Age differences in the default network at rest and the relation to self-referential processing

Abstract

Older adults show a 'positivity bias' in tasks involving emotion and self-referential processing. A critical network that is involved in self-referencing and shows age-related decline is the default network (DN). The purpose of the current study was to investigate age differences in pre- and post-task DN functional connectivity (FC) and signal variability, and to examine whether they are predictive of the positivity bias in self-referencing. We measured FC and within-subject variability of the DN in resting-state scans preceding and following tasks involving personality judgements on the self and a close other. Older adults endorsed more positive traits than younger adults on both tasks. FC was weaker post-task in older vs younger adults, and younger adults had greater variability than older adults in DN nodes. Younger adults with higher post-task DN variability had more negative self-ratings. For both age groups, greater FC in the DN was associated with more negative self-ratings. Neither FC nor variability was related to other ratings, despite the potential for self-processing when making other judgements. Our findings suggest that ageing leads to reduced FC and variability in the DN, which is most apparent after task, and may be one mechanism underlying the positive bias with age.

Keywords: ageing; default network; positivity effect; resting state; self-referential processing.

Fig. 1

The first LV pattern showing positive correlations between DN seed regions and resting-state brain regions, across age groups and time points. (a) Spatial FC pattern using the vmPFC seed (b) spatial FC pattern using the PCC seed. Activity in warm-colored regions (positive BSRs) indicates positive correlations between brain scores and the DN seed region (c) Correlations across time with the vmPFC. FC in younger adults was strengthened post-task relative to pre-task, while older adults maintained similar levels of FC. (d) Correlations across time with the PCC. Young and older adults show stable positive correlations between the PCC and DN regions. Older adults had weaker correlations compared with younger adults at post-task rest, using the vmPFC and PCC seed.

Fig. 2

(a) Correlations and (b) s.d. between DN regions at pre- and post-task rest for younger and older adults (error bars represent the within subject standard error of the mean and asterisk represent a significant difference across age groups at P < 0.05).

Fig. 3

Scatter plot showing the DN post-task resting-state correlations (between the vmPFC and PCC) for each participant and the proportion of yes responses for self-ratings. Although the age groups are indicated with different symbols, the regression line has been plotted for both age groups combined, because the interaction with age was not significant. In the graph, larger values of the rating indicate more endorsements of positive traits.

Fig. 4

Scatter plot showing the DN post-task resting-state variability (s.d. of the vmPFC and PCC) for each participant and the proportion of yes responses for self-ratings. Because the interaction with age was significant, the regression line has been plotted separately for young and older adults. In the graph, larger values of the rating indicate more endorsements of positive traits.