Changes in the brain intrinsic organization in both on-task state and post-task resting state

Abstract

The dynamic and robust characteristics of intrinsic functional connectivity of coherent spontaneous activity are critical for the brain functional stability and flexibility. Studies have demonstrated modulation of intrinsic connectivity within local spatial patterns during or after task performance, such as the default mode network (DMN) and task-specific networks. Moreover, recent studies have compared the global spatial pattern in different tasks or over time. However, it is still unclear how the large-scale intrinsic connectivity varies during and after a task. To better understand this issue, we conducted a functional MRI experiment over three sequential periods: an active semantic-matching task period and two rest periods, before and after the task respectively (namely, on-task state and pre-/post-task resting states), to detect task-driven effect on the dynamic large-scale intrinsic organization in both on-task state and post-task resting state. Three hierarchical levels were investigated, including (a) the whole brain small-world topology, (b) the whole pairwise functional connectivity patterns both within the DMN and between the DMN and other regions (i.e., the global/full DMN topography), and (c) the DMN nodal graph properties. The major findings are: (1) The large-scale small-world configuration of brain functional organization is robust, regardless of the behavioral state changing, while it varies adaptively with significantly higher local efficiency and lower global efficiency during the on-task state (P < 0.05, Monte-Carlo corrected); (2) The DMN may be essentially engaged during both task and post-task processes with adaptively varied spatial patterns and nodal graph properties. The present study provides further insights into the robustness and plasticity of the brain intrinsic organization over states, which may be the basis of memory and learning in the brain.

Introduction

Spontaneous brain activity, utilizing the majority of brain energy in specific organizations over different states or time (Raichle and Mintun, 2006), may substantially account for the dynamic, robust intrinsic functional architecture of the brain (for a review, see Fox and Raichle, 2007). This activity is also related to the establishment of early cortical patterns (Price et al., 2006, Sur and Leamey, 2001) and the development of intrinsic functional networks over ages (Fair et al., 2008, Fair et al., 2009). From an evolutionary perspective at a much larger temporal scale, the brain organization has evolved to a marvelous highly-complex system, with supporting dynamic and effective integration of specialized local information processing (Sporns and Zwi, 2004) and a broad flexibility of cognitive processes (Sporns et al., 2004). So, it is important to investigate the brain intrinsic organization of coherent spontaneous activity for understanding how the brain works.

By functional connectivity (Friston et al., 1993), studies have identified many intrinsic spatial patterns in coherent low-frequency blood oxygen level dependent (BOLD) fluctuations of functional MRI (fMRI) during a continuous resting state. For instance, local spatial patterns have been identified among anatomically separated regions in neuroanatomical systems, including the motor (Biswal et al., 1995, Lowe et al., 1998), auditory (Cordes et al., 2001), visual (Lowe et al., 1998), language (Hampson et al., 2002), attention (Fox et al., 2006) and default-mode systems (the default mode network, DMN) (Fox et al., 2005, Greicius et al., 2003). The local spatial patterns can provide insights into the intrinsic functional architecture of the human brain (Fox and Raichle, 2007). On the other hand, for the global spatial pattern over the whole brain, topological properties such as the small-world characteristics (high clustering coefficient and short characteristic path distance) and power-law (or truncated power-law) degree distribution have been demonstrated (for reviews, see Bassett and Bullmore, 2006, Bullmore and Sporns, 2009, Sporns et al., 2004) in both large-scale functional (Achard et al., 2006, Eguíluz et al., 2005, Salvador et al., 2005a, Salvador et al., 2005b, Van den Heuvel et al., 2008) and structural (Hagmann et al., 2007, He et al., 2007, Iturria-Medina et al., 2008) networks. The small-world organization can support both functional segregation and integration (for reviews, see Bassett and Bullmore, 2006, Sporns and Zwi, 2004, Sporns et al., 2004), which are two fundamental organizational principles of the cerebral cortex (Friston, 2002, Tononi et al., 1998, Zeki and Shipp, 1988). It can also facilitate rapid adaptive reconfiguration of neuronal assemblies in support of the cognitive state changing (Bassett and Bullmore, 2006).

Studies have also demonstrated that the intrinsic spatial patterns of coherent spontaneous activity could be modulated during a task (Bianciardi et al., 2009, Calhoun et al., 2008, Fransson, 2006, Friston and Büchel, 2000, Hampson et al., 2002, Jiang et al., 2004, Liu et al., 1999, Lowe et al., 2000), or after task performance (Hasson et al., 2009, Lewis et al., 2009, Peltier et al., 2005, Tambini et al., 2010, Waites et al., 2005). We suggest that both on-task and subsequent resting (post-task) states could be called “task-driven” states. Of all the local intrinsic spatial patterns, the DMN is uniquely a set of brain regions remaining more active at rest than during task performance in an organized fashion (Fox et al., 2005, Greicius et al., 2003), and is thought to mediate processes that are important for the resting state (Raichle et al., 2001). Several local spatial patterns within the DMN have been demonstrated to be changed in both task-driven states, including an on-task state (Fransson, 2006) and a post-task resting state (Hasson et al., 2009, Tambini et al., 2010, Waites et al., 2005). At a larger temporal scale, studies have also suggested that intrinsic brain functional networks would be developing over ages in specific ways (Fair et al., 2008, Fair et al., 2009). By contrast, some other studies reported no significant change in a local DMN spatial pattern based on a seed region of posterior cingulate cortex (PCC) during task performance (Greicius et al., 2003, Hampson et al., 2006), and nor in a task-specific network after task performance (Albert et al., 2009).

On the other hand, for the variability of the global spatial pattern across states, a pioneer work by Bassett et al. (2006) using magnetoencephalograph (MEG) suggests that small-world properties are not sensitive to a visually cued finger tapping task. However, there are two limitations of that study: (1) they did not design sequential sessions of task and rest; (2) MEG sensors may be problematic in network node definitions, because the sensors may detect spatially overlapping signals (Ioannides, 2007, Rubinov and Sporns, 2010). There are other different studies investigating changes in large-scale brain network topology over time during learning (Bassett et al., 2011), or with respect to different working memory tasks (Ginestet and Simmons, 2011), or over ages at a large temporal scale (Fair et al., 2009, Meunier et al., 2009, Wang et al., 2010). It appears that the dynamic characteristics of intrinsic functional connectivity of coherent spontaneous activity are critical for the brain functional stability and flexibility. Hence, it is necessary to investigate the robustness and plasticity of the large-scale intrinsic organization of coherent spontaneous activity in terms of the “small-world” and the DMN topological properties together in both task-driven states.

The present study focused on a sequential procedure of pre-task resting, on-task and post-task resting states by fMRI, and investigated task-driven effect on the dynamic large-scale intrinsic functional organization in both on-task state and post-task resting state (see Fig. 1A). Three hierarchical levels of topologies were investigated, including (a) the whole brain small-world topology, (b) the whole pairwise functional connectivity patterns both within the DMN and between the DMN and other regions (i.e., the global/full DMN topography), and (c) the DMN nodal graph properties (see Fig. 1B). Small-world analysis could macroscopically characterize the balanced coordination between the local specialization and global integration of parallel information processing (functional segregation and integration). The DMN topological analysis could provide insights into lower-level dynamic topological properties in the large-scale intrinsic organization. Thus, the specific question examined in this study is what topological changes would occur in the large-scale intrinsic organization when the brain evolves from a pre-task-resting state to both task-driven states, in terms of (1) the small-world configuration and (2) the global DMN topography and nodal properties.

To answer this question, we conducted an fMRI experiment that recorded BOLD signals over three sequential periods: an active semantic-matching task period and two rest periods, before and after the task respectively. Under the three states, we constructed three groups of brain functional networks for each subject using a prior anatomical automatic labelling (AAL) atlas (see Table 1, 45 for each cerebral hemisphere, Tzourio-Mazoyer et al., 2002) in the low-frequency BOLD signals (0.01–0.08 Hz). Finally, the three-level topologies in terms of the small-world configuration, the global DMN topography and its nodal properties were investigated across states, and their differences between the task-driven states and pre-task resting state were further statistically evaluated. This three-level topological analysis could provide further insights into the robustness and plasticity of the intrinsic organization during or after a task.