Academic journal article Perception and Psychophysics

Cross-Trial Priming in Visual Search for Singleton Conjunction Targets: Role of Repeated Target and Distractor Features

Academic journal article Perception and Psychophysics

Cross-Trial Priming in Visual Search for Singleton Conjunction Targets: Role of Repeated Target and Distractor Features

Article excerpt

Kristjánsson, Wang, and Nakayama (2002) demonstrated that visual search for conjunctively defined targets can be substantially expedited ("primed") when target and distractor features are repeated on consecutive trials. Two experiments were conducted to examine whether the search response time (RT) facilitation on target-present trials results from repetition of target-defining features, distractor features, or both. The experiments used a multiple conjunctive search paradigm (adapted from Kristjánsson et al., 2002), in which the target and distractor features were varied (i.e., repeated) independently of each other across successive trials. The RT facilitation was numerically largest when both target and distractor features were repeated, but not significantly larger than that when only distractor features were repeated. This indicates that cross-trial priming effects in conjunctive visual search result mainly from the repetition of distractor, rather than target, features.

In the visual search paradigm, observers' task is to discern the presence of a target element among varying numbers of nontarget (distractor) elements. How efficiently a target can be detected has been found to depend on specific properties of the search display (see Wolfe, 1998, for a review)-in particular, whether the target differs from the distractors in a single feature or in a conjunction of different features, each of which is separately present in the distractors. However, even in the latter case, search can be surprisingly efficient. Since Treisman's proposal of her influential feature integration theory (FIT; Treisman & Gelade, 1980), several mechanisms have been considered to account for how efficient detection of feature conjunction targets may be achieved. Although most accounts assume some form of top-down guidance of visual search (based on knowledge of the target-defining features), recent work has pointed to the potential role of bottom-up priming mechanisms (see the review below). The present experiments strengthen the case for bottom-up priming, and also show that priming is based mainly, though not exclusively, on the repetition of abstractor rather than target features.

Efficient Visual (Conjunction) Search

FIT (Treisman & Gelade, 1980) characterizes target detection as a two-stage process. At the first, "preattentive" stage, the basic perceptual features of objects (e.g., color) are coded by dimension-specific modules of analyzers operating across the visual field. Spatiotopic feature maps are formed by each module, registering an object's feature value(s) within a particular dimension (e.g., red, green, etc., within the color module). At the second stage, spatial attention focuses on a position within a master map of locations (which signals where registered features occur within the field, but not their individual values) in order to retrieve and combine, within and across dimensions, the various features recorded at that position in the feature maps. This process leads to the creation of a temporary object representation referred to as an object file, which can be used to access stored knowledge for object recognition.

FIT proposes two distinct types of search behavior, depending on whether the target to be detected in a visual search display is defined by a unique feature in a given dimension (e.g., the only red element among blue elements) or a unique conjunction of features (e.g., the only red and vertical element among red horizontal and blue vertical elements). In a single-feature search, activity from a single feature map will signal the presence of the target, enabling it to be detected rapidly and independently of the number of display elements. This type of search behavior is referred to as parallel. In contrast, in feature conjunction search, serial focal-attentional inspection of element locations on the master map is necessary in order to bind the features together correctly, leading to search RTs that increase linearly with the number of elements. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.