SEarch via Recursive Rejection (SERR): A Connectionist Model of Visual Search

doi:10.1006/cogp.1993.1002

Cognitive Psychology

Volume 25, Issue 1, January 1993, Pages 43-110

https://doi.org/10.1006/cogp.1993.1002 Get rights and content

Abstract

In studies of visual search, a general distinction is often made between the processes involved when detection of a target is unaffected by the number of distractors in the field and those involved when search time increases linearly as a function of the number of distractors present. In the former case, processes are said to be "pre-attentive" and to operate in parallel across the visual field; in the latter, processing is said to require focal attention and to be spatially serial. In this paper, we present a connectionist model which performs visual search in parallel across a window defining the model′s functional field. Elements in the field are allowed to group, using simple principles of similarity and spatial proximity. Search operates via the recursive rejection of areas of field where stable and unambiguous grouping has been achieved. Performance of the model is unaffected by the number of distractors present when the distractors form a single group. As the number of competing distractor groups increases, there is an increased likelihood that targets are missed. Setting a response criterion to balance miss rates generates serial increases in search time as a function of the number of distractors. These results are shown to match the functions produced when human subjects search displays varying in the number of distractor groups. The implications of the model are discussed, and the results of five experiments are presented that test novel predictions derived from the model.

References (0)

Cited by (212)

Dwelling on distractors varying in target-distractor similarity
2019, Acta Psychologica
Present day models of visual search focus on explaining search efficiency by visual guidance: The target guides attention to the target's position better in more efficient than in less efficient search. The time spent processing the distractor, however, is set to a constant in these models. In contrast to this assumption, recent studies found that dwelling on distractors is longer in more inefficient search. Previous experiments in support of this contention all presented the same distractors across all conditions, while varying the targets. While this procedure has its virtues, it confounds the manipulation of search efficiency with target type. Here we use the same targets over the entire experiment, while varying search efficiency by presenting different types of distractors. Eye fixation behavior was used to infer the amount of distractor dwelling, skipping, and revisiting. The results replicate previous results, with similarity affecting dwelling, and dwelling in turn affecting search performance. A regression analysis confirmed that variations in dwelling account for a large amount of variance in search speed, and that the similarity effect in dwelling accounts for the similarity effect in overall search performance.
Continuous to discrete: Ensemble-based segmentation in the perception of multiple feature conjunctions
2018, Cognition
Although objects around us vary in a number of continuous dimensions (color, size, orientation, etc.), we tend to perceive the objects using more discrete, categorical descriptions (e.g., berries and leaves). Previously, we described how continuous ensemble statistics of simple features are transformed into categorical classes: The visual system tests whether the feature distribution has one or several peaks, each representing a likely “category”. Here, we tested the mechanism of segmentation for more complex conjunctions of features. Observers discriminated between two textures filled with lines of various lengths and orientations, which had same distributions between the textures, but opposite directions of correlations. Critically, feature distributions could be “segmentable” (only extreme feature values and a large gap between them) or “non-segmentable” (both extreme and middle values with smooth transition are present). Segmentable displays yielded steeper psychometric functions indicating better discrimination (Experiment 1). The effect of segmentability arises early in visual processing (Experiment 2) and is likely to be provided by global sampling of the entire field (Experiment 3). Also, rapid segmentation requires both feature dimensions having a “segmentable” distribution supporting division of the textures into categorical classes of conjunctions. We propose that observers select items from one side (peak) of one dimension and sample mean differences along a second dimension within the selected subset. In this scenario, subset selection is a limiting factor (Experiment 4) of texture discrimination. Yet, segmentability provided by the sharp feature distributions seems to facilitate both subset selection and mean comparison.
Priming of Pop-out does not provide reliable measures of target activation and distractor inhibition in selective attention: Evidence from a large-scale online study
2018, Vision Research
Lamy, Antebi, Aviani, and Carmel (2008) reported that in a simple search task where participants located an odd coloured circle, the inter-trial relations could be used to derive robust and independent measures of target activation and distractor inhibition. When a target feature repeated there was a benefit, and when the previous target feature became the distractor feature there was a cost. These two measures correlated and were taken to reflect a measure of target activation. When the distractor feature repeated there was a benefit and when the previous distractor feature became the current target feature there was a cost, these two measures correlated and were taken to reflect a measure of distractor inhibition. In the current study we examined the same colour search task online on a large group of 312 participants. The results revealed significant effects of target and distractor repetition and switching. However, the correlations reported by Lamy et al. (2008) were non-significant. Instead we found the correlations between the two measures of repetition and the two measures of switching.
Inhibition drives configural superiority of illusory Gestalt: Combined behavioral and drift-diffusion model evidence
2016, Cognition
Citation Excerpt :
One reason for the stronger effect of closure in distractors than in the target might simply derive from the fact that, in the typical CSE paradigm, there are more (most often three) distractors as compared to only a single target. In fact, visual search experiments show that, as set size increases, grouped distractors usually bring about a strong modulation of search efficiency (Conci et al., 2007a, 2007b), suggesting that the benefit of grouping in distractors increases as the number of candidate target configurations becomes larger (see also below and Humphreys & Müller, 1993). In this view, the ‘emergence’ of a configural target thus appears to be a by-product of the efficient suppression of a grouped array of distractors.
Illusory Kanizsa figures demonstrate that a perceptually completed whole is more than the sum of its composite parts. In the current study, we explored part/whole relationships in object completion using the configural superiority effect (CSE) with illusory figures (Pomerantz & Portillo, 2011). In particular, we investigated to which extent the CSE is modulated by closure in target and distractor configurations. Our results demonstrated a typical CSE, with detection of a configural whole being more efficient than the detection of a corresponding part-level target. Moreover, the CSE was more pronounced when grouped objects were presented in distractors rather than in the target. A follow-up experiment systematically manipulated closure in whole target or, respectively, distractor configurations. The results revealed the effect of closure to be again stronger in distractor, rather than in target configurations, suggesting that closure primarily affects the inhibition of distractors, and to a lesser extent the selection of the target. In addition, a drift–diffusion model analysis of our data revealed that efficient distractor inhibition expedites the rate of evidence accumulation, with closure in distractors particularly speeding the drift toward the decision boundary. In sum, our findings demonstrate that the CSE in Kanizsa figures derives primarily from the inhibition of closed distractor objects, rather than being driven by a conspicuous target configuration. Altogether, these results support a fundamental role of inhibition in driving configural superiority effects in visual search.
Directing memory content to attentional templates: The finiteness effect of predictive information
2023, Frontiers in Psychology
Investigating the role of spatial filtering on distractor suppression
2023, Attention, Perception, and Psychophysics

View all citing articles on Scopus

View full text

Regular ArticleSEarch via Recursive Rejection (SERR): A Connectionist Model of Visual Search

Abstract

Regular Article
SEarch via Recursive Rejection (SERR): A Connectionist Model of Visual Search