Identify each characteristic as being associated with either high agency or high communion.

Given the central role of social processes in the conceptual definition of communion and unmitigated communion (Helgeson, 1994; Helgeson & Fritz, 1998), a social psychophysiological approach (Smith & Gerin, 1998) is of obvious importance in testing this model. One challenge in studying the social psychophysiology of CVR is the identification of key conceptual dimensions of social processes (Smith, Gallo, & Ruiz, in press). A great variety of social factors alter CVR and an organizing framework for these influences could facilitate systematic and cumulative investigation of the mechanism (Smith et al., in press). The interpersonal circumplex (Fig. 1) is a useful heuristic in this regard (Gallo, Smith, & Kircher, 2000; Smith, Limon, Gallo, & Ngu, 1996). It allows many classes of influences on CVR—traits, interpersonal contexts, social stimuli, and social behaviors—to be mapped into a common conceptual space, including the psychosocial variables that place individuals at risk for CVD. For example, the personality trait of chronic hostility in combination with relevant social stressors is associated with increased CVR (Houston, 1994; Suls & Wan, 1993), as is exposure to hostile social stimuli such as harassment (e.g., Bongard, Pfeiffer, al'Absi, Hodapp, & Lninenkemper, 1997; Gallo et al., 2000). These findings demonstrate that stimuli and traits around the hostile pole of the circumplex are associated with greater CVR.

Similarly, recent evidence has also shown that behavioral expressions of dominance increase CVR (Smith et al., 1989, Smith et al., 1990, Smith et al., 1996, Smith et al., 1997). Some research suggests that at least among men the trait of dominance is associated with greater CVR, as is exposure to dominant social stimuli (Newton, Bane, Flores, & Greenfield, 1999). However, the association of individual differences in dominance with CVR is inconsistent in the small number of related studies (Rejeski et al., 1989, Rejeski et al., 1990; Saab, Matthews, Stoney, & McDonald, 1989). Supportive stimuli can be placed on the friendly pole of the circumplex (Trobst, 2000) and have been found to attenuate CVR (Gerin, Milner, Chawla, & Pickering, 1995; Kamarck, Annunziato, & Amateau, 1995; Lepore, 1995; Uchino & Garvey, 1997), providing that the social stimulus is not perceived as an evaluative threat (Allen, Blascovich, Tomaka, & Kelsey, 1991). Although it is known that supportive social stimuli can attentuate CVR (Kamarck, Peterman, & Raynor, 1998; Lepore, 1998), the cardiovascular effect of enacting or expressing social support is not well established (cf. Davis & Matthews, 1996).

As noted above, recent theory and research have shown that trait agency is closely related to dominance, whereas trait communion is closely related to friendliness (Lippa, 1995; Wiggins, 1991; Wiggins & Broughton, 1991). In fact, Wiggins and Trapnell (1996) have argued that agency and communion may be more accurate descriptions of the overarching psychological dimensions of the circumplex. Further, circumplex-based measures have been used to evaluate similarities and differences among communion, agency, and their unmitigated forms (Fritz & Helgeson, 1998; Helgeson & Fritz, 1999).

Research on the pursuit of agency (e.g., effort to influence others, achievement striving, competition, etc.) suggests that these activities evoke heightened CVR among both men and women, relative to control tasks (e.g., Matthews, Davis, Stoney, Owens, & Caggiula, 1991; Smith et al., 1989, Smith et al., 1990, Smith et al., 1997). Some, but not all, studies suggest that agency stressors evoke greater CVR among men than among women (e.g., Brown & Smith, 1992; Lash, Gillespie, Eisler, & Southard, 1991; Stoney, Davis, & Matthews, 1987). Studies of communion stressors are much less common. In some studies, communion stressors (e.g., disagreement with a spouse; being supportive and accommodating to a customer) have a larger effect on CVR among women than among men (Smith et al., 1996; Smith, Gallo, Ngu, Goble, & Stark, 1998). However, most studies of communion stressors have compared them to agency stressors rather than carefully chosen control conditions (Davis & Matthews, 1996; Smith et al., 1996), limiting conclusions regarding the specific effects of communion striving. That is, communion stressors could have a variety of effects on CVR relative to appropriate controls for simple speaking and other artifactual influences on cardiovascular responses and some of these effects could be similar to those evoked by agency tasks. Hence, simple comparisons between agency and communion tasks obscure the specific effects of communion stressors.

In Helgeson's (1994) model, the value or importance women place on communion goals should lead them to exert effort in such tasks. Effortful engagement, in turn, would increase CVR (Wright & Kirby, 2001). However, experience in communion-related activities may lead women to have confidence in their abilities in such tasks and perceptions of higher ability (and, as a result, lower task difficulty) would tend to produce less effortful engagement and CVR—especially when the standards for successful task performance are clear and appraised as attainable. If the standards for success in a communion task are ambiguous or very difficult, the importance of the task would heighten women's engagement and CVR (Wright & Kirby, 2001).

Blascovich and Tomaka (1996) have proposed a model that provides a different account of when motives such as communion increase versus decrease CVR, by examining patterns of cardiovascular response. Stress-induced increases in blood pressure reflect either (or both) increases in cardiac output (CO) (i.e., volume of blood pumped by the heart) or total peripheral resistance (TPR) (i.e., constriction vs. dilation of the vasculature) (Sherwood et al., 1990). Increases in heart rate reflect either (or both) increased sympathetic stimulation of the heart (i.e., pre-ejection period; PEP) or decreases in parasympathetic dampening (i.e., respiratory sinus arrhythmia; RSA) of heart rate (Cacioppo, Uchino, & Bernston, 1994). Hence, similar levels of blood pressure or heart rate reactivity can reflect quite distinct physiological processes and the use of impedance cardiography and related techniques can assess these potential determinants of CVR (Cacioppo et al., 1994; Sherwood et al., 1990).

Blascovich and Tomaka (1996) maintain that differences in task appraisal produce distinct patterns of these responses. If an individual views a task as important and within their coping ability, this challenge appraisal results in increases in CO and sympathetic stimulation of the heart (PEP) and less TPR. In contrast, if an individual views a task as important but taxing or exceeding their coping ability, this threat appraisal results in relatively greater increases in TPR. Hence, whereas traditional models have emphasized levels of CVR, the model of Blascovich and Tomaka (1996) suggests that the pattern of cardiovascular responses and related determinants is more informative. Several studies support this view (e.g., Tomaka et al., 1993, Tomaka et al., 1999). In the context of women's responses to agency versus communion stressors, it is possible that the former could evoke a threat response, whereas the latter could evoke the pattern associated with challenge.

Given the potential importance of situational communion stressors in women's health, we examined women's CVR in response to three tasks—an agency stressor, a communion stressor, and an appropriate control. Use of a low agency/low communion control condition otherwise equal to the agency and communion tasks in relevant artifacts that affect CVR (e.g., speaking, social interaction) permitted the examination of the specific effects of these stressors. Given Helgeson's (1994) predictions regarding the central role of communion in health risk for women, it was expected that women would display increased CVR in response to the communion stressor. We also expected to replicate our prior findings that the agency stressor would heighten CVR in young women (Smith et al., 1989, Smith et al., 1997). Thus, comparison of the agency stressor versus control task and communion stressor versus control task represented two a priori, directional contrasts (Wampold, Davis, & Good, 1990). Direct comparisons between the agency and communion tasks are of potential interest (Davis & Matthews, 1996; Smith et al., 1996), especially since they could reveal the distinct patterns of psychological and cardiovascular responses described by Blascovich and Tomaka (1996). However, the comparisons between agency and communion tasks should be considered secondary and exploratory in our analysis plan.

As discussed above, agency and communion also represent individual differences. Therefore, we examined the effect of trait agency and communion (i.e., masculine and feminine sex roles) on CVR. Newton et al. (1999) found that trait agency was inversely, but non-significantly related to CVR in women, perhaps because of a small sample size. Consistent with Helgeson's (1994) theory, we predicted that this association would be significant in our larger study.

Given the difficulty of modeling psychosocial risk processes in the laboratory, it is essential to assess the effects of the related manipulations. Interpersonal methods are useful in this regard (Smith et al., in press). The interpersonal circumplex not only provides a description of key dimensions for organizing the social psychophysiology of CVR, it also provides standardized, well-validated assessment procedures. We used two such instruments, the Revised Interpersonal Adjective Scale (IAS-R; Wiggins, Trapnell, & Phillips, 1988) and the Impact Message Inventory (IMI; Kiesler, Schmidt, & Wagner, 1997) to assess participants' construal of their own and their interaction partners' dominance and friendliness. Compared to the control condition, the agency stressor was expected to be seen as involving greater dominance and less friendliness. In contrast, the communion stressor task was expected to evoke greater perceptions of friendliness overall and perceptions of the partner as somewhat submissive. To assess aspects of challenge and threat, we also assessed task appraisals and affective responses.

Participants

Sixty female participants were recruited from introductory and lower level psychology courses at a large university. Women were told that the purpose of the study was to examine the effects of social interaction on cardiovascular responses. Upon arrival, they were randomly assigned to one of the three experimental conditions.

Baseline period

Participants completed self-report measures (described below), and physiological monitoring devices were attached. Four band electrodes were placed in the configuration