A network-level test of the role of the co-activated default mode network in episodic recall and social cognition

Abstract

Resting-state network research is extremely influential, yet the functions of many networks remain unknown. In part, this is due to typical (e.g., univariate) analyses independently testing the function of individual regions and not examining the full set of regions that form a network whilst co-activated. Connectivity is dynamic and the function of a region may change based on its current connections. Therefore, determining the function of a network requires assessment at this network-level. Yet popular theories implicating the default mode network (DMN) in episodic memory and social cognition, rest principally upon analyses performed at the level of individual brain regions. Here we use independent component analysis to formally test the role of the DMN in episodic and social processing at the network level. As well as an episodic retrieval task, two independent datasets were employed to assess DMN function across the breadth of social cognition; a person knowledge judgement and a theory of mind task. Each task dataset was separated into networks of co-activated regions. In each, the co-activated DMN, was identified through comparison to an a priori template and its relation to the task model assessed. This co-activated DMN did not show greater activity in episodic or social tasks than high-level baseline conditions. Thus, no evidence was found to support hypotheses that the co-activated DMN is involved in explicit episodic or social tasks at a network-level. The networks associated with these processes are described. Implications for prior univariate findings and the functional significance of the co-activated DMN are considered.

Keywords: Default mode network; Episodic memory; Independent component analysis; Resting-state networks; Social cognition.

Fig. 1

The spatial overlap between the a priori template obtained from Jackson et al. (2019) and the component identified to be the DMN in each task is shown. The components identified in the Episodic Task (E9), Social—Person Knowledge Task (P18) and Social—Theory of Mind Task (T8) datasets are shown in red. The template is shown in green. Overlap is shown in yellow.

Fig. 2

The task relation of the DMN component identified in each task is shown. While only the medial surface is presented here, these components are the same as shown in Fig. 1. Conditions with activity significantly different to rest (P < .05 after Bonferroni adjustment) are highlighted with an asterisk. Differences between conditions are demonstrated by a black line between the condition bars. A. The Episodic Task DMN component is displayed in violet. The involvement of the component in the episodic (violet) and baseline conditions (grey) is displayed against rest. The DMN component is significantly less engaged in the episodic condition than the control task. B. The DMN component identified in the Social—Person Knowledge Task is displayed in yellow. The activity of the component in the social semantic (yellow), non-social semantic (orange) and baseline conditions (grey) are shown compared to rest. The DMN component is not significantly related to the task model. C. The DMN component identified in the Social—Theory of Mind Task is shown in cyan. The activity of this component in the social motion (cyan) and non-social motion conditions (grey) are shown against rest. The DMN component was equally deactivated in both social and non-social conditions. The co-activated DMN was not found to be significantly more active in the condition of interest (episodic/social processing) than the baseline condition for any task.

Fig. 3

The components with greater involvement in the episodic than the baseline condition are displayed. Each component is shown in violet in the top panel and the task relation of the component is shown in the lower graph. The graph shows the effect size of each component in the episodic (violet) and baseline (grey) conditions compared to rest. Conditions with activity significantly different (P < .05 after Bonferroni adjustment) to rest are highlighted with an asterisk. A tilde is shown if there is a trend (P < .1 after Bonferroni adjustment) towards this difference. Differences between conditions are demonstrated by a black line between the condition bars.

Fig. 4

The single component with greater involvement in the theory of mind than the baseline condition is displayed. The spatial extent of the component is shown in cyan in the top panel and the task relation of the component is shown in the lower graph. The graph shows the effect size of each component in the social motion (cyan) and non-social motion conditions (grey) compared to rest. Conditions with activity significantly different (P < .05 after Bonferroni adjustment) to rest are highlighted with an asterisk. Differences between conditions are demonstrated by a black line between the condition bars.