Visual illusions, solid/outline invariance and non-stationary activity patterns

When making decisions under uncertainty, human subjects do not always act as rational decision makers, but often resort to one or more mental “shortcuts”, or heuristics, to arrive at a decision. How do such “top-down” processes affect real-world decisions that must take into account empirical, “bottom-up” sensory evidence? Here we use recognition of camouflaged objects by expert viewers as an exemplar case to demonstrate that the effect of heuristics can be so strong as to override the empirical evidence in favor of heuristic information, even though the latter is random. We provided the viewers a random number that we told them was the estimate of a drone reconnaissance system of the probability that the visual image they were about to see contained a camouflaged target. We then showed them the image. We found that the subjects’ own estimates of the probability of the target in the image reflected the random information they were provided, and ignored the actual evidence in the image. However, when the heuristic information was not provided, the same subjects were highly successful in finding the target in the same set of images, indicating that the effect was solely attributable to the availability of heuristic information. Two additional experiments confirmed that this effect was not idiosyncratic to camouflage images, visual search task, or the subjects’ prior training or expertise. Together, these results demonstrate a novel aspect of the interaction between heuristics and sensory information during real-world decision making, where the former can be strong enough to veto the latter. This ‘heuristic vetoing’ is distinct from the vetoing of sensory information that occurs in certain visual illusions.

Tetrahydrocannabinol (cannabis) and lysergic acid diethylamide (LSD) are psychomimetic agents that induce impairment of sensory perception. Illusions and hallucinations are mostly visual. Most frequently the visual phenomena occur in conjunction with drug abuse. Three previously drug-addicted patients were examined for either persisting or spontaneously recurrent visual phenomena. Two patients complained of persisting visual illusions (vibrations, dyskinetopsia and impaired depth perception) during more than 12 months after an excessive use of cannabis. The third patient was a multiple drug abuser (LSD for 6 years) and complained of visual hallucinations and palinopsia following heavy ethanol intake, 20 years after stopping the use of any drug. Results from neuro-ophthalmic and neurological examinations were normal for the first two patients. The third patient presented abnormal visual fields with preserved visual acuity; electroencephalography was abnormal, suggesting an underlying toxic encephalopathy. Persistent visual illusions or hallucinations can occur during several months after an intake of cannabis. Flash-back phenomena are frequent amongst LSD abusers. They rarely occur at long times after the last intake (20 years in the present case); when they do so, precipitating factors are often present (ethanol, medication, anesthesia). Such phenomena reflect the cortical dysfunctions that can be induced by illegal substances.

Introduction: The present study sets out to determine which sensory system mostly influences self-motion perception when visual and vestibular cues are in conflict. We paired caloric vestibular stimulation that signaled motion in either the clockwise or counter-clockwise direction with a visual display that indicated self-rotation in either the same or opposite directions.Methods: In Experiment 1 (E1), caloric vestibular stimulation was used to produce vestibular circular vection. In Experiment 2 (E2), a virtual optokinetic drum was used to produce visual circular vection in a VR headset. Vection speed, direction, and duration were recorded using a potentiometer knob the participant controlled in E1 and E2. In Experiment 3 (E3), visual and vestibular stimuli were matched to be at approximately equal speeds across visual and vestibular modalities for each participant setting up Experiment 4 (E4). In E4, participants observed a moving visual pattern in a virtual reality (VR) headset while receiving caloric vestibular stimulation. Participants rotated the potentiometer knob while attending to visual–vestibular stimuli presentations to indicate their perceived circular vection. E4 had two conditions: 1) A congruent condition where calorics and visual display indicated circular vection in the same direction; 2) an incongruent condition where calorics and visual display indicated circular vection in opposite directions.Results and discussion: There were equal reports of knob rotation in the direction consistent with the visual and vestibular self-rotation direction in the incongruent condition of E4 across trials. There were no significant differences in knob rotation speed and duration in both conditions. These results demonstrate that the brain appears to weigh visual and vestibular cues equally during a visual–vestibular conflict of approximately equal speeds. These results are most consistent with the optimal cue integration hypothesis.

The processing and perception of visual stimuli are greatly dependent on the context in which those stimuli are presented. While this is not always (and indeed is probably not supposed to be) evident when viewing real-world scenes, the abundance of contextual effects is revealed for example by the vast number of different visual illusions. Context can profoundly change the way the visual system processes single visual stimuli. In contrast masking, surround suppression, and crowding, distracting stimuli impair the observer's ability to detect, identify, and make other judgements about visual stimuli. These perceptual phenomena presumably reflect the operation of neural mechanisms that adjust the operating range of the visual system according to the prevailing conditions, or act to reduce responses to redundant stimulus features. This work first investigates the relationship between masking, which is caused by stimuli that are spatially superimposed on a target stimulus, and surround suppression, which is caused by spatially displaced stimuli. This was done using a temporal masking paradigm and measuring the time-courses of masking and surround suppression. The results reveal how the effects of these suppressive interactions are delicately dependent on stimulus timing and suggest that surround suppression acts extremely fast, having an effect on the early part of the neural response. Finally, contextual effects such as surround suppression and crowding are often considered to be automatic processes that occur over fixed areas in the visual field. The final part of this work shows how changing the spatial layout, and consequently the perceptual organization, of the stimuli can dramatically reduce surround suppression and crowding effects, even in conditions where the current models would predict strong effects. Surround suppression and crowding seem to operate only within perceptually homogeneous areas, while stimuli that are perceptually separate, the ones that stand out, are not affected.

2 - The human visual system described through visual illusions

Stability and Intermittency in Large-Scale Coupled Oscillator Models for Perceptual Segmentation

Purpose:Contrast time‐density curves may help differentiate malignant tumors from normal tissues or benign tumors. Repetitive scans using conventional CT or cone beam CT techniques, which Result in unacceptably high dose, may not achieve the desired temporal resolution. In this study we describe and demonstrate a 4D imaging technique for imaging and quantifying contrast flows requiring only one or two 360° scans.Methods:A dual‐gantry system is used to simultaneously acquire two projection images at orthogonal orientations. Following the scan, each or both of the two 360° projection sets are used to reconstruct an average contrast enhanced image set which is then segmented to form a 3D contrast map. Alternatively, a pre‐injection scan may be made and used to reconstruct a pre‐injection image set which is subtracted from the post‐injection image set to form the 3D contrast map. Each of the two 360° projection sets is divided into 12 subsets, thus creating 12 pairs of 30° limited angle projection sets, each corresponding to a time spanning over 1/12 of the scanning time. Each pair of the projection sets are reconstructed as a time specific 3D image set with the maximum likelihood estimation iterative algorithm using the contrast map as the constraint. As a demonstration, a 4D abdominal phantom was constructed from clinical CT images with blood flow through the normal tissue and a tumor modeled and imaging process simulated.Results:We have successfully generated a 4D image phantom, and calculated the projection images. The time density curves derived from the reconstructed image set matched well with the flow model used to generate the phantom.Conclusion:Dual‐gantry image acquisition and constrained iterative reconstruction algorithm may help to obtain time‐density curves of contrast agents in blood flows, which may help differentiate malignant tumors from normal tissues or benign tumors.

Uncertainty in visual processes predicts geometrical optical illusions

Mapping transient hypoxia from in situ activation of 15O by photon beams: A simulation study

The strength of certain visual illusions, including contrast-contrast and apparent motion, is weakened in individuals with schizophrenia. Such phenomena have been interpreted as the impaired integration of inhibitory and excitatory neural responses, and impaired top-down feedback mechanisms. To investigate whether and how these factors influence the perceived contrast-contrast and apparent motion illusions in individuals with schizophrenia, we propose a two-layer network, with top-down feedback from layer 2 to layer 1 that can model visual receptive fields (RFs) and their inhibitory and excitatory subfields. Our neural model suggests that illusion perception changes in individuals with schizophrenia can be influenced by altered top-down mechanisms and the organization of the on-center off-surround receptive fields. Alteration of the RF inhibitory surround and/or the excitatory center can replicate the difference of illusion precepts between individuals with schizophrenia within certain clinical states and normal controls. The results show that the simulated top-down feedback modulation enlarges the difference of the model illusion representations, replicating the difference between the two groups. We propose that the heterogeneity of visual and in general sensory processing in certain clinical states of schizophrenia can be largely explained by the degree of top-down feedback reduction, emphasizing the critical role of top-down feedback in illusion perception, and to a lesser extent on the imbalance of excitation/inhibition. Our neural model provides a mechanistic explanation for the modulated visual percepts of contrast-contrast and apparent motion in schizophrenia with findings that can explain a broad range of visual perceptual observations in previous studies. The two-layer motif of the current model provides a general framework that can be tailored to investigate subcortico-cortical (such as thalamocortical) and cortico-cortical networks, bridging neurobiological changes in schizophrenia and perceptual processing.

Purpose: : Over the preceding decade, onboard volumetric imaging utilizing cone beam CT (CBCT) has become an indispensible tool for image‐guided radiotherapy. A novel method has been developed to deform the planning CT to maximize the spatial congruence between the digitally reconstructed radiographs (DRRs) and CBCT projections. This method maps the planning CT images to the current anatomy using a sparse subset of CBCT projections. Methods: A finite element modeling methodology was developed to automatically generate adaptive meshes for a given region of interest without the need for segmentation. In this methodology, more control points are generated at organs boundaries, special tissues, contour boundaries, or high intensity gradients. The displacement of boundaries and important features is directly represented by the displacement of control points, instead of interpolation on regular grids. A splatting method was developed to generate DRRs of the deformed CT. These DRRs were compared with corresponding 2D projections from CBCT scans; as a Result, the displacement vector fields on the CT were optimized iteratively to obtain a final deformed CT image. A digital phantom was used to evaluate this method. Results: Typically 60 projections were needed to reconstruct a set of 3D images while several hundreds of projections were required for a conventional CBCT reconstruction. The feature‐based meshing method led to better results than either regular orthogonal or regular tetrahedral meshes. The image intensity results of the feature‐based meshing method were nearly identical to those of the voxel‐based method (normalized cross correlation efficient (NCC) ∼ 0.9858 vs 0.9872), while the feature‐based meshing method led to much better consistency of displacement vector fields relative to ground truth than voxel‐based method (NCC ∼ 0.8265 vs 0.6261). Conclusion: This feature‐based meshing method is capable of reconstructing volumetric images with significantly fewer projections while retaining the image quality of planning CT. This research is supported by CPRIT Individual Investigator Award RP110329.

Visual perception is often conceptualized as the analysis of static images which are acquired during fixations. These images are then spatially decomposed into their basic components, which form the fundament of visual perception. Understanding these spatial mechanisms has been a central aim of pattern vision, and has led to a successful and widely adopted standard model of spatial vision. The spatial view on pattern vision, however, overlooks a key aspect of visual processing: the eyes are never still. Even during fixations, involuntary eye movements incessantly modulate the visual input. These eye movements challenge the assumption that visual processing can occur independently of motion. In recent years, empirical evidence has accumulated which suggests that fixational eye movements, particularly ocular drift, actively shape visual processing. Building on these findings, this thesis argues that ocular drift is essential to how the visual system encodes spatial structure, even in the absence of external motion. It calls for a shift from static to active, spatiotemporal models of pattern vision. To support this shift, it integrates computational modeling, psychophysics, and new experimental tools to explore how ocular drift influences edge and pattern perception. The first part of this thesis revisits foundational assumptions in spatial vision. In a first study, it presents a proof-of-concept model that extends a standard spatial vision model by ocular drift and temporal processing. The active model components facilitate edge extraction, but also reveal limitations in current datasets which cannot clearly distinguish between static and active accounts of pattern vision. In response, the next two studies introduce new software tools and a benchmark dataset that specifically target the spatial frequency selective mechanisms underlying pattern vision. Using this dataset, the fourth study explicitly contrasts spatial and active accounts of pattern vision. The results show that incorporating ocular drift improves predictions of human edge sensitivity and reveals that traditional models may rely on compensatory biases to account for the absence of these eye movements. The final study introduces a behavioral task in which participants trace edges in natural scenes. This approach enables the study of pattern vision in more naturalistic contexts while maintaining the analytical rigor of traditional psychophysics through signal detection theory. The resulting dataset supports both standard analyses and investigations into individual differences and the visual features that guide edge perception in real-world settings. Altogether, this thesis integrates insights from active perception into a mechanistically grounded framework of pattern vision. It redefines early visual processing as an active, embodied process shaped by the observer’s own movements. These findings challenge static models of early visual processes and advocate for a broader paradigm shift that places motion, context, and environmental interaction at the core of perception. Finally, it lays a foundation for future research into active models of vision, offering both theoretical direction and practical tools to study visual perception under more natural conditions.

To evaluate a magnetic resonance imaging (MRI) technique that integrates time-resolved angiography with stochastic trajectories (TWIST) view sharing and Dixon for a breast dynamic contrast-enhanced (DCE)-MRI application. Simulation study: K-space data at six timepoints (1 pre-, 5 postcontrast) were generated by performing Fourier transform on a digital "phantom" with 3-9 mm enhancing lesions and three types of enhancement curves (persistent, plateau, washout). Images were reconstructed with and without TWIST. Clinical study: Six TWIST-Dixon image sets (one pre-, five postcontrast) were acquired in 18 patients on a 3T scanner, followed by one conventional image set. The last TWIST-Dixon and the conventional images were scored for seven criteria: perceived signal-to-noise ratio (P.SNR), visualization of anatomy, fat suppression (FS) accuracy, FS uniformity, ghosting artifact, edge ringing artifact, and overall image quality (IQ). Simulation study: With proper TWIST parameters (pA ≥33%, pB ≥50%), the enhancement underestimation was 5% or less for tumor size ≥5 mm. Clinical study: TWIST-Dixon images have significantly better scores in all criteria except for ghosting artifacts, where the difference was not significant. With proper parameters, TWIST-Dixon provides higher perceived SNR, more accurate fat suppression, and better overall image quality for breast DCE-MRI without sacrificing accuracy in the enhancement estimation.

To determine if semiautomated kinetic perimetry (SKP) is reproducible and comparable to Goldmann manual kinetic perimetry (GVF). Glaucoma patients were recruited to perform visual field testing using GVF and SKP. Specific isopters were tested, quantified, and compared. Visual field patterns were analyzed for shape and defect. Ten patients (16 eyes) underwent visual field studies using SKP and GVF, and 8 patients completed a second SKP on a different day. Individual isopter areas were similar between GVF and SKP, although 60% of isopters were larger on SKP by an average of 15%. This was statistically significant for the smaller isopters, I4e (P=0.02) and I2e (P=0.05). Retesting with SKP on a separate day, showed similar isopter areas (P values=0.3 to 1.0), however, the exact location of isopters in degrees from central fixation tended to vary with the smaller test object sizes. Isopter position in degrees from the central axis agreed in at least 3 quadrants in approximately 65% of fields compared. SKP visual field defects and patterns were similar between test strategies. SKP and GVF testing produce similar visual field results in glaucoma patients, and SKP testing seems to be reliable and reproducible in this population. However, overlapping isopters, typically associated with nonorganic vision loss, and jagged isopters were sometimes observed in SKP visual fields. Further study of SKP is needed to explore these findings.

Purpose: To adapt a standardized technique for assessment of visual fields to the study of patients with birdshot chorioretinopathy and to describe visual field patterns and changes over time for affected individuals. Methods: A retrospective chart review was performed of patients with a diagnosis of birdshot chorioretinopathy who had undergone at least two automated visual field studies. A masked evaluation of each visual field was performed using standardized criteria. Subjective symptoms and visual acuity measurements were compared to visual field data from the same examination dates. Serial visual field studies were evaluated for changes in the findings over time. Results: Six patients met inclusion criteria. A total of 104 visual field studies (ranging from 2–19 studies per eye) were evaluated. Visual field abnormalities occurred in all tested eyes and were observed even in eyes with a Snellen visual acuity of 20/15. Discordant patterns were observed in the two eyes of some individual patients. On serial testing, multiple patterns of visual field abnormality were observed in the same eye over time. Conclusions: A standardized protocol for systematic evaluation of visual field measurements can be used to characterize disturbances of vision in patients with birdshot chorioretinopathy. Abnormalities of visual fields may occur even when central visual acuity is preserved.