Abstract

Methods

Results

Attention patterns for emotion faces

We conducted a repeated measures GLM with time (baseline, post-mood induction, and post-delay period), emotion cue (sad, angry, and happy), maltreatment (CTSPC physical abuse scale score), systolic blood pressure reactivity (SBPR), trait rumination (CRSS score), and all interactions as predictors of attention on the dot probe task. ² All predictor variables were Z-scored.

We predicted that maltreated children with high physiological reactivity to experience of a sad emotional state would show heightened attention for sad faces immediately after the sad mood induction. Additionally, we predicted that those reactive maltreated children who also tend to ruminate would continue to show heightened attention for sad faces over time. Contrary to predictions, there was no five-way interaction among time point, emotion cue, maltreatment, SBPR, and trait rumination on attention patterns (see Table 1 for results from the omnibus model).

Table 1

Results of repeated measures GLM with time, emotion cue, maltreatment (PA), systolic blood pressure reactivity (SBPR), trait rumination (RUM), and all interactions as predictors of attention. Interactions involving time × emotion not included (all n.s. at p > .05).

	F	Partial eta-squared
Time	1.01	.019
Time × PA	0.14	.003
Time × RUM	1.32	.025
Time × SBPR	0.08	.001
Time × PA × RUM	4.40	.079*
Time × PA × SBPR	0.45	.009
Time × RUM × SBPR	2.47	.046
Time × PA × RUM × SBPR	1.91	.036
Emotion	1.82	.034
Emotion × PA	6.27	.109*
Emotion × RUM	3.09	.057*
Emotion × SBPR	3.25	.060*
Emotion × PA × RUM	4.27	.077*
Emotion × PA × SBPR	7.84	.133*
Emotion × RUM × SBPR	3.62	.066*
Emotion × PA × RUM × SBPR	2.81	.052*
PA	1.76	.033
RUM	2.43	.045
SBPR	0.00	.000
PA × RUM	0.11	.002
PA × SBPR	0.01	.000
RUM × SBPR	0.25	.005
PA × RUM × SBPR	0.00	.000

Open in a separate window

^*p < .05

Changes in attention to emotion faces

Only the interaction between maltreatment and trait rumination on attention for all emotion faces changed across time points, F(2, 102) = 4.40, p = .02, partial eta-squared = .08. We further explored this interaction by examining the relationship between maltreatment and rumination at each time point. Children on average did not show biased attention toward or away from emotion faces at any time point. However, there was a significant interaction between maltreatment and trait rumination on attention for emotion faces after the delay period, B = 9.88, p = .04, partial eta-squared = .08.

To further investigate this interaction, we used the simple slopes technique (Aiken and West, 1991). We examined the simple slope at one SD above the mean maltreatment score and when number of physically abusive experiences was zero. Children who experienced higher levels of maltreatment and who had higher levels of trait rumination showed a greater attention bias toward emotion faces overall after the delay period, B = 19.01, p = .007, partial eta-squared = .14. In contrast, children who experienced no maltreatment did not differ in attention for emotion faces as a function of trait rumination, B = 3.75, p = .24.

Changes in attention to sad faces

Although the omnibus analyses indicated changes in attention over time did not vary based on emotion cue type, we had particular interest in the chronometry of attention patterns for sad cues and repeated analyses on attention for sad faces alone. The relationship between maltreatment and attention for sad faces changed across time, F(2, 102) = 5.33, p = .006, partial eta-squared = .10, see Figure 1. We further explored this interaction by examining the influence of maltreatment on attention for sad faces at each time point. Children on average did not show an attention bias toward sad faces at any time point. However, after the delay period, children who had experienced more maltreatment showed more attention toward sad faces, B = 20.36, p = .012, partial eta-squared = .12. In sum, maltreated children showed heightened attention to sad faces after the delay period, and those maltreated children who ruminate showed generalized heightened attention to emotion faces after the delay period.

Open in a separate window

Figure 1

Attention patterns for sad faces at baseline, immediately after induction of a sad affective state, and after a delay period. Positive attention bias scores indicate attention toward sad faces. No and high physical abuse represent points at zero physical abuse and one SD above the mean physical abuse score, respectively.

Stable attention patterns

Across time points, children on average did not show a bias toward or away from any emotion cue. However, there was a trend for a significant four-way interaction among emotion cue, maltreatment, SBPR, and trait rumination, F(2, 102) = 2.81, p = .07, partial eta-squared = .05. To further explore this interaction, we examined the relationships among maltreatment, SBPR, and trait rumination for each emotion cue separately.

Attention patterns for sad faces were of particular interest in the present study. Across time points, there was a significant interaction between maltreatment experience and trait rumination on attention patterns for sad faces, F(1, 51) = 5.81, p = .02, partial eta-squared = .10, see Figure 2. Children who had experienced high levels of maltreatment (1 SD above the mean) showed a greater attention bias toward sad faces if they had higher levels of trait rumination, B = 19.03, p = .013, partial eta-squared = .11. In contrast, children who had no maltreatment experience did not differ in attention for sad faces as a function of trait rumination, B = 0.72, p = 0.98.

Open in a separate window

Figure 2

Interaction of maltreatment and trait rumination on attention bias to sad faces across all dot probe tasks. Positive attention bias scores indicate attention toward sad faces. No and high physical abuse represent points at zero physical abuse and one SD above the mean physical abuse score, respectively. Low and high trait rumination represent points one SD below and above the mean, respectively.

For happy faces, like sad faces, there was a significant maltreatment by trait rumination interaction across all time points, B = −10.48, p = .03, partial eta-squred = .09. For children with higher levels of maltreatment (1 SD above the mean), there was a trend for those with higher levels of trait rumination to avoid happy faces overall, B = −12.40, p = .08, partial eta-squared = .06. In contrast, children with no maltreatment experience did not differ in attention for happy faces as a function of trait rumination, B = 3.78, p = .25. In sum, maltreated children who ruminate show more attention for sad faces and tend to show less attention for happy faces across time points. For anger faces, attention patterns did not differ as a function of maltreatment, SBPR, trait rumination, or their interaction.

Discussion and Conclusion

The current study was designed to test whether the experience of child maltreatment contributes to the preferential attention to sad cues that is associated with risk for depression. Specifically, we examined if individual differences in emotion reactivity to and regulation of a sad emotional state predicted which maltreated children exhibited depression-relevant attention biases to sad cues. We did so immediately after experience of a sad emotion and after a delay period when emotional state should have returned to baseline.

Children with histories of maltreatment showed preferential attention to sad, emotion-congruent cues only after they experienced a sad emotional state. Those maltreated children who engage in rumination showed a generalization of this effect, such that they showed preferential attention for all emotion cues after they experienced a sad emotional state. Additionally, maltreated children who engage in rumination showed a more stable pattern of heightened attention to sad cues across time points and a trend toward avoidance of happy cues across time points. Attention to sad cues did not vary as a function of emotional state for these children. This pattern cannot be attributed to negative affect or depression: Current negative affect and depressive symptoms in the past year did not influence attention biases and effects remained when controlling for current depressive symptoms.

The results of the current study expand on previous research indicating that maltreated children show heightened attention for anger cues to suggest that maltreated children also show heightened attention for sad, depression-relevant cues in certain contexts. Specifically, children who experienced higher levels of maltreatment did not immediately respond to a sad emotional state by showing a sad attention bias, but showed heightened attention for sad cues after a brief delay period. This finding is consistent with the idea that early experience may predispose individuals to depression when they experience a stressor. Additionally, the current results suggest that maltreated children who ruminate show a more stable pattern of heightened attention for sad cues.

These attention patterns could serve as a risk mechanism. Those maltreated children who have difficulty regulating emotional states pay more attention to sad cues and may also pay less attention to happy cues. These children may engage in even more rumination about their current social environment in response to these social cues, which may exacerbate sad moods when they occur. This could lead to a perpetual loop of maladaptive coping that exacerbates and prolongs sad moods.

Analyses of attention for anger cues did not replicate previous findings of an association between maltreatment and heightened attention for anger cues (e.g., Pine et al., 2005; Pollak, et al., 2005; Shackman, et al., 2007). However, much of the previous research has used either physiological indices of attention (such as skin conductance or ERP) or measures of visual attention at a shorter stimulus presentation than in the current study (500ms as compared to 1500ms). Attention patterns at 1500ms stimulus presentation likely reflect top-down control processes rather than vigilance. It may be that children with higher levels of maltreatment were vigilant to anger cues, but that they did not dwell on these cues and this pattern was not captured with the present methods.

The precise influence of trait rumination on attention biases to sad cues in this experiment is unclear. Although trait rumination predicted attention patterns, it is not possible to determine if trait ruminators actually engaged in rumination during the study. Future research would benefit from including state-level measures of rumination and manipulating rumination in studies of attention biases.

We did not observe the previously demonstrated relationship between children’s depressive symptoms and sad attention bias (e.g., Hankin, Gibb, Abela, & Flory, 2010). However, no children in the current study met criteria for a major depressive episode in the past year and depressive symptoms were low. Although children’s symptoms of depression in the past year were not related to attention patterns, we are unable to rule out the possibility that children in the current study had experienced a depressive episode prior to the past year. Therefore, it is not possible to determine if the results of the current study represent putative risk mechanisms that occur prior to development of depression. Replication of the current results in a sample of children who have never experienced a depressive episode is needed.

We assessed maltreatment using parent self-report, which is likely to elicit social desirability pressure to under-report abusive behaviors. However, the self-report measure used in the current study has been shown to be a valid measure of maltreatment and is associated with court-substantiated cases of maltreatment. Additionally, social desirability influences work against our hypotheses, and any abusive behaviors that parents report on this measure are likely to be an underestimate. Although children in the current study experienced a range of physical abuse, more children had lower exposure to abuse and results were influenced by a relatively small number of individuals with high levels of physical abuse. Replication of the current results with larger sample sizes of children with high levels of maltreatment exposure is needed.

Despite extensive evidence that maltreated children are at heightened risk for depression, there has been minimal empirical exploration of which individuals who experience maltreatment go on to develop depression. The present data suggest that maltreated children show patterns of heightened attention to sad cues that emerge a brief time after induction of a sad emotional state. Additionally, maltreated children with difficulties with regulation of sad affect, as indexed by ruminative approaches to regulation, show a more stable pattern of preferential attention to sad cues. These patterns of attention have been previously observed in individuals at-risk for and with diagnoses of depression, and may reflect a mechanism linking maltreatment with development of depression for some individuals.

Acknowledgements

Abbreviation

Footnotes

Contributor Information

Sarah E. Romens, Sarah Romens, Department of Psychology, University of Wisconsin – Madison, Waisman Center 1500 Highland Avenue, Room 399, Madison, WI 53705; ude.csiw@mherbes..

Seth D. Pollak, Seth Pollak, Department of Psychology, University of Wisconsin – Madison, Waisman Center, 1500 Highland Avenue, Room 399, Madison, WI 53705; ude.csiw@kallops..

References