Allan Dobbins

Two stages of curve detection suggest two styles of visual computation

The problem of detecting curves in visual images arises in both computer vision and biological visual systems. Our approach integrates constraints from these two sources and suggests that there are two different stages to curve detection, the first resulting in a local description, and the second in a global one. Each stage involves a different style of computation: in the first stage, hypotheses are represented explicitly and coarsely in a fixed, preconfigured architecture; in the second stage, hypotheses are represented implicitly and more finely in a dynamically constructed architecture. We also show how these stages could be related to physiology, specifying the earlier parts in a relatively fine-grained fashion and the later ones more coarsely.

Endstopping and curvature

Hypercomplex or endstopped visual cortical neurons are usually supposed to be concerned with length or end point analysis. However, recent evidence demonstrates that endstopped neurons are curvature-selective, a connection that we explore here in some detail. A model of endstopped simple cells is developed and a variety of computational simulations examine the connection of the model to the reported length and orientation responses of endstopped neurons. Even and odd versions of the model are described, both of which are shown to be curvature-selective. Even-symmetric instances of the model respond well to thin curves over a range of curve orientation and curvature, independent of sign of curvature. In contrast, odd-symmetric instances respond to both thin and thick curves while exhibiting a more complex curvature-sign dependence--responding in a sign-selective fashion to curved lines but not to curved edges. Finally, the response of the endstopped model to curve singularities is explored, and the possible role of nonendstopped and endstopped cells in building curve descriptions is discussed.

The Organization Of Curve Detection: Coarse Tangent Fields And Fine Spline Coverings

We propose a new paradigm for curve detection in which an autonomous and atomic description is computed between measurements on the image and global curves. The description takes the form of a discrete tangent field, i.e.. an estimated representation of quantized tangents and curvatures at each possible trace point. We then present a new algorithm for inferring global curves through this tangent field. by inferring a covering of the global curve. The elements of the covering are short splines, each of which moves according to a potential distribution computed from the tangent field. Both stages of the algorithm are parallel and both bear some analogy to biological mechanisms for curve detection

Differential ambiguity reduces grouping of metastable objects

Two coaxial, ambiguously rotating objects tend to be perceived as corotating. Such grouping could be the consequence of bottom-up, cooperative interactions between the stimuli, or the top-down selection of object properties consistent with a model of the objects or scene. However, we find that the coupling between an ambiguous and unambiguous object is sharply reduced, presenting a challenge for both explanations of grouping. We describe experiments that support the idea that top-down feedback is necessary to select and stabilize a perceptual interpretation for ambiguous figures. Reduced coupling between objects of differing ambiguity can be explained if the feedback is global and proportional to ambiguity.

Competition in bistable vision is attribute-specific.

We employ ambiguous figures and rivalrous stimuli that have multiple ambiguous properties to show that the different attributes of an ambiguous stimulus can undergo independent switching dynamics. This suggests that competition is distributed and attribute-specific, consistent with the known functional segregation of visual processing. Conflicting evidence that binocular rivalry is an early or late visual process may be better understood as evidence for attribute-specific competition occurring at multiple stages of visual processing. Specifically, we show that whether perceptual selection during binocular rivalry is early and eye-based or late and percept-based depends on the particular ambiguous attributes of the rivalrous stimulus.

Asymmetries in Perception of 3D Orientation

Visual scene interpretation depends on assumptions based on the statistical regularities of the world. People have some preference for seeing ambiguously oriented objects (Necker cubes) as if tilted down or viewed from above. This bias is a near certainty in the first instant (∼1 s) of viewing and declines over the course of many seconds. In addition, we found that there is modulation of perceived orientation that varies with position—for example objects on the left are more likely to be interpreted as viewed from the right. Therefore there is both a viewed-from-above prior and a scene position-dependent modulation of perceived 3-D orientation. These results are consistent with the idea that ambiguously oriented objects are initially assigned an orientation consistent with our experience of an asymmetric world in which objects most probably sit on surfaces below eye level.

Perceiving surfaces in depth beyond the fusion limit of their elements.

An isolated dot appears double outside a small disparity range called Panums fusional area. In random-dot stereograms (RDSs), however, this doubling, or diplopia of dot elements is not evident at any disparity. Nevertheless, depth is perceived up to disparities that greatly exceed Panums fusional limit. Either one is unaware of dot diplopia at disparities exceeding Panums fusional limit or the fusion limit is extended. To examine these possibilities, we developed a novel RDS in which dichoptically color-coded dots have a distinctive color when fused, and return to their intrinsic colors when diplopic. We measured the fusion limit of dots in this RDS, and compared it to the patent stereopsis limit of the perceived surface in similar RDSs. We found that the fusional area of dots in the RDS was comparable to Panums fusional area. Furthermore, there was clear dissociation between the fusion limit and the patent stereopsis limit in the RDS. We conclude that the elements composing a surface are not necessarily fused when a large disparity surface is perceived in depth.

Relationship Between Alpha Rhythm and the Default Mode Network: An Eeg-fmri Study

Purpose: Reports of the relationship between the default mode network (DMN) and alpha power are conflicting. Our goal was to assess this relationship by analyzing concurrently obtained EEG/functional MRI data using hypothesis-independent methods. Methods: We collected functional MRI and EEG data during eyes-closed rest in 20 participants aged 19 to 37 (10 females) and performed independent component analysis on the functional MRI data and a Hamming-windowed fast Fourier transform on the EEG data. We correlated functional MRI fluctuations in the DMN with alpha power. Results: Of the six independent components found to have significant relationships with alpha, four contained DMN-associated regions: One independent component was positively correlated with alpha power, whereas all others were negatively correlated. Furthermore, two independent components with opposite relationships with alpha had overlapping voxels in the medial prefrontal cortex and posterior cingulate cortex, suggesting that subpopulations of neurons within these classic nodes within the DMN may have different relationships to alpha power. Conclusions: Different parts of the DMN exhibit divergent relationships to alpha power. Our results highlight the relationship between DMN activity and alpha power, indicating that networks, such as the DMN, may have subcomponents that exhibit different behaviors.

Recurrent global dynamics underlie perceptual grouping

Abstract Arrays of ambiguous objects often appear to group into a single interpretation. Examples include Attneaves triangles and the dynamic dot arrays of Ramachandran and Anstis. However, there are other displays in which perceptual grouping fails. We find that in a variety of such displays—which interpretations are seen, or whether grouping occurs at all—depends on whether the ambiguous property of the objects can be assigned values consistent with a single viewer position or motion. This implies that grouping depends on the visual system developing a 3-D scene model which constrains the relation of the viewer to objects in the scene.

Can Rotational Grouping Be Determined by the Initial Conditions

Objects rotating in depth with an ambiguous rotation direction frequently appear to rotate together. Corotation is especially strong when the objects are interpretable as having a shared axis. We manipulated the initial conditions of the experiment by having pairs of objects initially appear to be unambiguous, and then make either a sudden or gradual transition to ambiguous spin. We find that in neither case do coaxial counter-rotating objects persist in being perceived as counter-rotating. This implies that the perceptual constraint that favors coaxial corotation overrides the initial perceptual state of the objects.