Globally consistent depth sorting of overlapping 2D surfaces in a model using local recurrent interactions

A multi-scale neurodynamic implementation of incremental grouping

Incremental grouping is a process entailing serial binding of distal image elements into a unified object representation. At the neural level, incremental grouping involves propagation of the enhanced firing rate among feature-tuned neurons in the early visual cortex. Here, we developed a multi-resolution neural model of incremental grouping. In the model, propagation of the enhanced firing rate is achieved by computing the activity difference between two sets of units: attentional or A-units, whose firing rate is modulated by their horizontal collaterals, and non-attentional or N-units that receive only feedforward input. The activity difference is computed on dendrites that act as independent computational subunits. The proposed model employs multiple spatial scales to account for a variable speed of incremental grouping. In addition, the model incorporates the L-junction detection network that enables incremental grouping over L-junctions. Computer simulations show that the timing of attentional modulations in the model is comparable with neurophysiological measurements in monkey primary visual cortex.

Neural dynamics of spreading attentional labels in mental contour tracing

Behavioral and neural data suggest that visual attention spreads along contour segments to bind them into a unified object representation. Such attentional labeling segregates the target contour from distractors in a process known as mental contour tracing. A recurrent competitive map is developed to simulate the dynamics of mental contour tracing. In the model, local excitation opposes global inhibition and enables enhanced activity to propagate on the path offered by the contour. The extent of local excitatory interactions is modulated by the output of the multi-scale contour detection network, which constrains the speed of activity spreading in a scale-dependent manner. Furthermore, an L-junction detection network enables tracing to switch direction at the L-junctions, but not at the X- or T-junctions, thereby preventing spillover to a distractor contour. Computer simulations reveal that the model exhibits a monotonic increase in tracing time as a function of the distance to be traced. Also, the speed of tracing increases with decreasing proximity to the distractor contour and with the reduced curvature of the contours. The proposed model demonstrated how an elaborated version of the winner-takes-all network can implement a complex cognitive operation such as contour tracing.

A border-ownership model based on computational electromagnetism

The mathematical relation between a vector electric field and its corresponding scalar potential field is useful to formulate computational problems of lower/middle-order visual processing, specifically related to the assignment of borders to the side of the object: so-called border ownership (BO). BO coding is a key process for extracting the objects from the background, allowing one to organize a cluttered scene. We propose that the problem is solvable simultaneously by application of a theorem of electromagnetism, i.e., "conservative vector fields have zero rotation, or "curl." We hypothesize that (i) the BO signal is definable as a vector electric field with arrowheads pointing to the inner side of perceived objects, and (ii) its corresponding scalar field carries information related to perceived order in depth of occluding/occluded objects. A simple model was developed based on this computational theory. Model results qualitatively agree with object-side selectivity of BO-coding neurons, and with perceptions of object order. The model update rule can be reproduced as a plausible neural network that presents new interpretations of existing physiological results. Results of this study also suggest that T-junction detectors are unnecessary to calculate depth order.

A conceptual framework of computations in mid-level vision

If a picture is worth a thousand words, as an English idiom goes, what should those words-or, rather, descriptors-capture? What format of image representation would be sufficiently rich if we were to reconstruct the essence of images from their descriptors? In this paper, we set out to develop a conceptual framework that would be: (i) biologically plausible in order to provide a better mechanistic understanding of our visual system; (ii) sufficiently robust to apply in practice on realistic images; and (iii) able to tap into underlying structure of our visual world. We bring forward three key ideas. First, we argue that surface-based representations are constructed based on feature inference from the input in the intermediate processing layers of the visual system. Such representations are computed in a largely pre-semantic (prior to categorization) and pre-attentive manner using multiple cues (orientation, color, polarity, variation in orientation, and so on), and explicitly retain configural relations between features. The constructed surfaces may be partially overlapping to compensate for occlusions and are ordered in depth (figure-ground organization). Second, we propose that such intermediate representations could be formed by a hierarchical computation of similarity between features in local image patches and pooling of highly-similar units, and reestimated via recurrent loops according to the task demands. Finally, we suggest to use datasets composed of realistically rendered artificial objects and surfaces in order to better understand a model's behavior and its limitations.

The "side" matters: how configurality is reflected in completion

The perception of figure-ground organization is a highly context-sensitive phenomenon. Accumulating evidence suggests that the so-called completion phenomenon is tightly linked to this figure-ground organization. While many computational models have applied borderline completion algorithms based on the detection of boundary alignments, we point out the problems of this approach. We hypothesize that completion is a result of computing the figure-ground organization. Specifically, the global interactions in the neural network activate the "border-ownership" sensitive neurons at the location where no luminance contrast is given and this activation corresponds to the perception of illusory contours. The implications of this result to the general property of emerging Gestalt percepts are discussed.

Explaining seeing? Disentangling qualia from perceptual organization

Abstract Visual perception and integration seem to play an essential role in our conscious phenomenology. Relatively local neural processing of reentrant nature may explain several visual integration processes (feature binding or figure-ground segregation, object recognition, inference, competition), even without attention or cognitive control. Based on the above statements, should the neural signatures of visual integration (via reentrant process) be non-reportable phenomenological qualia? We argue that qualia are not required to understand this perceptual organization.

Grouping puts figure-ground assignment in context by constraining propagation of edge assignment

Figure-ground organization involves the assignment of edges to a figural shape on one or the other side of each dividing edge. Established visual cues for edge assignment primarily concern relatively local rather than contextual factors. In the present article, we show that an assignment for a locally unbiased edge can be affected by an assignment of a remote contextual edge that has its own locally biased assignment. We find that such propagation of edge assignment from the biased remote context occurs only when the biased and unbiased edges are grouped. This new principle, whereby grouping constrains the propagation of figural edge assignment, emerges from both subjective reports and an objective short-term edge-matching task. It generalizes from moving displays involving grouping by common fate and collinearity, to static displays with grouping by similarity of edge-contrast polarity, or apparent occlusion. Our results identify a new contextual influence on edge assignment. They also identify a new mechanistic relation between grouping and figure-ground processes, whereby grouping between remote elements can constrain the propagation of edge assignment between those elements. Supplemental materials for this article may be downloaded from http://app.psychonomic-journals.org/content/supplemental.

The role of remote closure in the perception of occlusion at junctions and illusory contours

Abstract. Perceived occlusion at T-junctions or illusory contours at implicit T-junctions are often modelled by using edge information without surface context. We explored the effect of closure on perceived occlusion at T-junctions. Two vertical lines separated by a gap each had six abutting horizontal lines on opposite sides forming T-junctions. These lines were either closed or not closed into pairs at the stem ends of the Ts. In experiment 1, closed T-junction stems gave a much stronger sense of occlusion at the vertical lines than unclosed ones, even though closure information was remote from the putative occlusion and local T-junction information remained constant. When the outer two T-junctions were converted to L-junctions, perceived occlusion considerably diminished. The effect of closure on illusory-contour strength for stimuli like those of experiment 1 but with the vertical lines omitted was explored in experiment 2. The two sets of horizontal lines, separated by a gap, were either closed or not closed into pairs at their outer ends. Illusory-contour strength along the vertical alignments was much greater for closed pairs. Line terminations on both sides of the gap enhanced illusory-contour strength, but whether they were collinear or not had little effect.

Interactions of motion and form in visual cortex - A neural model

In this work, we present a neural model simulating parts of the motion and the form pathway of the visual cortex. It is shown how the visual features motion, disparity, and form that are represented in a distributed way in areas V1, V2, and MT mutually interact at several levels. Thus, their information is shared without the need of explicit neural representation for each combination of features. In particular, we address the issue of 2D extrinsic motion cues generated at occlusions that have to be treated differently than 2D intrinsic motion features of the same object. We suggest that here information of the form channel, namely the indication of a junction, is necessary to achieve a correct percept in the motion pathway. Furthermore, we investigated the question of how a percept of either pattern or component motion is generated in a scenario of moving bars that only differs in the presence or absence of occlusions, like in the chopstick and the barberpole display. We propose different roles for various kinds of MT cells that are involved in the interactions with the form pathway, simulating purely integrative cells tuned to motion and to motion and stereo, but also contrast cells responding strongly when motion in the surround is in the opposite direction. The model simulations reproduce psychophysical and neurophysiological results of the chopstick as well as of the barberpole illusion. The temporal course of the dominant motion percept generated by the iterative interplay between motion and form pathway is in line with data of ocular following responses in primates and humans.