Attention II

Accentuation-Suppression and Scaling: Arousal Modulation of Attentional Processing

Thomas Alrik Sørensen and Claus Bundesen

The limitations of the visual short-term memory (VSTM) system have become an increasingly popular field of study. One line of inquiry has focused on the way attention selects objects for encoding into VSTM. Using the framework of the Theory of Visual Attention (TVA; Bundesen, 1990, Psychological Review, 97, 523-547) different components of attention were measured while arousal was systematically varied by manipulating temporal expectancies of the observer. The results showed increments in both overall speed of encoding into VSTM and efficiency of visual selection as the level of arousal was increased, whereas VSTM capacity showed some decrease with increasing arousal. Furthermore, attentional weights seemed to be modulated such that, as arousal increased, the attentional focus narrowed around fewer, highly relevant objects. Arousal seems to interact with attention in two different ways: Through a scaling mechanism modulating the decision bias of the observer and also through an accentuation-suppression mechanism that modulates the degree of subjective relevance of objects, contracting attention around fewer, highly relevant objects while suppressing less relevant objects. These mechanisms may afford an evolutionary advantage for an observer, ensuring that the most relevant objects are selected and encoded faster under high arousal, and making later processing more focused as the capacity of VSTM is decreased.

Attention allocation to regions perceived as holes

Marco Bertamini and Mai Salah Helmy

The reversal of figure-ground assignment for a closed region is the purest form of figure-ground organisation, and the two resulting percepts are that of an object and that of a hole. Both are non-accidental regions and can share an identical outline. In one series of experiments observers judged the shape of a contour that could be either the same or different from an irrelevant surrounding contour. We report that different (incongruent) inside and outside contours produce stronger interference when they form a single object-with-hole, compared to a hierarchical set of surfaces, or a single hole separating different surfaces (a trench). Therefore which surface owns the contour constrains the interference between shapes, and regions perceived as holes do not display object-like properties. In another series of experiments we used a probe presented near the contour, and we compared the speed of responses to probes presented near corners or along the straight edge. We replicated the corner enhancement effect (Cole et al. 2001, JEP:HPP, 27, 1356-1368) but only when the probe was perceived as located on the surface that owned the corner, thus even in this case holes did produce very different results compared to the same region perceived as an object.

The role of parietal cortex in updating changing visual representations

Christina Howard, Naheem Bashir and Glyn Humphreys

The dynamic nature of the world requires us to keep our visual representations up to date. Representations are more precise if they use a narrow time window over which information is sampled and averaged. Limited temporal precision can cause percepts of changing objects to lag by up to 250 ms in the past (Howard and Holcombe, 2008, Vision Research, 48, 1164-1180). The parietal lobe is involved in spatial vision, and has been implicated in temporal aspects of perception, such as onset-offset discrimination, apparent motion and temporal order judgements. However, the role of the parietal lobe in temporal precision of perception is still unresolved. We tested patients with parietal lobe damage on a continuous monitoring task. Patients attempted to keep track of the changing spatial frequency of a luminance grating. After a few seconds, the grating disappeared and patients reported its final spatial frequency. By comparing the reported value and the actual feature values over the last frames of the stimulus, patients’ perceptual lags were calculated. We find longer and more erratic perceptual lags in patients with parietal damage than in healthy adults. Performance remained poor for some patients even with slow 2.5 Hz change rates. We discuss the role of parietal cortex in spatio-temporal processing.

Visual illusions of size influence throwing performance in darts

Rouwen Canal-Bruland, Hemke van Doorn and John van der Kamp

Baseball players who hit the ball more often judge the ball to be bigger than their less successful counterparts [Witt and Proffitt, 2005, Psychological Science, 16(12), 937-938]. This phenomenon is dubbed action-specific perception. The processes underlying action-specific effects are largely unknown. To further specify the constraints under which performance-related alterations in perception occur, we conducted two experiments in which participants performed a dart throwing task towards a circle-shaped target. The target was either presented in isolation or surrounded by smaller or larger circles, inducing the well-known Ebbinghaus size illusion. In line with our predictions, results revealed that participants who perceived the target as being bigger (than it physically was) hit the target significantly more often than those who perceived it as being smaller. However, the typically reported correlation between perception and performance was absent. While the findings thus provide strong evidence that visual illusions influence target-related performance, they do not support an action-specific perception account.

Temporal integration in rapid serial visual presentation

Elkan Akyurek

Visual temporal integration is a perceptual process that combines successive sensory input across brief intervals, thereby merging it into a single coherent percept. It has been suspected that temporal integration might play a role in two-target rapid serial visual presentation (RSVP) tasks, particularly when the targets follow each other directly, that is, at Lag 1. To date, evidence for this idea has come from the analysis of report order errors, but these could not conclusively demonstrate that other mechanisms (such as prior entry) might not instead underlie performance at Lag 1. Using a novel design, in which the targets could be combined perceptually in a meaningful way such that the conjunction of the two target elements constituted a possible target stimulus itself, it was shown that observers indeed frequently reported the integrated percept. These reports were robust across a range of different visual targets. Further experiments tested the extent to which temporal integration might play a role across larger intervals and multiple successive targets. The results demonstrated that temporal integration of target-related information can occur across a relatively wide temporal range, and multiple successive stimuli, suggesting that temporal integration may play an important role not only in short intervals, but also in the construction of longer (attentional) event episodes.

Linking perceptual animacy to attention: Chasing detection among visually indistinguishable distractors

Hauke S. Meyerhoff, Markus Huff and Stephan Schwan

Perceptual animacy describes the tendency of human observers to interpret spatio-temporal contingencies in motions of simple shapes in anthropomorphic terms such as social causation or intention. Typically, perceptual animacy has been studied isolated from other psychological concepts such as attention. Recent studies established a chasing detection paradigm that allowed to study animacy psychophysically. This line of research has argued that a chase is perceived efficiently among distractors. Here, we present several variants of the chasing detection paradigm to investigate the link between perceptual animacy and attention. In Experiments 1a and 1b, we show that detection accuracy and response latency depend on the physical set size, suggesting that chasing detection is inefficient. Experiment 2 replicates these findings with an attentional set size manipulation. Finally, Experiment 3 combines the chasing detection task with an attentional cueing paradigm to directly guide attention. Valid cues enhanced chasing detection indicating that chasing detection requires effortful visual search through subsets of all possible items. Overall, our results suggest that human observers can visually search for a pair of chasing objects among otherwise indistinguishable distractors. However, there is no pop-out effect for chasing stimuli. Thus, attention plays a crucial role in chasing detection.

Automatic correspondence by surface features during visual tracking across spatiotemporal discontinuities

Frank Papenmeier, Hauke S. Meyerhoff, Georg Jahn and Markus Huff

Observers can track multiple moving objects among indistinguishable distractors. Previous research focused on the role of spatiotemporal features for tracking. In three experiments, we studied the role of surface features of tracked objects for establishing object correspondence during visual tracking. We presented three-dimensional scenes with balls moving on a floor plane. These balls were colored briefly before and after a spatiotemporal discontinuity (i.e., an abrupt scene rotation). Distinct colors matching the spatiotemporal information across the discontinuity caused higher tracking performance than homogeneous coloring. In contrast, swapping distinct colors during the discontinuity impaired performance. In the swap condition, participants mis-selected more distractors appearing in a former target color than distractors appearing in a former distractor color. Even when color never supported tracking in any trial and when we instructed participants to ignore color, we observed the tendency to select distractors in former target colors. This shows that surface features are encoded and utilized automatically during tracking. Introducing abrupt scene rotations disrupts the spatial configuration in the picture plane. This is no precondition for automatic color correspondence, which we replicated in a further experiment with abrupt zooms retaining the spatial configuration. We conclude that the tracking mechanism utilizes spatiotemporal and surface features.

TAM2: A Theory of Eye Movements During Categorical Visual Search

Gregory Zelinsky

Categorical search, the task of finding and recognizing categorically-defined targets, has long been neglected by search theory. Recently, we compared several computer vision models and found that one combining C2 features with color described categorical search guidance and recognition exceptionally well for a complex class of objects (VSS 2012). The present work integrates this feature with TAM, enabling the modeling of eye movements to categorical targets. TAM2 uses distance from an SVM classification boundary to obtain a probability map indicating evidence for the target category at each scene location. We tested TAM2 in 3 experiments: (1) a variable set size (6/13/20) present/absent search task using teddy bear targets and random category distractors, (2) a search task in which distractors had high, medium, or low visual similarity to the target category, and (3) a task identical to (1) using butterfly targets instead of bears to test generalizability. In each task TAM2 successfully described the guidance of gaze to categorical targets. Importantly, TAM2 supplements rather than supplants TAM; if a target preview is available and categorical search is unnecessary, TAM2 reverts back to TAM. It is only when target appearance is uncertain that TAM2 uses categorical target features retrieved from LTM to drive its search.