Abstract-The

No abstract

One of the most fascinating and impressive aspects of skilled performance is the ability of the experienced eye to encode at a glance the essence of briefly presented stimulus material, which is related to the domain of expertise (henceforth, domain-specific knowledge). For example, Kundel and Nodine (1975) showed expert radiologists X-ray films for 200 msec. The experts were able to detect and name 70% of the abnormalities in the films.Crucial to this process of rapid perception, particularly for visual displays that require multiple fixations for encoding, is the ability to encode large clusters of related information-that is, chunks-and to locate the most relevant areas, or identify the salient locations, on which to focus attention. In order to examine these early perceptual encoding processes, we required chess players at different skill levels to choose the best move for simple, tactically active chess positions while we monitored their eye-fixation patterns. If more skilled players can extract relational information about piece clusters more efficiently than less skilled players (e.g., in parallel, as seen in Reingold, Charness, Pomplun, & Stampe, 2001; Reingold, Charness, Schultetus, & Stampe, 2001), we hypothesize that skilled players' first few seconds of fixations will be characterized by a greater likelihood of fixating on empty squares (in order to maximize information extraction from surrounding piece-occupied squares). Also, when more skilled players fixate squares occupied by pieces, they should be more likely to fixate salient pieces.The most influential investigation of the perceptual aspects of skilled performance originated from the pioneering work on chess by de Groot (1946Groot ( /1978 and Simon (1973a, 1973b). De Groot presented chess positions briefly (2-15 sec) and then removed them from view. Even after such a brief exposure, the best chess players were able to reproduce the locations of the chess pieces almost perfectly (about 93% correct for positions containing about 25 pieces). More generally, performance in this task systematically varied as a function of skill. De Groot concluded that perception and memory were more important differentiators of chess expertise than was the ability to think ahead in the search for good moves. In a classic study, Simon (1973a, 1973b) replicated and extended de Groot's findings demonstrating that after viewing chess positions for 5 sec, chess masters were able to reproduce these positions much more accurately than less-skilled players. However, there was little difference as a function of expertise when random board configurations were used instead of game positions, indicating that the superior immediate memory performance of the skilled players was not attributable to the general superiority of their memory systems or processes (i.e., hardware aspects of memory). More recently, a very small but reliable advantage in recall for random configurations has been shown for expert players, although this is probably attributable to the occasional presenc...

The perception of objects in our visual world is influenced by not only their low-level visual features such as shape and color, but also their high-level features such as meaning and semantic relations among them. While it has been shown that low-level features in real-world scenes guide eye movements during scene inspection and search, the influence of semantic similarity among scene objects on eye movements in such situations has not been investigated. Here we study guidance of eye movements by semantic similarity among objects during real-world scene inspection and search. By selecting scenes from the LabelMe object-annotated image database and applying Latent Semantic Analysis (LSA) to the object labels, we generated semantic saliency maps of real-world scenes based on the semantic similarity of scene objects to the currently fixated object or the search target. An ROC analysis of these maps as predictors of subjects’ gaze transitions between objects during scene inspection revealed a preference for transitions to objects that were semantically similar to the currently inspected one. Furthermore, during the course of a scene search, subjects’ eye movements were progressively guided toward objects that were semantically similar to the search target. These findings demonstrate substantial semantic guidance of eye movements in real-world scenes and show its importance for understanding real-world attentional control.

Recent research on attentional guidance in real-world scenes has focused on object recognition within the context of a scene. This approach has been valuable for determining some factors that drive the allocation of visual attention and determine visual selection. This article provides a review of experimental work on how different components of context, especially semantic information, affect attentional deployment. We review work from the areas of object recognition, scene perception, and visual search, highlighting recent studies examining semantic structure in real-world scenes. A better understanding on how humans parse scene representations will not only improve current models of visual attention but also advance next-generation computer vision systems and human-computer interfaces.

We examined the flexibility of guidance in a conjunctive search task by manipulating the ratios between different types of distractors. Participants were asked to decide whether a target was present or absent among distractors sharing either colour or shape. Results indicated a strong effect of distractor ratio on search performance. Shorter latency to move, faster manual response, and fewer fixations per trial were observed at extreme distractor ratios. The distribution of saccadic endpoints also varied flexibly as a function of distractor ratio. When there were very few same-colour distractors, the saccadic selectivity was biased towards the colour dimension. In contrast, when most of the distractors shared colour with the target, the saccadic selectivity was biased towards the shape dimension. Results are discussed within the framework of the guided search model.

Recently, there has been great interest among vision researchers in developing computational models that predict the distribution of saccadic endpoints in naturalistic scenes. In many of these studies, subjects are instructed to view scenes without any particular task in mind so that stimulus-driven (bottom-up) processes guide visual attention. However, whenever there is a search task, goal-driven (top-down) processes tend to dominate guidance, as indicated by attention being systematically biased toward image features that resemble those of the search target. In the present study, we devise a top-down model of visual attention during search in complex scenes based on similarity between the target and regions of the search scene. Similarity is defined for several feature dimensions such as orientation or spatial frequency using a histogram-matching technique. The amount of attentional guidance across visual feature dimensions is predicted by a previously introduced informativeness measure. We use eye-movement data gathered from participants’ search of a set of naturalistic scenes to evaluate the model. The model is found to predict the distribution of saccadic endpoints in search displays nearly as accurately as do other observers’ eye-movement data in the same displays.

In three experiments, participants' visual span was measured in a comparative visual search task in which they had to detect a local match or mismatch between two displays presented side by side. Experiment 1 manipulated the dif®culty of the comparative visual search task by contrasting a mismatch detection task with a substantially more dif®cult match detection task. In Experiment 2, participants were tested in a single-task condition involving only the visual task and a dual-task condition in which they concurrently performed an auditory task. Finally, in Experiment 3, participants performed two dual-task conditions, which differed in the dif®culty of the concurrent auditory task. Both the comparative search task dif®culty (Experiment 1) and the divided attention manipulation (Experiments 2 and 3) produced strong effects on visual span size. q