Visual attention: the where,what, how and why of saliency

Attention influences the processing of visual information even in the earliest areas of primate visual cortex. There is converging evidence that the interaction of bottom-up sensory information and top-down attentional influences creates an integrated saliency map, that is, a topographic representation of relative stimulus strength and behavioral relevance across visual space. This map appears to be distributed across areas of the visual cortex, and is closely linked to the oculomotor system that controls eye movements and orients the gaze to locations in the visual scene characterized by a high salience.     

Feature-based attention in the frontal eye field and area V4 during visual search

When we search for a target in a crowded visual scene, we often use the distinguishing features of the target, such as color or shape, to guide our attention and eye movements. To investigate the neural mechanisms of feature-based attention, we simultaneously recorded neural responses in the frontal eye field (FEF) and area V4 while monkeys performed a visual search task. The responses of cells in both areas were modulated by feature attention, independent of spatial attention, and the magnitude of response enhancement was inversely correlated with the number of saccades needed to find the target. However, an analysis of the latency of sensory and attentional influences on responses suggested that V4 provides bottom-up sensory information about stimulus features, whereas the FEF provides a top-down attentional bias toward target features that modulates sensory processing in V4 and that could be used to guide the eyes to a searched-for target.     

The role of attentional priority and saliency in determining capacity limits in enumeration and visual working memory

Many common tasks require us to individuate in parallel two or more objects out of a complex scene. Although the mechanisms underlying our abilities to count the number of items, remember the visual properties of objects and to make saccadic eye movements towards targets have been studied separately, each of these tasks require selection of individual objects and shows a capacity limit. Here we show that a common factor--salience--determines the capacity limit in the various tasks. We manipulated bottom-up salience (visual contrast) and top-down salience (task relevance) in enumeration and visual memory tasks. As one item became increasingly salient, the subitizing range was reduced and memory performance for all other less-salient items was decreased. Overall, the pattern of results suggests that our abilities to enumerate and remember small groups of stimuli are grounded in an attentional priority or salience map which represents the location of important items.     

Eye movements modulate visual receptive fields of V4 neurons

The receptive field, defined as the spatiotemporal selectivity of neurons to sensory stimuli, is central to our understanding of the neuronal mechanisms of perception. However, despite the fact that eye movements are critical during normal vision, the influence of eye movements on the structure of receptive fields has never been characterized. Here, we map the receptive fields of macaque area V4 neurons during saccadic eye movements and find that receptive fields are remarkably dynamic. Specifically, before the initiation of a saccadic eye movement, receptive fields shrink and shift towards the saccade target. These spatiotemporal dynamics may enhance information processing of relevant stimuli during the scanning of a visual scene, thereby assisting the selection of saccade targets and accelerating the analysis of the visual scene during free viewing.     

Predicting the eye fixation locations in the gray scale images in the visual scenes with different semantic contents

In recent years, there has been considerable interest in visual attention models (saliency map of visual attention). These models can be used to predict eye fixation locations, and thus will have many applications in various fields which leads to obtain better performance in machine vision systems. Most of these models need to be improved because they are based on bottom-up computation that does not consider top-down image semantic contents and often does not match actual eye fixation locations. In this study, we recorded the eye movements (i.e., fixations) of fourteen individuals who viewed images which consist natural (e.g., landscape, animal) and man-made (e.g., building, vehicles) scenes. We extracted the fixation locations of eye movements in two image categories. After extraction of the fixation areas (a patch around each fixation location), characteristics of these areas were evaluated as compared to non-fixation areas. The extracted features in each patch included the orientation and spatial frequency. After feature extraction phase, different statistical classifiers were trained for prediction of eye fixation locations by these features. This study connects eye-tracking results to automatic prediction of saliency regions of the images. The results showed that it is possible to predict the eye fixation locations by using of the image patches around subjects’ fixation points.     

Toward a unified theory of visual area V4

Visual area V4 is a midtier cortical area in the ventral visual pathway. It is crucial for visual object recognition and has been a focus of many studies on visual attention. However, there is no unifying view of V4's role in visual processing. Neither is there an understanding of how its role in feature processing interfaces with its role in visual attention. This review captures our current knowledge of V4, largely derived from electrophysiological and imaging studies in the macaque monkey. Based on recent discovery of functionally specific domains in V4, we propose that the unifying function of V4 circuitry is to enable selective extraction of specific functional domain-based networks, whether it be by bottom-up specification of object features or by top-down attentionally driven selection.     

Psychophysical tests of the hypothesis of a bottom-up saliency map in primary visual cortex

A unique vertical bar among horizontal bars is salient and pops out perceptually. Physiological data have suggested that mechanisms in the primary visual cortex (V1) contribute to the high saliency of such a unique basic feature, but indicated little regarding whether V1 plays an essential or peripheral role in input-driven or bottom-up saliency. Meanwhile, a biologically based V1 model has suggested that V1 mechanisms can also explain bottom-up saliencies beyond the pop-out of basic features, such as the low saliency of a unique conjunction feature such as a red vertical bar among red horizontal and green vertical bars, under the hypothesis that the bottom-up saliency at any location is signaled by the activity of the most active cell responding to it regardless of the cell's preferred features such as color and orientation. The model can account for phenomena such as the difficulties in conjunction feature search, asymmetries in visual search, and how background irregularities affect ease of search. In this paper, we report nontrivial predictions from the V1 saliency hypothesis, and their psychophysical tests and confirmations. The prediction that most clearly distinguishes the V1 saliency hypothesis from other models is that task-irrelevant features could interfere in visual search or segmentation tasks which rely significantly on bottom-up saliency. For instance, irrelevant colors can interfere in an orientation-based task, and the presence of horizontal and vertical bars can impair performance in a task based on oblique bars. Furthermore, properties of the intracortical interactions and neural selectivities in V1 predict specific emergent phenomena associated with visual grouping. Our findings support the idea that a bottom-up saliency map can be at a lower visual area than traditionally expected, with implications for top-down selection mechanisms.     

Training Top-Down Attention Improves Performance on a Triple-Conjunction Search Task

Training has been shown to improve perceptual performance on limited sets of stimuli. However, whether training can generally improve top-down biasing of visual search in a target-nonspecific manner remains unknown. We trained subjects over ten days on a visual search task, challenging them with a novel target (top-down goal) on every trial, while bottom-up uncertainty (distribution of distractors) remained constant. We analyzed the changes in saccade statistics and visual behavior over the course of training by recording eye movements as subjects performed the task. Subjects became experts at this task, with twofold increased performance, decreased fixation duration, and stronger tendency to guide gaze toward items with color and spatial frequency (but not necessarily orientation) that resembled the target, suggesting improved general top-down biasing of search.     

Top-Down but Not Bottom-Up Visual Scanning is Affected in Hereditary Pure Cerebellar Ataxia

The aim of this study was to clarify the nature of visual processing deficits caused by cerebellar disorders. We studied the performance of two types of visual search (top-down visual scanning and bottom-up visual scanning) in 18 patients with pure cerebellar types of spinocerebellar degeneration (SCA6: 11; SCA31: 7). The gaze fixation position was recorded with an eye-tracking device while the subjects performed two visual search tasks in which they looked for a target Landolt figure among distractors. In the serial search task, the target was similar to the distractors and the subject had to search for the target by processing each item with top-down visual scanning. In the pop-out search task, the target and distractor were clearly discernible and the visual salience of the target allowed the subjects to detect it by bottom-up visual scanning. The saliency maps clearly showed that the serial search task required top-down visual attention and the pop-out search task required bottom-up visual attention. In the serial search task, the search time to detect the target was significantly longer in SCA patients than in normal subjects, whereas the search time in the pop-out search task was comparable between the two groups. These findings suggested that SCA patients cannot efficiently scan a target using a top-down attentional process, whereas scanning with a bottom-up attentional process is not affected. In the serial search task, the amplitude of saccades was significantly smaller in SCA patients than in normal subjects. The variability of saccade amplitude (saccadic dysmetria), number of re-fixations, and unstable fixation (nystagmus) were larger in SCA patients than in normal subjects, accounting for a substantial proportion of scattered fixations around the items. Saccadic dysmetria, re-fixation, and nystagmus may play important roles in the impaired top-down visual scanning in SCA, hampering precise visual processing of individual items.     

Alpha ERD and human visual selective attention

We compared the alpha band EEG depression (event-related desynchnization, ERD) level in two tasks, involving activation of different attentional processes: visual search for a deviant relevant stimulus among many similar ones and visual oddball. Control data for the visual search task consisted of simple viewing of several stimuli being of the same shape as the relevant stimulus in the search trials. Gaze position was verified by eye tracking method. We interpreted alpha band ERD as a correlate of activation of attentional processes. Fixating the target in visual search task caused a significantly larger ERD than fixating the same stimuli in control trials over all leads. We suppose this to be related with task and visual environment complexities. The frontal ERD domination may indicate attentional control over voluntary movements execution (top-down attention). The caudal ERD may be related with updating of visual information as a result of search process (bottom-up attention). Both relevant and irrelevant stimuli in the oddball task also induced alpha band ERD, but it was larger in response to relevant one and reached maximum level over occipital leads. Domination of caudal ERD in oddball task is supposed to indicate bottom-up attention processes.     

Oculomotor control and spatial processing.

In the past year research in the oculomotor system has concentrated on some hitherto neglected areas, and also caused a re-evaluation of several long-standing concepts. Careful studies of the translational (otolith) vestibulo-ocular reflex and the torsional system have demonstrated their importance. A re-evaluation of the role of the superior colliculus in the generation of saccades has provided evidence for its participation in the feedback process. New studies of the interaction of eye movements and visual processing have shown that the brain can compensate for the visual effects of eye movements and maintain a retinotopic representation of visual space for the saccadic system.     

Cell-type-specific synchronization of neural activity in FEF with V4 during attention

Shifts of gaze and shifts of attention are closely linked and it is debated whether they result from the same neural mechanisms. Both processes involve the frontal eye fields (FEF), an area which is also a source of top-down feedback to area V4 during covert attention. To test the relative contributions of oculomotor and attention-related FEF signals to such feedback, we recorded simultaneously from both areas in a covert attention task and in a saccade task. In the attention task, only visual and visuomovement FEF neurons showed enhanced responses, whereas movement cells were unchanged. Importantly, visual, but not movement or visuomovement cells, showed enhanced gamma frequency synchronization with activity in V4 during attention. Within FEF, beta synchronization was increased for movement cells during attention but was suppressed in the saccade task. These findings support the idea that the attentional modulation of visual processing is not mediated by movement neurons.     

Computational modelling of visual attention

Five important trends have emerged from recent work on computational models of focal visual attention that emphasize the bottom-up, image-based control of attentional deployment. First, the perceptual saliency of stimuli critically depends on the surrounding context. Second, a unique 'saliency map' that topographically encodes for stimulus conspicuity over the visual scene has proved to be an efficient and plausible bottom-up control strategy. Third, inhibition of return, the process by which the currently attended location is prevented from being attended again, is a crucial element of attentional deployment. Fourth, attention and eye movements tightly interplay, posing computational challenges with respect to the coordinate system used to control attention. And last, scene understanding and object recognition strongly constrain the selection of attended locations. Insights from these five key areas provide a framework for a computational and neurobiological understanding of visual attention.     

Temporal encoding of spatial information during active visual fixation

Humans and other species continually perform microscopic eye movements, even when attending to a single point. These movements, which include drifts and microsaccades, are under oculomotor control, elicit strong neural responses, and have been thought to serve important functions. The influence of these fixational eye movements on the acquisition and neural processing of visual information remains unclear. Here, we show that during viewing of natural scenes, microscopic eye movements carry out a crucial information-processing step: they remove predictable correlations in natural scenes by equalizing the spatial power of the retinal image within the frequency range of ganglion cells' peak sensitivity. This transformation, which had been attributed to center-surround receptive field organization, occurs prior to any neural processing and reveals a form of matching between the statistics of natural images and those of normal eye movements. We further show that the combined effect of microscopic eye movements and retinal receptive field organization is to convert spatial luminance discontinuities into synchronous firing events, beginning the process of edge detection. Thus, microscopic eye movements are fundamental to two goals of early visual processing: redundancy reduction and feature extraction.     

Concurrent TMS-fMRI and psychophysics reveal frontal influences on human retinotopic visual cortex

BACKGROUND: Regions in human frontal cortex may have modulatory top-down influences on retinotopic visual cortex, but to date neuroimaging methods have only been able to provide indirect evidence for such functional interactions between remote but interconnected brain regions. Here we combined transcranial magnetic stimulation (TMS) with concurrent functional magnetic resonance imaging (fMRI), plus psychophysics, to show that stimulation of the right human frontal eye-field (FEF) produced a characteristic topographic pattern of activity changes in retinotopic visual areas V1-V4, with functional consequences for visual perception. RESULTS: FEF TMS led to activity increases for retinotopic representations of the peripheral visual field, but to activity decreases for the central field, in areas V1-V4. These frontal influences on visual cortex occurred in a top-down manner, independently of visual input. TMS of a control site (vertex) did not elicit such visual modulations, and saccades, blinks, or pupil dilation could not account for our results. Finally, the effects of FEF TMS on activity in retinotopic visual cortex led to a behavioral prediction that we confirmed psychophysically by showing that TMS of the frontal site (again compared with vertex) enhanced perceived contrast for peripheral relative to central visual stimuli. CONCLUSIONS: Our results provide causal evidence that circuits originating in the human FEF can modulate activity in retinotopic visual cortex, in a manner that differentiates the central and peripheral visual field, with functional consequences for perception. More generally, our study illustrates how the new approach of concurrent TMS-fMRI can now reveal causal interactions between remote but interconnected areas of the human brain.     

Remapping for visual stability

Visual perception is based on both incoming sensory signals and information about ongoing actions. Recordings from single neurons have shown that corollary discharge signals can influence visual representations in parietal, frontal and extrastriate visual cortex, as well as the superior colliculus (SC). In each of these areas, visual representations are remapped in conjunction with eye movements. Remapping provides a mechanism for creating a stable, eye-centred map of salient locations. Temporal and spatial aspects of remapping are highly variable from cell to cell and area to area. Most neurons in the lateral intraparietal area remap stimulus traces, as do many neurons in closely allied areas such as the frontal eye fields the SC and extrastriate area V3A. Remapping is not purely a cortical phenomenon. Stimulus traces are remapped from one hemifield to the other even when direct cortico-cortical connections are removed. The neural circuitry that produces remapping is distinguished by significant plasticity, suggesting that updating of salient stimuli is fundamental for spatial stability and visuospatial behaviour. These findings provide new evidence that a unified and stable representation of visual space is constructed by redundant circuitry, comprising cortical and subcortical pathways, with a remarkable capacity for reorganization.     

Retinotopic maps, spatial tuning, and locations of human visual areas in surface coordinates characterized with multifocal and blocked FMRI designs

The localization of visual areas in the human cortex is typically based on mapping the retinotopic organization with functional magnetic resonance imaging (fMRI). The most common approach is to encode the response phase for a slowly moving visual stimulus and to present the result on an individual's reconstructed cortical surface. The main aims of this study were to develop complementary general linear model (GLM)-based retinotopic mapping methods and to characterize the inter-individual variability of the visual area positions on the cortical surface. We studied 15 subjects with two methods: a 24-region multifocal checkerboard stimulus and a blocked presentation of object stimuli at different visual field locations. The retinotopic maps were based on weighted averaging of the GLM parameter estimates for the stimulus regions. In addition to localizing visual areas, both methods could be used to localize multiple retinotopic regions-of-interest. The two methods yielded consistent retinotopic maps in the visual areas V1, V2, V3, hV4, and V3AB. In the higher-level areas IPS0, VO1, LO1, LO2, TO1, and TO2, retinotopy could only be mapped with the blocked stimulus presentation. The gradual widening of spatial tuning and an increase in the responses to stimuli in the ipsilateral visual field along the hierarchy of visual areas likely reflected the increase in the average receptive field size. Finally, after registration to Freesurfer's surface-based atlas of the human cerebral cortex, we calculated the mean and variability of the visual area positions in the spherical surface-based coordinate system and generated probability maps of the visual areas on the average cortical surface. The inter-individual variability in the area locations decreased when the midpoints were calculated along the spherical cortical surface compared with volumetric coordinates. These results can facilitate both analysis of individual functional anatomy and comparisons of visual cortex topology across studies.     

Distribution of attention modulates salience signals in early visual cortex

Previous research has shown that the extent to which people spread attention across the visual field plays a crucial role in visual selection and the occurrence of bottom-up driven attentional capture. Consistent with previous findings, we show that when attention was diffusely distributed across the visual field while searching for a shape singleton, an irrelevant salient color singleton captured attention. However, while using the very same displays and task, no capture was observed when observers initially focused their attention at the center of the display. Using event-related fMRI, we examined the modulation of retinotopic activity related to attentional capture in early visual areas. Because the sensory display characteristics were identical in both conditions, we were able to isolate the brain activity associated with exogenous attentional capture. The results show that spreading of attention leads to increased bottom-up exogenous capture and increased activity in visual area V3 but not in V2 and V1.     

Visual attention: spotlight on the primary visual cortex.

Visual search tasks appear to involve spatially selective attention to the target, but evidence for attentional modulation in the visual area with the most precise retinotopic organization V1 has been elusive. Recent imaging studies show that spatial attention can indeed enhance visual responses in human V1.     

Saccade generation by the frontal eye fields in rhesus monkeys is separable from visual detection and bottom-up attention shift

The frontal eye fields (FEF), originally identified as an oculomotor cortex, have also been implicated in perceptual functions, such as constructing a visual saliency map and shifting visual attention. Further dissecting the area's role in the transformation from visual input to oculomotor command has been difficult because of spatial confounding between stimuli and responses and consequently between intermediate cognitive processes, such as attention shift and saccade preparation. Here we developed two tasks in which the visual stimulus and the saccade response were dissociated in space (the extended memory-guided saccade task), and bottom-up attention shift and saccade target selection were independent (the four-alternative delayed saccade task). Reversible inactivation of the FEF in rhesus monkeys disrupted, as expected, contralateral memory-guided saccades, but visual detection was demonstrated to be intact at the same field. Moreover, saccade behavior was impaired when a bottom-up shift of attention was not a prerequisite for saccade target selection, indicating that the inactivation effect was independent of the previously reported dysfunctions in bottom-up attention control. These findings underscore the motor aspect of the area's functions, especially in situations where saccades are generated by internal cognitive processes, including visual short-term memory and long-term associative memory.