In this chapter, we will address the issue of eye movements or dynamic vision. We have seen the mechanisms of the vision of the movement and now we will see how all the information that comes to us from the outside world is processed, from a scenario that changes continuously.
types of eye movements
Surprisingly, when we analyze the movements of the eyes, most of the time we are not fixing or following an object continuously. Actually, what the eyes do is to "jump" from one point to another of the scene, until in the end, we look at something and keep the focus following the movement made by that object or adjusting it to our movement, if is that we are not stopped.
In general, we classify movements in three categories:
- Sacadas
- Slow tracking.
- Fixation.
To which should be added the vergencies as a different movement from the previous ones.
the saccades
The saccades are the fastest eye movements, which redirect the eyes to a new point on the stage that surrounds us.
During the draw, the brain is blind, it does not process visual information, it only does so when we are fixing an object. To achieve this, the visual perception system uses two mechanisms, the first is related to the resolution capacity of the retina, only the taint It has a high resolution power that allows you to send detail information to the brain, so, until an object is fixed, until it is not foveolized, the information sent by the peripheral retina is of a low quality, 10 times lower than only 20º of this one.
Eye movements in moving objects
When an object is fixed, whether we are moving or if we are still but with the head moving, to stabilize the object and the environment that surrounds us, we have two very powerful mechanisms, the Oculo-Vestibular Reflex (VOR) and the Optokinetic reflex (OKR).
With these systems, we can stabilize the environment and the objects and establish a slow and controlled movement of tracking that object, even if it is moving or we are moving or, both at the same time (within limits).
Frequency of eye movements
It is admitted that we make an average of 1 to 4 sacks per second, 4 being the maximum number, for example when we read. A saccade takes 30 ms and the time we maintain fixation is around 300 ms, which means that if we do not see during the saccade, it is equivalent to saying that we are blind for 1.5 hours a day.
The saccades are bilateral and symmetrical, it only changes something if it is an eye movement of vergence.
The saccades have very similar characteristics in all individuals. Its speed is not constant and varies according to the amplength of your journey, faster and longer as the ampaltitude, with an initial ascent phase and a subsequent saturation phase. They are triggered after the arrival of external stimuli such as a movement of something within the visual scene, the appearance of a new object but, in general, they are generated involuntarily and constantly when we do something that requires visual information, but without need to search for any particular object. For example, when we are walking, the eyes move in saccades, scanning the scene in front of us, processing information of minimal quality but sufficient to orient ourselves and not stumble, they are taken somewhat slower than the previous ones, 130 msec.
Operation of eye movements
In most cases, the saccades alternate the position, the direction of the gaze, from one side to the other, under the action of the muscle groups, (if the right medial rectus and left lateral rectus are activated, looking to the left , the next serve will be performed, to the right, activating the right and middle left laterals, movements with vertical directions are less frequent.
In real life there is a scan of the environment with amp30 degrees, of which, 20º correspond to the eyes, saccades and 10º to the movement of the head.
Control of eye movements
The control of eye movements is carried out at three levels:
- Cerebral cortex, responsible for the voluntary movements.
- Structures of the midbrain, especially the superior colliculus, responsible for the saccades.
- Brain stem, responsible for the motor system that directly controls the eye muscles.
The ocular muscles receive innervation from the nuclei of the brain stem, which in turn receive innervation from the premotor nuclei, also located in the stem, located above the former. The premotor nuclei receive innervation from the paramedian pontine reticular formation, which is responsible for saccades (burst units) and for maintaining the position of the eyes during object fixation (tonic units). At this level are also the reticular nuclei of the pontine tegmentum, responsible for fine tracking movements.
On the premotor nuclei also arrive innervations of the vestibular nucleus, responsible for the stabilization of the vision during the movements of the head. In the premotor nuclei is where the innervational patterns responsible for the different eye movements are organized.
visual orientation
At a higher level is the superior colliculus, responsible for visual orientation, not only of the eyes, but also of the head and trunk. It is responsible for saccades and saccade-related head movements. Visual signals arrive in its superficial region and auditory and somatosensory signals arrive in its deeper layers. The colliculus receives innervations from the cortex, from the visual frontal regions (frontal eye fields, FEF), from the lateral intraparietal area (LIP), and from V1. The FEF and LIP regions of each hemisphere would be interconnected through the thalamus.
It seems that the saccades are activated directly from the visual frontal regions or the superior colliculus although the FEF would represent a superior stage of control of the oculomotor system. The FEF would also be responsible for the fine tracking and tracking of objects.
The role played by the lateral intraparietal area is less clear. It seems that this would be more related to motor functions, coordination of movements of hands, head, etc., from visual stimuli.
Eye movements and movements of the head
On the visual frontal regions would arrive stimuli coming from the dorsolateral prefrontal cortex, a region associated with the elaboration of plans of action and inter-relation with the world that surrounds us.
When the head is moved, there is a compensatory movement of the eyes, opposite to the direction of the head, to stabilize the images. This phenomenon is achieved through the VOR and OKR systems.
VOR system
The VOR system, through the semicircular channels of the ear, detects the speed of the head and sends a compensating signal to the nuclei that control the extra-ocular muscles, via oculomotor nuclei.
The compensation reflex is practically the same as that of the head, with a gain equal to 1, very fast, of approximately 15 msec. If the rotation stimulus continues, as when we are in front of a rotating drum, when the amplitudinal movement of the eye exceeds the 50º, a saccade is produced where the eyes return to the starting position, initiating a new tracking movement, similar to a nistagmus.
OKR system
When the rotation speed of the head is not so high, the OKR is activated instead of the VOR.
It works according to the speed at which the image slides on the retina, generating a compensatory eye movement in the opposite direction, through the stimulation of the extraocular muscles.
Unlike VOR, eye movement is adjusted by a feedback mechanism that tries to keep the image aligned in the retina, in the macula. The VOR has no readjustment and the OKR tries to stabilize the scene so that the fine tracking control of the objects starts.
Generalizing, we can say that the OKR works to stabilize the scene, through areas ampLees of this, while fine movements, require a small target. Fine movements can operate up to speeds not exceeding 15º sec, and if it exceeds this speed, a saccade is fired that attempts to correct the position and restart the fine tracking movement.
When 100º sec are exceeded, a constant saccadic movement begins. When there is constant fine tracking, a gain of 0.9 is produced, which means that if an object moves at 10º sec, the eyes follow it at 9º sec, with a displacement error on the retina of 1º sec. This explains the impossibility of fixing an object in the exact place that corresponds to it and explains conflictive situations such as the error of the referees when whistling an offside.
the vergences
When it comes to adjusting the vision at different distances, it is when the vergences are activated.
The vergences are slow and fine movements, with a latency of 120 msec approx. They are activated after a retinal (macular) disparity and can be accompanied by accommodation stimulation, refocusing of the image.
Eye movements in everyday life
One of the most important issues in visual perception refers to the scanning and fixation patterns of a scene, and these patterns are similar in most individuals.
eye fixation points
To study the first point, patterns of scanning and fixation, one of the most used examples is the reading and inspection of frames.
The first studies showed that there was a great disparity between the fixation points of both eyes and their path over the scene. By means of the new eye tracker systems (recording of eye movements), the opposite was evidenced, despite certain disparities between the two eyes, in most cases a great parallel is observed, making Hering's law good.
Regarding the pattern of inspection of the scene, a certain concordance was also found, there was a fixation on the most relevant points of the scene, faces, figures, etc. If the scene is analyzed with different observers, the differences are greater in the inspection pattern but again a certain similarity is observed, fixing faces, figures and other remarkable elements of the scene.
Another interesting aspect is that the eyes make several saccades per second, that is to say that the observer remains for a short time fixing a certain point of the scene, jumps and returns to that point several times.
Eye movements in a scene
There are two schools of thought that try to explain the pattern of inspection of the scene.
- Bottom-up (bottom-up): In which the eyes move according to outgoing stimuli from the scene that reach the retina, without almost intervening cognitive aspects.
- top down: Where the movements are not so much the result of the properties of the scene but of mental phenomena, as the purpose of looking for a certain object or carrying out a certain action, clearly of a cognitive nature.
Currently, the tendency is to think of a mixed mechanism, where outgoing stimuli (bottoom up) and cognitive aspects (top-down) are combined.
In this line of study, we sought to know what makes a stimulus "outgoing". The first thing is to assume that something will be salient or important, when it demands our attention. We must differentiate between what catches our attention and is identified in some part of our peripheral retina, but does not trigger a saccade to fix it with the macula and the stimulus that does cause foveolization with the saccade.
Which stimuli cause eye movements and which do not
The first is the most frequent in daily life and follows a different mechanism from the second.
In the first case we would speak of a preattentive process and in the second fully attentive. The difference between both processes becomes clear when we experiment with the detection of basic stimuli in a scene with distractors. In most cases, individuals quickly locate the differential element, in less than 10 msec, less time than necessary to perform a fovealization movement, it is an automatic “pop-out” mechanism as described by Treisman in 1988.
If the number of distractors increases but the reaction time, of detection, remains constant, it is a pure, peripheral, bottom-up process and, if the number of distractors implies an increase in reaction time, attention expenditure, means that Cognitive mechanisms come into play and we are facing a top-down process.
It is currently assumed that in real life we function with a pop-out mechanism (in parallel) and only in certain circumstances do we go on to fovealize something through a post-attention sequential scrutiny.
How we see the world
The preattention mechanism works in parallel, that is, different low-level stimuli from the scene arrive simultaneously at the retina, including color, brightness, orientation, etc. Each of these stimuli produces a map of characteristics and each map is combined with the others in an additive way, until forming a single outgoing map with several hot spots that can attract attention and generate a saccadic movement to go to foveolize one of them. those points (Itti and Koch, 2000). One of the problems posed by this theory is that if there are clearly "hot" points that attract attention, we would always go to those points and not appreciate the rest of the scene. To solve this problem, it is speculated that there is a transient hot spot inhibition mechanism, so that the eyes can move freely over the scene capturing low-level, non-salient stimuli.
It does not seem that it is totally like that, but it is true that when we analyze a scene, we usually look at certain points, more or less recursively (hot spots), although the order of eye movements is different each time or when comparing different individuals (Tatler and Vincent, 2008, 2009).
The location of objects generates eye movements
Another important aspect in the detection of outgoing stimuli is the location of these in the scene.
We are aware that when they are located in central regions, the detection level is higher than when they are in the periphery, therefore, the idea that there is a system of certain inhibition or filtering of the most powerful outgoing stimuli, to allow the eyes to take inspection outlets throughout the entire scene takes more body. These movements also follow certain "norms", usually from amplitud not very large and more frequently on the horizontal axis (less the vertical and less the oblique).
Opposite the bottom-up conception of salient stimuli, the top-down hypothesis is situated, which is confirmed by studies that show that the movement of the eyes changes when going from simply observing one thing to when we ask the observer to look for something determined at the scene. At this time, the eye tracker shows a completely different pattern of movements, which shows that these movements are determined by a top-down mental process.
Eye movements, search or observe
The most generalized idea is that in situations where there is no strong influence of the mind (search for something, etc.), the bottoom-up mechanism predominates, whereas when a mental process starts, the top-down mechanisms predominate.
One of the points that still raises doubts is the fact of how we recognize a certain object, when we capture some of the salient aspects in the peripheral retina or when the saccade that foveolizes it is produced.
Eye movements in day to day
It seems that in situations where there is no clear mental guidance, the vision would follow a mechanism of master maps formed by the low level maps resulting from the outgoing stimuli that start from the peripheral retina. A map of average quality is generated, enough for us to have a vision that guides us in the scene, minimizing the eye movements to foveolize (Vincent, 2007).
Fovealization appears in concrete tasks guided by an idea or mental process, generally search for something and, conditioned by previous cognitive experiences (Torralva and Oliva 2003 and 2006). If we look for a jug of water in a photo of a kitchen, we will first inspect the countertop before the floor, it is more likely that it is in that area of the kitchen in the photo.
I want to see something
The role of thought, the mind, in guiding eye movements, seems to be increasingly relevant. There are experiences made with the eye tracker in a scene where the outgoing stimuli have been filtered, such as veiling of faces, etc. and it is observed that there is no reduction in the fixation of hot spots, such as the eyes, nose and mouth, although it is not appreciate correctly. These results would be in line with the fact that most of these salient points, such as the eyes on a face, have a certain cognitive basis.
With regard to faces, in humans and only in humans, they are detected according to pop-out patterns, <10 msec, even in scenes with a ampThe group of distractors (Hershler, 2005).
Ocular fixation time in scenes
Another point of interest is the ocular movement with respect to the time we look or inspect a scene.
It is not clear that in the initial phases of recognition the bottom-up mechanisms, based on outgoing stimuli, "send" and that as time passes they lose effect. What does seem more accepted is the time in which we maintain fixation at some point in the scene. In the initial phases it would be around 250 msec and later it will lengthen, until it reaches 350 or 400 msec. It is as if we first carried out a quick general inspection and then we were analyzing in more detail some of the points scrutinized.
How we see the world around us
If we ask any individual how we see, he will surely respond that we see continuously and completely, that is, like when we watch a movie. The reality is very different.
The vision process is subjective, incomplete and discontinuous. The image that we process, in most of the time, does not come from a macular fixation but from the peripheral retina or, in the best of cases, paramacular, that is, a low-level image, only when we foveolize is when we really we perceive the details of the outside world and that occupies a minimum time in the global computation of the period of vision throughout the day. On the other hand, in a third of the time, during saccadic movements, our vision is blind and, finally, the image of the outside world that we form in our psyche is highly influenced by cognitive and emotional aspects, which implies a concept of subjective vision, not objective, where what an individual "sees" or how he sees it, will be different from how another individual sees it.
discontinuous vision
That the vision is discontinuous we see it in the fact that oneself is not conscious or cannot perceive the saccades of the eyes. If we are placed in front of the mirror we are not able to see this type of eye movement, however it is something evident when we check it in the record made by an eye-tracker. The continuity of vision is a mere illusion.
Incomplete vision
The incomplete vision is something that until the nineties was not admitted, until then it was believed that the vision involved an analysis of the scene according to the method "point by point", detailed analysis of each space and that its addition allowed to reconstruct the scene as if it were a painting, is what is known as "picture in the mind".
Studies by various authors, among them O'Regan (2000), showed that when we observe a scene, a painting or a photo, if during the blind phase of a shoot, we change something of that scene, even if it is quite significant, in the majority of cases we will not appreciate it, we will continue to perceive the scene as in its beginning.
One possible inference from this fact is that we do not see through an internal representation of the world around us. For O'Regan, what happens is that we form an idea of the environment that surrounds us, at a low level, just to orient ourselves and, only in certain circumstances, do we detect the details of a part of that environment, when we are interested in something in concrete and foveolated it.
The vision and the degree of attention
The degree of attention that we put in each moment is a fundamental factor when it comes to explaining what we see. It is very different the details that we will remember if we walk along a path, that if we walk on an unknown path trying to remember where we are going if we know that we should repeat it another day as hiking guides.
For Rensink, 2000, as shown in the original figure of his work, rather than a picture-like image (in two dimensions), what we do is to build a virtual representation in 3D with very variable levels of detail, only the most significant ones for each purpose (orientation, artistic, look for mushrooms, etc).
Eye movements related to language
Rensink's virtual representation would be directly linked to the idea of integrated representation of Altman and Kamide, 2007, who point out the relationship between visual information and language. If we leave an observer in one room and then move him to another where there is nothing, if in the first there was a window, when we talk about the window or a window in general, it is found that in most cases he makes a movement of the eyes towards the position where the window was in the first room, as if we were looking for it within a virtual construction in our mind, where space and vision combine.
This idea of integrating language and vision is situated more in the dynamic conception of visual perception, moving away from the static pictorial conception of yesteryear. Spoken information and spatial representation generate a much more effective virtual environment.
Imagine spaces and eyes
The virtual space can be of two types:
Egocentric: We represent what we would see before us, like when we look for a specific object in a room, we look at objects smaller than our body.
allocentric: The virtual space is bigger, the whole room or the apartment where we are, in general it is a representation bigger than us (Hayhoe, 2008)
Another important aspect is how we retain or how much we retain, the information we are using from the environment at a given time.
When we do something that requires visual information, we have said that we tend to build a virtual representation of the environment and what we have in front of us, the object we manipulate, that virtual scenario is not formed with all the details of reality, only those aspects that are useful for us to perform that particular task.
Visual retention time
The problem that arises now is not only what elements enter the mental representation, but also how they appear and disappear, since we have said that the virtual representation is dynamic, not static pictorial. Tatler, 2001, found that if an individual who is performing a certain type of task, such as serving coffee, is told in the middle of the task to turn on the lights (press the switch on the wall ), and at the end of this action he is asked to describe what he remembers, he will give a lot of details about the wall but very few about the initial task of being a café, which only a few seconds ago occupied his main attention.
The conclusion reached is that we only use the information from the environment, the information that is minimally necessary to carry out the tasks that we are carrying out at a precise moment and that this information is changing rapidly. What we use is instantaneous, only while it is useful to us, so when we change to something else, that data is removed from the working memory, it is as if it is not interesting to save it, it is no longer useful in the virtual representation and keeping it can interfere with new incoming points and lead to error, in addition to the energy expense of maintaining more data than necessary.
Visual movements in planned tasks
When we advance in the previous idea of virtual representation, as something dynamic, it reaches its peak when it comes to tasks with an important motor activity, such as playing tennis.
In every motor action, two levels can be differentiated:
- Specification of the task.
- Execution of the task.
Scheduled tasks
The specification of the task assumes everything that is related to the planning of a task, the list of actions necessary to achieve the proposed objective.
It is necessary to differentiate between routine tasks and those that are not.
In rural areas of India, families in charge of a blind minor frequently isolate and deprive him/her of the care and attention they provide to their other children; such situation becomes even more severe among lower-caste families, orphans and if the blind child is a girl. daily works no detailed supervision is needed, we usually do it semiautomatically, and a competitive process of schemes that guide our activity in the development of a task is carried out quickly and with little energy expenditure.
In rural areas of India, families in charge of a blind minor frequently isolate and deprive him/her of the care and attention they provide to their other children; such situation becomes even more severe among lower-caste families, orphans and if the blind child is a girl. non-routine tasks the mechanisms of sequential schemes take place, which require more time and are typical of non-routine tasks. In the latter case, the tasks are marked by the supervisory attention system, which assumes the awareness of the steps to perform in a task. This system is located in the dorsal prefrontal cortex, while task planning is performed in the premotor cortex and the basal ganglia.
