When there is no external task to perform, thoughts often flow in a spontaneous and unconstrained manner. These ‘undirected’ forms of thought are a prevalent aspect of waking mental activity (Christoff, 2012). Our laboratory has begun to examine the neural basis of such spontaneous thought. In an initial fMRI study (Christoff, Ream, & Gabrieli, 2004), we found evidence of considerable contribution from regions in the medial temporal lobe, visual cortex, and the rostrolateral PFC, suggesting that long-term memory, visual imagery, and introspection may form the basis of spontaneous thought. Despite an upsurge of scientific interest in mind-wandering in subsequent years, studies on this phenomenon were limited to indirect methods – often relying on retrospective reports or using performance errors on a sustained attention task as a proxy for mind-wandering. Recognizing this, we undertook the first investigation of mind-wandering that employed an online experience sampling approach to index this phenomenon during fMRI scanning. This study (Christoff, Gordon, Smallwood, Smith, & Schooler, 2009), which combined objective measures of mind wandering (errors in task performance), subjective self-reports, and brain imaging, suggested that default network regions and executive network regions (e.g., dorsolateral prefrontal cortex, rostrolateral prefrontal cortex, and dorsal anterior cingulate cortex) are concurrently engaged during mind wandering—a finding predicted by behavioral theories of off-task thought.
In subsequent work, we argued that dreaming can be considered a form of mind wandering or spontaneous thought: the neural substrate of dreaming is highly similar to the default mode network regions strongly implicated in mind wandering, and subjective content reports from both waking ‘daydreams’ and night-time dreaming contain many strong similarities (Fox, Nijeboer, Solomonova, Domhoff, and Christoff, 2013). Expanding on the results of (Christoff et al., 2009), which suggested involvement of high-level PFC regions in mind wandering, we have highlighted a number of constructive interactions between metacognition and spontaneous thought processes (Fox and Christoff, 2014). We also published a meta-analytic review of functional neuroimaging research on mind-wandering, which revealed activation spanning the default mode network, frontoparietal control network, and other non-DMN regions (Fox et al. 2015).
We have also examined the neural dynamics surrounding the onset of a spontaneous thought. To do this, we have employed the introspective reports of highly-trained meditators (an approach referred to as ‘neurophenomenology’). In an fMRI study in this vein (Ellamil et al. 2016) we demonstrated that the medial temporal lobe subcomponent of the default network, involving regions such as the hippocampus, peak in activation prior to the onset of a spontaneous thought, while the frontoparietal control network regions, such as dorsolateral prefrontal cortex, peak following the onset of a spontaneous thought. In an EEG study using the same paradigm (Girn et al. 2017), we found a similar progression of connectivity that begins with default mode network and salience network regions, with the subsequent incorporation of control network regions in the period immediately preceding reports of thought onset.
In our most recent work, we have emphasized the limitations of the hitherto dominantly used approach to study spontaneous thought. This approach has typically understood spontaneous thought based on its content – for example, based on whether it is related or unrelated to one’s current task or immediate environment. In our novel Dynamic Framework of Thought (Christoff et al. 2016), published in Nature Reviews Neuroscience, we have argued that these static content-based measures of thought are unable to capture a core aspect of spontaneous thought: its dynamics. In this framework, we define spontaneous thought as “a mental state, or a sequence of mental states, that arises relatively freely owing to an absence of strong constraints on the contents of each state and on the transitions from one mental state to another.” Our framework shifts the emphasis towards the dynamics of the thinking process, and away from the content of a given set of thoughts. To empirically capture dynamic qualities of thought, we have recently introduced a novel experience sampling measure by probing participants with the question: “Is your mind moving about freely?” Using this measure, we have demonstrated that freely moving thought is empirically dissociable from task-unrelatedness (Mills et al. 2017), a finding that underscores the importance of using dynamic measures for the study of spontaneous thoughts. This framework explicitly links thought dynamics to fluctuations in the functional interactions of large-scale brain networks, and offers a novel way to understand clinically significant alterations in spontaneous thought and their neural basis.