Attention in the Animal Brain

In most contexts attention refers to our ability to concentrate our perceptual experience on a selected portion of the available sensory information, and, in doing so, to achieve a clear and vivid impression of the environment. To evaluate something that seems as fundamentally introspective as attention, cognitive science research usually uses a measure of behavioral performance that is correlated with attention. To examine brain mechanisms of attention, the correlations are extended another level by measuring the activity of neurons during different 'attentive' behaviors. Although this article focuses exclusively on attentive processes in the visual system, attentive processing occurs within each sensory system (see auditory attention) and more generally in most aspects of cognitive brain function. Our understanding of the neuronal correlates of attention comes principally from the study of the influence of attentive acts on visual processing as observed in animals. The selective aspects of attention are apparent in both of vision's principal functions, identifying objects and navigating with respect to objects and surfaces.

Attention is a dynamic process added on top of the passive elements of selection provided by the architecture of the visual system. For foveate animals, looking or navigating encompasses the set of actions necessary to find a desired goal and place it in foveal view. The selective aspects of attention within this context deal primarily with decisions about information in the peripheral field of view. Seeing or identifying objects encompasses a more detailed analysis of centrally available information. In this context the selective aspects of attention deal primarily with the delineation of objects and the integration of their parts that lead to their recognition (see OBJECT RECOGNITION, ANIMAL STUDIES).

Although the retinae encode a wide expanse of the visual environment, object analysis is not uniform across the visual field but instead is concentrated in a small zone called the field of focal attention (Neisser 1967). Under most circumstances this restricted zone has little to due with acuity limits set by the receptor density gradient in the retina but is due to an interference between objects generated by the density of the visual information. Focal attention encompasses the dynamic phenomena that enable us to isolate and examine objects under the conditions of interference. The selective aspect of attention raises an important question, namely, how many things can be attended to at one time? Interestingly, the answer varies, and depends at least in part on the level of analysis that is necessary to distinguish between the things that are present. What is clear is that the moment-to-moment analytic capacity of the visual system is surprisingly limited.

An examination of the physiology and anatomy of visual processes in animals, especially primates, provides us with key pieces of the attention puzzle. Visual information is dispersed from primary visual cortex through extrastriate cortex along two main routes. One leads ventrally toward anterior temporal cortex, the other dorsally into parietal association cortex. The ventral stream progression portrays a system devoted to object analysis, and in anterior temporal areas, represents a stage where sensory processes merge with systems associated with object recognition and memory (Gross 1992). The dorsal stream emphasizes the positions of surfaces and objects and in parietal areas represents a stage where the sensory and motor processes involved in the exploration of the surrounding space become intertwined (Mountcastle 1995). The different emphasis in information processing within parietal and temporal areas is also apparent with respect to the influences of attentional states. Both the sensitivity to visual stimuli and the effective receptive field size are more than doubled for parietal visual neurons during an attentive fixation task, whereas under similar conditions the receptive fields of inferior temporal cortical neurons are observed to collapse around a fixation target.

As visual processing progresses in both the dorsal and ventral streams, there is an accompanying expansion in the receptive field size of individual neurons and a corresponding convergence of information as an increasing number of objects fit within each receptive field. Despite the convergence, an inseparable mixture of information from different objects does not occur in part because of the competitive, winner-take-all nature of the convergence and in part because of attentive selection. Directing attention to a particular object alters the convergent balance in favor of the attended object and suppresses the neural response to other objects in the neuron's receptive field (Moran and Desimone 1985; Treue and Maunsell 1996).

Within the ventral stream and at progressively higher levels of object analysis, the competitive convergence of the system forces a narrowing of processing by selecting what information or which objects gain control of the neural activity. The connectivity of the visual system is not simply a feedforward system but a highly interconnected concurrent network. Whatever information wins the contention at one level is passed forward and backward and often laterally (Van Essen and DeYoe 1995). These factors heavily constrain the neural activity, limiting the activation of neurons at each subsequent level to a progressively restricted set of stimulus configurations. As increasingly higher levels of analytic abstraction are attained in the ventral stream, the receptive field convergence narrows the independence of parallel representations until in the anterior temporal lobe the neuronal receptive fields encompass essentially all of the central visual field. Because directing attention emphasizes the processing of the object(s) at the attended location, the act of attending both engages the ventral stream on the object(s) at that location and dampens out information from the remaining visual field (Chelazzi et al. 1993). These circumstances parallel the capacity limitation found in many forms in vision and may constitute the performance limiting factor. Capacity limits may vary depending upon the level of convergence in the cortical stream that must be reached before a discriminative decision about the set of observed objects can be achieved (Merigan, Nealey, and Maunsell 1993).

When attention is to be shifted to a new object, as we read the next word or reach for a cup or walk along a path, information must be obtained from the periphery as to the spatial layout of objects. The dorsal stream appears to provide this information. The convergence of information within the neuronal receptive fields generates sensitivity to large surfaces, their motion and boundaries and the positions of objects within them, without particular sensitivity to the nature of the objects themselves. The parietal visual system is especially sensitive to the relative motion of surfaces such as that generated during movement through the environment. An object to which attention is shifted is usually peripherally located with respect to the object currently undergoing perceptual analysis. For parietal cortical neurons, maintained attention on a particular object results in a heightened visual sensitivity across the visual field and, in contrast to the temporal stream, a suppressed sensitivity to objects currently at the locus of directed attention (Motter 1991).

The transition between sensory information and MOTOR CONTROL is subject to a clear capacity limitation -- competition between potential target goals must be resolved to produce a coherent motor plan. When target goals are in different locations, spatially selective processing can be used to identify locations or objects that have been selected as target goals. In the period before a movement is made, the neural activity associated with the visual presence of the object at the target site evolves to differentiate the target object from other objects. This spatially selective change, consistent with an attentive selection process, has been observed in parietal cortex as well as the motor eye fields of frontal cortex and subcortical visuomotor areas such as the superior colliculus (Schall 1995).

How early in the visual system are attentive influences active? If attention effectively manipulates processing in the earliest stages of vision, then the visual experiences we have are in part built up from internal hypotheses about what we are seeing or what we want to see. Two sets of physiological observations suggest these important consequences of selective attention do occur. First, directed attention studies and studies requiring attentive selection of stimulus features have shown that the neural coding of objects can be completely dominated by top-down attentive demands as early as extrastriate cortex and can bias neuronal processing even in primary visual cortex. Second, after arriving in primary visual cortex, visual information spreads through the cortical systems within 60-80 msec. The effects of selective attention develop in parallel during the next 100 msec in extrastriate occipital, temporal, and parietal cortex and the frontal eye fields, making it not only difficult to pinpoint a single decision stage but also making it likely that a coherent solution across areas is reached by a settling of the network (Motter 1997).

The detailed physiological insights gained from animal studies complement the imaging studies of ATTENTION IN THE HUMAN BRAIN that have probed higher-order cognitive functions and attempted to identify the neural substrates of volitional aspects of attention. Together these sets of studies have provided new views of several classic phenomena in attention including capacity limitations and the temporal progression of selection.

References

Chelazzi, L., E. K. Miller, J. Duncan, and R. Desimone. (1993). A neural basis for visual search in inferior temporal cortex. Nature 363:345-347.

Gross, C. G. (1992). Representation of visual stimuli in inferior temporal cortex. Philosophical Transactions of the Royal Society, London, Series B 335:3-10.

Merigan, W. H., and J. H. R. Maunsell. (1993). How parallel are the primate visual pathways? Annual Review of Neuroscience 16:369-402.

Merigan, W. H., T. A. Nealey, and J. H. R. Maunsell. (1993). Visual effects of lesions of cortical area V2 in macaques. Journal of Neuroscience 13:3180-3191.

Moran, J., and R. Desimone. (1985). Selective attention gate visual processing in the extrastriate cortex. Science 229:782-784.

Motter, B. C. (1991). Beyond extrastriate cortex: the parietal visual system. In A. G. Leventhal, Ed., Vision and Visual Dysfunction, vol. 4, The Neural Basis of Visual Function. London: Macmillan, pp. 371-387.

Motter, B. C. (1998). Neurophysiology of visual attention. In R. Parasuraman, Ed., The Attentive Brain. Cambridge, MA: MIT Press.

Mountcastle, V. B. (1995). The parietal system and some higher brain functions. Cerebral Cortex 5:377-390.

Neisser, U. (1967). Cognitive Psychology. New York: Appleton-Century-Crofts.

Schall, J. D. (1995). Neural basis of saccade target selection. Reviews in the Neurosciences 6:63-85.

Treue, S., and J. H. R. Maunsell. (1996). Attentional modulation of visual motion processing in cortical areas MT and MST. Nature 382:539-541.

Van Essen, D. C., and E. A. DeYoe. (1995). Concurrent processing in the primate visual cortex. In M.S. Gazzaniga, Ed., The Cognitive Neurosciences. Cambridge, MA: MIT Press, pp. 383-400.

Attention in the Animal Brain

See also

References

Further Readings