Identification and categorization in visual search

Seven experiments were addressed to the general question of whether the identification of letters and numbers is a more rapid process than the categorization of such stimuli. Subjects were required to make a single response if a target stimulus specified by name (e.g., “A,” “2”) or designated by category class alone (e.g., “letter,” “number”) was presented in a trial. The principal findings were: (1) identification reaction times (RTs) were faster than categorization RTs: (2) RTs for targets shown without a context were faster than RTs for targets shown in the context of other stimuli; (3) identification RTs for targets shown in the context of stimuli from a different conceptual-taxonomic category were faster than RTs for targets shown in the context of stimuli from the same category only when target-context stimulus discriminability differencet were optimized. The results were interpreted in terms of a two-stage processing model in which context face,ors affect the duration of an initial encoding-scanning stage and search instruction (effective memory size) factors affect the duration of the memory comparison stage.     

Integration of information in visual search

Two experiments studied how information about the nontarget items in a visual search task is used for the control of the search. The first experiment used the detection of “hurdle” stimuli to demonstrate that efficient memory representations of the context items can be established within each particular trial. This finding is explained by a model for the short-term integration of context information. The second experiment which varied the complexity of the local but not the global context provided some information about the nature of the integration operations involved. In its final version the model postulates two stages of processing with independent mechanisms of integration. Spatial integration at the first stage deletes repetitions within small samples. Temporal integration at the second stage stores and primes the memory representations of the context items over larger intervals. It is assumed that transient temporal integration within trials is mediated by the same mechanism that underlies permanent temporal integration between trials.     

Category and word search: generalizing search principles to complex processing

This research examines how the major phenomena of visual search for single characters generalize to word search and word-category search. Experiment 1 examined word and category visual search when the target and distractor sets had a varied mapping (VM) across trials. Reaction time was a linear function of the number of comparisons with a positive slope of 48 msec per word, 92 msec per category. The VM reaction time data indicated a self-terminating comparison process, and there was little or no improvement with practice. Experiment 2 examined search with a consistent mapping (CM) between targets and distractors. Category search slope dropped to 2 msec, and the function became nonlinear. Word search slope dropped to 18 msec, but the function became nonlinear. Word search slope dropped to 18 msec, but the function was still linear. Experiment 3 examined category detection carried out concurrently with serial recall of digits, allowing assessment of search performance under high workload. High workload caused a severe performance reduction in VM category search, and this decrement did not decrease with practice. High workload reduced initial performance in CM category search, but this decrement was eliminated with practice. The present category search results are similar to previous letter search results. Four principles of search are discussed in the context of a theory of automatic/control processing.     

The relative contribution of scene context and target features to visual search in scenes

Many experiments have shown that knowing a targets visual features improves search performance over knowing the target name. Other experiments have shown that scene context can facilitate object search in natural scenes. In this study, we investigated how scene context and target features affect search performance. We examined two possible sources of information from scene context—the scenes gist and the visual details of the scene—and how they potentially interact with target-feature information. Prior to commencing search, participants were shown a scene and a target cue depicting either a picture or the category name (or no-information control). Using eye movement measures, we investigated how the target features and scene context influenced two components of search: early attentional guidance processes and later verification processes involved in the identification of the target. We found that both scene context and target features improved guidance, but that target features also improved speed of target recognition. Furthermore, we found that a scenes visual details played an important role in improving guidance, much more so than did the scenes gist alone.     

Toward an understanding of age-related memory and visual search effects.

Young and old Ss were tested in 3 experiments conducted to explore factors leading to age-related performance differences in consistent mapping (CM) and varied mapping (VM) search tasks. The separate and combined influences of memory scanning and visual search on age-related search effects were examined. In both CM letter and CM semantic category search, age interacted with comparison load in visual and hybrid memory-visual search conditions, whereas differential age effects were not present in pure memory search. For VM search, age effects were present only in pure memory search. These data support the view that the separation of type of search training (CM and VM) as well as memory and visual search components is critical for predicting age-related performance differences. The dissociation of the pattern of age effects in memory and visual search suggests that memory and visual search involve different processing mechanisms.     

What top-down task sets do for us: an ERP study on the benefits of advance preparation in visual search

When target-defining features are specified in advance, attentional target selection in visual search is controlled by preparatory top-down task sets. We used ERP measures to study voluntary target selection in the absence of such feature-specific task sets, and to compare it to selection that is guided by advance knowledge about target features. Visual search arrays contained two different color singleton digits, and participants had to select one of these as target and report its parity. Target color was either known in advance (fixed color task) or had to be selected anew on each trial (free color-choice task). ERP correlates of spatially selective attentional target selection (N2pc) and working memory processing (SPCN) demonstrated rapid target selection and efficient exclusion of color singleton distractors from focal attention and working memory in the fixed color task. In the free color-choice task, spatially selective processing also emerged rapidly, but selection efficiency was reduced, with nontarget singleton digits capturing attention and gaining access to working memory. Results demonstrate the benefits of top-down task sets: Feature-specific advance preparation accelerates target selection, rapidly resolves attentional competition, and prevents irrelevant events from attracting attention and entering working memory.     

Differential central resource demands of memory scanning and visual search: The role of consistent and varied mapping

In this paper we compare the effect of mapping on the central resource requirements of two search processes. With consistent mapping of items to the role of either target or distractor, search becomes very efficient; with varied mapping (each item serves as both a target and a distractor in the same experiment), search is less efficient. One interpretation of this effect of manipulating the consistency of mapping is that search under varied mapping requires limited capacity central resources but the requirement for these resources is reduced with consistent mapping. We tested this interpretation by varying the consistency of mapping in memory scanning (Experiments 1 and 2) and in visual search (Experiments 3 and 4) as the second of two tasks in the psychological refractory period (PRP) paradigm. Responses were location based, rather than present/absent judgements, and a target was present on every trial. The results indicated that there was less of a requirement for central resources with consistent mapping than with varied mapping in both memory scanning and in visual search; however, the effect of the mapping manipulation on central resource requirements was different for memory scanning than for visual search. Memory scanning was not done in parallel with a resource-demanding stage of a second task under varied mapping, but was done in parallel with this stage under consistent mapping. Part of visual search under both types of mapping was done in parallel with a resource-demanding stage of a second task and part was not; consistent mapping reduced the duration of the stage of processing that could not be carried out in parallel. We conclude that under consistent mapping, switching among items in memory becomes less controlled and therefore has less of a requirement for central resources than under varied mapping. In visual search, however, consistent mapping improves performance by facilitating the comparison of the target item to items in the search array.     

Visual search at different spatial scales

Treisman and Gelade's (1980) feature-integration model claims that the search for separate ("primitive") stimulus features is parallel, but that the conjunctions of those features require serial scan. Recently, evidence has accumulated that parallel processing is not limited to these "primitive" stimulus features, but that combinations of features can also produce parallel search. In the experiments reported here, the processing of feature conjunctions was studied when the stimulus features of a combination were at different spatial scales. The patterns in the search array were composed of three cross-shaped or T-shaped (local) elements, which formed an oblique bar (the global pattern) 45 deg or 135 deg in orientation. When the target and distractors differed from each other at one spatial scale only (either in the bar orientation or in the shape of the local elements), target detection was independent of the number of distractors, i.e., the search was parallel. In the conjunction task, in which the target and distractors were defined as the combinations of the bar orientation and the element shape, i.e., both spatial scales were relevant, the detection of the target required slow serial scrutiny of the search array. It is possible that the conjunction search could not be performed in parallel because switches between the two scales (or spatial frequency channels) are linked to attention and the task required the use of both scales in order to find the target.     

Retrieval search for category and thematic information in memory by children and adults

Retrieval search for the last target word of four noun category (Bus-Airplane-Car-Train) and thematic (Suitcase-Coat-Ticket-Train) stimuli by children and adults was examined in four experiments. Stimulus classification at acquisition was ensured by orienting questions. Associative search structures were externalized more or less at retrieval by providing cues reinstating the whole (Bus-Airplane-Car) or only part (Bus-Airplane, or Bus) of the associative context, cues with the same or changed (Car-Bus-Airplane) word order, or cues representing the stimulus classification (Vehicle). The results suggest that the associative structures of memory help constrain and mediate retrieval search for stimulus information, and that these structures as well as representations of the stimulus events in memory seem to differ for thematic and category events, and for children and adults, contributing to search differences.     

Visual search and foraging compared in a large-scale search task

It has been argued that visual search is a valid model for human foraging. However, the two tasks differ greatly in terms of the coding of space and the effort required to search. Here we describe a direct comparison between visually guided searches (as studied in visual search tasks) and foraging that is not based upon a visually distinct target, within the same context. The experiment was conducted in a novel apparatus, where search locations were indicated by an array of lights embedded in the floor. In visually guided conditions participants searched for a target defined by the presence of a feature (red target amongst green distractors) or the absence of a feature (green target amongst red and green distractors). Despite the expanded search scale and the different response requirements, these conditions followed the pattern found in conventional visual search paradigms: feature-present search latencies were not linearly related to display size, whereas feature-absent searches were longer as the number of distractors increased. In a non-visually guided foraging condition, participants searched for a target that was only visible once the switch was activated. This resulted in far longer latencies that rose markedly with display size. Compared to eye-movements in previous visual search studies, there were few revisit errors to previously inspected locations in this condition. This demonstrates the important distinction between visually guided and non-visually guided foraging processes, and shows that the visual search paradigm is an equivocal model for general search in any context. We suggest a comprehensive model of human spatial search behaviour needs to include search at a small and large scale as well as visually guided and non-visually guided search.     

Effects of secondary tasks on search performance

Summary Two experiments studied the effect of secondary task load on continuous visual search. Two effects were observed when a secondary task was applied. Search speed (which is assumed to reflect the efficiency of nontarget rejection) was impaired, whereas target detection distance (which presumably reflects the efficiency of target detection) increased. These findings are explained in the framework of a two-process model of visual search which assumes that two operations are performed simultaneously during each fixation, that is, automatic detection and controlled search. The secondary task interferes with the operation of controlled search which reacts to this impairment by way of a compensatory slow-down of search speed. This gain in time also benefits the operation of automatic detection which converts the temporal gain into a spatial extension of the area under its control.     

The effects of search criteria and retrieval cue availability on memory for words

The effects of search criteria upon recall performance following a word classification task were investigated. Rival hypotheses were tested, one based on the number of cues available at retrieval and the other on the amount of processing carried out during classification. After being presented with two-component category names, subjects classified words on the basis of either category membership or the number of semantic components the words shared with a target category, and then they received an unexpected recall test. Experiment 1 favored an amount-of-processing explanation, but it was felt that redundancy in the stimulus materials might invalidate this conclusion. Experiment 2 confirmed this suspicion and clearly supported an explanation in terms of retrieval cue availability. The notion of partial semantic processing of words is discussed, and it is argued that the number of congruent encodings formed during classification can be used as an index of elaboration and subsequent memory performance.     

The cost of categorization in visual search: Incomplete processing of targets and field items

A partial processing hypothesis is proposed to account for performance under a visual search condition where target and field items belong to the different conceptual categories, letter and digit (between-category search), as compared to a condition in which they belong to the same category (within-category search). This hypothesized mechanism implies that less information is registered and/or retained in between- than in within-category search. This prediction was tested and confirmed in three experiments. The results indicate that both targets and field items are processed less deeply in between- than in within-category search.     

Prioritization in visual search: visual marking is not dependent on a mnemonic search

Visual marking (VM) refers to our ability to completely exclude old items from search when new stimuli are presented in our visual field. We examined whether this ability reflects an attentional scan of the old items, possibly allowing observers to apply inhibition of return or maintain a memory representation of already seen locations. In four experiments, we compared performance in two search conditions. In the double-search (DS) condition, we required participants to pay attention to a first set of items by having them search for a target within the set. Subsequently, they had to search a second set while the old items remained in the field. In the VM condition, the participants expected the target only to be in the second (new) set. Selection of new items in the DS condition was relatively poor and was always worse than would be expected if only the new stimuli had been searched. In contrast, selection of the new items in the VM condition was good and was equal to what would be expected if there had been an exclusive search of the new stimuli. These results were not altered when differences in Set 1 difficulty, task switching, and response generation were controlled for. We conclude that the mechanism of VM is distinct from mnemonic and/or serial inhibition-of-return processes as involved in search, although we also discuss possible links to more global and flexible inhibition-of-return processes not necessarily related to search.     

Lesioning a connectionist model of visual search: selective effects on distractor grouping.

A connectionist model of visual search is presented, in which search is determined by patterns of grouping between distractors and between the target and the distractors. Grouping is based on conjunctive relations between simple form elements (corners, line-end terminators), and uses principles of similarity and spatial proximity. In its normal, 'unlesioned' state, the model simulates the search data generated by human subjects when they search for simple form conjunctions amongst either homogeneous or heterogeneous distractors (Humphreys & Muller, in press). In this paper, the performance of the model is examined after the model is subject to various types of 'lesion'. 'Lesioning' is produced either by increasing the internal noise on the activation functions governing the interactions between processing units, or by eliminating processing units from different loci in the model. Increasing the internal noise within the model, or lesioning units within 'high-level' processing stages, can generate the selective effects on search found in a detailed single case study of an agnosic patient (Humphreys, Riddoch, Quinlan, Price & Donnelly, this volume): Namely, there is selective disruption of search for a form conjunction amongst homogeneous distractors relative to search for the same target amongst heterogeneous distractors. This selective effect can be attributed to the 'lesion' disrupting grouping between distractors, which normally facilitates search with homogeneous distractors. The simulations demonstrate that (1) search processes can be disrupted by adding internal noise to search functions, (2) that there can be selective effects on grouping processes, and (3) that these selective effects can be associated with different types of lesion to different stages in the model. The implications of the simulations for understanding visual processing impairments in agnosia are discussed.     

A two-stage detection model applied to skilled visual search by radiologists

Harvard Medical School and Peter Bent Brigham Hospital, Boston, Massachusetts 02115 The model treats the detection of targets in a visual search task as a concatenation of two serial detection stages. Preattentive visual mechanisms in the initial stage function as a filter, selecting specific features of a visual pattern for the observer’s explicit attention and final cognitive evaluation. The model uses bivariate normal distributions to represent the decision variables for the two serial stages, assuming different parameters for the target and nontarget features in a test set. The model is applied to the detection performance of radiologists interpreting chest x-rays under various conditions of search. It accounts for the substantial improvement in radiologists’ ability to distinguish between target and nontarget test features when they had to search the x-ray images, compared to their performance without visual search. A change in the ROC curve between two different search tasks could be interpreted as a shift in the selection cutoff used by the preattentive filter.     

The role of memory and restricted context in repeated visual search

Previous studies have shown that the efficiency of visual search does not improve when participants search through the same unchanging display for hundreds of trials (repeated search), even though the participants have a clear memory of the search display. In this article, we ask two important questions. First, why do participants not use memory to help search the repeated display? Second, can context be introduced so that participants are able to guide their attention to the relevant repeated items? Experiments 1-4 show that participants choose not to use a memory strategy because, under these conditions, repeated memory search is actually less efficient than repeated visual search, even though the latter task is in itself relatively inefficient. However, when the visual search task is given context, so that only a subset of the items are ever pertinent, participants can learn to restrict their attention to the relevant stimuli (Experiments 5 and 6).     

Visual search time based on stochastic serial and parallel processings

Two stochastic latency models were developed in order to account for random fluctuations of the search time that the S consumed when he was asked to search for a single target number or one of two (or three) target numbers through the sets of small two-digit numbers randomly distributed on a white square. It was shown that the search time distribution for a single target number can be adequately described in terms of McGill’s two-step serial processing model. The parallel processing model whose structure is exponential was found to fit approximately the search time distributions for the first target number that the S detects among the two (or three) target numbers predesignated. This model, however, does not satisfactorily handle the long search times.     

Task demands determine the specificity of the search template

When searching for an object, an observer holds a representation of the target in mind while scanning the scene. If the observer repeats the search, performance may become more efficient as the observer hones this target representation, or "search template," to match the specific demands of the search task. An effective search template must have two characteristics: It must reliably discriminate the target from the distractors, and it must tolerate variability in the appearance of the target. The present experiment examined how the tolerance of the search template is affected by the search task. Two groups of 18 observers trained on the same set of stimuli blocked either by target image (block-by-image group) or by target category (block-by-category group). One or two days after training, both groups were tested on a related search task. The pattern of test results revealed that the two groups of observers had developed different search templates, and that the templates of the block-by-category observers better captured the general characteristics of the category. These results demonstrate that observers match their search templates to the demands of the search task.     

Words,shape, visual search and visual working memory in 3‐year‐old children

Do words cue children's visual attention, and if so, what are the relevant mechanisms? Across four experiments, 3‐year‐old children (N = 163) were tested in visual search tasks in which targets were cued with only a visual preview versus a visual preview and a spoken name. The experiments were designed to determine whether labels facilitated search times and to examine one route through which labels could have their effect: By influencing the visual working memory representation of the target. The targets and distractors were pictures of instances of basic‐level known categories and the labels were the common name for the target category. We predicted that the label would enhance the visual working memory representation of the target object, guiding attention to objects that better matched the target representation. Experiments 1 and 2 used conjunctive search tasks, and Experiment 3 varied shape discriminability between targets and distractors. Experiment 4 compared the effects of labels to repeated presentations of the visual target, which should also influence the working memory representation of the target. The overall pattern fits contemporary theories of how the contents of visual working memory interact with visual search and attention, and shows that even in very young children heard words affect the processing of visual information.