Imagine driving to work along the same route you take each day. Your mind wanders from one thing to the next: the staff meeting in the afternoon, plans for the weekend, a gift you need to buy for a friend. Suddenly, a car cuts you off, and these thoughts immediately vanish—all of your attention focuses on maneuvering the steering wheel to avoid a collision. Although momentarily flustered, you—and your thoughts—return to the same wandering pattern a minute or two later.

As we go about our waking lives, our stream of consciousness typically cycles through many such alternations between introspection and outward attention throughout the day. It appears that the back-and-forth dance between these inward and outward mental states may be fundamental to brain function. A new study, led by neuroscientist Zirui Huang of the Center for Consciousness Science at the University of Michigan, suggests that the shifting balance between a network responsible for awareness of the environment and another responsible for awareness of self may be a defining feature of consciousness.

The evidence comes from the absence of this pattern of brain activity in people rendered unresponsive, whether by anesthesia or a neuropathologic condition. As well as advancing our understanding of consciousness, the work could lead to the development of techniques to monitor it, either prior to surgery or during the treatment of people with disorders of consciousness, such as vegetative or “locked-in” patients.

Over the past two decades, neuroscientists have identified a network of brain regions responsible for various kinds of introspection, from mind wandering to recollection and planning. The concept of “background” brain activity began drawing attention when neurologist Marcus Raichle and his colleagues at Washington University in St. Louis showed that the organ’s energy consumption rose by less than 5 percent when performing a focused mental task, suggesting it is never really idle. In 2001 Raichle coined the term “default mode” to describe this activity. Converging lines of evidence then led to identification of regions comprising the default mode network (DMN), which underlies this self-directed cognition.

Activity in the DMN is “anticorrelated” with activity in the so-called dorsal attention network (DAT): the more active one of the two networks is, the less active the other tends to be. Activity in the DAT corresponds to attention directed outward, while the DMN underlies consciousness of self. This arrangement provides a potential account of our conscious experience in terms of a reciprocal balance between two opposing neural networks. “It’s not an either-or thing; you’re just tipping a balance,” Raichle says. “We slide back and forth, but they’re both there to some degree.”

A portion of this research has remained controversial because of a method used to clean noise from brain scan data that some researchers argue will always generate anticorrelated patterns as an artifact of processing the data. In the study, published Wednesday in Science Advances, Huang and his colleagues avoided the issue by adopting an approach that did not use this processing method. They instead took advantage of machine-learning techniques to classify brain activation patterns into eight groups. Two of them corresponded to the DMN and DAT, and six are related to other known networks underlying brain functions: the sensory and motor network, the visual network, the ventral attention network, the frontoparietal network and two networks representing cross-brain states of activation and deactivation.

To capture the brain activity, the team used a technique called resting-state functional magnetic resonance imaging (rsfMRI). Rather than averaging activity over long periods, which is typically done when using rsfMRI to estimate how well-connected regions are, the researchers wanted to investigate how moment-to-moment brain activation unfolds over time.

They showed that the organ rapidly cycles through different states, corresponding to each of the eight networks, with some transitions being more probable than others—which Huang describes as a “temporal circuit. ” Notably, the brain passes through intermediate states between DMN and DAT activation rather than flipping instantaneously between these two extremes, which represents the highest-level cognitive processes.

The researchers scanned 98 participants, who were either lying still but conscious or in an unresponsive state. The latter was caused by propofol or ketamine anesthesia or by a neuropathological condition known as unresponsive wakefulness syndrome — a vegetative condition resulting from brain injury. All of these unresponsive states had one thing in common: the DMN and DAT were “isolated” from the constant flitting between networks of the temporal circuit, and they virtually never activated.

Each type of unresponsiveness varied in terms of the molecular mechanisms, neural circuits and experiences involved (those under ketamine anesthesia reported hallucinations, for instance). These observations could indicate that the absence of DMN-DAT activity is common to any form of diminished consciousness and that its presence may be a necessary feature of full consciousness. “What [the researchers are] suggesting here is: if you mess with that balance, you see a cost in consciousness,” says Raichle, who was not involved in the study. “It’s an interesting way to frame [DMN-DAT activity], and it’s descriptive of our consciousness. But does it explain it? I’m not sure.”

In another experiment, the researchers showed that playing a sound increased activation of the ventral attention network (which redirects our attention to unexpected stimuli) and suppressed activation of the DMN in conscious  participants but not in unresponsive ones. A final control experiment assessed network activation in a database of brain scans of psychiatric patients. The scientists found no difference between this group and conscious participants in terms of DMN and DAT activity, showing that its loss is specific to reduced responsiveness, not any form of disordered cognition.

There also were differences among the various unresponsive states. For instance, participants given ketamine more frequently entered cross-brains states of activation and deactivation. This pattern was also seen in schizophrenic patients’ scans, suggesting hyperactivated patterns may correspond to hallucinatory experiences common to both ketamine use and schizophrenia. “If all the processors share information everywhere in the brain, I guess you may lose the difference between yourself and the environment,” Huang says. “Everything occurs at once, and you have distortions of your mental content.”

The work could potentially be used to develop measures of consciousness for assessing the efficacy of treatments for disorders of consciousness or for online monitoring of anesthesia. “Once we see the two networks are diminished, we think individuals aren’t aware of their environment,” Huang says. Measures to gauge whether an individual is conscious or not could assist physicians in the surgical suite. He next plans to investigate the neural mechanisms that regulate these transitions in the temporal circuit comprising these brain networks—an exploration of what orchestrates the dancing dynamics of conscious activity.