**1. Introduction**

Pain is a unique and unorthodox sense. We reflexively respond to pain without being conscious of it, ye<sup>t</sup> the conscious experience of pain can be overwhelming, debilitating, and chronic, depending on the cognitive, emotional, and allostatic context. Although the spinal reflexes mediate the rapid, unconscious avoidance of nociceptive stimuli, the experience of pain relies on the central nervous system's amplification and abstraction of nociception. In chronic pain, the healthy, adaptive relationship between experiential pain and nociception becomes disrupted. Pain-inducing contextual factors often replace pain-inducing nociception.

Two general categories of chronic hypersensitivity to pain exist. Allodynia is pain resulting from hypersensitivity to normal innocuous somatosensory (non-painful) stimuli. For example, pain resulting from brushing a hand with a feather [1]. Hyperalgesia is increased sensitivity, intensity, or duration of painful stimuli [1,2]. Both of these conditions can arise from hyperexcitability of the dorsal horn of the spinal cord (i.e., where pain fibers first synapse). However, it is thought these conditions may also be "centralized," presumably via inappropriate neuronal computations in a circuit comprised of the anterior cingulate cortex (ACC), insular cortex, and thalamus. For example, a burn can transiently cause a reduction of the pain threshold for mechanical stress, but the cognitive and a ffective elements of a burn can induce chronic hyperalgesia. Allodynia and hyperalgesia occur with spontaneous pain in chronic pain disorders.

Chronic and acute pain have sparked a growing interest in recent years. Chronic pain is widespread. In 2008, an estimated 100 million adults in the United States (U.S.) were affected by chronic pain. Total health care expenses attributable to pain, along with the amount associated with lower worker productivity, cost the U.S. between \$560 and \$635 billion per year [3]. This estimate is higher than the annual cost of heart disease (\$309 billion), cancer (\$243 billion), or diabetes (\$188 billion) [3,4]. Pain treatment is complex and multidimensional, often involving medical providers, pain clinics, and state governments regulating narcotic drugs [5]. Narcotics, specifically opioids, are a poor treatment solution for two significant reasons—tolerance and lethality. Opioid use is associated with tolerance, as larger doses are required to achieve the same level of pain relief, ye<sup>t</sup> high doses of opioids can be lethal. In 2016, 64,000 people died from drug overdoses in the U.S.—42,000 of those deaths were opioid related [6]. Opioids are not a viable long-term treatment for chronic pain. One potential solution is to design less systemic, more targeted interventions to disrupt pain signals within the brain, for example with electrical stimulation.

The use of intracranial electrical stimulation began in the mid-1900s, with the development of stereotactic frames. These devices enabled early psychologists/physiologists/psychiatrists to explore the effects before lesioning the areas of interest for a therapeutic effect [7]. Heath, a psychiatrist [8], implanted electrodes in schizophrenic patients to study the effects of intracranial stimulation. He discovered that electrical stimulation resulted in euphoria and analgesia, especially with stimulation of the septal area [9–14]. In 1954, deep brain stimulation (DBS) rat experiments alleviated pain with stimulation targeted to the septal nuclei, mammillothalamic tract, and cingulate cortex [14–16]. After these promising animal studies, stimulation of the human septum (intending to treat pain) proved less successful, with only one out of six patients with terminal cancer experiencing reduced pain [16,17].

The motivation to treat pain using electrical stimulation increased with the introduction of Melzack and Wall's gate theory in the 1960s [18,19]. Gate theory is the notion that non-painful inputs close nerve "gates" through the activation of Aβ fibers. Gate closure disrupts a painful input from traveling along signaling pathways to the central nervous system, a process that is the basis for spinal cord stimulation (SCS). SCS is currently approved by the U.S. Food and Drug Administration (FDA) to treat chronic pain of the trunk and limbs, intractable low back pain, leg pain, and pain from failed back surgery syndrome. In Europe, SCS is also approved for refractory angina pectoris and peripheral limb ischemia [20]. Success using SCS suggests that stimulation of other targets within the central nervous system may also modulate pain signaling. Positive results demonstrated by cortical stimulation support this. A recent pooled effect from 12 trials found motor cortex stimulation improves pain by 65.1% in postradicular plexopathy, 46.5% in trigeminal neuropathy, 35.2% after stroke, 34.1% in phantom limbs, and 29.8% in plexus avulsion [21]. Motor cortex stimulation works by affecting the activity in the thalamic nuclei and somatosensory regions. It modulates a vast network of structures, including the cerebellum, striatum, ventral posterolateral nucleus, and other thalamic areas. A more targeted approach is highly favorable to decrease the negative side effects associated with electrically stimulating all of these structures.

Whereas the path from Melzack and Wall's gate theory to present-day SCS is well defined, the path of DBS through the late 1900s meanders. After the disappointing results of septal DBS, studies continued on a wide array of potential anatomical brain targets in rodents and humans throughout the mid-to-late 1900s [22–28]. When studies translated from animals to human studies, they varied in electrode numbers, stimulation parameters, and anatomical targets, leading to inconsistent conclusions [17]. The FDA requested industry intervention to provide further data on safety and efficacy [14,29]. The first Medtronic study failed to show that half the patients had at least 50% pain relief [17]. The second trial failed because of a lack of enrollment (further details provided in Section 2). In 1989, the FDA rejected using DBS for chronic pain treatment [29].

To design safe and effective targeted therapies, such as DBS, it is critical to understand more about the circuits involved in pain processing in time and space and the potentially distributed nature of pain encoding in the human brain. The peripheral and spinal circuitry involved in pain is well characterized [30]. Here, we instead focus on the cerebral localization of pain networks, with an eye towards the development of targeted neurostimulation therapies for chronic pain [17]. The structures discussed in this review—such as the somatosensory cortex, thalamus, insula, and the ACC—consistently correlate with painful stimuli [17,31–34]. Neuroimaging studies routinely demonstrate activation in these areas, as shown in Figure 1 [35]. Therefore, we discuss the processing of conscious perception of pain in the cerebrum, the historical electrical stimulation of these regions, if pertinent, and the future application of DBS to modulate this activity.

**Figure 1.** Overlay of three color-coded pain-related terms. "Chronic pain" is blue, "painful" is yellow, and "pain" is red. Functional magnetic resonance imaging (fMRI) studies of these terms are visualized in coronal (**a**), axial (**b**), and sagittal (**c**) axes from the Neurosynth database (neurosynth.org), showing consistent activation of the anterior cingulate cortex (ACC), thalamus, insula, and brainstem.
