Novel and non-routine tasks often require information processing and behavior to adapt from moment to moment depending on task requirements and current performance. This ability to adapt is an executive function that is referred to as cognitive control. Patients with moderate-to-severe traumatic brain injury (TBI) have been reported to exhibit impairments in cognitive control and functional magnetic resonance imaging (fMRI) has provided evidence for TBI-related alterations in brain activation using various fMRI cognitive control paradigms. There is some support for greater and more extensive cognitive control-related brain activation in patients with moderate-to-severe TBI, relative to comparison subjects without TBI. In addition, some studies have reported a correlation between these activation increases and measures of injury severity. Explanations that have been proposed for increased activation within structures that are thought to be directly involved in cognitive control, as well as the extension of this over-activation into other brain structures, have included compensatory mechanisms, increased demand upon normal processes required to maintain adequate performance, less efficient utilization of neural resources, and greater vulnerability to cognitive fatigue. Recent findings are also consistent with the possibility that activation increases within some structures, such as the posterior cingulate gyrus, may reflect a failure to deactivate components of the default mode network (DMN) and that some cognitive control impairment may result from ineffective coordination between the DMN and components of the salience network. Functional neuroimaging studies examining cognitive control-related activation following mild TBI (mTBI) have yielded more variable results, with reports of increases, decreases, and no significant change. These discrepancies may reflect differences among the various mTBI samples under study, recovery of function in some patients, different task characteristics, and the presence of comorbid conditions such as depression and posttraumatic stress disorder that also alter brain activation. There may be mTBI populations with activation changes that overlap with those found following more severe injuries, including symptomatic mTBI patients and those with acute injuries, but future research to address such dysfunction will require well-defined samples with adequate controls for injury characteristics, comorbid disorders, and severity of post-concussive symptoms.
Traumatic brain injury (TBI) is a neurological insult of major public health significance with over 1.7 million new injuries each year among Americans under the age of 35 (1). Numerous studies, most of which have been conducted with moderate-to-severe TBI due to blunt head trauma, have reported findings consistent with a mixed and highly heterogeneous neuropathology that may include multifocal or diffuse axonal injury, as well contusions and other focal lesions (2). Additional injury may occur as a result of edema, herniation, hemorrhage, ischemia, inflammation, and excitotoxic processes (2, 3). Structures and connections of the frontal and limbic regions have been said to be especially vulnerable to these various pathological processes (3, 4). Executive functions are highly dependent on the integrity of this neural substrate, and it is not surprising that such functions, including cognitive control, are often impaired following TBI (5–7).
Cognitive control allows for flexibility in human thought and behavior and may be defined as the ability to pursue task-related goals in the presence of conditions that include conflicting information or interference, prepotent response alternatives, or the need to interrupt or switch an ongoing activity (8–10). A common factor in all of these situations is the top-down direction, or biasing, of cognition and this is necessary for information processing and behavior to adapt from moment to moment depending on task requirements and performance (8, 11, 12). Cognitive control relies upon the active maintenance of neural activity associated with the internal representation of goals and task-related rules or contingencies (11–13). However, it is a complex construct that likely includes multiple component processes, some of these processes overlap with those of other executive functions (e.g., working memory), and it contributes to performance on various high level cognitive tasks, including those representing domains such as attention, memory, and language (8–10, 14).
Although prefrontally guided top-down direction is critical for cognitive control and other executive functions, the prefrontal cortex (PFC) is only one of several structures that contribute to cognitive control (15). Another important structure is the anterior cingulate cortex, which is thought to monitor performance and internal bodily states associated with task-related reward conditions, to determine whether task performance is adequate, and to signal to the dorsolateral PFC when mental effort or top-down direction needs to be increased (11, 15–17). Some anterior cingulate functions, including the detection of states associated with reward and expected outcomes, likely depend on distant connections with structures such as the insula (17). These various structures may be vulnerable to disconnection associated with diffuse axonal injury and other TBI-related neuropathology (18–20).
Functional magnetic resonance imaging (fMRI) provides an indirect measure of neural activity and has the potential to reveal changes in brain function associated with neuropathology, including alterations following TBI (21, 22). One powerful application of this method is the use of fMRI paradigms to examine brain activation during cognitive tasks (22), including those which place a demand upon executive functions such as cognitive control. This type of research has the potential to reveal relationships between specific cognitive impairments and dysfunction within the underlying neural substrate, to provide a neuroimaging marker that may contribute to differential diagnosis, and to lead to the development of methods to track changes in brain activity associated with recovery and treatment (23). Cognitive control is a high level function that is critical for the completion of many complex and non-routine tasks (8, 11). Despite the importance of this topic and the incredible potential offered by fMRI research, only a few studies have examined changes in cognitive control-related activation following TBI, and these have often suffered from various methodological limitations. The purpose of this article is to provide an overview of that existing research, to discuss findings that contribute to our understanding of how cognitive control may be impaired following TBI, and to provide some suggestions to improve future research and increase its relevance.
Although fMRI research has also investigated working memory and other executive functions following TBI (24, 25), the current review will focus on cognitive control by examining fMRI studies that have specifically addressed the top-down direction of cognition and related cognitive control processes (e.g., performance monitoring). This research has employed fMRI paradigms adapted from common clinical measures of cognitive control, such as the Stoop Test (26), as well as experimental procedures developed specifically for the purpose of acquiring fMRI data [e.g., Ref. (27)]. Studies using paradigms that assess other functions, such as working memory or attention, are also included within this review if they had incorporated procedures to investigate top-down control [e.g., Ref. (28)]. Some had utilized a block design approach [e.g., Ref. (29)], whereas others had employed event-related fMRI [e.g., Ref. (30)]. A major feature of block design fMRI paradigms is that this method combines images acquired across an entire block of trials, which then prevents the separation of images acquired within a block to examine activation relative to different types of stimuli or responses (31). Event-related designs have the advantage of allowing the examination of images at the trial level, including the ability to isolate correct or incorrect responses, but these designs typically have less statistical power (31). It is also possible to capitalize upon some of the advantages of both approaches by employing a mixed design (32). […]