Mean Eye Position Research Articles

An Object-Based Bayesian Framework for Top-Down Visual Attention

We introduce a new task-independent framework to model top-down overt visual attention based on graph-ical models for probabilistic inference and reasoning. We describe a Dynamic Bayesian Network (DBN) that infers probability distributions over attended objects and spatial locations directly from observed data. Probabilistic inference in our model is performed over object-related functions which are fed from manual annotations of objects in video scenes or by state-of-the-art object detection models. Evaluating over ∼3 hours (appx. 315,000 eye fixations and 12,600 saccades) of observers playing 3 video games (time-scheduling, driving, and flight combat), we show that our approach is significantly more predictive of eye fixations compared to: 1) simpler classifier-based models also developed here that map a signature of a scene (multi-modal information from gist, bottom-up saliency, physical actions, and events) to eye positions, 2) 14 state-of-the-art bottom-up saliency models, and 3) brute-force algorithms such as mean eye position. Our results show that the proposed model is more effective in employing and reasoning over spatio-temporal visual data.

Estimation of the degree of autism spectrum disorder by the slow phase of optokinetic nystagmus in typical adults

Atypical eye movement patterns demonstrated by individuals with autism spectrum disorder (ASD) have the potential to serve as biomarkers for ASD diagnosis. However, instead of estimating individual differences in the degree of ASD from those patterns, many researchers have compared ASD groups with typical development groups. This study investigates the relationship between the Autism-spectrum Quotient (AQ) scores in typical adults, which can evaluate the degree of the traits associated with ASD, as well as the properties of optokinetic nystagmus (OKN), including the gain of the slow phase, the peak velocity and duration of the fast phase, the frequency, and the mean eye position of OKN. A random dot pattern that moved in one direction was presented on the display, and the participants' eye movements were measured. The results showed a negative correlation between subjects' AQ scores and the gain of slow-phase OKN. In addition, the correlations between subjects' AQ scores and the properties of OKN fast phase were not significant. These results indicate that the gain of slow-phase OKN could be a biomarker that estimates individual differences in the degree of ASD, reflected in our findings which considered AQ scores in typical adults.

Use of the Bispectral Index to Predict Eye Position of Children during General Anesthesia.

PurposeTo assess the relationship between eye position and anesthesia depth using the bispectral index (BIS) value, a parameter derived from electroencephalography data.MethodsWe investigated the relationship between BIS value and eye position in 32 children who underwent surgery for epiblepharon under general anesthesia. BIS values were recorded continuously throughout the procedure (from induction to awakening). Eye positions were video-recorded and analyzed after surgery. The vertical position of each eye was scored according to its height in relation to the medial canthus. An eye position in which the upper eyelid covered one-third of the cornea was defined as a significant ocular elevation.ResultsThe BIS value correlated inversely with the end-tidal concentration of each anesthetic agent, whereas it correlated positively with the eye elevation score (eye position = 0.014 × BIS + 0.699, p = 0.011). The mean eye position score was significantly greater in patients whose BIS values were over 65. Eleven patients (34.4%) had significant ocular elevation; their mean concurrent BIS value was 61.6. Two of these patients had elevation during surgery and 9 had elevation during emergence from anesthesia.ConclusionsWe found that high BIS values were correlated with low levels of anesthetic concentration and high eye position, suggesting that BIS monitoring may be useful for predicting eye position during anesthesia. Particular attention must be given to eye position during ophthalmic surgery. Anesthesia depth can be maintained by assuring that the BIS value remains below 65.

Microsaccades enable efficient synchrony-based visual feature learning and detection

Fixational eye movements are common across vertebrates, yet their functional roles, if any, are debated [1]. To investigate this issue, we exposed the Virtual Retina simulator [2] to natural images, generated realistic drifts and microsaccades using the model of ref. [3], and analyzed the output spike trains of the parvocellular retinal ganglion cells (RGC). We first computed cross-correlograms between pairs of RGC that are strongly excited by the image corresponding to the mean eye position. Not surprisingly, in the absence of eye movements, that is when analyzing the tonic (sustained) response to a static image, these cross-correlograms are flat. Adding some slow drift (~20 min/s, self-avoiding random walk) creates long timescale (>1s) correlations because both cells tend to have high firing rates for central positions. Adding microsaccades (~0.5° in 25ms, that is ~20°/s) creates short timescale (tens of ms) correlations: cells that are strongly excited at a particular landing location tend to spike synchronously shortly after the landing. What do the patterns of synchronous spikes represent? To investigate this issue, we fed the RGC spike trains to neurons equipped with spike timing-dependent plasticity (STDP) and lateral inhibitory connections, as in ref. [4]. Neurons self-organized, and each one selected a set of afferents that consistently fired synchronously. We then reconstructed the corresponding visual stimuli by convolving the synaptic weight matrices with the RGC receptive fields. In most cases, we could easily recognize what was learned (e.g. a face), and the neuron was selective (e.g. only responded for microsaccades that landed on a face). Without eye movements, or with only the drift, the STDP-based learning failed, because it needs correlations at a timescale roughly matching the STDP time constants [5]. Microsaccades are thus necessary to generate a synchrony-based coding scheme. More specifically, after each microsaccade landing, cells that are strongly excited by the image corresponding to the landing location tend to fire their first spikes synchronously. Patterns of synchronous spikes can be decoded rapidly – as soon as the first spikes are received – by downstream “coincidence detector” neurons, which do not need to know the landing times. Finally, the required connectivity to do so can spontaneously emerge with STDP. As a whole, these results suggest a new role for microsaccades – to enable efficient visual feature learning and detection thanks to synchronization – that differs from other proposals such as time-to-first spike coding with respect to microsaccade landing times.

Asymmetries in attention as revealed by fixations and saccades.

Neurologically normal individuals devote more attention to the left side; an asymmetry known as pseudoneglect, which reflects right hemisphere involvement in visuospatial attention. The role of eye movements in attentional asymmetries has received little consideration, particularly in terms of the greyscales task. Stimulus length, elevation, and presentation duration were manipulated, while monitoring eye movements during the greyscales task. Region of interest analyses compared time spent examining each quadrant of the stimulus. Further, saccades were examined in conjunction with fixations to gain an understanding of overall eye movement patterns. Scatterplots combining x-and y-coordinates illustrate mean eye position. Results demonstrated a comparison strategy was used, where the dark portions of each rectangle were fixated. Mean eye position was within the lower left quadrant. The left visual field was inspected most for the baseline condition. Interestingly, the lower visual field was examined most when duration, length, or elevation was manipulated. Eye movement patterns provide a possible explanation for why correlations are y not observed between visuospatial tasks. Different strategies, based on specific-task demands, are likely to be used, which in turn, engage separate aspects of visuospatial attention.

What/Where to Look Next? Modeling Top-Down Visual Attention in Complex Interactive Environments

Several visual attention models have been proposed for describing eye movements over simple stimuli and tasks such as free viewing or visual search. Yet, to date, there exists no computational framework that can reliably mimic human gaze behavior in more complex environments and tasks such as urban driving. In addition, benchmark datasets, scoring techniques, and top-down model architectures are not yet well understood. In this paper, we describe new task-dependent approaches for modeling top-down overt visual attention based on graphical models for probabilistic inference and reasoning. We describe a dynamic Bayesian network that infers probability distributions over attended objects and spatial locations directly from observed data. Probabilistic inference in our model is performed over object-related functions that are fed from manual annotations of objects in video scenes or by state-of-the-art object detection/recognition algorithms. Evaluating over approximately 3 h (approximately 315 000 eye fixations and 12 000 saccades) of observers playing three video games (time-scheduling, driving, and flight combat), we show that our approach is significantly more predictive of eye fixations compared to: 1) simpler classifier-based models also developed here that map a signature of a scene (multimodal information from gist, bottom-up saliency, physical actions, and events) to eye positions; 2) 14 state-of-the-art bottom-up saliency models; and 3) brute-force algorithms such as mean eye position. Our results show that the proposed model is more effective in employing and reasoning over spatio-temporal visual data compared with the state-of-the-art.

Velocity storage mechanism in zebrafish larvae

The optokinetic reflex (OKR) and the angular vestibulo-ocular reflex (aVOR) complement each other to stabilize images on the retina despite self- or world motion, a joint mechanism that is critical for effective vision. It is currently hypothesized that signals from both systems integrate, in a mathematical sense, in a network of neurons operating as a velocity storage mechanism (VSM). When exposed to a rotating visual surround, subjects display the OKR, slow following eye movements frequently interrupted by fast resetting eye movements. Subsequent to light-off during optokinetic stimulation, eye movements do not stop abruptly, but decay slowly, a phenomenon referred to as the optokinetic after-response (OKAR). The OKAR is most likely generated by the VSM. In this study, we observed the OKAR in developing larval zebrafish before the horizontal aVOR emerged. Our results suggest that the VSM develops prior to and without the need for a functional aVOR. It may be critical to ocular motor control in early development as it increases the efficiency of the OKR.

Central fixations with rightward deviations: saccadic eye movements on the landmark task

Neurologically normal individuals show an attentional bias toward the left side, which results from right hemisphere activation during visuospatial tasks. The strength of this bias is influenced by various factors, such as line length, vertical elevation and presentation time. What remains unknown is how participants gather information via saccadic eye movements during task performance and how this relates to their responses. Eye movements were recorded while participants performed the landmark task. Fixations and saccades were both analysed to gain a complete understanding of eye movement patterns. Fixations tended to focus on the centre of the line, with few left-right differences. Saccades were examined by creating histograms illustrating all x-coordinates which were examined over the course of each trial. Interestingly, mean eye position varied with participant response, with an overall tendency to look to the right of centre. Results are consistent with prior research, which has primarily looked at fixations and demonstrates the necessity of examining saccades as well as fixations in order to see how eye movement patterns relate to pseudoneglect.

Direction‐dependent visual cortex activation during horizontal optokinetic stimulation (fMRI study)

Looking at a moving pattern induces optokinetic nystagmus (OKN) and activates an assembly of cortical areas in the visual cortex, including lateral occipitotemporal (motion-sensitive area MT/V5) and adjacent occipitoparietal areas as well as ocular motor areas such as the prefrontal cortex, frontal, supplementary, and parietal eye fields. The aim of this functional MRI (fMRI) study was to investigate (1) whether stimulus direction-dependent effects can be found, especially in the cortical eye fields, and (2) whether there is a hemispheric dominance of ocular motor areas. In a group of 15 healthy subjects, OKN in rightward and leftward directions was visually elicited and statistically compared with the control condition (stationary target) and with each other. Direction-dependent differences were not found in the cortical eye fields, but an asymmetry of activation occurred in paramedian visual cortex areas, and there were stronger activations in the hemisphere contralateral to the slow OKN phase (pursuit). This can be explained by a shift of the mean eye position of gaze (beating field) in the direction of the fast nystagmus phases of approximately 2.6 degrees, causing asymmetrical visual cortex stimulation. The absence of a significant difference in the activation pattern of the cortical eye fields supports the view that the processing of eye movements in both horizontal directions is mediated in the same cortical ocular motor areas. Furthermore, no hemispheric dominance for OKN processing was found in right-handed volunteers.

A review of the effects of space flight on the asymmetry of vertical optokinetic and vestibulo-ocular reflexes

Prolonged microgravity during orbital flight is a unique way to modify the otolith inputs and to determine the extent of their contribution to the vertical vestibulo-ocular reflex (VOR) and optokinetic nystagmus (OKN). This paper reviews the data collected on 10 astronauts during several space missions and focuses on the changes in the up-down asymmetry. Both the OKN elicited by vertical visual stimulation and the active VOR elicited by voluntary pitch head movements showed an asymmetry before flight, with upward slow phase velocity higher than downward slow phase velocity. Early in-flight, this asymmetry was inverted, and a symmetry of both responses was later observed. An upward shift in the vertical mean eye position in both OKN and VOR suggests that these effects may be related to otolith-dependent changes in eye position which, in themselves, affect slow phase eye velocity.

Modulation of spatial attention with unidirectional field motion: an implication for the shift of the OKN beating field

During optokinetic nystagmus (OKN) the mean eye position of gaze (the beating field) shifts in the direction of the fast phases. The function of this shift may be to re-orient the eyes in the direction of self-motion which optic flow implies (in-coming field). This idea leads to the hypothesis that visual attention may be directed toward the In-coming field. In Experiment 1, subjects detected a visual flash presented against unidirectional field motion. The OKN beating field was shifted toward the In-coming field, and manual reaction times were shorter when the target appeared in the In-coming field. Experiment 2 revealed that this In-coming field advantage occurred even when OKN (and thus the mean eye-position shift) was suppressed. Subsequent experiments showed that the In-coming field advantage is not due to a local motion interaction (Experiment 3), survives subject's voluntary allocation of attention (Experiment 4), and develops over less than 320 ms after the onset of the motion field (Experiment 5). These results suggest that unidirectional field motion tends to automatically shift visual attention toward the In-coming field.

Changes in horizontal oculomotor behaviour coincide with a shift in visual motion perception.

During full-field rotation of the visual field, subjects commonly experience an initial perception of object-motion which 'switches' to a perception of self-motion. We studied the characteristics of the horizontal optokinetic nystagmus responses evoked by a moving visual stimulus in these two perceptual states over a range of stimulus velocities. During self-motion perception mean eye position was found to shift more in the direction of the newly appearing stimulus elements with a slight reduction in slow phase gain in comparison to the nystagmus evoked during object-motion perception. The results may reflect a modified strategy of spatial attention with increased emphasis on anticipatory eye movements during visually induced self-motion perception.

Latent and Congenital Nystagmus in Down Syndrome

Although nystagmus has been reported in Down syndrome (DS), it has been poorly characterized, because most investigators have relied on clinical observations rather than on eye movement recordings. This study was conducted to investigate nystagmus in DS, using quantitative measurements of eye movements. Ocular motility and visual functions were examined in 26 unselected adults with DS and compared with those in an age-matched group of 35 subjects with other causes of mental retardation. The eye movements of those with clinically evident nystagmus were recorded with the infrared technique. We also recorded the eye movements of a child with DS and nystagmus. Nystagmus was identified in six (23%) adults with DS and in none in the control group. All six patients showed latent/manifest latent nystagmus (LMLN), prominent with the covering of one eye, and esodeviations of 10 to 30 prism diopters. Eye movement recordings confirmed LMLN with its exponentially decaying waveform. Frequencies ranged from 2 to 5 Hz and amplitudes from 5 degrees to 20 degrees. While attempting to fixate straight ahead in the absence of visual cues, three subjects exhibited shifts in the mean eye position. In contrast with the findings in adults, the only child with DS examined had both congenital nystagmus and LMLN waveforms. The predominant type of nystagmus in the study subjects with DS is LMLN. The high prevalence of LMLN may reflect abnormal integration of visuospatial information that is typical of DS. The concurrent presence of congenital nystagmus in a child but only LMLN in the adults with DS raises the possibility of age-related waveform changes or could reflect sample variation.

Clinical significance of cervico-ocular reactions

Cervically induced eye movements consist of a nystagmus and a deviation of the mean eye position (shift). They show in relation to different neck torsion velocities maximum reactions at slow velocities. The clinical significance of these cervico-ocular reactions is discussed controversially. Therefore we investigated 40 healthy subjects without any neurootological findings, who, in addition, underwent a manual examination. The neck torsion test was performed automatically with quantifiable stimulus parameters and a complete head fixation by means of individual dental casts. It could be shown that at a constant chair velocity of 5 degrees/s every healthy subject exhibits cervical nystagmus and/or shift deviations. In comparison a group of 30 patients with an upper cervical spine syndrome also showed similar cervico-ocular reactions without significant difference. It can be concluded that a muscle hypertonus in the deep neck region does not lead to pathological, cervically induced eye movements and that the cervical nystagmus itself is not a pathognomonic sign for cervical, proprioceptive vertigo.

Properties of superior vestibular nucleus neurons projecting to the cerebellar flocculus in the squirrel monkey.

1. Properties of superior vestibular nucleus (SVN) neurons and their projection to the cerebellar flocculus were studied in alert squirrel monkeys by using chronic unit and eye movement recording and microstimulation techniques. Twenty-three cells were antidromically activated from the ipsilateral flocculus, and seventeen of these were also orthodromically activated from the ipsilateral VIIth nerve at monosynaptic latencies. Only 1 of these 23 units was also inhibited by flocculus stimulation. According to their response properties, 9 of the cells were pure vestibular, 2 were vestibular-pause, and 12 were position-vestibular cells. The mean eye position sensitivity of these position-vestibular cells was significantly lower than that of cells projecting to the oculomotor nucleus (OMN). No eye movement-only neurons were antidromically activated from the flocculus. No cells could be antidromically activated from both the oculomotor nucleus and the flocculus.

A "beat-to-beat" interval generator for optokinetic nystagmus.

An analysis of optokinetic responses was used to derive an iterative model that reproduces the duration of nystagmus slow phases and eye position control during optokinetic nystagmus. Optokinetic nystagmus was recorded with magnetic search coils from red-eared turtles (Pseudemys scripta elegans) during monocular, random dot pattern stimulation at constant velocities ranging from 0.25-63 degrees/s. The beat-to-beat behavior of slow phase durations was consistent with the existence of an underlying neural clock, termed the basic interval generator, that is based on an integrate-to-fire neuron model. This hypothetical basic interval generator produces an interval that is the product of the duration of the previous interval and a mean 1 truncated normal variate with variance sigma 2. Data analyses indicated that the initial value of the interval generator during a period of nystagmus, termed tau 0, is proportional to the inverse square root of slow phase eye velocity. Further, if the eye was deviated in the slow phase direction (re mean eye position) when the slow phase began, the slow phase duration was consistent with a single cycle of the basic interval generator. However, if the eye was deviated in the fast phase direction, the distribution of the durations of the ensuing slow phases indicated that a proportion of the slow phases were produced by more than one cycle of the basic interval generator. This phenomenon is termed "skipping a beat" and occurs with probability Ps. Finally, the amplitude of fast phases behaved as a linear function of eye position at the fast phase onset and the product of tau 0 and slow phase eye velocity. A computer simulation reproduced the observed distribution of slow phase durations, the proportion of fast phases in the fast phase and slow phase directions and the distribution of eye positions at the onset and end of fast phases. This novel model suggests that both timing and eye position information contribute to the alternation of nystagmus fast and slow phases.

Cervico-ocular eye movements in relation to different neck torsion velocities.

Rotation of the trunk with a fixed head causes isolated stimulation of cervical afferent input. Eleven patients with marked cervico-ocular reactions (COR) were selected. A neck torsion test (NTT) was performed with the help of a special apparatus. The resulting cervically induced eye movements consisted of two components. In addition to a cervical nystagmus, with its slow, compensatory phase, a deviation of the mean eye position (shift) in an anticompensatory direction, opposite to trunk rotation was observed. The COR in relation to different neck torsion velocities distinctly showed a tuning curve with maximum reactions between 4 degrees/s and 9 degrees/s. Above these values, a strong decrease in COR resulted. The optimal working range vastly differs from that of the vestibulo-ocular reflex (VOR).

Effects of light and dark environments on macaque and human fixational eye movements

Eye position of two macaques and two humans was recorded while they detected the unpredictable dimming of a fixation spot in a dark or a light environment. Fixational saccades often had complex waveforms that resulted from clustering of two or more saccadic displacements with no intervening drift periods. In the dark, all subjects had low frequencies of saccade clusters (0.15-0.61/sec). Three of the subjects increased the frequency of saccade clusters and decreased the magnitude of the displacements when the task was performed in the normally lighted laboratory. The higher saccade frequencies in the light did not automatically result in greater dispersion of eye position. One of the humans, who had the lowest saccade frequency, was relatively unaffected by the stimulus conditions. The change in stimulus conditions had a more pronounced effect on the fixational eye movements of the macaques than the humans. In the dark, the macaques made saccades mostly down and nasal, but in the light the saccadic displacements occurred over a wider range of angular directions. Mean eye position was higher in the dark than in the light. The humans altered the direction of their saccadic displacements very little and did not change their mean eye position when they switched from a dark to a lighted environment. The influence of a lighted environment is interpreted as an interaction between foveolar and peripheral retinal inputs. The results suggest that fixational saccades may have more than one role, perhaps including stimulation of pathways originating in the peripheral retina. These peripheral field inputs have a stronger effect on the fixational control system of macaques than of humans.

Modulation of Mean Eye Position during Vestibular Induced Eye Movements of the Rhesus Monkey

Vestibular induced nystagmus in the dark was measured during sinusoidal rotational stimulation in three male monkeys using electrooculography. A periodic modulation of mean eye position was observed. This periodic modulation was quantified by least squares fitting through the nystagmus a sinusoid whose frequency was the same as that of the stimulus. The magnitude and the phase of the best fitting sinusoid were used to determine gain and phase delay at various combinations of stimulus amplitudes and frequency. Results indicate that the mean eye position in the alert animals is predominantly modulated in the direction of the fast component of nystagmus. The magnitude of this modulation of mean eye position is independent of the amplitude of the stimulus. The phase relationship of this modulation is dependent on the frequency of stimulation. In drowsy animals, the modulation of mean eye position is predominantly in the direction of the slow component of nystagmus. We conclude that the modulation of mean eye position during vestibular induced nystagmus operates as an automatic gain control system which is sensitive to the animal's level of alertness. The quantitative data in this report are useful in refining current models of vestibular nystagmus.

Contributions to the study of “Blindsight”—II. The role of specific practice for saccadic localization in patients with postgeniculate visual field defects

This study investigated the effect of practice on saccadic localization in the perimetrically blind field region by patients with postgeniculate brain damage. The analysis of the saccadic responses of three patients with homonymous visual field defects showed that in the first test sessions they did not choose between different locations but, rather, preferred a consistent magnitude of eye movements. In these sessions no correspondence between target and mean eye position was found. After the patients were forced to change final eye position in each trial, a significant location performance was observed. It seems, therefore, that the choice between different locations is a necessary, but obviously not a sufficient, prerequisite for the ability to locate targets within the perimetrically blind hemifield. Forcing the patient to frequently choose between different locations within his perimetrically blind hemifield may help him to use his spared localization ability even though he can never see the target.