Trains of Thought: Distinctive Characteristics of Mind-Wandering
John Locke observed that "[W]hilst we are awake, there will always be a train of ideas succeeding one another in our minds" (Locke, 1689/ 1979, II.19, section 1). We can distinguish these trains of thoughts along at least two dimensions: (a) types of links between thoughts (i.e., how the individual items in a train of thought are linked together) and (b) types of thoughts (i.e., the kinds of thought elements that are linked together in the train). Mind-wandering involves discursive links that connect episodic thought, and each is discussed in turn.
Mind-wandering involves trains of thought that are linked together in a substantially discursive fashion; thoughts appear to be drawn somewhat randomly from some relevant domain (Irving, 2015). Contrast this with "focused" trains of thought that arise during certain other tasks: doing mental arithmetic with three-digit numbers, planning next week's dinners, or solving the Tower of Hanoi problem. In these tasks, the mind stays focused on a particular topic and follows a structured sequence of operations. A frontoparietal brain network implements a number of critical functions for sustaining these sequences. For example, the network performs executive functions, such as maintaining the instructions for producing the task-specific sequences (often referred to as task sets) (Baddeley 1996; Monsell, 2003). It is also the source of attentional signals for selecting certain representational contents to be sustained and manipulated (Corbetta & Shulman, 2002; Woldorff et al., 2004), as well as monitoring and control signals that regulate the flow of thought so that it does not deviate from prescribed lines (Miller, 2000; Miller & Cohen, 2001).
There is evidence that mind-wandering also activates certain regions of this same frontoparietal network (Christoff et al., 2009; Fox, Spreng, Ellamil, Andrews-Hanna, & Christoff, 2015; Smallwood, Brown, Baird, & Schooler, 2012; Teasdale et al., 1995). This finding may initially seem to be at odds with the claim that the discursivity of mind-wandering is to be contrasted with the organized sequencing of focused modes of thought. The seeming tension, however, can be reconciled by noting that the frontoparietal network in question performs a number of distinct storage, executive, attentional, monitoring, and control functions. Mind-wandering and focused modes of thought might engage distinct processes that fall under these headings, or, even if the very same processes are engaged, they might be engaged in substantially different ways.
To illustrate these possibilities, consider that mind-wandering appears to involve conscious awareness of trains of discursive thoughts (as was noted in "The Extended CLS Model" section). According to a number of leading theories (Baars, 1997; Carruthers, 2015; De Brigard & Prinz, 2010; Dehaene & Naccache, 2001; Prinz, 2012), the deployment of attention is required for conscious awareness; when attention targets certain mental representations, those representations are selectively strengthened, rendering them conscious. Focused trains of thought also involve conscious awareness, in this case of tightly sequenced thoughts that are organized by task sets and sustained against distraction by cognitive control signals. For focused trains of thought, then, in addition to using attention for conscious awareness, additional frontoparietal resources are required, including executive resources to maintain the task sets (Baddeley, 1996; Baddeley, Chincotta, & Adlam, 2001) and monitoring and control resources to generate cognitive control signals (Miller, 2000; Miller & Cohen, 2001). If this understanding is correct, mind-wandering and focused trains of thought should both engage frontoparietal regions, but they should exhibit distinct, though potentially partially overlapping, neural signatures. Moreover, focused thought, because it engages more extensive frontoparietal resources, would plausibly activate frontoparietal regions more vigorously. To test these predictions, it would be particularly useful to perform studies that examine the neural underpinnings of mind-wandering and focused trains of thought in the same individuals, thus allowing head-to-head comparison of their respective neural profiles.
The claim that mind-wandering is discursive might also, at first pass, seem to fit poorly with evidence that mind-wandering is strongly influenced by personal goals. For example, Morsella and colleagues (Morsella, Ben-Zeev, Lanska, & Bargh, 2010) found that when people are told they will later face a quiz about certain geography facts, their subsequent mind-wandering was concerned with geography 70% of the time. In comparison, control subjects, who were told of an upcoming geography test and then immediately after told they would not in fact have to take it, subsequently thought about geography only 10% of the time. Other studies using experience sampling (Baird, Smallwood, & Schooler, 2011) and diary methods (Klinger & Cox, 1987) also corroborate the connection between mind-wandering and personal goals.
Does this body of evidence contradict the claim that mind-wandering is discursive? No, the two claims can be reconciled. The key to seeing this is to note that interleaved learning requires the specification of a domain of examples. Recall that in the extended CLS architecture, the discursivity of mind-wandering is explained through the computational advantages of interleaved learning: When the elements from some domain of learning examples are repeatedly rebroadcast (and the order of broadcast is substantially varied), this greatly facilitates deep learning of the meaningful patterns in the learning set. The present proposal is that personal goals can influence mind-wandering by biasing the selection of a domain of examples from which interleaved sampling subsequently occurs. For example, if a person is concerned with doing well on a geography test, then the relevant domain of examples might consist of the set of one's geography-relevant experiences, and mind-wandering will discursively draw from this set. If an important relationship, perhaps with a spouse, is threatened, then the relevant domain of examples might include experiences with that person, or experiences with significant others more broadly. There is no need to suppose the person intentionally or explicitly selects a certain domain of examples for the purposes of mind-wandering. Rather, the proposal is that one's personal goals have an automatic and nonconscious biasing effect in selecting a relevant domain of examples from which mind-wandering discursively samples.
It is interesting to speculate that rumination represents a limiting case of this phenomenon. If personal goals bias and constrain the domain from which mind-wandering mechanisms sample, it might be possible for certain goals, especially goals that are relatively strong and specific, to constrain this domain excessively. The result will be that one's spontaneous thoughts take on the repetitive, per- severative quality that is characteristic of rumination. As was noted earlier, trains of topic-constrained thought can also be generated via the maintenance of task sets and the generation of cognitive control signals, as occurs during focused activities such as planning, deliberation, and problem-solving. On the current proposal, however, the mechanism that maintains trains of topical thought during rumination is quite different. Rumination has the spontaneous quality of mind-wandering; ruminative thoughts simply pop into one's stream of consciousness unbidden. The maintenance of topical focus is achieved, it is proposed, by the biasing effects of personal goals; they constrain the domain of experiences from which mindwandering mechanisms discursively sample sufficiently tightly that there is little space for the mind to wander to a variety of internal episodes. Some tentative initial evidence for this "constrained mindwandering" model of rumination comes from a number of neuroimaging studies that link hyperactivity of the default network, a brain network centrally implicated in mind-wandering, with ruminative thoughts (Berman et al., 2011; Kross, Davidson, Weber, & Ochsner, 2009) as well as depression (Nejad, Fossati, & Lemogne, 2013; Sheline et al., 2009), a disorder characterized by rumination.