This investigation tested the incremental utility of cognitive retraining as a component within a program designed to prevent child maltreatment. High-risk families (N = 96) were randomly assigned to a control condition, home visitation modeled after the Healthy Start program (unenhanced home visitation), or home visitation that included a cognitive component (enhanced home visitation). Mothers were identified late during pregnancy or soon after birth, and their participation continued for 1 year. Lower levels of harsh parenting were found among mothers in the enhanced home visitation condition than among those in the unenhanced home visitation or control conditions. Prevalence of physical abuse (percentage of mothers who were abusive) during the first year was 26% in the control condition, 23% in the unenhanced home visitation condition, and 4% in the enhanced home visitation condition. Benefits were greatest in families that included a medically at-risk child. A linear pattern of benefits was found for child health; as program features were added, benefits for child health increased.

The paradoxical use of punitive force by adults with low perceived power was explored in teaching interactions. Punitive force was measured by the intensity of physical effort used in operating controls that displayed negative feedback to child trainees. Women who differed in perceived power were assigned to (a) situations in which they had high, low, or ambiguous control and (b) responsive, unresponsive, or ambiguously responsive children. Women with low perceived power--when given ambiguous control--were more likely than other women to (a) use high levels of punitive force and (b) show elevated levels of autonomic arousal. Arousal, in turn, partially mediated the relationship between perceived power and use of force. Women with low perceived power were also more likely to attribute intentionality to children whose behavior was ambiguous. Results were interpreted as having implications for violence within adult-child relationships.

We would like to express our appreciation to all the community agencies and individuals involved in the Family Program, along with all the home visitors and families who participated in this study. Services provided within the program were funded by the State of California. Our thanks also go to Veronica Barraza, who skillfully managed the data set, and to Deborah Holmes, who provided clinical supervision to home visitors.

Changes in children's attentional engagement were assessed as a function of their exposure to "teachers" who differed in perceived power and the communication style that is associated with perceived power. In Study 1, "teachers" (women assigned to an instructional role) were selected on the basis of their perceived power; low-power women were more likely than high-power women to display communication ambiguity. Children responded to low-power women with low levels of (1) autonomic orienting (consistent with low attention) and (2) high errors on a cognitively demanding task (mental arithmetic). Attentional disengagement was found to be mediated by the ambiguous communication style of low-power adults. In Study 2, the "teacher" was a confederate who systematically varied the facial and vocal ambiguity of her instructions. Children showed the lowest levels of orienting and the highest level of errors when the "teacher" was ambiguous in both face and voice. Results were interpreted as showing that adult ambiguity (naturally occurring or experimentally produced) leads to reductions in children's attentional engagement.

We measured the ways that women with varying degrees of perceived power respond physiologically to children's elevated vocal pitch (F 0 : fundamental frequency), a social dependence/immaturity cue. Listeners believed either that they would provide instructions or make judgments about the children they heard. As predicted, women with low perceived power in caregiving relationships (i.e., who attributed greater power to children than to self) were highly reactive to children's pitch properties-in particular, when they anticipated providing instructions. When expecting to provide instructions to children with higher F 0 voices, women with low perceived power showed elevations in cortisol levels and heart rate (consistent with defensive mobilization for threat). In all other pairings of women and children, cortisol and heart rate levels held relatively constant or declined. In addition, women with low perceived power showed better recall of messages from children with higher F 0 voices than lower F 0 voices. Implications are drawn for interaction patterns that foster caregiving stress and conflict.