new time variable and the hormone measures must be normally distributed, otherwise a Spearman correlation analysis was performed. If a linear association was found, time-correction was done by using the
unstandardized residuals as new data for the hormone measures. If no significant association was found between T and/or CORT concentrations and the time of sample collection, no time-correction was necessary.
fMRI data acquisition and processing
All acquisition and processing of fMRI data is the same as in Van’t Veer et al., 2019. Images were obtained on a 3T Philips Achieva MR unit equipped with a 32-channel SENSE
(Sensitivity encoding) head coil, at the Leiden University Medical Centre. A block design with 295 T2*-weighted whole-brain echoplanar images was used for the functional scan, TR (repetition time) = 2200 ms. Scan duration was 11 min and we used the following acquisition parameters: TE (echo time) = 30 ms, flip angle = 80°, FOV (field of view) in mm = 220 × 220 × 115 (RL, AP, FH, respectively), voxel size: 2.75 × 2.75 × 3.025 mm with a 10% interslice gap, 38 transverse slices. To allow for steady-state tissue magnetization, functional volumes 1-5 were dummies.
Processing of images was performed using FEAT (FMRI Expert Analysis Tool) Version 5.0.8, part of FSL (FMRIB’s Software Library). The boundary-based registration algorithm (Greve and Fischl, 2009) was used to register the functional data to the high-resolution structural image of each participant. Registration of the high-high-resolution structural image to standard space was done using FLIRT (FMRIB’s linear image registration tool) (Jenkinson et al., 2002). Linear registration was applied because non-linear registration did not perform well for some participant data (Van’t Veer et al., 2019). Motion correction was done using MCFLIRT (motion correction FLIRT) (Jenkinson et al., 2002). Non-brain tissue removal was carried out using BET (Brain Extraction Tool) (Jenkinson et al., 2002).
Volumes were smoothed using a 5 mm full-width-at-half-maximum Gaussian kernel and the entire 4D dataset was grand-mean intensity normalized by a single multiplicative factor. In addition, a 90 s cut-off high-pass filter was added (Gaussian-weighted least-squares straight line fitting, sigma = 45.0 s). Time-series statistical analysis was carried out using FILM (FMRIB’s improved linear model) with local autocorrelation correction (Woolrich et al., 2001), temporal derivatives and double-Gamma HRF convolution for the regressors of interest. Regressors were the onsets of blocks of three videos belonging to the two conditions (threat vs neutral). To check for any confounding factors, the following were added as
extraneous variables (EVs): presentation of the pictures, in-scanner questions and movement [motion parameters and additional pre-computed motion outliers] (Van’t Veer et al., 2019). ROI (region of interest) analysis
A ROI analysis was conducted for the bilateral amygdalae. Mean Z-values averaged over all weighted voxels within the bilateral amygdalae were extracted for each participant and condition (threat vs. neutral), and were subsequently analysed in SPSS.
Statistical analysis
First, a dependent t-test was done among the mean amygdala z-values of both conditions (threat and neutral). If assumptions were not met, a related-samples Wilcoxon signed rank test was conducted instead. Second, the participants’ amygdala values for the two conditions, infant-threatening and neutral videos, were transformed into one amygdala difference score variable (mean z-values threat – mean z-values neutral). If assumptions were met, a Pearson correlation analysis was conducted among the amygdala difference scores and (possibly time-corrected) T and CORT concentrations. If assumptions were not met, a Spearman correlation analysis was used instead.
Results
Effect of threat on brain activation
A Shapiro-Wilk’s test, a visual inspection of the histograms, normal Q-Q plots and box plots showed that the mean amygdala values were not normally distributed for the
infant-threatening video condition (p=.03). However, the mean amygdala values were approximately normally distributed for the neutral video condition (p>.05). A Related-Samples Wilcoxon Signed Rank test (p=.04) showed that the median of differences between
threatening and neutral video conditions was not equal to 0. This indicates that there is an effect of threat on amygdala activation.
Brain activation difference related to endogenous hormone levels
Salivary T. As stated earlier in the method section, a new time difference variable was made,
the difference (in minutes) between the standard time (8:00) and the time of sample collection (e.g. 11:00). After transforming the left skewed distribution of time difference using the natural logarithm, a visual inspection of the histogram, normal Q-Q plot and box plot showed that the time difference data was approximately normally distributed, however the Shapiro-Wilk’s test said otherwise (p=.04). After transforming the T concentrations using the natural logarithm, a Shapiro-Wilk’s test (p>.05) and a visual inspection of the histogram, normal Q-Q plot and box plot showed that the salivary T concentrations were approximately normally distributed. Inspection of the residuals plot showed no pattern forming and the regression coefficients were significant (p=.01, R2= .31), supporting a linear association between T
concentrations and sampling time. Considering the transformed time difference data did not meet the assumption of normality, a Spearman correlation analysis was conducted. The Spearman test showed a significant negative correlation between T and time, r(17)= -.54, p=.018. This indicates that the later in the day saliva is collected, T concentrations tend to be lower. To correct for this, the unstandardized residuals from the linear regression analysis were used as corrected T data. A second Shapiro-Wilk’s test showed that the time-corrected T data was normally distributed (p>.05). As stated in the method section, a second new variable was made called amygdala difference; the difference between mean z-values of the amygdala while participants watched infant-threatening videos and neutral videos. A Shapiro-Wilk’s test showed that the amygdala difference data was approximately normally distributed (p= 0.66). As stated above, the time-corrected T data was also normally
distributed. Therefore, a Pearson correlation analysis was conducted among the time-corrected T concentrations and mean amygdala z-value differences (threat-neutral). No significant correlation was found, r(17)= .220, p=.36.
Salivary CORT. After transforming the CORT concentrations using the natural logarithm, a
Shapiro-Wilk’s test (p>.05) and a visual inspection of the histogram, normal Q-Q plot and box plot showed that the salivary CORT concentrations were approximately normally distributed. Inspection of the residual plot showed no pattern forming and the coefficients were not significant (p=.83, R2=.003), suggesting no linear or non-linear association between
CORT concentrations and sampling time. A Spearman correlation analysis supported this, r(17)=.11, p=.65, hence no time correction was necessary. A Pearson correlation analysis was conducted among the ln-transformed CORT concentrations and mean amygdala z-value differences (threat-neutral). No significant correlation was found, r(17)=.20 , p=.42.
Discussion
In line with prior results by Van’t Veer et al. and our own expectations, we found increased activations in the amygdala for infant-threatening situations compared to neutral situations. We focused specifically on the processing of infant-threatening situations, not on processing
of threat to the participant himself or another adult. The task was designed to evoke the neural process of perceiving an infant in need of protection for which a behavioural action might be required and imagined.
The primary interest in this study was identifying associations of T and CORT with expectant father’s amygdala response to infant-threatening situations. Contrary to our expectations, no associations were found between basal T or basal CORT and expectant fathers’ amygdala response to infant threat. Taken together, these results suggest that basal T and basal CORT do not affect or relate to the neural processing of protective behaviour in fathers. However, with a small sample size, caution must be applied during interpretation.
A possible explanation for the non-existent T relation could be that only a small percentage of males has a high enough T concentration to draw a significant effect. In a small sample size, the chance of including enough individuals with high enough T levels is
reduced. The T concentrations measured in this study were on the low end of the broad normal range (25-250 pg/mL), compared to prior research (Ellison et al., 2002; Penton-Voak and Chen, 2004). This is likely because, in this study, 75% of the saliva samples were
collected between 14:00 and 19:00, at which point endogenous T and CORT levels have decreased significantly since waking (Dabbs, 1990; Nicolson, 2004; Brambilla et al., 2009; Sjörs, Ljung and Jonsdottir, 2014). Individual differences between participants often lead to large ranges in CORT concentrations, even in studies with large sample sizes (Cohen et al., 2006). In studies on this topic, hormone measurements should be taken at the same time of day (Penton-Voak and Chen, 2004), or even better, within the same amount of time after waking, on multiple days (Ellison et al., 2002). Collecting samples on only one day, instead of multiple days, has implications for the reliability of measurement, specifically for cortisol (Adam and Kumari, 2009). This measurement method could aid in nearing correct levels of trait characteristics (Adam and Kumari, 2009). These samples would be procured with at-home saliva collection kits.
There needs to be a partial shift in focus from parental sensitivity to parental protection in order to gain more understanding of the neural correlates related to infant protective behaviour. One limitation to this study is that the neural effect of the infant-threatening videos on the participant is not definitively related to protective behaviour. One way to find more evidence for this relation is by looking into more robust methods of
inducing protective behaviour. The participant really needs to believe their child is in danger. Of course, ethically this is not easy. However, perhaps a more realistic, 3D view of the infant-threatening situation by means of virtual reality (VR) glasses could aid in bringing
participants’ neural reaction closer to that of a real situation. Heartrate measurements and analysis of neural correlates related to movement initiation, like the premotor cortex, could also be taken into account (Churchland, Santhanam and Shenoy, 2006).
References
Ackermann, S. et al. (2012) ‘Testosterone levels in healthy men are related to amygdala
reactivity and memory performance’, Psychoneuroendocrinology, 37(9), pp. 1417–1424. doi: 10.1016/j.psyneuen.2012.01.008.
Adam, E. K. and Kumari, M. (2009) ‘Assessing salivary cortisol in large-scale,
epidemiological research’, Psychoneuroendocrinology, 34(10), pp. 1423–1436. doi: 10.1016/j.psyneuen.2009.06.011.
Alyousefi-Van Dijk, K. et al. (2019) ‘Vasopressin differentially affects handgrip force of
expectant fathers in reaction to own and unknown infant faces’, Frontiers in Behavioral Neuroscience, 13(May), pp. 1–10. doi: 10.3389/fnbeh.2019.00105.
Van Anders, S. M., Tolman, R. M. and Volling, B. L. (2012) ‘Baby cries and nurturance
affect testosterone in men’, Hormones and Behavior. Elsevier Inc., 61(1), pp. 31–36. doi: 10.1016/j.yhbeh.2011.09.012.
Atzil, S. et al. (2012) ‘Synchrony and specificity in the maternal and the paternal brain:
Relations to oxytocin and vasopressin’, Journal of the American Academy of Child and Adolescent Psychiatry. Elsevier Inc., 51(8), pp. 798–811. doi: 10.1016/j.jaac.2012.06.008.
Bakermans-Kranenburg, M. J. and van IJzendoorn, M. H. (2017) ‘Protective parenting:
neurobiological and behavioral dimensions’, Current Opinion in Psychology. Elsevier Ltd, 15, pp. 45–49. doi: 10.1016/j.copsyc.2017.02.001.
Birger, M. et al. (2003) ‘Aggression: The testosterone-serotonin link’, Israel Medical
Association Journal, 5(9), pp. 653–658.
Bos, P. A. et al. (2018) ‘Prenatal and postnatal cortisol and testosterone are related to parental
caregiving quality in fathers, but not in mothers’, Psychoneuroendocrinology. Elsevier, 97(January), pp. 94–103. doi: 10.1016/j.psyneuen.2018.07.013.
Brambilla, D. J. et al. (2009) ‘The effect of diurnal variation on clinical measurement of
serum testosterone and other sex hormone levels in men’, Journal of Clinical Endocrinology and Metabolism, 94(3), pp. 907–913. doi: 10.1210/jc.2008-1902.
Churchland, M. M., Santhanam, G. and Shenoy, K. V. (2006) ‘Preparatory activity in
premotor and motor cortex reflects the speed of the upcoming reach’, Journal of Neurophysiology, 96(6), pp. 3130–3146. doi: 10.1152/jn.00307.2006.
Cohen-Bendahan, C. C. C. et al. (2015) ‘Explicit and implicit caregiving interests in
expectant fathers: Do endogenous and exogenous oxytocin and vasopressin matter?’, Infant Behavior and Development. Elsevier Inc., 41, pp. 26–37. doi: 10.1016/j.infbeh.2015.06.007.
Cohen, S. et al. (2006) ‘Socioeconomic status, race, and diurnal cortisol decline in the
Coronary Artery Risk Development in Young Adults (CARDIA) Study’, Psychosomatic Medicine, 68(1), pp. 41–50. doi: 10.1097/01.psy.0000195967.51768.ea.
Dabbs, J. M. (1990) ‘Salivary testosterone measurements: Reliability across hours, days, and
weeks’, Physiology and Behavior, 48, pp. 83–86. doi: 10.1016/0031-9384(90)90265-6.
Donaldson, Z. R. and Young, L. J. (2008) ‘Oxytocin, vasopressin, and the neurogenetics of
sociality’, Science, 322(5903), pp. 900–904. doi: 10.1126/science.1158668.
Edelstein, R. S. et al. (2017) ‘Prospective and dyadic associations between expectant
parents’ prenatal hormone changes and postpartum parenting outcomes’, Developmental Psychobiology, 59(1), pp. 77–90. doi: 10.1002/dev.21469.
Ellison, P. T. et al. (2002) ‘Population variation in age-related decline in male salivary
testosterone’, Human Reproduction, 17(12), pp. 3251–3253. doi: 10.1093/humrep/17.12.3251.
Finegood, E. D. et al. (2016) ‘Psychobiological influences on maternal sensitivity in the
context of adversity.’, Developmental Psychology, 52(7), pp. 1073–1087. doi: 10.1037/dev0000123.
responses to infant cries in new fathers’, Hormones and Behavior, 42(4), pp. 399–413. doi: 10.1006/hbeh.2002.1840.
Gao, W., Stalder, T. and Kirschbaum, C. (2015) ‘Quantitative analysis of estradiol and six
other steroid hormones in human saliva using a high throughput liquid chromatography-tandem mass spectrometry assay’, Talanta. Elsevier, 143, pp. 353–358. doi:
10.1016/j.talanta.2015.05.004.
Gettler, L. T. et al. (2011) ‘Longitudinal evidence that fatherhood decreases testosterone in
human males’, Proceedings of the National Academy of Sciences of the United States of America, 108(39), pp. 16194–16199. doi: 10.1073/pnas.1105403108.
Greve, D. N. and Fischl, B. (2009) ‘Accurate and robust brain image alignment using
boundary-based registration’, NeuroImage. Elsevier Inc., 48(1), pp. 63–72. doi: 10.1016/j.neuroimage.2009.06.060.
Hancock, K. J., Lawrence, D. and Zubrick, S. R. (2014) ‘Higher maternal protectiveness is
associated with higher odds of child overweight and obesity: A longitudinal Australian study’, PLoS ONE, 9(6). doi: 10.1371/journal.pone.0100686.
Herman, J. P. et al. (2012) ‘Neural regulation of the stress response: Glucocorticoid
feedback mechanisms’, Brazilian Journal of Medical and Biological Research, 45(4), pp. 292–298. doi: 10.1590/S0100-879X2012007500041.
Hurtado, A. M. and Hill, K. R. (1992) ‘Paternal effect on offspring survivorship among ache
and hiwi hunter-gatherers: Implications for modeling pair-bond stability’, in Father-Child Relations: Cultural and Biosocial Contexts, pp. 31–55. doi: 10.4324/9780203792063.
Van IJzendoorn, M. H., Juffer, F. and Duyvesteyn, M. G. C. (1995) ‘Breaking the
Intergenerational Cycle of Insecure Attachment: A Review of the Effects of Attachment‐ Based Interventions on Maternal Sensitivity and Infant Security’, Journal of Child
Psychology and Psychiatry, 36(2), pp. 225–248. doi: 10.1111/j.1469-7610.1995.tb01822.x.
Jenkinson, M. et al. (2002) ‘Improved Optimization for the Robust and Accurate Linear
Registration and Motion Correction of Brain Images’, NeuroImage, 17(2), pp. 825–841. doi: 10.1006/nimg.2002.1132.
Lesser, J. et al. (2009) ‘Parental Protectiveness and Unprotected Sexual Activity among
Latino Adolescent Mothers and Fathers’, AIDS Education and Prevention, 21(5 supplement b), pp. 88–102. doi: 10.1521/aeap.2009.21.5_supp.88.
Ljubijankić, N. et al. (2008) ‘Daily fluctuation of cortisol in the saliva and serum of healthy
persons’, Bosnian Journal of Basic Medical Sciences, 8(2), pp. 110–115. doi: 10.17305/bjbms.2008.2962.
Mehta, P. H. et al. (2017) ‘Hormonal underpinnings of status conflict: Testosterone and
cortisol are related to decisions and satisfaction in the hawk-dove game’, Hormones and Behavior. Elsevier Inc., 92, pp. 141–154. doi: 10.1016/j.yhbeh.2017.03.009.
Mills-Koonce, W. R. et al. (2009) ‘Psychophysiological Correlates of Parenting Behavior in
Mothers of Young Children’, Developmental Psychobiology, 51(8), pp. 650–661. doi: 10.1002/dev.20400.
Nicolson, N. A. (2004) ‘Childhood parental loss and cortisol levels in adult men’,
Psychoneuroendocrinology, 29(8), pp. 1012–1018. doi: 10.1016/j.psyneuen.2003.09.005.
Penton-Voak, I. S. and Chen, J. Y. (2004) ‘High salivary testosterone is linked to masculine
male facial appearance in humans’, Evolution and Human Behavior, 25(4), pp. 229–241. doi: 10.1016/j.evolhumbehav.2004.04.003.
Richter, L., Griesel, R. D. and Manegold, J. (2004) The importance of caregiver-child
interactions for the survival and healthy development of young children, WHO Library Cataloguing-in-Publication Data.
Rilling, J. K. and Mascaro, J. S. (2017) ‘The neurobiology of fatherhood’, Current Opinion
Sherman, G. D. et al. (2017) ‘Sex differences in cortisol’s regulation of affiliative behavior’,
Hormones and Behavior, 92, pp. 20–28. doi: 10.1016/j.yhbeh.2016.12.005.
Sjörs, A., Ljung, T. and Jonsdottir, I. H. (2014) ‘Diurnal salivary cortisol in relation to
perceived stress at home and at work in healthy men and women’, Biological Psychology, 99, pp. 193–197. doi: 10.1016/j.biopsycho.2014.04.002.
Storey, A. E. et al. (2011) ‘The effects of social context on the hormonal and behavioral
responsiveness of human fathers’, Hormones and Behavior. Elsevier Inc., 60(4), pp. 353– 361. doi: 10.1016/j.yhbeh.2011.07.001.
Taylor, S. E. et al. (2000) ‘Biobehavioral responses to stress in females: Tend-and-befriend,
not fight-or-flight.’, Psychological Review, 107(3), pp. 411–429. doi: 10.1037/0033-295X.107.3.411.
Thijssen, S. et al. (2018) ‘Effects of vasopressin on neural processing of infant crying in
expectant fathers’, Hormones and Behavior. Elsevier, 103(December 2017), pp. 19–27. doi: 10.1016/j.yhbeh.2018.05.014.
Thompson, R. R. et al. (2006) ‘Sex-specific influences of vasopressin on human social
communication’, Proceedings of the National Academy of Sciences of the United States of America, 103(20), pp. 7889–7894. doi: 10.1073/pnas.0600406103.
Van’t Veer, A. E. et al. (2019) ‘Exploring the neural basis for paternal protection: An
investigation of the neural response to infants in danger’, Social Cognitive and Affective Neuroscience, 14(4), pp. 447–457. doi: 10.1093/scan/nsz018.
Veer, I. M. et al. (2012) ‘Endogenous cortisol is associated with functional connectivity
between the amygdala and medial prefrontal cortex’, Psychoneuroendocrinology. doi: 10.1016/j.psyneuen.2011.12.001.
Weisman, O., Zagoory-Sharon, O. and Feldman, R. (2014) ‘Oxytocin administration,
salivary testosterone, and father-infant social behavior’, Progress in
Neuro-Psychopharmacology and Biological Psychiatry. Elsevier Inc., 49, pp. 47–52. doi: 10.1016/j.pnpbp.2013.11.006.
Witte, A. M. et al. (2019) ‘Fathers today: Design of a randomized controlled trial examining
the role of oxytocin and vasopressin in behavioral and neural responses to infant signals’, BMC Psychology. doi: 10.1186/s40359-019-0356-2.
Woolrich, M. W. et al. (2001) ‘Temporal autocorrelation in univariate linear modeling of
