Opponent motion interactions in the perception of transparent motion

Multi-stability with ambiguous visual stimuli in Drosophila orientation behavior

It is widely accepted for humans and higher animals that vision is an active process in which the organism interprets the stimulus. To find out whether this also holds for lower animals, we designed an ambiguous motion stimulus, which serves as something like a multi-stable perception paradigm in Drosophila behavior. Confronted with a uniform panoramic texture in a closed-loop situation in stationary flight, the flies adjust their yaw torque to stabilize their virtual self-rotation. To make the visual input ambiguous, we added a second texture. Both textures got a rotatory bias to move into opposite directions at a constant relative angular velocity. The results indicate that the fly now had three possible frames of reference for self-rotation: either of the two motion components as well as the integrated motion vector of the two. In this ambiguous stimulus situation, the flies generated a continuous sequence of behaviors, each one adjusted to one or another of the three references.

Vision is considered an active process in humans and higher animals in which the stimulus is interpreted by the subject and can be perceived in different ways if it is ambiguous. We aimed to find out whether this also holds for lower animals, such as the fruit fly Drosophila melanogaster. To provide ambiguity, we exposed flies to transparent motion stimuli in a flight simulator and found their behavior to be multi-stable. These results show that the visual system of the fly can separate the individual components of a transparent motion stimulus, and that this kind of stimulus is ambiguous to the fly. The extent to which the fly shows component selectivity in its behavior depends on several properties of the stimulus, like pattern contrast and element density. The alternations between the different behaviors exhibit a stochasticity reminiscent of the temporal dynamics in human multi-stable perception.

Motion-defined surface segregation in human visual cortex

Surface segregation provides an efficient way to parse the visual scene for perceptual analysis. Here, we investigated the segregation of a bivectorial motion display into transparent surfaces through a psychophysical task and fMRI. We found that perceptual transparency correlated with neural activity in the early areas of the visual cortex, suggesting these areas may be involved in the segregation of motion-defined surfaces. Two oppositely rotating, uniquely colored random dot kinematograms (RDKs) were presented either sequentially or in a spatially interleaved manner, displayed at varying alternation frequencies. Participants reported the color and rotation direction pairing of the RDKs in the psychophysical task. The spatially interleaved display generated the percept of motion transparency across the range of frequencies tested, yielding ceiling task performance. At high alternation frequencies, performance on the sequential display also approached ceiling, indicative of perceived transparency. However, transparency broke down in lower alternation frequency sequential displays, producing performance close to chance. A corresponding pattern mirroring the psychophysical data was also evident in univariate and multivariate analyses of the fMRI BOLD activity in visual cortical areas V1, V2, V3, V3AB, hV4, and V5/MT+. Using gray RDKs, we found significant presentation by frequency interactions in most areas; differences in BOLD signal between presentation types were significant only at the lower alternation frequency. Multivariate pattern classification was similarly unable to discriminate between presentation types at the higher frequency. This study provides evidence that early visual cortex may code for motion-defined surface segregation, which in turn may enable perceptual transparency.

Perceptual separation of transparent motion components: the interaction of motion, luminance and shape cues

Transparency is perceived when two or more objects or surfaces can be separated by the visual system whilst they are presented in the same region of the visual field at the same time. This segmentation of distinct entities on the basis of overlapping local visual cues poses an interesting challenge for the understanding of cortical information processing. In psychophysical experiments, we studied stimuli that contained randomly positioned disc elements, moving at two different speeds in the same direction, to analyse the interaction of cues during the perception of motion transparency. The current work extends findings from previous experiments with sine wave luminance gratings which only vary in one spatial dimension. The reported experiments manipulate low-level cues, like differences in speed or luminance, and what are likely to be higher level cues such as the relative size of the elements or the superposition rules that govern overlapping regions. The mechanism responsible for separation appears to be mediated by combination of the relevant and available cues. Where perceived transparency is stronger, the neural representations of components are inferred to be more distinguishable from each other across what appear to be multiple cue dimensions. The disproportionally large effect on transparency strength of the type of superposition of disc suggests that with this manipulation, there may be enhanced separation above what might be expected from the linear combination of low-level cues in a process we term labelling. A mechanism for transparency perception consistent with the current results would require a minimum of three stages; in addition to the local motion detection and global pooling and separation of motion signals, findings suggest a powerful additional role of higher level separation cues.

Non-Bayesian contour synthesis

Recent research has witnessed an explosive increase in models that treat percepts as optimal probabilistic inference. The ubiquity of partial camouflage and occlusion in natural scenes, and the demonstrated capacity of the visual system to synthesize coherent contours and surfaces from fragmented image data, has inspired numerous attempts to model visual interpolation processes as rational inference. Here, we report striking new forms of visual interpolation that generate highly improbable percepts. We present motion displays depicting simple occlusion sequences that elicit vivid percepts of illusory contours (ICs) in displays for which they play no necessary explanatory role. These ICs define a second, redundant occluding surface, even though all of the image data can be fully explained by an occluding surface that is clearly visible. The formation of ICs in these images therefore entails an extraordinarily improbable co-occurrence of two occluding surfaces that arise from the same local occlusion events. The perceived strength of the ICs depends on simple low-level image properties, which suggests that they emerge as the outputs of mechanisms that automatically synthesize contours from the pattern of occlusion and disocclusion of local contour segments. These percepts challenge attempts to model visual interpolation as a form of rational inference and suggest the need to consider a broader space of computational problems and/or implementation level constraints to understand their genesis.

Perceptual costs for motion transparency evaluated by two performance measures

Transparency perception is recognized as one of the important phenomena to understand the computational mechanism of early visual system. Transparency perception indicates that a simple theory reconstructing a single-valued field of a visual attribute, such as an optical-flow field, cannot model the neural mechanism for the human visual system and raises a fundamental issue of how visual attributes are represented and detected in the brain. It is considered that one of the important cues to reveal the neural encoding mechanism for overlapping surfaces is the perceptual cost in transparency perception. It has been known that the perceptual performance in motion transparency is worse than that expected from single motion perception. This perceptual "cost" would reflect the encoding strategy for transparent motions. Here we present a systematic study comparing the perceptual costs in motion transparency evaluated by two performance measures. The result showed that the properties of the perceptual costs varied with the performance measures. The perceptual cost evaluated by the motion detection threshold became smaller as a directional difference between overlapping motions increased, whereas the cost examined with the precision of directional judgments became worse. A computational analysis suggests that these contradictory results cannot be explained by a simple population coding model for motion directions.

Opponent-motion mechanisms are self-normalizing

In the ultimate stage of the Adelson-Bergen motion energy model [Adelson, E. H., & Bergen, J. (1985). Spatiotemporal energy models for the perception of motion. Journal of the Optical Society of America, 2, 284-299], motion is derived from the difference between directionally opponent energies E(L) and E(R). However, Georgeson and Scott-Samuel [Georgeson, M. A., & Scott-Samuel, N. E. (1999). Motion contrast: A new metric for direction discrimination. Vision Research, 39, 4393-4402] demonstrated that motion contrast-a metric that normalizes opponent motion energy (E(L)-E(R)) by flicker energy (E(L)+E(R))-is a better descriptor of human direction discrimination. In a previous study [Rainville, S. J. M., Makous, W. L., & Scott-Samuel, N. E. (2002). The spatial properties of opponent-motion normalization. Vision Research, 42, 1727-1738], we used a lateral masking paradigm to show that opponent-motion normalization is selective for flicker position, orientation, and spatial-frequency. In the present study, we used a superposition masking paradigm and compared results to lateral masking data, as the two masking types activate local and remote normalization mechanisms differentially. Although selectivity for flicker orientation and spatial frequency varied across observers, bandwidths were similar across lateral and superimposed masking conditions. Additional experiments demonstrated that normalization signals are pooled over a spatial region whose aspect ratio and size are consistent with those of local motion detectors. Together, results show no evidence of remote normalization signals predicted by broadband inhibitory models [(e.g.) Heeger, D. J. (1992). Normalization of cell responses in cat striate cortex. Visual Neuroscience, 9, 181-197; Foley, J. M. (1994). Human luminance pattern-vision mechanisms: Masking experiments require a new model. Journal of the Optical Society of America A-Optics and Image Science, 11, 1710-1719] but support a local normalization process whose spatial properties are inherited from low-level motion detectors.

The efficiency of speed discrimination for coherent and transparent motion

Transparent motion involves the integration and segmentation of local motion signals. Previous research found a cost for processing transparent random dot motions relative to single coherent motions. However, this cost can be the result of the increased complexity of the transparent stimuli. We investigated this possibility by measuring the efficiency of transparent and coherent motions. Since efficiency normalises human performance to that of an ideal observer in the same task, performance can be compared fairly across tasks. Our task, identical in both transparent and coherent conditions, was to discriminate the fastest speed between two opposite motion directions. In two experiments where we varied dot density and speed, we confirmed the cost in human sensitivity for transparent motion but also found a cost for the ideal observer. The outcome was a consistent residual cost in efficiency for transparent motion. This result points to a processing limitation for transparent motion analogous to previously suggested inhibitory mechanisms between opposite directions of motion. Furthermore, we found that both transparent and coherent motion efficiencies decreased as dot density increased. This latter result stresses the importance of the correspondence problem and suggests that local motion signals are integrated over large areas.

A gain-control model relating nulling results to the duration of dynamic motion aftereffects

Strength of the motion aftereffect (MAE) is most often quantified by its duration, a high-variance and rather 'subjective' measure. With the help of an automatic gain-control model we quantitatively relate nulling-thresholds, adaptation strength, direction discrimination threshold, and duration of the dynamic MAE (dMAE). This shows how the nulling threshold, a more objective two-alternative forced-choice measure, relates to the same system property as MAE-durations. Two psychophysical experiments to test the model use moving random-pixel-arrays with an adjustable luminance signal-to-noise ratio. We measure MAE-duration as a function of adaptation strength and compare the results to the model prediction. We then do the same for nulling-thresholds. Model predictions are strongly supported by the psychophysical findings. In a third experiment we test formulae coupling nulling threshold, MAE-duration, and direction-discrimination thresholds, by measuring these quantities as a function of speed. For the medium-to-high speed range of these experiments we found that nulling thresholds increase and dMAE-durations decrease about linearly, whereas direction discrimination thresholds increase exponentially with speed. The model description then suggests that the motion-gain decreases, while the noise-gain and model's threshold increase with speed.

Exogenously cued attention triggers competitive selection of surfaces

It has been reported that when an endogenous cue directs attention to a brief translation of one of two superimposed surfaces, observers reliably report the direction of that translation as well as the direction of a second translation of the cued surface. In contrast, if the uncued surface translates second, direction judgments are severely impaired for several hundred milliseconds. We replicated this result, but found that the impairment survived the removal of the endogenous cue. The impairment is therefore not due to endogenously cued attention. Instead, a brief translation of one surface acts as an exogenous cue that triggers an automatic selection mechanism, which suppresses processing of the other surface. This study provides a clear case of exogenous cueing of surface-based attention. We relate these results to identified competitive selection mechanisms in visual cortex.

An oscillatory correlation model of visual motion analysis

We describe and evaluate a model of motion perception based on the integration of information from two parallel pathways: a motion pathway and a luminance pathway. The motion pathway has two stages. The first stage measures and pools local motion across the input animation sequence and assigns reliability indices to these pooled measurements. The second stage groups locations on the basis of these measurements. In the luminance pathway, the input scene is segmented into regions on the basis of similarities in luminance. In a subsequent integration stage, motion and luminance segments are combined to obtain the final estimates of object motion. The neural network architecture we employ is based on LEGION (locally excitatory globally inhibitory oscillator networks), a scheme for feature binding and region labeling based on oscillatory correlation. Many aspects of the model are implemented at the neural network level, whereas others are implemented at a more abstract level. We apply this model to the computation of moving, uniformly illuminated, two-dimensional surfaces that are either opaque or transparent. Model performance replicates a number of distinctive features of human motion perception.

The spatial properties of opponent-motion normalization

The final stage of the Adelson-Bergen model [J. Opt. Soc. Am. A 2 (1985) 284] computes net motion as the difference between directionally opposite energies E(L) and E(R). However, Georgeson and Scott-Samuel [Vis. Res. 39 (1999) 4393] found that human direction discrimination is better described by motion contrast (C(m))--a metric where opponent energy (E(L)-E(R)) is divided by flicker energy (E(L)+E(R)). In the present paper, we used a lateral masking paradigm to investigate the spatial properties of flicker energy involved in the normalization of opponent energy. Observers discriminated between left and right motion while viewing a checkerboard in which half of the checks contained a drifting sinusoid and the other half contained flicker (i.e. a counterphasing sinusoid). The relative luminance contrasts of flicker and motion checks determined the checkerboard's overall motion contrast C(m). We obtained selectivity functions for opponent-motion normalization by measuring C(m) thresholds whilst varying the orientation, spatial frequency, or size of flicker checks. In all conditions, performance (percent correct) decayed lawfully as we decreased motion contrast, validating the C(m) metric for our stimuli. Thresholds decreased with check size and also improved as we increased either the orientation or spatial-frequency difference between motion and flicker checks. Our data are inconsistent with Heeger-type normalization models [Vis. Neurosci. 9 (1992) 181] in which excitatory inputs are normalized by a non-selective pooling of inhibitory inputs, but data are consistent with the implicit assumption in Georgeson and Scott-Samuel's model that flicker normalization is localized in orientation, scale, and space. However, our lateral masking paradigm leaves open the possibility that the spatial properties of flicker normalization would be different if opponent and flicker energies spatially overlapped. Further characterization of motion contrast will require models of the spatial, temporal, and joint space-time properties of mechanisms mediating opponent-motion and flicker normalization.

Direction repulsion in unfiltered and ring-filtered Julesz textures

Perceived directions of motion were measured for each of two superposed two-dimensional dynamic random patterns consisting of unfiltered or ring-filtered dense random-check (Julesz) textures. One pattern always moved in a cardinal direction (up, down, left, or right), and the other texture always moved in an oblique direction separated from the cardinal component by 20 degrees-80 degrees. Several cardinal/oblique speed ratios were tested. In Experiment 1, the textures were unfiltered. In Experiment 2, the textures were ring filtered and had the same center frequency (1, 2, or 4 cpd). In Experiment 3, a 1-cpd ring-filtered texture was paired with a 2-, 4-, or 8-cpd texture. Subjects consistently misperceived the directions of component motion in these experiments; the angular separation of movement of the two textures was perceptually exaggerated, a phenomenon referred to as direction repulsion (Marshak & Sekuler, 1979). The results show that (1) direction repulsion occurs across at least a fourfold range of spatial frequencies and a sixfold range of speed ratios, (2) direction repulsion varies systematically with speed ratio, and (3) across most conditions, direction repulsion is anisotropic--direction repulsion is more evident in the oblique directions than in the cardinal directions. These findings suggest that the spatiotemporal range of inhibitory interactions involved in motion transparency is much greater than previously appreciated.