Bridging Brain, Mind, and Behavior: 2005 Research Awards

Medical Research Council, London, United Kingdom
Principal Investigator: John Duncan, $449,729 over four years.
Neural mechanisms of attention studied with cross-species fMRI and single cell physiology

We constantly experience the spectacular power of selective attention. Arguing with a friend in the corridor, we fail to notice passing colleagues, people entering and leaving offices, the passage of time. This powerful selectivity has been documented in countless behavioural experiments - a person, for example, who listens carefully to a speech message coming from one loudspeaker knows little or nothing of what is said in another. Selective attention of this sort touches on some of the fundamental problems of human cognition - on how we produce focused, coherent lines of thought and behaviour, and on consciousness itself.

Visual selective attention has also become one of the success stories of cognitive neuroscience. Beyond primary visual cortex, visual information is distributed to a network of cortical areas, together making up the classical cortical visual system. In different parts of this system, neurons work towards different goals - object recognition, visuomotor control, spatial navigation and so on. In many and perhaps most visual areas, experiments in awake, behaving monkeys show the filtering power of selective attention. When the monkey pays attention to some visual input, the neural representation of this input is strong and sustained. When attention is focused elsewhere, the neural signal is attenuated. Such results suggest a direct parallel to the everyday experience of attentional filtering: Central awareness of important information, and loss of much of the remainder. The experiments open the door to a detailed physiological account of brain mechanisms underlying a key aspect of higher cognitive function.

At the same time, we increasingly suspect that monkey studies like these show only one part of the picture. The clue comes from functional magnetic resonance imaging (fMRI) in the human brain. The strength of monkey electrophysiology is the detail with which a neural system can be examined; the strength of fMRI is its ability to measure whole brain activity, and thus to give a broad brush picture of brain processes associated with cognitive functions. In recent experiments, we find that attention does far more than filter processing in the classical visual system. In addition, an attended event is associated with concurrent activity in specific regions of parietal cortex, the frontal lobe, and subcortical structures. For many parts of this attention circuit, equivalent events in the monkey brain are unknown. To begin work on the physiology of this circuit, we propose linked experiments using human fMRI, single unit physiology, and as a direct bridge between the two, fMRI in the awake, behaving animal.

A typical experiment in our human fMRI laboratory works like this. Streams of visual stimuli are shown at two or more locations. The subject is told to attend to stimuli in some locations and to ignore others; in the simplest case, the subject is asked just to watch attended events, with no task to perform or decisions to be made. To make all events equivalent in terms of sensory input, the subject's eyes are always fixed on the centre of a display screen; attended and unattended events occur at various locations around this central fixation point. Using fMRI, we contrast the brain's response to two kinds of input events - attended events, which are consciously seen, and unattended events, which usually are not. As expected, we see that attended events produce stronger activity in large sections of the classical visual system - in particular occipitotemporal regions involved in high-level object recognition. As in the monkey experiments, such results suggest that attention filters neural representations in much of the cortical visual system. Outside the visual system, however, there is a much more extensive pattern of cortical and subcortical activity associated with attended events. Included in this pattern is activity in the parietal lobe (along the intraparietal sulcus), the lateral frontal lobe (posterior part of the inferior frontal sulcus, spreading down to frontal operculum/insula), the medial frontal lobe (anterior cingulate/supplementary motor area), the basal ganglia and the thalamus.

Viewed in terms of human fMRI, this attention circuit has a striking familiarity. As we and others have shown, a very similar activity pattern is associated with a wide range of different kinds of mental activity, suggesting something of central importance to human thought and behaviour. Viewed in terms of monkey physiology, however, this broad attention circuit is essentially unknown territory. We do not know if monkeys too have such broad attentional activity; if they do, we have essentially no physiological information concerning different circuit components and their activity. In our work, we begin to explore this unknown territory - to define the components of an attentional circuit in the behaving animal, and to begin detailed investigation of their physiological properties.

The key link from human to animal studies is monkey fMRI. In our collaborating laboratory (Max-Planck, Tbingen), fMRI has been developed for the awake monkey sitting upright in a vertical-bore magnet. Monkeys are trained to sit still and to tolerate the fMRI environment. In our experiments, monkeys will be trained using methods essentially identical to those in human studies. Streams of stimuli will appear in several locations around a central fixation point. By pre-cueing, animals will be trained to focus attention on one set of locations, waiting for occasional target stimuli but otherwise simply watching the attended stimulus sequence. As in the human studies, the experiments will contrast brain activity associated with attended and unattended stimulus events. In this way, the experiments will ask whether monkeys too show an attention circuit corresponding to human findings. In the parietal lobe, we shall ask what corresponds to extensive human activation along the intraparietal sulcus. In the frontal lobe, we shall seek for correspondences on the lateral frontal surface, in the operculum/insula, and on the medial frontal surface. Subcortically, we shall be able to examine activity of the basal ganglia and the thalamus. Especially for small subcortical structures, the high spatial resolution of the monkey scanner will allow the attention circuit to be defined in substantially greater detail than human experiments can achieve. As these experiments progress, equivalent human studies (MRC, Cambridge) will be used as a direct check on cross-species equivalence.

While the strength of fMRI is breadth of spatial coverage, the strength of electrophysiology is detail. With the results from Tbingen as a guide, in the Oxford electrophysiology lab we shall move on to an expanded view of selected circuit components. From the perspective of monkey physiology, there are several candidates in the human data whose involvement in visual attention is very much of a surprise. In the intraparietal sulcus of the monkey, for example, is a set of subregions with well-known specialization for different aspects of spatial behaviour and visuomotor control. Why should such regions be active for conscious perception of a new visual event - even when attention is fixed on one spatial location and no response choices need be made? Our experiments will examine the physiological properties of parietal neurons in this situation, and examine how such responses differentiate attended and unattended events. In the frontal lobe, similarly, little is known concerning several parts of the human circuit and their potential attentional role. For the anterior cingulate/supplementary motor area, for example, most physiological data concern a role in task and action control. Again, existing data suggest little that could be relevant to visual selective attention. More generally, our experiments will ask how components of an attentional circuit respond to visual events, and how they contribute to the development and maintenance of an attentional focus.

Undoubtedly, monkey experiments have been of great value in beginning to uncover the neural mechanisms underlying cognitive function. For visual attention, however, we think that modulation of activity in the classical visual system is just one part of a much broader set of brain events. With awake monkey fMRI, we can now relate detailed electrophysiology to the kind of circuit overview obtained from human imaging experiments. In our view, the time is right to explore the big picture of brain events underlying selective, conscious perception.