 |
Neuronal Mechanisms of Attention
Summary: John Maunsell is interested in understanding how attention influences the representation of sensory information in cerebral cortex, and how these changes improve behavioral performance.
Attention plays a critical role in perception. At any moment, we can give full attention to only a tiny fraction of the sensory information reaching our brain. Attention to a particular location enhances detection, improves discrimination, and speeds responses at that position relative to others. Often attention makes the difference between seeing something or missing it entirely. These pronounced changes in performance are associated with changes in the way the brain processes sensory information. My laboratory investigates how attention changes the way that individual neurons represent visual information, and how those changes affect behavior.
We use microelectrodes to record the electrical signals of neurons in the visual regions of the cerebral cortex of monkeys. The monkey's visual system has been extensively studied, and its structure and function are very similar to the structure and function of the human visual system. The part of cerebral cortex that serves vision is divided into dozens of separate areas, each of which has its own representation of the visual scene and contains neurons that are specialized for representing a particular sort of visual information. For example, some areas are specialized to represent motion, and each neuron within those areas responds only to stimuli moving in a particular direction, regardless of the color, size, or shape of the moving stimuli. Neurons in other cortical areas are specialized to represent other types of visual information, such as the orientation of edges or stimulus color. Cortical areas also differ in the complexity of the sensory information that they represent: while some cortical areas contain neurons that respond well to any contour or edge, neurons in other areas respond only to specific, complicated patterns, forms, or objects. For example, neurons in some parts of visual cortex are active only when the subject views faces.
The responses of neurons within these specialized cortical areas depend not only on the signals coming from the eyes but also on signals related to attention, which come from other parts of the brain. Neurophysiological studies from many laboratories have shown that neurons respond differently when monkeys shift their attention from one stimulus to another, even when those shifts in attention are made without moving their eyes. Most neurons respond more strongly to a stimulus when the animal pays attention to it. Thus, attention boosts the neuronal signals representing the part of the scene that the viewer considers interesting, while suppressing signals related to the rest.
In a series of experiments, we have shown that attention adjusts the strength of neuronal responses dynamically but does not greatly alter the selectivity of neurons for particular stimuli. Attention is not all-or-none, and by controlling the amount of attention that an animal directed toward different stimuli, we have shown that a neuron's responses to a stimulus vary in proportion to the amount of attention directed to the stimulus (which was measured by testing the animal's behavioral sensitivity to that stimulus). Attention can change the strength of neuronal responses within a fraction of a second as the animal focuses more or less on different parts of the visual scene.
While attention changes the strength of neuronal responses considerably, it has little effect on their selectivity. As mentioned above, each sensory neuron is most sensitive to a particular visual attribute, such as color, orientation, or direction of motion, and each responds strongly only to a particular range of stimuli that matches its selectivity (e.g., a particular range of colors or orientations). We examined whether attention to stimulus orientation affects orientation-selective neurons by restricting their responses to a narrower range of orientations. Attention did not change the range of orientations to which a neuron responded. We have also shown that attention does not systematically change neuronal tuning for stimulus contrast. Thus, attention changes the strength of neuronal responses without changing what they respond to.
In addition to our experiments on attention, we have also begun asking how the rest of the brain accesses the visual representations in cortex to guide behaviors. One approach to this question is to ask how tightly the signals of different neurons are linked to a specific behavior. Previously we showed that when animals must detect noisy motion signals, there are trial-to-trial correlations between the responses of some neurons and the probability that the animal will detect the motion. This correlation suggests that these neurons contribute to the detection of the stimulus. These measurements were made with the direction of motion aligned to the preferred direction of each neuron tested. We have recently examined how this correlation varies when the movement is aligned to directions that are not the neuron's preferred direction. We have found that the correlation between a neuron's responses and behavioral performance exists only for a narrow range of directions around the neuron's preferred direction. This suggests that the behavioral response is based on signals from a limited number of neurons: those with preferred directions close to the direction of motion that the animal is trying to detect. Thus, behavioral decisions appear to use a mechanism that can selectively access signals from small groups of cortical neurons that provide the best sensory information for performing the task.
In other experiments we are examining whether certain levels of visual cortex are more important for generating visual perceptions. Many studies have shown that the activity of neurons in the higher stages of cortex correlates more strongly with perceptions and decisions. On the other hand, when people have their cortex electrically stimulated in certain medical evaluations, they are most likely to report seeing something if the earliest levels of visual cortex are electrically stimulated. We used microelectrodes to activate tiny groups of neurons in different cortical areas of monkeys to find out whether animals are more sensitive to these perturbations when they are made in earlier or later visual areas. No visual stimulus was presented, so the animal's detection was based entirely on the neuronal signal introduced by the electrode.
We trained monkeys to report whenever they detected the microstimulation, and then tested different sites in cortex by varying the strength of the microstimulation to find the threshold stimulus level that the animal needed to detect stimulation reliably. We measured thresholds for many sites in several areas that spanned all levels of visual cortex. Consistent with observations in humans, monkeys detected stimulation of the first state of visual cortex immediately, but not stimulation of later stages. However, with extended training the animals became able to detect stimulation in any area, with only slightly higher thresholds in later stages. This suggests that small perturbations in neuronal activity in any level of cortex can be roughly comparable in their ability to produce percepts. These observations do not support the idea that later stages of cortex have privileged access to perception.
Our work should lead to an increased understanding of the role of attention in creating the representations that underlie sensation and perception, and the neuronal mechanisms by which the activity of millions of cortical neurons leads to unified perception.
A grant from the National Institutes of Health provided partial support for these projects.
Last updated: April 30, 2008
|
|
|
|
