The reliability of eye-head coordination
First experiment: The effect on eye-head coordination strategy of adding a dynamic peripheral visual scene
In this first experiment, eye-head coordination measurements were made within a virtual environment and compared to those obtained, for the same subjects, when using the VPS system. The choice of a virtual reality simulator was due to the need for rigorous control of peripheral vision stimulation conditions. The results of this study have already been published in this magazine (Faubert et al., 2007) and will therefore be only briefly referred to here. This first experiment is nevertheless an opportunity to present the methods and equipment used in all the other studies described below, particularly the virtual reality simulator.
Ten subjects were involved in this study. They were instructed to look, as naturally as possible, moving their eyes and their head, at points of light presented at 40° eccentricity randomly to the right and left of the central target, which appeared systematically prior to the display of each peripheral target. The stimulation sequences were identical to those used in the VPS system. The subjects were seated in a virtual reality simulator. Superimposed over the visual targets was a cloud of balls moving by expansion or contraction creating an optical flow over a wide dimension of the field of perception. The balls were perceived as having a 10cm diameter, with a density of 4 balls /m². Four different speeds were considered: 1.67m/s, 3.34m/s, 6.68m/s and 13.36 m/s. The eye-head coordination coefficient was measured in this virtual environment by means of a magnetic sensor (Flock-of-birds) placed on liquid crystal spectacles worn by the subjects. In the control situation, no peripheral environment was superimposed over the luminous targets, thus simulating the measurement as performed by the VPS. For each test, the average value and the standard deviation of the gain were calculated. The gain is the ratio between rotation of the head and the angular eccentricity of the target. Since our targets were situated at 40°, a gain of 1 therefore represents a 40° rotation of the head.
|Picture 2 – An optical flow is presented in the virtual reality simulator, superimposed over the 40° targets.|
The results showed that the subjects' behaviours remained stable whatever the stimulation conditions in the peripheral environment. Thus, an analysis of variance performed on the data showed that the periphery did not have any significant effect on the average eye-head gain, or on the standard deviation in this measurement (for more details please refer to the previous article, Faubert et al., 2007).
Second experiment: Is being seated or standing liable to change the way we move our head and our eyes to observe the environment?
The previous experiment gave us an initial indication of the average stability of eye-head coordination when a large dynamic environment is added to the presentation of the usual local targets. Over and above this observation, several questions remain: even if average behaviour does not change, does each individual retain a similar strategy from one circumstance to another? The second question refers to the recent arrival, amongst vision professionals, of new eye-head coordination measurement tools. Indeed, the Visioffice® has, over the past few years, become an essential system in the sales approach used by opticians (see illustration below). The Visioffice can come as a tower and therefore mean that eye-head coordination has to be measured in a standing position. This was unquestionably new and it was therefore necessary, prior to the launch of this new system, to ensure that the measurements were not impacted by the subject's change in position in front of the stimuli. The question may at first appear trivial but previous work had indicated that the answer to it was not. McCluskey and Cullen (2007) showed, in fact, that monkeys moved their heads more when sitting in front of the screen than in a standing position. No study had been used to make this type of measurement in Humans. We therefore carried out the experiments presented below.
|Picture 3 - The Visioffice, the new eye-head coordination measurement system used by vision professionals. Since this system is sometimes in the form of a tower (see illustration opposite) the coefficient measurement is taken when the subject is standing, and not seated, as was previously the case with the VPS.|
This experiment was carried out in the same environment as the previous one. An optical flow was therefore presented, or not, when the eye-head coordination measurements were taken, using the same stimulation conditions as those planned with the VPS system. Twenty subjects were recruited for this study. They were separated into 2 groups: young adults, aged between 20 and 30 years (N=10) and presbyopes, aged over 45 (N=10). The instructions were identical to those given for the previous experiment and measurements were made in both seated and standing positions.
|Picture 4 – An optical flow is presented in the virtual reality simulator, superimposed over the targets at 40° that the subject is to look at. In this example, the subject is standing. He will carry out this same sequence in a seated position, in the same position as the one usually used with the VPS.|
Results again showed that average eye-head behaviour, in all stimulation conditions remained eminently stable. Statistical analysis thus showed us that changes in body position and peripheral environment did not have any significant effect on the amplitude of the head movement, i.e. the gain (see figure below), whether for young adults or people with presbyopia.
|Picture 5 - Amplitude of head rotation, measured in the form of gain, depending on the conditions under which the information is presented and on the position of the subjects, either seated or standing. These results shown that the gain remains very stable for all the situations envisaged, whether for young or older individuals.|
Beyond confirmation of the reliability of the average behaviour in terms of changes in stimulation conditions, a correlation analysis showed that individual gains were significantly similar between sitting and standing positions. The graph below illustrates this very strong correlation for the group of young individuals (r2=0.88); a similar level of correlation (significant) is obtained for the group with presbyopia. These data therefore confirm the high degree of reliability of the measurement method which, even when several elements are changed, leads to a value that is very similar on average and for each individual.
|Picture 6 - Correlation between the measurements made in the same position (standing in the graph opposite) with the VPS and in the virtual reality simulator; this example is given for the group of young subjects; the same levels of correlation were obtained for the other comparisons, 2 -2.|
Third experiment: Does eye-head coordination strategy vary when peripheral targets are modified?
When talking with Eye Care professionals about measuring eye and head coordination, a question came up regularly: current tools define the strategy using relatively simple peripheral targets, DELs (Electroluminescent Diodes); is the measurement obtained in this way representative of habitual behaviour, the same as that when one looks at letters, faces, cars, etc.? In order to give some elements of response to this question, we therefore carried out a study in which the physical characteristics of the peripheral targets were modified. Thus, and in order to get closer to more "ecological" stimulation conditions, we used targets that were no longer merely visual, but also included sound. Scientific literature in this field told us that this was a relevant question. Siegmund and colleagues (1987) showed, for example, that the combination of auditory and visual stimuli modified eye-head exploration behaviour, reducing the latency of head rotation, compared to circumstances in which the targets were solely visual or auditory. However, no previous study had investigated the effect of these uni- v. multi-sensorial stimuli on the amplitude of head rotation. This was therefore the work we undertook in the study presented below. In addition to the multi-sensorial nature of the stimuli, we also varied the shape of the visual targets.
This experiment used a different set-up from the one used in previous work. A new tool has been specifically developed for the requirements of this protocol. Contrary to the existing systems (i.e. VPS and Visioffice), it allows to change the parameters of the peripheral and central stimuli, in order to diversify the contexts of eye-head coordination measurements and thus specify the ways this individual strategy operates when surrounding targets change. This system comprises 3 stimulation units. Each one has a digital counter, a DEL and a micro speaker. This combination means that visual targets of various shapes can be generated together with spatially corresponding auditory stimuli. In this experiment, the peripheral targets could be points of light (i.e. as with the VPS and Visioffice), or letters, with or without 70dB white noise, as an auditory stimulus. Six peripheral target conditions were therefore proposed in this experiment, in random order: DELs only, DELs only presented in a non-random order, letters only, white noise only, DELs with white noise, letters with white noise. In the second situation, the non-random aspect of the appearance of the targets, on the right and on the left, was managed in the following way: in the same way as all the experiments done in this field, the sequence began with presentation of the central DEL, the first peripheral could then appear randomly either on the right or on the left. The central target was then lit again and then the second peripheral target was presented on the opposite side to the first one. Thanks to this method, the spatial uncertainty on the location of the peripheral target was drastically reduced. Indeed, when a peripheral target was presented on one side, the next one would necessarily be located on the other. Forty-five subjects took part in this study. They were separated into 2 groups: young adults aged between 20 and 30 (N=30) and presbyopes aged 45 and over (N=15).
The first result of this study is illustrated below. In the same way as in previous studies, the average eye-head gain remained highly stable, whatever the parameters of the peripheral targets. Only the circumstance of auditory stimulation alone, for the group of presbyopes, presented a significant difference in average gain compared to the others. This result could be explained by presbycusis, which typically occurs at the age of around 45-50 and which, by reducing sensitivity thresholds, may lead to an increase in head rotation in order to improve spatial localisation of the source, since the auditory stimulation was given in total darkness. Additional work would be necessary in order to verify this hypothesis.
|Picture 7 – Average gain (y axis) in percentage rotation of the head, depending on the parameters of the peripheral targets. The blue histograms correspond to the young adult group and the red histograms to the subjects aged 45 and over.|
The results also suggested that the great individual stability in eye-head gains, despite changes in stimulation conditions, discussed previously, was no longer present in the context of this last protocol. Thus we observed that, for certain individuals, eye-head behaviour was very similar whether the targets were uni- or multi-modal and with DELs or letters. On the other hand, other subjects presented a great variability in their eye-head gain, depending on the conditions. This result is illustrated below. This study therefore sheds new light and demonstrates that the coordination of eyes and head rotations to look at peripheral targets can be, for some people, extremely reliable in terms of changes in stimulation conditions. For others, this behaviour is less well anchored and can vary when the target parameters (or even the peripheral environment) are modified. As with eye-head gain, there exists a continuum of stability in this behaviour. This is then a new element within our knowledge of this eye-head coordination strategy, enabling us to continue with optimisation of presbyope adaptation to their personalised progressive lenses.
|Picture 8 – Average gain (y axis) in percentage of head rotation and typical difference in gain (errors bars) for the 30 young adults involved in this experiment (x axis). The same inter-individual differences were found in the group of presbyopes.|
Since 2003, individuals with presbyopia can obtain progressive lens equipment that is personalised according to the eye-head coordination strategy they personally use to look at peripheral targets. The many positive comments made about this product have established the relevance of this new parameter in the choice of the best suited optical solution for each particular individual. This success comes notably from the quality of the measurements of this individual behaviour. The series of studies presented in this article highlights the great reliability of the way in which eye-head behaviour is defined by VPS or Visioffice. The experiments show that the environment in which the observer is placed, as well as his posture in front of the visual targets, can be changed drastically, without modifying the individual's strategy. Nevertheless, these existing systems do have their limits. The last study highlights the differences in eye-head behaviour stability when visual targets and/or the task performed changes. Some people's behaviour is very reliable in the face of any alteration made to stimulation conditions. For others, on the contrary, eye-head behaviour is less well anchored and may vary when the parameters of the targets, task, environment, instructions, etc. are modified. A new tool, with several measurement conditions and tasks, could define this "anchoring" or "stability" of the eye-head coordination strategy on an individual basis and thus improve knowledge of the visuomotor behaviour specific to every wearer and future wearer of progressive lenses.
To observe a target on the periphery, the eyes and head move to position the gaze on the zone of visual interest. This fixation strategy, over and above its apparent simplicity, requires precise coordination, of both eyes and head control systems. Several studies showed that the propensity to involve the head in this strategy of moving and fixating the eyes is specific to each individual. Whilst some people look at a peripheral target moving only the eyes, others make large head rotations.
Fuller (1992) was one of the first to demonstrate these inter-individual differences. ESSILOR has been working in this field for over 10 years, in collaboration with the University of Montreal (Simonet, Bonnin, 2003). Initial work resulted in the design of a new range of products that are personalised according to this propensity to move the eyes or the head more in order to observe a peripheral target. These are the Varilux IPSEO progressive lenses. In addition to the conceptual importance of this approach, a way had to be found to measure individual behaviour, whilst respecting the constraints linked to clinical usage of this tool. The success of this new lens therefore depended on the quality of measurements. The VPS system was designed for this purpose (see illustration below).
|Picture 1 - The VisionPrint system (VPS), a system used by vision professionals to
measure an individual's eye-head coordination strategy.
This system has been extensively distributed and is used in a very large number of ECP (Eye Care Professional) stores. Although progressive lenses that are personalised according to the individual eye-head coordination strategy have demonstrated their efficiency, some professionals have questioned the principle of defining the eye-head coefficient with the VPS. Indeed, to what extent is a value established on a given day, in specific circumstances, representative of the "normal" or "habitual" behaviour of the individual? To answer this question, a series of experiments was performed by the ESSILOR R&D teams, and in a laboratory at the University of Montreal directed by Jocelyn Faubert, holder of the NSERC-ESSILOR Industrial Research Chair, a Chair whose aim is to increase theoretical and practical knowledge of the functioning of visuomotor coordination in general and of the eye-head pairing in particular. The aim of this article is therefore to offer an overview of this work and demonstrate the relevance of individual measurement of visual exploration, performed with the VPS and other similar systems.