Ambiguous Motion Signals Research Articles

A Possible Role for End-Stopped V1 Neurons in the Perception of Motion: A Computational Model

We present a model of the early stages of processing in the visual cortex, in particular V1 and MT, to investigate the potential role of end-stopped V1 neurons in solving the aperture problem. A hierarchical network is used in which the incoming motion signals provided by complex V1 neurons and end-stopped V1 neurons proceed to MT neurons at the next stage. MT neurons are categorized into two types based on their function: integration and segmentation. The role of integration neurons is to propagate unambiguous motion signals arriving from those V1 neurons that emphasize object terminators (e.g. corners). Segmentation neurons detect the discontinuities in the input stimulus to control the activity of integration neurons. Although the activity of the complex V1 neurons at the terminators of the object accurately represents the direction of the motion, their level of activity is less than the activity of the neurons along the edges. Therefore, a model incorporating end-stopped neurons is essential to suppress ambiguous motion signals along the edges of the stimulus. It is shown that the unambiguous motion signals at terminators propagate over the rest of the object to achieve an accurate representation of motion.

The interaction between integration and segmentation neurons for motion perception

The main challenge for understanding motion processing in the visual system is that local motion signals do not accurately represent the direction of the whole object. This is known as the The small receptive fields of individual neurons can only measure motion components orthogonal to their preferred direction. Neuro-physiological studies show that neurons in brain area MT (V5) have a major role in dealing with this problem [1]. We develop a model to overcome the aperture problem through excitatory and inhibitory interconnections between MT neurons. The model contains two levels of rate-based neuron. The first level is a model of V1 with two sets of neurons: complex neurons and hyper-complex end-stopped neurons. End-stopped neurons respond to the end-points of the stimulus , thereby representing an unambiguous estimation of the direction of motion [2]. The response of V1 neurons is transmitted to the second level, the model of MT neurons. MT neurons are categorized as integration and segmentation neurons. The role of integration neurons is to integrate local motion signals to solve the aperture problem; the role of segmentation neurons is to detect discontinuities in the input stimulus to inhibit integration neurons [3]. Excitatory recurrent connections between integration neurons propagate end-stopped information from the object terminators. As each spatial location contains neurons selective to different directions, inhibitory connections are added between the neurons such that there will be only one neuron active at each spatial location. Long-range inhibitory connections are also added between more distant MT neurons selective to other directions. Segmentation neurons send inhibitory connections to integration neurons to prevent the propagation of motion signals outside the border of the input stimulus. Inhibitory connections between segmentation neurons are defined based on center-surround interactions; this suppresses activity when the motion signals in the center and surround are in the same directions; ensuring that their activity will not prevent the propagation of motion signals when there is no discontinuity. The results of simulations show that the activity of V1 neurons along the edges is much stronger than the activity of the neurons at the end-points. Activation of end-stopped neurons in V1 suppresses ambiguous motion signals along the edges of the stimulus. Therefore, unambiguous information at the end-points wins the competition over the neurons with ambiguous motion signals. The final output of the model shows that MT neurons achieve an accurate estimation of motion using the information provided by the end-stopped neurons. We developed a model of MT neurons for motion perception based on the available neuro-physiological data. These neurons receive their input from neurons in V1. The results show the successful function of these neurons in the perception of motion of a single moving bar. In this model, we also investigated the role of end-stopped neurons in overcoming the aperture problem. The results suggest that MT neurons are not able to solve the aperture problem without the vital role played by end-stopped cells.

Asymmetric global motion integration in drifting Gabor arrays.

We examined how ambiguous motion signals are integrated over space to support the unambiguous perception of global motion. The motion of a Gaussian windowed drifting sine grating (Gabor) is consistent with an infinite number of grating velocities. To extract the consistent global motion of multi-Gabor arrays, the visual system must integrate ambiguous motion signals from disparate regions of visual space. We found an interaction between spatial arrangement and global motion integration in this process. Linear arrays of variably oriented Gabor elements appeared to move more slowly, reflecting suboptimal integration, when the direction of global translation was orthogonal to the line as opposed to along it. Circular arrays of Gabor elements appeared to move more slowly when the global motion was an expansion or contraction rather than a rotation. However, there was no difference in perceived speed for densely packed annular arrays for these global motion pattern directions. We conclude that the region over which ambiguous motion is integrated is biased in the direction of global motion, and the concept of the association field, held to link like elements along a contour, needs to be extended to include global motion computation over disparate elements referencing the same global motion.

Human neural responses involved in spatial pooling of locally ambiguous motion signals

Early visual motion signals are local and one-dimensional (1-D). For specification of global two-dimensional (2-D) motion vectors, the visual system should appropriately integrate these signals across orientation and space. Previous neurophysiological studies have suggested that this integration process consists of two computational steps (estimation of local 2-D motion vectors, followed by their spatial pooling), both being identified in the area MT. Psychophysical findings, however, suggest that under certain stimulus conditions, the human visual system can also compute mathematically correct global motion vectors from direct pooling of spatially distributed 1-D motion signals. To study the neural mechanisms responsible for this novel 1-D motion pooling, we conducted human magnetoencephalography (MEG) and functional MRI experiments using a global motion stimulus comprising multiple moving Gabors (global-Gabor motion). In the first experiment, we measured MEG and blood oxygen level-dependent responses while changing motion coherence of global-Gabor motion. In the second experiment, we investigated cortical responses correlated with direction-selective adaptation to the global 2-D motion, not to local 1-D motions. We found that human MT complex (hMT+) responses show both coherence dependency and direction selectivity to global motion based on 1-D pooling. The results provide the first evidence that hMT+ is the locus of 1-D motion pooling, as well as that of conventional 2-D motion pooling.

Perceptual entrainment of individually unambiguous motions

When two bistable spheres defined by dots oscillating around a common axis (coaxial condition) are placed near each other, their motions become entrained such that they appear to rotate in the same direction. When the dots in the two spheres oscillate around parallel axes (non-coaxial condition), entrainment is much reduced, suggesting that this phenomenon is driven by interactions between the global stimulus representations rather than between the locally ambiguous motion signals. In order to examine where these interactions may be arising, we created highly unambiguous spheres by introducing size and contrast cues on the dots. We find that coupling is strong for the two nearly unambiguous spheres rotating in opposite directions of motion in the coaxial condition but, importantly, that this effect is subject to attentional control. Forcing observers to withdraw their attention from the rotating spheres by performing a demanding task at fixation significantly reduces the amount of coupling, arguing against mediation by low-level interactions between the spheres. These results reveal a disjunct between the local motion signals that are unambiguous and the global percept, implicating a process that requires active suppression of reliable visual information. We propose that this is a dynamic process that requires feedback from higher levels to bind opposite motions into a unified directional percept.

A Model of Motion Transparency Processing with Local Center-Surround Interactions and Feedback

Motion transparency occurs when multiple coherent motions are perceived in one spatial location. Imagine, for instance, looking out of the window of a bus on a bright day, where the world outside the window is passing by and movements of passengers inside the bus are reflected in the window. The overlay of both motions at the window leads to motion transparency, which is challenging to process. Noisy and ambiguous motion signals can be reduced using a competition mechanism for all encoded motions in one spatial location. Such a competition, however, leads to the suppression of multiple peak responses that encode different motions, as only the strongest response tends to survive. As a solution, we suggest a local center-surround competition for population-encoded motion directions and speeds. Similar motions are supported, and dissimilar ones are separated, by representing them as multiple activations, which occurs in the case of motion transparency. Psychophysical findings, such as motion attraction and repulsion for motion transparency displays, can be explained by this local competition. Besides this local competition mechanism, we show that feedback signals improve the processing of motion transparency. A discrimination task for transparent versus opaque motion is simulated, where motion transparency is generated by superimposing large field motion patterns of either varying size or varying coherence of motion. The model's perceptual thresholds with and without feedback are calculated. We demonstrate that initially weak peak responses can be enhanced and stabilized through modulatory feedback signals from higher stages of processing.

Visual competition between ambiguous and unambiguous motion signals in grating patterns

Visual competition between ambiguous and unambiguous motion signals in grating patterns

Rotating dotted ellipses: Motion perception driven by grouped figural rather than local dot motion signals

Unlike the motion of a continuous contour, the motion of a single dot is unambiguous and immune to the aperture problem. Here we exploit this fact to explore the conditions under which unambiguous local motion signals are used to drive global percepts of an ellipse undergoing rotation. In previous work, we have shown that a thin, high aspect ratio ellipse will appear to rotate faster than a lower aspect ratio ellipse even when the two in fact rotate at the same angular velocity [Caplovitz, G. P., Hsieh, P. -J., & Tse, P. U. (2006) Mechanisms underlying the perceived angular velocity of a rigidly rotating object. Vision Research, 46(18), 2877–2893]. In this study we examined the perceived speed of rotation of ellipses defined by a virtual contour made up of evenly spaced dots. Results Ellipses defined by closely spaced dots exhibit the speed illusion observed with continuous contours. That is, thin dotted ellipses appear to rotate faster than fat dotted ellipses when both rotate at the same angular velocity. This illusion is not observed if the dots defining the ellipse are spaced too widely apart. A control experiment ruled out low spatial frequency “blurring” as the source of the illusory percept. Conclusion Even in the presence of local motion signals that are immune to the aperture problem, the global percept of an ellipse undergoing rotation can be driven by potentially ambiguous motion signals arising from the non-local form of the grouped ellipse itself. Here motion perception is driven by emergent motion signals such as those of virtual contours constructed by grouping procedures. Neither these contours nor their emergent motion signals are present in the image.

Laminar cortical dynamics of visual form and motion interactions during coherent object motion perception

How do visual form and motion processes cooperate to compute object motion when each process separately is insufficient? Consider, for example, a deer moving behind a bush. Here the partially occluded fragments of motion signals available to an observer must be coherently grouped into the motion of a single object. A 3D FORMOTION model comprises five important functional interactions involving the brain's form and motion systems that address such situations. Because the model's stages are analogous to areas of the primate visual system, we refer to the stages by corresponding anatomical names. In one of these functional interactions, 3D boundary representations, in which figures are separated from their backgrounds, are formed in cortical area V2. These depth-selective V2 boundaries select motion signals at the appropriate depths in MT via V2-to-MT signals. In another, motion signals in MT disambiguate locally incomplete or ambiguous boundary signals in V2 via MT-to-V1-to-V2 feedback. The third functional property concerns resolution of the aperture problem along straight moving contours by propagating the influence of unambiguous motion signals generated at contour terminators or corners. Here, sparse 'feature tracking signals' from, for example, line ends are amplified to overwhelm numerically superior ambiguous motion signals along line segment interiors. In the fourth, a spatially anisotropic motion grouping process takes place across perceptual space via MT-MST feedback to integrate veridical feature-tracking and ambiguous motion signals to determine a global object motion percept. The fifth property uses the MT-MST feedback loop to convey an attentional priming signal from higher brain areas back to V1 and V2. The model's use of mechanisms such as divisive normalization, endstopping, cross-orientation inhibition, and long-range cooperation is described. Simulated data include: the degree of motion coherence of rotating shapes observed through apertures, the coherent vs. element motion percepts separated in depth during the chopsticks illusion, and the rigid vs. nonrigid appearance of rotating ellipses.

Integration of form and motion within a generative model of visual cortex

One of the challenges faced by the visual system is integrating cues within and across processing streams for inferring scene properties and structure. This is particularly apparent in the inference of object motion, where psychophysical experiments have shown that integration of motion signals, distributed across space, must also be integrated with form cues. This has led several to conclude that there exist mechanisms which enable form cues to ‘veto’ or completely suppress ambiguous motion signals. We describe a probabilistic approach which uses a generative network model for integrating form and motion cues using the machinery of belief propagation and Bayesian inference. We show, using computer simulations, that motion integration can be mediated via a local, probabilistic representation of contour ownership, which we have previously termed ‘direction of figure’. The uncertainty of this inferred form cue is used to modulate the covariance matrix of network nodes representing local motion estimates in the motion stream. We show with results for two sets of stimuli that the model does not completely suppress ambiguous cues, but instead integrates them in a way that is a function of their underlying uncertainty. The result is that the model can account for the continuum of bias seen for motion coherence and perceived object motion in psychophysical experiments.

Neural models of motion integration and segmentation

A neural model is developed of how motion integration and segmentation processes compute global motion percepts. Figure-ground properties, such as occlusion, influence which motion signals determine the percept. For visible apertures, a line's extrinsic terminators do not specify true line motion. For invisible apertures, a line's intrinsic terminators create veridical feature tracking signals, which are amplified before they propagate across space and are integrated with ambiguous motion signals within line interiors. This integration process is the result of several processing stages: directional transient cells respond to image transients and input to a directional short-range filter that selectively boosts feature tracking signals. Competitive interactions further boost feature tracking signals and create speed-selective receptive fields. A long-range filter gives rise to true directional cells by pooling signals over multiple orientations and opposite contrast polarities. A distributed population code of speed tuning realizes a size–speed correlation, whereby activations of multiple spatially short-range filters of different sizes are transformed into speed-tuned cell responses. These mechanisms use transient cell responses, output thresholds that covary with filter size, and competition. The model reproduces empirically derived speed discrimination curves and simulates data showing how visual speed perception and discrimination are affected by stimulus contrast.

Neural dynamics of motion integration and segmentation within and across apertures

A neural model is developed of how motion integration and segmentation processes, both within and across apertures, compute global motion percepts. Figure–ground properties, such as occlusion, influence which motion signals determine the percept. For visible apertures, a line's terminators do not specify true line motion. For invisible apertures, a line's intrinsic terminators create veridical feature-tracking signals. Sparse feature-tracking signals can be amplified before they propagate across position and are integrated with ambiguous motion signals within line interiors. This integration process determines the global percept. It is the result of several processing stages: directional transient cells respond to image transients and input to a directional short-range filter that selectively boosts feature-tracking signals with the help of competitive signals. Then, a long-range filter inputs to directional cells that pool signals over multiple orientations, opposite contrast polarities, and depths. This all happens no later than cortical area MT. The directional cells activate a directional grouping network, proposed to occur within cortical area MST, within which directions compete to determine a local winner. Enhanced feature-tracking signals typically win over ambiguous motion signals. Model MST cells that encode the winning direction feed back to model MT cells, where they boost directionally consistent cell activities and suppress inconsistent activities over the spatial region to which they project. This feedback accomplishes directional and depthful motion capture within that region. Model simulations include the barberpole illusion, motion capture, the spotted barberpole, the triple barberpole, the occluded translating square illusion, motion transparency and the chopsticks illusion. Qualitative explanations of illusory contours from translating terminators and plaid adaptation are also given.

Motion integration across differing image features

To interpret the projected image of a moving object, the visual system must integrate motion signals across different image regions. Traditionally, researchers have examined this process by focusing on the integration of equally ambiguous motion signals. However, when the motions of complex, multi-featured images are measured through spatially limited receptive fields, the resulting motion measurements have varying degrees of ambiguity. In a series of experiments, we examine how human observers interpret images containing motion signals of differing degrees of ambiguity. Subjects judged the perceived coherence of images consisting of an ambiguously translating grating and an unambiguously translating random dot pattern. Perceived coherence of the dotted grating depended upon the degree of concurrence between the velocities of the grating terminators and dots. Depth relationships also played a critical role in the motion integration process. When terminators were suppressed with occlusion cues, coherence increased. When dots and gratings were presented at different depth planes, coherence decreased. We use these results to outline the conditions under which the visual system uses unambiguous motion signals to interpret object motion.

The perception of globally coherent motion

How do human observers perceive a coherent pattern of motion from a disparate set of local motion measures? Our research has examined how ambiguous motion signals along straight contours are spatially integrated to obtain a globally coherent perception of motion. Observers viewed displays containing a large number of apertures, with each aperture containing one or more contours whose orientations and velocities could be independently specified. The total pattern of the contour trajectories across the individual apertures was manipulated to produce globally coherent motions, such as rotations, expansions, or translations. For displays containing only straight contours extending to the circumferences of the apertures, observers' reports of global motion direction were biased whenever the sampling of contour orientations was asymmetric relative to the direction of motion. Performance was improved by the presence of identifiable features, such as line ends or crossings, whose trajectories could be tracked over time. The reports of our observers were consistent with a pooling process involving a vector average of measures of the component of velocity normal to contour orientation, rather than with the predictions of the intersection-of-constraints analysis in velocity space.