Visual search is a type of perceptual task requiring attention that typically involves an active scan of the visual environment for a particular object or feature (the target) among other objects or features (the distractors). Visual search can take place with or without eye movements. The ability to consciously locate an object or target amongst a complex array of stimuli has been extensively studied over the past 40 years. Practical examples of using visual search can be seen in everyday life, such as when one is picking out a product on a supermarket shelf, when animals are searching for food among piles of leaves, when trying to find a friend in a large crowd of people, or simply when playing visual search games such as Where's Wally?
Much previous literature on visual search used reaction time in order to measure the time it takes to detect the target amongst its distractors. An example of this could be a green square (the target) amongst a set of red circles (the distractors). However, reaction time measurements do not always distinguish between the role of attention and other factors: a long reaction time might be the result of difficulty directing attention to the target, or slowed decision-making processes or slowed motor responses after attention is already directed to the target and the target has already been detected. Many visual search paradigms have therefore used eye movement as a means to measure the degree of attention given to stimuli. However, eyes can move independently of attention, and therefore eye movement measures do not completely capture the role of attention.
Search types Feature search
Feature search (also known as "disjunctive" or "efficient" search) is a visual search process that focuses on identifying a previously requested target amongst distractors that differ from the target by a unique visual feature such as color, shape, orientation, or size. An example of a feature search task is asking a participant to identify a white square (target) surrounded by black squares (distractors). In this type of visual search, the distractors are characterized by the same visual features. The efficiency of feature search in regards to reaction time (RT) and accuracy depends on the "pop out" effect, bottom-up processing, and parallel processing. However, the efficiency of feature search is unaffected by the number of distractors present.
The "pop out" effect is an element of feature search that characterizes the target's ability to stand out from surrounding distractors due to its unique feature. Bottom-up processing, which is the processing of information that depends on input from the environment, explains how one utilizes feature detectors to process characteristics of the stimuli and differentiate a target from its distractors. This draw of visual attention towards the target due to bottom-up processes is known as "saliency." Lastly, parallel processing is the mechanism that then allows one's feature detectors to work simultaneously in identifying the target.
Conjunction search
Conjunction search (also known as inefficient or serial search) is a visual search process that focuses on identifying a previously requested target surrounded by distractors possessing no distinct features from the target itself. An example of a conjunction search task is having a person identify a green X (target) amongst distractors composed of purple Xs (same shape) and green Os (same color). Unlike feature search, conjunction search involves distractors (or groups of distractors) that may differ from each other but exhibit at least one common feature with the target. The efficiency of conjunction search in regards to reaction time (RT) and accuracy is dependent on the distractor-ratio and the number of distractors present. As the distractors represent the differing individual features of the target more equally amongst themselves (distractor-ratio effect), reaction time(RT) increases and accuracy decreases. As the number of distractors present increases, the reaction time (RT) increases and the accuracy decreases. However, with practice the original reaction time (RT) restraints of conjunction search tend to show improvement. In the early stages of processing, conjunction search utilizes bottom-up processes to identify pre-specified features amongst the stimuli. These processes are then overtaken by a more serial process of consciously evaluating the indicated features of the stimuli in order to properly allocate one's focal spatial attention towards the stimulus that most accurately represents the target.
In many cases, top-down processing affects conjunction search by eliminating stimuli that are incongruent with one's previous knowledge of the target-description, which in the end allows for more efficient identification of the target. An example of the effect of top-down processes on a conjunction search task is when searching for a red 'K' among red 'Cs' and black 'Ks', individuals ignore the black letters and focus on the remaining red letters in order to decrease the set size of possible targets and, therefore, more efficiently identify their target.
Real world visual search In everyday situations, people are most commonly searching their visual fields for targets that are familiar to them. When it comes to searching for familiar stimuli, top-down processing allows one to more efficiently identify targets with greater complexity than can be represented in a feature or conjunction search task. In a study done to analyze the reverse-letter effect, which is the idea that identifying the asymmetric letter among symmetric letters is more efficient than its reciprocal, researchers concluded that individuals more efficiently recognize an asymmetric letter among symmetric letters due to top-down processes. Top-down processes allowed study participants to access prior knowledge regarding shape recognition of the letter N and quickly eliminate the stimuli that matched their knowledge. In the real world, one must use prior knowledge everyday in order to accurately and efficiently locate objects such as phones, keys, etc. among a much more complex array of distractors. Despite this complexity, visual search with complex objects (and search for categories of objects, such as "phone", based on prior knowledge) appears to rely on the same active scanning processes as conjunction search with less complex, contrived laboratory stimuli, although global statistical information available in real-world scenes can also help people locate target objects. While bottom-up processes may come into play when identifying objects that are not as familiar to a person, overall top-down processing highly influences visual searches that occur in everyday life. Familiarity can play especially critical roles when parts of objects are not visible (as when objects are partly hidden from view because they are behind other objects). Visual information from hidden parts can be recalled from long-term memory and used to facilitate search for familiar objects.
Reaction time slope It is also possible to measure the role of attention within visual search experiments by calculating the slope of reaction time over the number of distractors present. Generally, when high levels of attention are required when looking at a complex array of stimuli (conjunction search), the slope increases as reaction times increase. For simple visual search tasks (feature search), the slope decreases due to reaction times being fast and requiring less attention. However, the use of a reaction time slope to measure attention is controversial because non-attentional factors can also affect reaction time slope.
Visual orienting and attention
One obvious way to select visual information is to turn towards it, also known as visual orienting. This may be a movement of the head and/or eyes towards the visual stimulus, called a saccade. Through a process called foveation, the eyes fixate on the object of interest, making the image of the visual stimulus fall on the fovea of the eye, the central part of the retina with the sharpest visual acuity.
There are two types of orienting: Exogenous orienting is the involuntary and automatic movement that occurs to direct one's visual attention toward a sudden disruption in his peripheral vision field. Attention is therefore externally guided by a stimulus, resulting in a reflexive saccade. Endogenous orienting is the voluntary movement that occurs in order for one to focus visual attention on a goal-driven stimulus. Thus, the focus of attention of the perceiver can be manipulated by the demands of a task. A scanning saccade is triggered endogenously for the purpose of exploring the visual environment.
Visual search relies primarily on endogenous orienting because participants have the goal to detect the presence or absence of a specific target object in an array of other distracting objects.
Early research suggested that attention could be covertly (without eye movement) shifted to peripheral stimuli, but later studies found that small saccades (microsaccades) occur during these tasks, and that these eye movements are frequently directed towards the attended locations (whether or not there are visible stimuli). These findings indicate that attention plays a critical role in understanding visual search.
Subsequently, competing theories of attention have come to dominate visual search discourse. The environment contains a vast amount of information. We are limited in the amount of information we are able to process at any one time, so it is therefore necessary that we have mechanisms by which extraneous stimuli can be filtered and only relevant information attended to. In the study of attention, psychologists distinguish between pre-attentive and attentional processes. Pre-attentive processes are evenly distributed across all input signals, forming a kind of "low-level" attention. Attentional processes are more selective and can only be applied to specific preattentive input. A large part of the current debate in visual search theory centres on selective attention and what the visual system is capable of achieving without focal attention.
Theory Feature integration theory (FIT)
A popular explanation for the different reaction times of feature and conjunction searches is the feature integration theory (FIT), introduced by Treisman and Gelade in 1980. This theory proposes that certain visual features are registered early, automatically, and are coded rapidly in parallel across the visual field using pre-attentive processes. Experiments show that these features include luminance, colour, orientation, motion direction, and velocity, as well as some simple aspects of form. For example, a red X can be quickly found among any number of black Xs and Os because the red X has the discriminative feature of colour and will "pop out." In contrast, this theory also suggests that in order to integrate two or more visual features belonging to the same object, a later process involving integration of information from different brain areas is needed and is coded serially using focal attention. For example, when locating an orange square among blue squares and orange triangles, neither the colour feature "orange" nor the shape feature "square" is sufficient to locate the search target. Instead, one must integrate information of both colour and shape to locate the target.
Evidence that attention and thus later visual processing is needed to integrate two or more features of the same object is shown by the occurrence of illusory conjunctions, or when features do not combine correctly. For example, if a display of a green X and a red O are flashed on a screen so briefly that the later visual process of a serial search with focal attention cannot occur, the observer may report seeing a red X and a green O.