The reorienting system of the human brain: from environment to theory of mind

Abstract

Survival can depend on the ability to change a current course of action to respond to potentially advantageous or threatening stimuli. This "reorienting" response involves the coordinated action of a right hemisphere dominant ventral frontoparietal network that interrupts and resets ongoing activity and a dorsal frontoparietal network specialized for selecting and linking stimuli and responses. At rest, each network is distinct and internally correlated, but when attention is focused, the ventral network is suppressed to prevent reorienting to distracting events. These different patterns of recruitment may reflect inputs to the ventral attention network from the locus coeruleus/norepinephrine system. While originally conceptualized as a system for redirecting attention from one object to another, recent evidence suggests a more general role in switching between networks, which may explain recent evidence of its involvement in functions such as social cognition.

Figure 1. Focusing Attention and Reorienting Attention Recruit Interacting Networks

(Left panel) Focusing attention on an object produces sustained activations in dorsal fronto-parietal regions in the intraparietal sulcus, superior parietal lobule, and frontal eye fields, as well as visual regions in occipital cortex (yellow and orange colors) but sustained deactivations in more ventral regions in supramarginal gyrus and superior temporal gyrus (TPJ) and middle and inferior prefrontal cortex (blue and green colors). (Right panel) When an unexpected but important event evokes a reorienting of attention, both the dorsal regions and the formerly deactivated ventral regions are now transiently activated.

Figure 2. Definition of Dorsal and Ventral Networks from Activation Data and Putative Interactions

(Top panel) Results from a meta-analysis of activation data. Regions in blue are consistently activated by central cues, indicating where a peripheral object will subsequently appear or what is the feature of an upcoming object. Regions in orange are consistently activated when attention is reoriented to an unexpected but behaviorally relevant object. (Bottom panel) Model for the interaction of dorsal (blue) and ventral (orange) networks during stimulus-driven reorienting. Dorsal network regions FEF and IPS send top-down biases to visual areas and via MFG to the ventral network (filtering signal), restricting ventral activation to behaviorally important stimuli. IPS-FEF are also important for exogenous orienting. Overall, the dorsal network coordinates stimulus-response selection. Conversely, when a salient stimulus occurs during stimulus-driven reorienting, the ventral network sends a reorienting signal to the dorsal network through MFG.

Figure 3. Functional Connectivity Defines Separate Dorsal and Ventral Networks

(Top panel) Four dorsal frontoparietal regions from the meta-analysis of activation studies shown in Figure 2 were used as seeds in an FC analysis of resting-state data. The map indicates regions that showed significant positive correlations with three (red) or four (yellow) of the seed regions. The dorsal network is largely reproduced in the resting-state FC maps. Regions that show significant negative correlations with three (green) or four (blue) of the seed regions are also shown and roughly reproduce the default network, possibly indicating a push-pull relationship between the two networks. (Bottom panel) Five ventral regions from Figure 2 were used as seeds for an FC analysis. Regions showing consistent positive correlations largely reproduce the ventral network, but negative correlations in default regions are not observed. The black arrow indicates that posterior MFG near the inferior frontal sulcus appears to be connected to both networks.

Figure 4. Ventral Network Activity Is Restricted to Task-Relevant Stimuli

(A) Exogenous orienting does not activate TPJ. Subjects saw a color singleton (exogenous cue) or a heterogeneous colored array (neutral cue) followed by a target (cue and cue period in light blue). Behavioral performance was speeded when the target location matched the singleton location, even though the cue and target locations were random. The time course of the BOLD signal shown in the graph indicates that R FEF (see “A” in the surface-rendered brain) showed a larger response for exogenous than neutral cues. In contrast, SMG showed a small deactivation during the cue period, followed by a small activation when the cue period ended (Kincade et al., 2005). (B) Salient irrelevant distracters influence the dorsal, not ventral, system. Subjects categorized the orientation of a line within a singleton shape (the circle). Salient but irrelevant singleton color distracters that impaired behavioral performance activated dorsal region SPL (see “B” in brain) rather than TPJ (de Fockert et al., 2004). (C) Unattended stimuli only activate TPJ if they are task relevant, not if they are irrelevant, even though they have high sensory salience. A task-relevant unattended letter activated angular gyrus and inferior frontal gyrus (see “C” in brain), but no responses were seen to the unattended but highly salient checkerboard. The angular gyrus response may reflect the combined activation of dorsal (IPS/SPL) and TPJ regions (Indovina and Macaluso, 2007). (D) Distracters only activate TPJ if they share features with a target, indicating a strong effect of task relevance. Subjects identified red foveal letters while ignoring irrelevant peripheral letters. Peripheral letters that matched the target color interfered with performance and activated TPJ, while non-target-colored letters had no effect (“D” in brain) (Serences et al., 2005).

Figure 5. TPJ Activity Is Suppressed during Focused Attention

(A) Subjects searched a rapid serial visual presentation (RSVP) display for a target digit. The number of distracter frames containing only letters prior to the target frame containing the target was varied. The graph shows the time course of activity in dorsal regions IPS and FEF and ventral region TPJ under conditions in which the target appeared near the end of the trial. In TPJ, a deactivation to the letter distracters was followed by an activation when the digit was presented or the trial was terminated. Interestingly, the deactivation to the letters was significantly greater when the subsequent digit was detected than when it was missed. Conversely, IPS and FEF showed sustained activations during search (Shulman et al., 2003). (B) Subjects encoded a visual display that they had to remember and then match to a probe display. During the retention interval, TPJ showed a deactivation (purple disk in the surface-rendered brain) that increased with the number of display items that had to be retained (Todd et al., 2005). (C) The statistical map shows regions with sustained activity as subjects searched through letter distracters in the RSVP experiment (see panel [A]), including dorsal attention regions IPS and FEF (red/orange in surface-rendered brain) but also regions in anterior insula and anterior cingulate that form a putative task-control network (Dosenbach et al., 2006). These regions may send top-down signals (see arrows) to the ventral network, which showed sustained deactivations during search (blue/green in surface-rendered brain), restricting its input to task-relevant objects (Shulman et al., 2003).

Figure 6. Interaction of Dorsal and Ventral Attention Networks

(A) The surface-rendered brains show the damaged right hemisphere regions (in dark gray) of a stroke patient with spatial neglect. The bottom graph shows the time course of BOLD activity in undamaged regions of IPS, with the green and red lines indicating, respectively, the time series for the indicated left and right IPS regions. Time courses from these regions are shown for a healthy subject, for the stroke patient immediately following the stroke, and for the same patient following recovery. While the healthy subject and the recovered stroke patient show highly correlated interhemispheric IPS activity, the same patient immediately after the stroke shows activity that is much less correlated. Therefore, damage to ventral regions, possibly including white matter tracts, impairs physiological interactions between undamaged dorsal regions (He et al., 2007a). (B) The surface-rendered brains show ventral (left) and dorsal (right) regions that are activated when the completion of a symphonic movement is detected. The time courses indicate that ventral activations (red lines) preceded the dorsal activations (blue lines), while a Granger Causality analysis of these regions indicated that ventral activity predicted dorsal activity (Sridharan et al., 2007).

Figure 7. Common Activation of TPJ during Reorienting, Task Transitions, and Social Cognition

(A) Regions in green show transient activity at the transition between a rest period and the onset of an extended block of trials (see time course in inset, indicating that both onsets and offsets are often observed). “Start cue” activity is observed within some ventral and dorsal regions, indicating the involvement of both networks in task transitions (Dosenbach et al., 2006). (B) A meta-analysis of activations across studies measuring reorienting and various aspects of social cognition. Largely similar TPJ activity is observed across paradigms, with perhaps a more posterior extension of activity in the social cognition paradigms (Decety and Lamm, 2007). (C) A within-subject comparison of reorienting and ToM paradigms revealed that both activated very similar TPJ regions (Mitchell, 2007). The bar graph shows the magnitude of the TPJ activation in the two paradigms. A large deactivation was observed in the control “false-photograph” condition with a significantly smaller deactivation in the experimental ToM “false-beliefs” condition. The reorienting paradigm yielded event-related activations that were larger during trials with invalid than valid cues. The blue and orange regions are taken from the meta-analysis of the dorsal and ventral networks in Figure 2. (D) Gaze perception activates superior temporal regions. The graph shows the time course of activity when another person makes or averts eye contact with the observer. Mutual gaze enhances the activation (Pelphrey et al., 2004).

Figure 8. Relationship between Activity in TPJ and Locus Coeruleus/Noradrenergic System

The surface-rendered brains show fMRI BOLD activations and deactivations relative to when subjects are fixating in an otherwise blank field (i.e., the baseline, left panel), when searching through letter distracters of an RSVP display (middle panel, ventral network is deactivated, dorsal network is activated), and when detecting a digit target in the display (right panel, both networks are activated along with other regions) (Shulman et al., 2003). The bottom panel shows spiking activity in monkey locus coeruleus neurons during analogous periods: an inattentive period in which a task is poorly performed and tonic activity is high, an attentive period in which the task is performed well and tonic activity is decreased, and target detection, which produces a phasic increase in activity (Usher et al., 1999). The inset trace shows event-related potentials recorded from the scalp of a human when a target is detected in a completely separate experiment, with the large positive deflection indicating the P300.