Search Results

This chapter reviews the features of the visual and oculomotor systems that are particularly important for understanding active vision. First, the chapter describes the inhomogeneity of the visual projections and the consequences of the resulting inhomogeneity on visual abilities. Human vision has a high resolution fovea at the centre and visual ability falling off quickly into peripheral vision. Second, the evidence for multiple types of parallel processing within the visual and oculomotor system is reviewed. Third, the basic characteristics of the oculomotor system are described and different types of eye movement are identified, followed by a more detailed description of saccadic eye movements: the fast ballistic eye movements that move the fovea to point at regions of interest.

This final chapter is concerned with how information is combined across saccades. The first section focuses on evidence for the detection of displacement during saccades or whether trans-saccadic fusion occurs. It is concluded that there is little evidence that detailed visual information is integrated at all across saccades. The chapter then provides an alternative account for the problem of integration across saccades and visual stability. In the conclusion to this chapter, the active vision cycle is described along with a description of the limitations of the active vision approach and points to some possible future directions.

Attention is crucial in vision, to avoid having the system overloaded by too much information. Some objects pop out, when distinguished from other objects around, and some do not. While objects on the fovea are attended most carefully, objects in the periphery are also attended to determine where to make the next saccade. The task being performed and the instructions given to the subject affect this. Attention modulates the response of cells in all parts of the visual and eye movement systems, overtly when an eye movement is made, and covertly when an eye movement is not made. Lesions of sensory areas lead to sensory losses; lesions of eye movement areas lead to failure to make eye movements; and lesions of areas in between, such as parietal cortex, lead to neglect, with unconscious but not conscious awareness in the area represented by the lesion.

There are eye movements to place objects of interest onto the fovea (saccades and vergence), and to keep it there (smooth pursuit, fixation, optokinetic, and vestibular). Signals converge onto the parietal cortex to determine what is of interest, and onto the frontal eye fields and superior colliculus to determine the size and direction of the movement to be made. Both sensory and motor signals are represented in all these areas, and saccades, smooth pursuit, and vergence signals are found in all of them. The basal ganglia act to release the system from fixation, and the cerebellum controls adaptation of the size of saccades. Different locations at these higher areas give the size and direction of the movement. This is translated into horizontal and vertical components in the brainstem, and strength of signal to give the size of the movement.

Recent experiments have proven that the vestibulo-ocular reflex (VOR) does not simply sum with vestibular eye movement commands during combined eye-head gaze saccades. Instead, if VOR gain is measured during combined eye-head gaze shifts, the gain is found to be a function of amplitude, decreasing from near unity during small-amplitude saccades to near zero at large amplitudes. In spite of this observation, large-amplitude saccades remain precise even if the head is perturbed during the movement. Since the VOR is not functional during these perturbed saccades, the eye trajectory does not change in response to this perturbation. Instead, the maintenance of saccadic accuracy is made by the changes in the movement duration.

Rapid, accurate and well coordinated saccades of the two eyes are a prerequisite for clear vision particularly during reading. The Iris group has studied fixation stability and saccadic parameters in good and bad readers. Specific impairments in these are found in dyslexic children, namely fixation instability due to disconjugate drifts, reduced binocular coordination of saccades, and increased latencies. These eye movement features normally develop slowly, only achieving adult quality by the age of 12. The saccadic and vergence control systems which are separate in the brainstem need to interact precisely with each other. Thus defects in the vergence system may influence the quality of the binocular coordination of saccades. In dyslexics difficulty with achieving accurate binocular coordination can be seen as a sort of micro-dyspraxia, related to cerebellar and magnocellular misfunction. The next step is to develop appropriate training techniques and evaluate their benefits for reading.

The complexity of the cortical interactions required to read are astounding. Not least are the interactions that occur within the visual cortex and beyond in the first 300ms or so of seeing a word. It has been speculated that the dorsal visual pathway plays a vital role in this early visual network by providing a preattentive spatial code for the features of letters and words and providing a spatial navigation mechanism for guiding saccades across the line of text. While a large literature has accumulated to implicate the dorsal pathway in reading, and deficits in the dorsal pathway have been demonstrated to be associated with reading failure, the next challenge is to explore the efficacy of visual training as a technique in reading remediation.

Saccades are fast conjugate eye movements that move both eyes quickly in the same direction, so that the image of an object of interest is brought on the foveae. Saccades can be made not only toward visual targets, but also toward auditory and tactile stimuli, as well as toward memorized targets. Saccades can be generated reflexively, and they are responsible for resetting the eyes back to the mid-orbital position during vestibulo-ocular or optokinetic stimulation. Saccades need to be fast to get the eyes on the target as soon as possible. They also need to be fast because our eyes act like cameras with slow shutters—vision is so blurred during saccades that the eyes have to move quickly to minimize the time during which no clear image is captured on the foveae. Indeed, saccades are the fastest type of eye movements, and they are among the fastest movements that the body can make. Saccade speed is not under voluntary control but depends on the size of the movement, with larger saccades attaining higher peak velocities. It has been estimated that we make more than 100,000 saccades per day. The burst neuron circuits in the brainstem provide the necessary motor signals to the extraocular muscles for the generation of saccades. There is a division of labor between the pons and the midbrain, with the pons primarily involved in generating horizontal saccades and the midbrain primarily involved in generating vertical and torsional saccades. However, because eye movements are a component of cognitive and purposeful behaviors in higher mammals, the process of deciding when and where to make a saccade occurs in the cerebral cortex. Not only does the cortex exert control over saccades through direct projections to the burst neuron circuits, it also acts via the superior colliculus. The superior colliculus is located in the midbrain and consists of seven layers: three superficial layers and four intermediate/ deep layers. The three superficial layers receive direct inputs from both the retina and striate cortex, and they contain a retinotopic representation of the contralateral visual hemifield.

Smooth pursuit consists of conjugate eye movements that allow both eyes to smoothly track a slowly moving object so that its image is kept on the foveae. For example, smooth pursuit eye movements are used when you track a child on a swing. Only animals with foveae make smooth pursuit eye movements. Rabbits, for instance, do not have foveae, and their eyes cannot track a small moving target. However, if a rabbit is placed inside a rotating drum painted on the inside with stripes so that the rabbit sees the entire visual field rotating en bloc, it will track the stripes optokinetically. Humans have both smooth pursuit and optokinetic eye movements, but pursuit predominates. When you track a small, moving object against a detailed stationary background, such as a bird flying against a background of leaves, the optokinetic system will try to hold your gaze on the stationary background, but it is overridden by pursuit. Pursuit works well at speeds up to about 70°/sec, but top athletes may generate pursuit as fast as 130°/sec. Pursuit responds slowly to unexpected changes—it takes about 100 msec to track a target that starts to move suddenly, and this is why we need the faster acting vestibulo-ocular reflex (VOR) to stabilize our eyes when our heads move. However, pursuit can detect patterns of motion and respond to predictable target motion in much less than 100 msec. Pursuit cannot be generated voluntarily without a suitable target. If you try to pursue an imaginary target moving across your visual field, you will make a series of saccades instead of pursuit. However, the target that evokes pursuit does not have to be visual; it may be auditory (e.g., a moving, beeping pager), proprioceptive (e.g., tracking your outstretched finger in the dark), tactile (e.g., an ant crawling on your arm in the dark), or cognitive (e.g., tracking a stroboscopic motion in which a series of light flashes in sequence, even though no actual motion occurs.

This chapter reviews some behavioural aspects of saccades recorded in alert, head-fixed, trained cats. The first part deals with characteristics of oblique saccades and their trajectory. The stretching of the smaller component of oblique saccades is mainly achieved by a prolongation of the deceleratory phase. The decelerator phase of centrifugal saccades can be controlled much more accurately in direction than the acceleratory period, and the generation of straight trajectories is more frequent when the eye returns to the primary eye position. Direction of the movement is thus a major factor that accounts for the trajectory of cats' saccades. Another factor is the initial position of the eye. These factors modify the dynamics of the components of oblique saccade and the co-ordination between them. The second part of the chapter is devoted to a behavioral analysis of slow corrective movements that follow the visually guided eye saccades. In the cat, the end of the saccadic eye displacement, or the intersaccadic period between a principal saccade and the following corrective saccade, often shows some long-lasting slow movements. Most of the saccadic eye shifts of cats end abruptly, but some of them are followed by a large displacement different from the preceding high-velocity movement. These drifts are target directed, and have been observed in trained and untrained cats. Some of these movements are obviously driven by a system reducing the residual error at the end of the saccade, referred to as “slow correcting movements.”

In this chapter a series of experiments are described where large-amplitude gaze shifts were recorded in the head-free condition, and eye saccades in the head-fixed condition. The amplitude of head-fixed saccades was chosen in order to correspond to the amplitude of the eye movements in the orbit in head-free condition. By comparing head-free and head-fixed eye saccades of equivalent size, vestibulo-ocular reflex (VOR) gain changes during large gaze shifts involving both the eye and the head can be indirectly measured. Quantification of the VOR gain variations during large gaze shifts has been done by a systematic comparison of saccade dynamics in head-free and head-fixed human subjects, under certain conditions, such as identical eye saccade amplitude and starting position. This analysis is only possible when the head velocity attains a sufficient level, and is thus reserved for the terminal part of the gaze shift. In addition, the validity of the estimation of the VOR gain and of the hypotheses proposed to explain the observations has been tested by model simulations.

In this chapter, the authors addressed the question of the degree to which gaze saccades executed under natural conditions are modulated by concomitant head movements by creating a “quasi-natural” situation. Head-fixed saccades were compared to head-free saccades. The first experiment in this study cannot prove or disprove the validity of Bizzi's classical summation hypothesis for human gaze saccades confined to the oculomotor range. It rather demonstrates that the presence of a classical VOR is irrelevant for the velocity of such gaze saccades since the concurrent head movement starts too late and is too slow to impart to the saccade a significant gain in velocity in the event of a disconnected VOR. The second experiment confirms that the factor of greater importance for the dynamics of head-free saccades is the orbital eccentricity and direction of the EiH contribution to saccades.

The objective of this chapter is to present the effects of voluntary control and stimulus conditions on three types of oculomotor behavior that were previously regarded as different subsystems. Examples are given in this chapter to illustrate that performance is determined by stimulus conditions and voluntary processes. The primary contribution of head movements to saccadic gaze shifts is a reduction of the duration of the gaze shift that results in different characteristics of the velocity profile. These experiments aim to study interactions between saccades and vergence or pursuit and vergence, and to give more insight into the degree to which the performance of the head motor systems and oculomotor behavior is determined by processing in modular subsystems.

In this chapter, the modification of saccades executed during active head or passive body rotation is investigated. An experimental setup was designed, which confined saccades to populations with amplitudes less than 30°. It was found out that due to the nonlinear relationship of amplitude and peak velocity of saccades and a constant contribution of the head rotation velocity, the gaze analysis will show different velocity profiles because of the different saccades types. There are also two proposed mechanisms to explain the duration decrease of about 30% for the total amplitude range studies. The first one is that the saccadic burst rate is changed by superimposing the VOR activity on it. The second is that the saccadic burst modifies an ongoing eye movement.

This chapter focuses on the discussion of the characteristics of neck muscle command signals and the descriptions of the aspects of head movement trajectories, eye saccades, and the coordination in visually triggered oblique gaze shifts by trained cats. The methods in the eye and head movement recording and the training procedure and experimental paradigm are presented. The results of the activity are explained in detail in this chapter, including (1) the timing of the neck muscle discharge as a function of the orientation of the head movement and (2) the temporal aspects of eye-head component coordination.

It has been shown from recent studies that concurrent head velocity affects gaze shift kinematics due to an inhibition of the VOR during the saccade in man. The goal of this study is to quantify the VOR-saccade interactions during small gaze shifts. A procedure of passive head rotation is thus designed to be able to quantify VOR gain associate with gaze shifts of even smaller amplitude. The main findings are summarized in the gaze velocity profiles. It was found that when saccadic response is executed when the head is moving in the same direction, gaze velocity is faster than when the head is fixed. Another finding is that the gaze shift duration is inversely related to gaze shift velocity.

Young infants can orient themselves toward visual targets by means of eye saccades and head movements. In this study, a population of “at-risk” infants is tested for visual preferences to evaluate the significance of a behavioral examination for such a population. A control population of fifty normal infants was also examined. It is concluded that the test of visual preferences discriminates the control group and the at-risk group after five months of age. The results have shown that there are different capabilities in selective visual discrimination for at-risk and control infants. The differences in responses just reflect the differences in later intellectual abilities.

When reading a text, the eyes mainly move forward by going from one word to the next, although in some instances they return to previous regions of the text. This chapter investigates why and when regressive saccades occur, and where they send the eyes. It reveals that regressive saccades are determined by three types of processes including visuomotor, lexical, and higher-level syntactic and semantic processes, although the former two may be predominant. Most regressions are initiated quickly in response to prior oculomotor and processing events and independently of incoming information. The likelihood of later-triggered regressions may, however, also rely on the processing of currently available word information, although such regressions would not all result from process-related inhibition. The length of regressive saccades is not determined randomly; rather, regressions aim at specific word locations, sending the eyes to one of the prior words, or the currently fixated word.

Readers execute regressions to move the eyes to a previously-read segment of text. Short-range regression and similar size forward directed saccades differ in that interword regressions are neither influenced by word length nor launch site. This chapter proposes that these regressions are controlled by relatively precise representations of word location. Longer-range regressions are relatively rare during reading. Recent results indicate that they are spatially inaccurate and that readers generally execute more than one regression to find the location of a relatively distant regression target. This chapter suggests that linguistic knowledge contributes to the guidance of these regressions.