The sample consisted of sixteen married couples.  The mean age of the husbands was 33 and the wives 31. The couples were all from Madison, Wisconsin. They were recruited by newspaper advertisements.

The study consisted of two parts.  First, participants were given questionnaires to assess their level of satisfaction and happiness in their marriage.  In addition, personality traits were assessed.  Finally, they were asked to do a test run in an fMRI scanner.  This was done to make sure that they were comfortable with the scanner and to show them how it works.

The second visit took place one week later.  The wives were put in an fMRI with an electrode attached to one ankle.  While in the fMRI the women were shown 12 non-threat indicating images (safety cues) and 12 threatening images (threat cues) - that is, images that when shown, indicated that an electrical shock was possible. The safety cue was a blue zero on a black background; the threat cue was a red x on a black background. After the cue was shown, there was up to a 10-second period when a shock could be administered.  During the experiment, two shocks were administered.  After the threat period was over, there was a rest period when the women were simply shown a black screen. During the rest period, they were asked for their rate their level of arousal and their feelings of unpleasantness on a 1 - 5 scale. 

This experiment was a repeated measures design.  Each woman would undergo the tests in the fMRI under three conditions: holding their husband's hand, holding the hand of a stranger (whom they did not meet until after the experiment), or not holding anyone's hand.  The conditions were counterbalanced to control for order effects.

Results showed that the women's subjective sense of unpleasantness and arousal was lowest when holding their husband's hand. In addition, the brain's "threat response" was lowest when holding their husband's hand.  The threat response was strongest when no hand was held. Not only this, but there was a negative correlation between the reported marital quality and the threat response - that is, the higher the reported marital quality, the lower the brain's threat response. It appears that social support is key to resilience.

The following video shows you a summary of the study.

The study was highly controlled.  The study was counterbalanced so that order effects such as habituation - that is, getting used to the shocks - was most likely not a confounding variable.  This means that the study had high internal validity.

The use of subjective ratings of unpleasantness was used to correlate with brain activity.  This helped to improve construct validity.

In addition to the high level of control, the study was highly artificial and thus lacks ecological validity.

The study had a very small sample. The sample consisted of 15 Caucasian (white) and 1 Asian female. The men were not tested in the scanner. This means that we cannot generalize to males.  In addition, since being in a "satisfying marriage" was a requirement for being in the study, this study cannot be generalized to people who are in unhappy relationships.  Although there was a correlation between marital satisfaction and the brain's response to a threat, all of the participants had good levels of marital satisfaction.

In addition, women with a history of mental health issues, such as depression, were excluded from the sample.  This also limits our ability to generalize the findings.

The actual mechanisms for why there is less activity in specific parts of the brain are not clear.  The original study hypothesizes that oxytocin or endorphins may play a role, but this could not be determined from the original study.

In this study, it was not the amygdala that showed a difference in activity when holding a spouse's hand.  One of the key parts of the brain that was activated was the anterior cingulate cortex [ACC]. When the ACC is active, the amygdala is not.

You may want to know more about how the anterior cingulate cortex can help with anxiety regulation.  This study by Zeidan et al (2013) looks at the role of the ACC in regulating anxiety in people who meditate.