Attention is a broad concept that is founded on the observation that people are limited in their ability to do many things at once, a limitation that has two basic causes.  First, the brain has a limited capacity to perceive and respond to events in the environment and therefore must concentrate on those that are most important given the person’s goals. Second, for behavior to be effective, it must be coherent and organized, and not directed toward contradictory purposes. For both reasons, the brain has developed powerful mechanisms for selecting and responding to important information in the environment, while ignoring or putting in the background events of less immediacy. Much of the research in this lab concerns the different neural systems that allow selective perception and responding, which are all grouped under the heading, ‘selective attention’.
            A helpful organizing principle concerns the sources and sites of selective attention. Sources refer to the brain systems that represent a person’s goals (i.e. when you’re hungry, finding an orange in the refrigerator), and send out ‘top-down’ signals to sensory areas so that events in the environment that meet the goal are analyzed with a high priority. Sites refer to the sensory-perceptual areas in the brain where the analysis of sensory events is influenced by these top-down signals.  Dorsal areas in frontal cortex and parietal cortex are key components in a brain system that sends out top-down signals reflecting the person’s current goals.  When we have a particular goal in mind, these areas are involved in determining what parts of the environment are carefully analyzed, i.e. what parts of the environment we attend to. For example, studies in our lab have shown that when a cue is given to attend to a particular part of space, areas in dorsal fronto-parietal cortex show enhanced activity and this activity can be observed in those parts of cortex that are selective for the cued region of space.  Moreover, this enhanced activity occurs for as long as attention is maintained at the location. Similar signals can be observed in low-level occipital areas, which are thought to be the ‘site’ of attentional effects.  These results immediately raise an important question: is there any evidence that the presumptive ‘source’ signals in dorsal fronto-parietal cortex cause the presumptive ‘site’ signals in occipital cortex? A recent study confirmed that during voluntary attention to a spatial location in the absence of a stimulus, the relationship between fronto-parietal cortex and occipital cortex was indeed asymmetric.  This work, a collaboration between our lab and the lab of Steven Bressler at South Florida University, used Granger Causality to show that under these conditions, occipital cortex was modulated by fronto-parietal cortex, not the reverse.
           While it has been known for a long time that parts of visual cortex are organized into a series of maps in which different regions of space are organized in an orderly topographic fashion, this principle also applies to higher-level dorsal fronto-parietal areas.  When we attend to a location in space, how is information in the corresponding part of the map given a high priority?  Studies in our lab by Chad Sylvester have shown that the priority is not specified by the absolute neural activity in that part of the map, but by the activity in that part of the map relative to the activity in the corresponding part of the map located in the opposite hemisphere.  Therefore, what matters is the balance of activity, a coding principle that eliminates effects of noise that are common over a map.  This principle was demonstrated both for lower-level maps in sensory cortex and for higher-level maps in dorsal fronto-parietal cortex.
           Chad’s work has also shown that activity in these lower level maps depends not simply on where we are attending in space, but also on the nature of the task we are trying to accomplish.  When it is important to detect an object that is difficult to see, top-down signals from fronto-parietal cortex depress activity in parts of the map that are unlikely to contain the object so that irrelevant noise does not interfere with our perception. When the object is easy to see but the task is to distinguish some fine detail in the object, low-level noise in other parts of the map will not interfere with perception.  Correspondingly, under these conditions, these parts of the map do not show as much depressed activity.
           The above work has examined how attending to a location influences neural activity in maps so that information at that location is given a high priority and influences subsequent behavior.  But survival can also depend on being sensitive to important information that occurs outside of where we are currently attending, as when a predator unexpectedly appears. We have discovered that a second neural system, based in ventral parietal and frontal cortex, serves as an alerting or switching signal that is particularly well activated when important events occur outside the focus of attention.  Activation of this system, in conjunction with activation of the above-described dorsal fronto-parietal network, shifts our attention to the new event. The ‘importance’ of an event, which determines whether this system is activated, may depend on enduring biological features but also on the changing requirements of a task. When we are searching for our car in a parking lot, any car of the appropriate color may trigger this system and shift our attention.  Work in this laboratory has helped to identify this second attentional system, which had not previously been described, and define some of its properties.  For example, when we are focused on a task at a particular location, this system is deactivated, preventing shifts of attention that might interfere with the task. Of course, if the system were simply shut down, it would not be possible to reorient attention to an important stimulus.  Therefore, the deactivation instead appears to reflect a filtering of the input to this system so that it is only activated by important stimuli.
           Although we talk of dorsal and ventral fronto-parietal systems, why should they be thought to act as a ‘system’? One criterion is that regions within a system are coactivated during tasks, i.e. activation of one area that is timelocked to a task or to a particular event in a task occurs in conjunction with timelocked activation in the other. Interestingly, this principle can be extended even to situations in which subjects are at rest and not performing a task, and the measured activity reflects ‘intrinsic’ or ‘spontaneous’ rather than task-evoked activity. In fact, the same areas that are co-activated during tasks show high correlation of activity over time in the resting state.  The interpretation of resting-state activity is an important topic that is currently under investigation, but this correlated activity provides a powerful tool for defining the presence of distinct, coherent brain networks. Using this tool, we have been able to show that the same dorsal and ventral attention networks that were initially defined by task-evoked co-activations, can also be identified through analyses of resting-state activity.