Neurodynamics of Visual Search

The Psychophysics of Visual Attention in Search Tasks

The visual system cannot process simultaneously the immense amount of information conveyed in a complex natural scene. To cope with this problem, attentional mechanisms are needed to select relevant scene information. Evidence for attentional mechanisms in visual processing comes mainly from psychophysical experiments using visual search tasks such as those developed by Anne Treisman. In visual search tasks, subjects examine a display containing randomly positioned items to detect a previously defined target. Items in the display that are different from the target are distracters. The main phenomenology can be understood from the dependence of the measured reaction time on the number of items in the display. There are two main types of search displays: feature search or “pop out” and conjunction or serial search. In a feature search task, the target differs from the distracters on a single feature (e.g., only in its color). In this case, search times are independent of the number of distracters. A typical example of pop-out search is the detection of a red bar within an array of differently tilted green bars. The result can be trivially explained with the activation of only parallel processes, and therefore the unique feature defining the target pops out. In a conjunction search task, the target is defined by a conjunction of features; each distracter shares at least one of those features with the target. Conjunction search experiments show that search time increases linearly with the number of distracters, implying a serial process. An example of conjunction search is the detection of a red vertical bar within a display containing vertical green or tilted red bars as distracters; that is, there is only one item sharing simultaneously (a conjunction of) the two features defining the target, but each distracter shares one feature with the target.

The classical hypothesis accounting for attentional selection in vision is that attention enhances the responses of neurons representing stimuli at a single relevant location in the visual field. This enhancement model is related to Hermann von Helmholtz's spotlight metaphor for focal attention. In this metaphor, a spotlight of attention illuminates part of the visual field; stimuli in the spotlight are processed in higher detail, while information outside the spotlight is filtered out. In this classical view, an object searched for in a cluttered scene is found by rapidly shifting the spotlight from one object to the next until the target is found. In this view, attention is based on explicit serial mechanisms. Based on these concepts, Anne Treisman proposed the so-called feature integration theory of visual selective attention. This theory is based on numerous psychophysical experiments on visual search and offers an interpretation of the binding problem, which asks how features detected by different cortical feature [Page 563]maps are assembled into objects. The feature integration theory distinguishes two processing stages. In the first stage, called preattentive, processing done in parallel across the visual field extracts primitive features (e.g., colors, orientation) without integrating them. In the second, attentive, stage information from limited parts of the field is integrated.

...