Research report
Fixation errors and timing in sequences of memory-guided saccades

https://doi.org/10.1016/S0166-4328(97)00160-5Get rights and content

Abstract

We analyzed the relation between position and amplitude errors during the performance of sequences of saccades to previously memorized target positions in complete darkness. Although a complete compensation (on the average) for fixation errors was observed, groups of successive saccades could be identified which showed propagation of position errors. These groups are characterized by a long fixation time prior to the first saccade and short fixations prior to the remaining saccades. These findings indicate that sequences of eye movements can be performed in two different modes: (1) extraretinal information about the actual eye position is used to correct fixation errors; (2) pre-programmed groups of saccades with limited length can be performed with fixed amplitudes neglecting the actual eye position. These groups tended to occur predominantly at the end of a sequence.

Introduction

It is widely believed that visual information plays the dominant role in error correction in saccadic eye movements (e.g. [2]). If, however, no visual information is available, what happens? To answer this question, we examined sequences of memory-guided saccades that had to be performed in the dark. To perform sequences of saccadic eye movements in the absence of visible targets, information about the desired movements has to be stored. It is possible that either a sequence of motor commands could be stored during memorization, or the target positions themselves.

There are different ways to regard the execution of eye movement sequences. Execution could be considered an open loop mode in which the single movements are performed with a fixed amplitude, regardless of the actual eye position. Alternatively, it could be viewed as a closed loop mode in which feedback information about the actual eye position is considered when the next component of the sequence has to be executed, thus providing the possibility of correcting errors. Such a tight linkage between amplitude and position control, and open and closed loop modes is only true, if the information that has to be passed to the burst generator(s) to make the eyes move, is indeed the desired displacement. This idea is a key feature of most models of the saccadic system (overview in [2]). Theoretically, an open loop position control could also be possible [6].

Our study addressed the following questions:

  • 1.

    Which information is stored in order to perform sequences of saccades to previously presented target positions: saccadic amplitudes or target positions?

  • 2.

    Is the execution based on pre-established motor programs or can feedback information about what has happened thus far influence the performance?

Since the oculomotor system is an imprecise system, i.e. the actual movement amplitude can differ somewhat from the intended amplitude, fixation errors occur. Certain predictions can be made about these errors. If motor commands (movement amplitudes) are stored, there should be a positive correlation between successive position errors due to error propagation. If target positions are stored and information about the actual eye position is used, the oculomotor system could correct position errors by using this information to generate the next saccadic command. Such a mechanism would lead to a compensatory correlation between position errors and the deviations of the amplitude of the following saccade from the nominal value as defined by the target step.

Some of the literature has focussed on position errors. In a previous study, Bock et al. [3] investigated error propagation during sequences of memory-guided saccades. The subjects repeatedly had to perform a fixed sequence of saccades without visual feedback. Assuming that a homogeneous model, i.e. a model with a single uniform mechanism, can explain the phenomenon, they pooled their data, calculated the correlation between successive position errors, and found a linear regression coefficient of about 0.4. They concluded that about 40% of a position error would be propagated and about 60% would be corrected. They also assumed that an extraretinal signal providing information about the actual eye position is used to correct position errors. Bridgeman et al. [4] conducted experiments on estimating the position of visual targets and pointing to perceived positions. They found a gain of 0.87 for the extraretinal signal. These findings suggest that a better error correction should have been possible. The reason for this discrepancy is, however, unclear.

Other researchers have examined the saccade sequence itself. Zingale et al. [11], for example, discovered that sequences of saccades have systematic characteristics in their time structure. The most striking pattern is a longer pause before the last three saccades of a sequence that separates different groups of saccades. Consequently, the fixation times in sequences of saccades are not homogeneously distributed. This suggests that a homogeneous model is not appropriate for describing the processes underlying the execution of sequences of saccades. These studies prompted us to investigate whether spatial accuracy is related in some way to timing, especially, grouping effects.

Section snippets

Experimental paradigm

Ten healthy subjects (mean age: 28, range: 24–37) were seated in a darkened room and told to memorize sequences of target positions while fixating a central laser spot. The targets were presented on a white, unstructured, half-cylindrical screen (distance from subject: 1.2 m) using a second laser projection system. All targets and the fixation spot lay on an imaginary horizontal line. The positions were randomly chosen from a range of 25° to the left of the fixation point to 25° to the right of

General observations

Twenty-four percent of the total 838 sequences presented to the subjects had to be discarded due to eye movements during the memorization period or because of missing responses (less saccades than targets). Thus, 641 sequences were analyzed. Although most subjects had the impression that they were not able to repeat the longer sequences correctly, the analysis showed that the subjects performed even the sequences consisting of five targets rather well. Since varying the presentation,

Error components

These errors that we term position errors, which can be measured, are not only errors caused by the oculomotor system. They can also have other causes (e.g. perception and memorization errors). Unfortunately, it is not possible to separate these components. Nevertheless, the correlations and regressions calculated here predominantly represent the correction or propagation of oculomotor errors. Since we know that feedback information can play a role in oculomotor control, we expect the

Acknowledgements

We wish to thank our subjects for their patience and Stefan Glasauer, Kevin O'Regan, and Gert Hauske for their helpful comments on the topic. Furthermore, we would like to thank Judy Benson and Michelle Seiche for their help with improving the English. The study was supported by the BMBF (FKZ 01KL9001) and the DFG (GRK 267/1-96).

References (11)

There are more references available in the full text version of this article.

Cited by (24)

  • Error compensation in random vector double step saccades with and without global adaptation

    2016, Vision Research
    Citation Excerpt :

    We then examined error compensation in a slow-paced (high pre-programming) and fast-paced (low pre-programming) DSS task. Ditterich, Eggert, and Straube (1998) suggested that high pre-programming should reduce error compensation in sequences because saccades are then executed based on fixed amplitudes. In this case better error compensation should be observed in our fast condition.

  • Revisiting corrective saccades: Role of visual feedback

    2013, Vision Research
    Citation Excerpt :

    In addition, recent evidence in an adaptation study indicates that visual error signals processing leading to the generation of corrective saccades differs between reactive and voluntary saccades (Panouilleres et al., 2011). Ditterich, Eggert, and Straube (1998) also suggested that sequences of memory-guided saccades can be performed in a preprogramming mode. Similarly, Sharika, Ramakrishnan, and Murthy (2008) demonstrated that motor preparation for the second corrective saccade may proceed in parallel with the preparation of the first erroneous saccade using a modified double-step redirect task.

  • Learning deferred imitation of long spatial sequences

    2011, Behavioural Brain Research
    Citation Excerpt :

    Thus, the absence of strong correlations in the current experiment is probably due to the feature that the vast majority of all direction changes were larger than 90°. This is an important difference to the study of Ditterich et al. [34] who observed larger correlations (up to 0.55) between successive fixation errors. The very small average correlation coefficients reported in the current study (0.37) were as small as the correlation coefficients (0.37) observed by Bock [18].

  • The planning of a sequence of saccades in pro- and antisaccade tasks: Influence of visual integration time and concurrent motor processing

    2008, Brain Research
    Citation Excerpt :

    Indeed, the fact that the within-object is not updated after the first saccade as a function of the eye position indicates that the saccade is preplanned before the first saccade on the basis of the object size (Vergilino and Beauvillain, 2000). The two saccades of the sequence can be seen as an entire unit planned before the first saccade and executed as a whole (Ditterich et al., 1998; Inhoff, 1986; Zingale and Kowler, 1987), and the oculocentric representation should be supported by a non-visual motor memory maintaining the future movement to be executed (Fuster, 1996). For between-object saccades, the motor vector of the saccade seems to be computed after the first saccade relative to the new eye position to aim for a spatial location on the second object.

  • The use of recurrent signals about adaptation for subsequent saccade programming depends on object structure

    2006, Brain Research
    Citation Excerpt :

    Furthermore, the effect of phase was significant for the one-object condition (F(1,5) = 75.99, P < 0.0005) but not for the two-object condition (F(1,5) = 1.14, NS). The analysis of the relationship between first and second saccade end position (Ditterich et al., 1998) is another convincing argument for the presence or lack of second saccade compensation. The updating hypothesis predicts that the second saccade should compensate for the adaptive modification of the first saccade end position.

View all citing articles on Scopus
View full text