Figure 12. Scatter plot of the binocular versus the monocular instability obtained from overlap prosaccade trials. The left panels depict the data from subjects between 7 and 17 years of age (N= 129), the right panel shows the data from older subjects 22 to 45 years of age (N=97). No correlations can be seen.

3.Development of Saccade Control

In the last section we have considered the stability of fixation as an important condition for perfect vision. Earlier, we have mentioned, that saccades are necessary for vision. This might sound as a contradiction. The real requirement is, that one should be able to generate sequences of saccades and fixations without an intrusion of unwanted saccades and without loosing the convergence of the two eyes when they are in register for a given distance. Therefore, both components of gaze control should function normally. This section will show that saccade control has to be subdivided into subfunctions described by different variables. We have to find out first, what these subfunctions are an d how they can be assessed.

3.1. The Optomotor Cycle and the Components of Saccade Control

The control of saccades has been investigated for about 40 years and still we do not understand the system completely. Specialization of visual scientists and oculomotorists has prevented that the two fields so closely related have b een investigated by corresponding combined research projects for a long time. The oculomotor research groups were interested in the eye movement as a move-

ment. Their interest begins when the eyes begin to move and it stops when the eyes stop to move. The visual groups on the other hand, were interested in time periods when the eyes do not move. They required their subjects to fixate a small spot, while being tested for their visual functions. Only when the interest was concentrated on the time just before the movements, there was a chance to learn more about the coordination of saccades and visual processes.

The time before a saccade is easily defined by the reaction time: one asks a subject to maintain fixation at one spot of light straight ahead and to make a fast eye movement to an other spot of light, as soon as it appeared. Under these conditions the reaction time is in the order of 200 ms. This is a value, which one can find in a student handbook.

However, there were several problems. The first was: why is this time so long? While this was a question all from the beginning there was no answer until 1983/84, when the express saccade was discovered in monkeys [Fischer and Boch, 1983] and in human observers [Fischer and Ramsperger, 1984]. The express saccades is the reflex movement to a suddenly presented light stimu lus after an extremely short reaction time (70-80 ms in monkeys and 100-120 ms in human observers). The reflex needs an intact superior colliculus [Schille r et al. 1987].

The Fig. 17 shows in its lower part a distribution of saccadic reaction times. It exhibits 3 modes: one at about 100 ms, the next at about 150 ms, and the third at about 200 ms. It was evident from these observation, that there was not only one reaction time with a (large and unexplained) scatter. Rather the reaction time spectrum indicated that there must be at least 3 different presaccadic processes that determine the beginning of a saccade, each taking its own time in a serial way. Depending on how many of the presaccadic processes are completed already before the occurrence of the target stimulus, the reaction time can take one out of three values, each with a certain amount of scatter [Fischer et al. 1995].

It became clear that the shortest reaction time was 100 ms (not 200 ms) and this was much easier to explain by the time the nerve impulses needed to be generated in the retina (20 ms), to travel to the cortex (10 ms), to the centres in the brain stem (10 ms) and finally to the eye muscles (15 ms). Another 5 ms elapse before the eye begins to move. A much shorter time remains, that was attributed to a central computation time to find the correct size of the saccade to be programmed. One has to know at this point, that saccades are pre-programmed movements: during the last 80 ms before a saccade actually starts, one cannot

The Fig. 13 shows a scheme, which summarizes and takes into account the different findings: the stop-function by fixation alternates with the reflex, the go-function. These two together built up a stop-and-go traffic of fixation s and saccades. What remains open, was the question of how it was possible to interrupt this automatic cycling. The answer came from observations of the frontal lobe functions: patients who lost parts of their frontal lobe at one side, were unable to suppress the reflex in a simple tasks, called the antisaccade task [Guitto n et al. 1985]. This task requires the subject to make a saccade to one side, when the stimulus is presented at the opposite side. The task became very popular during recent years, but it was used already many years ago [Hallet, 1978].

3.2.Methods and Definition of Variables

The fundamental aspects of saccade control as described by the optomotor cycle have been discovered by using two fundamental tasks, which are surprisingly similar but give insight into different aspects of the optomotor cycle. They have been used to quantitatively measure the state of the system of saccade control. We describe these methods and define the variables first. Then we will s ee some of the results obtained by the these methods.

The two tasks are called the overlap prosaccade task and the gap antisaccade task. The words pro and anti in their names refer to the instructions that the subject is given. The words overlap and gap describe the timing of the presentation of the fixation point. The Fig. 14 shows the sequence of frames for gap an d for overlap conditions.

In both tasks a small light stimulus is shown, which the subjects is asked to fixate. This stimulus is called the fixation point.

In overlap trials a new stimulus is added left or right t of the fixation point. The subjects is asked to make a saccade to this new stimulus, the target stimulus, as soon as it appears. Both, the fixation point and the target are vis ible throughout the rest of the trial: they overlap in time. This overlap condition and the task to look towards ('pro') the stimulus explain the complete name of the task: overlap prosaccade task.

The gap condition differs from the overlap condition in only one aspect: the fixation point is extinguished 200 ms before the target stimulus is presented . The time span from extinguishing the fixation point to the onset of the new targe t stimulus is called gap. In addition to this physical difference the instruction for the subject is also changed: the subject is required to make a saccade in the

direction opposite ('anti') of the stimulus: when the stimulus appears left, the subject shall look to the right and vice versa. Therefore the complete name of this task is: gap antisaccade task.

The prosaccade task with overlap condition allows to find too slow or too fast reaction times and to measure their scatter. The presence of a fixation p oint should prevent the occurrence of too many reflexes, the appearance of a new stimulus should allow a timely generation of a saccadic eye movement.

The antisaccade task with gap condition challenges the fixation system to maintain fixation and the ability to generate a saccade against the direction of the reflex.

+

Gap

Figure 14. The figure shows the sequence of frames for overlap and f or gap conditions. The horizontal arrows indicate, in which direction the saccade should be made: to the stimulus in the prosaccade task, in the direction opposite to the stimulus, in the antisaccade task.

Now we can define variables to describe the state of the saccade control system. First of all, one has to keep in mind, that these variables may be different for left versus right stimulation. Because left/right directed saccades are generated by the right/left hemisphere, side differences should not be much of a surprise. However, for the general definition of the variables to be us ed in the diagnosis, the side differences do not need to be considered at this point. The Fig. 15illustrates the definition of the variables described below. Time run s from left to right. The stimulus is indicated by the thick black line. Because its presentation is identical in both conditions, it is drawn only once in the middle. The fixation point is shown by the thin black line. In the case of an overlap trial, the fixation remains visible, in the case of a gap trial the fixation point is extinguished 200 ms before. In addition the figure shows schematically trace s

of eye movements, which help to understand the definition of the variables. Th e upper case shows a trace from an overlap trial. Usually one saccade is made and it contributes its reaction time, SRT. Below one sees two examples of traces. One shows a correct antisaccade, which contributes its reaction time, ANTISRT. The other trace depicts a trial with a direction error that was corrected a little later. It contributes the reaction time of the error, Pro-SRT and the correction time, CRT (in case the error was corrected). While these variables can be taken from every single trail, some other variables are determined by the analysis of the complete set of 200 trials: the percentage of express saccades from all overlap trials, the percentage of errors from all gap trials and the percentage of corrections among the errors.

Stimulus and Eye Movement Events

Figure 15. The schematic drawing of eye movement traces illustrates the defin i- tion of the different variables describing the performance of the prosaccade task with overlap conditions and the antisaccade task with gap conditions.

List of variables:

From the overlap prosaccade task the following mean values and the scatter are used:

•SRT: saccadic reaction time in ms from the onset of the target to the beginning of the saccade

•% expr: the percentage of express saccades, i.e. reaction times between

80 and 130 ms.

From the gap antisaccade task:

•A-SRT: the reaction time of the correct antisaccades

•Pro-SRT: the reaction time of the errors

•CRT: the correction time

•%err: the percentage of errors

•%corr: the percentage of corrections among the errors

Note that the percentage of trials, in which the subject missed to reach the opposite side within the time limit in the trial (700 ms from stimulus presentation) can be calculated as

pmis = perr · (100 − pcorr)/100

The latter variable combines errors rate and correction rate.

3.3. Prosaccades and Reflexes

The optomotor system has a number of reflexes for automatic reactions to different physical stimulations. The best known reflex is the vestibular-o cular reflex, which compensates head or body movements to stabilize the direction of gaze on a fixated object: the eyes move smoothly in the direction opposite to the head movement in order to keep the currently fixated object in the fovea. Similarly, is it possible to stabilize the image of a moving object by the optokinetic reflex. Both reflexes have little or nothing to do with reading.

The saccadic reflex is a reaction of the eyes to a suddenly appearing light stimulus. It was discovered only in 1983/84 by analysing the reaction times of the saccades in a situation, where the fixation point was extinguished shortly (200 ms gap) before a new target stimulus was presented. It was known at that time that under these gap conditions the reaction times were considerably shorter as compared to those obtained under overlap conditions [Saslow, 1967]. When the gap experiment of Saslow was repeated years later, it became evident that among the well know reactions around 150 ms after target onset there was a

separate group of extremely short reactions at about 100 ms, the express saccade [Fischer and Ramsperger, 1984].

The Fig. 16 shows the distribution of reaction times from a single subject. One clearly sees two peaks. The first peak consists of express sacca des, the second represents the fast regular saccades.

Figure 16. The figure shows the distributions of reaction times from a single subject. One clearly sees two peaks. The first represents the express saccades, the second the fast regular saccades.

The Fig. 17 shows the difference in the distributions of reaction times when gap and overlap trials were used. The separate peaks in the distributions indicate, that saccades can be generated at distinctly different reaction times depending on the preparatory processes between target onset and the beginning of the saccade. In the gap condition there is time during the gap to complete one or even two pre-saccadic processes. Therefore the chances of generation of express saccades is high. In the overlap condition it is the target stimulus, which triggers the preparatory processes and therefore the chances of express saccade are low.

If one leaves the fixation point visible throughout the trial (overlap condition) the reaction times are considerably longer, even longer as compared with the gap=0 condition (not shown here).

The consistent shortening of reaction time by introducing a temporal gap between fixation point offset and target onset was surprising, becau se the role of fixation and of the fixation point as a visual stimulus was unknown. The effec t is called the gap-effect and has been investigated in numerous studies of different

Figure 17. The figure shows the difference in the distributions of reaction times when gap and overlap trials were used. Note the separate peaks in the distributions.

research groups all around the world since 1984. The effect of the gap on the reaction time is strongest if the gap lasts approximately 200 Milliseconds. An overview and a list of publications can be found in an overview article [Fischer and Weber, 1993].

Today it is clear, that the main reason for the increase in reaction time under overlap conditions is due to an inhibitory actions of a separate subsystem in the control of eye movements, the fixation system. It is activated by a fovea l stimulus, which is being used as a fixation point and it inhibits the subsystem which generates saccades. If this stimulus is removed early enough, the inhibition is removed by the time the target occurs and a saccade can be generated immediately, i.e. after the shortest possible reaction time.

Note, that the effect of the gap is not a general reduction of reaction times,