Anschliessend klicken Sie bitte hier, um die Seite neu zu laden.
Beispiel
Diskussionen sind dermassen vielgestaltig (und oft verhältnismässig umfangreich), dass es schwierig ist, ein «typisches» Beispiel anzuführen. Trotzdem hier ein Beispiel eines englischen Artikels mit einigen Hinweisen.
von Wartburg, R., Wurtz, P., Pflugshaupt, T., Nyffeler, T., Lüthi, M. & Müri, R.M. (2007). Size matters: Saccades during scene perception. Perception, 36, 355-365.
4 DiscussionThe present study was designed to investigate the relation between saccade amplitude and image size during scene perception. Our meta-analysis as well as our new experimental data revealed a nearly perfect positive correlation of mean as well as median saccade amplitude with image size. This contrasts with earlier review articles in which the conclusion was reached that mean saccade amplitude is independent of image size (eg Henderson and Hollingworth 1998; Rayner and Pollatsek 1992). |
Hier wird nochmals kurz erwähnt, was gemacht wurde, und vor allem was das wichtiste Resultat ist. |
|
One important question is whether this is an effect of image size as such, or an effect of object spacing, which was also affected by our image-scaling procedure. Two arguments speak in favour of the view that image size is the more important factor. First, the results reported by Enoch (1959) which thus far is the only study that has directly addressed the question of how image size affects saccade amplitude are very similar (cf figure 1 and figure 5d). In that study, image size was not manipulated by scaling, but by placing paper masks in front of the image, a procedure which changes image size, but not object spacing. The high correspondence of Enoch's values with our results is even more striking as his experimental conditions and eye-tracking equipment were very different from ours. He used grey-scale experimental maps as stimuli, which were quite different from the colour scene images we used, and his corneal reflection ophthalmograph certainly had a different ability to detect small saccades than did our equipment. The second argument for the view that image size is more important a factor than object spacing comes from our meta-analysis of previous studies. It indicates a clear linear relationship between image size and mean saccade amplitude, independent of factors such as image type and content (and thus object spacing), viewing task, technical equipment, and whether images are scaled or just cropped. In consequence, mean and median saccade amplitude can reliably be predicted by a simple linear equation expressing saccade amplitude as a function of image size. |
Das zentrale Resultat wird nun in Beziehung zu früheren Studien gesetzt, um eine Begründung für die erhobenen Resultate zu liefern. |
|
This finding might suggest that it was reasonable to indicate saccade amplitudes in percentages of image size, not degrees. In reading studies, this is a customary practice as, over a large range, saccade size is constant if measured as a number of character spaces, not visual angle (Legge et al 1985; Morrison and Rayner 1981; for a review, see Rayner 1998). In other words, if a text display is scaled to different sizes, the saccade amplitudes relative to display size remain constant. However, this does not seem to be the case for saccade amplitudes in scene perception. Our analysis of median saccade amplitude relative to image size shows that the smaller a stimulus image is, the larger are the relative saccades. This might be an effect of the fact that, in an image that has been scaled down to a small size, more information is foveally available, and thus longer saccades can be made; or, in other words, fewer intra-object saccades are necessary. Another reason for this observation may be that, as image size is reduced, details tend to disappear on a digital display. If those details are frequently fixated in large images, but cannot be discriminated in a smaller image, other image features farther apart may be selected for fixation. Presumably, this is not a specific effect of digital displays. Even if image size is manipulated by modifying viewing distance of an image, small details would 'disappear' owing to limitations of visual acuity. |
Ein weiterer Gesichtspunkt wird diskutiert. Beachten Sie die Verwendung von «hedge words» |
|
As to the overall saccade amplitude distribution, it might seem at first sight that there is an identical modal value at approximately 0.9 deg for all four image sizes. At a closer look, however, it becomes clear that this mode is an artifact of the saccade parsing algorithm. When its parameters are tuned for higher sensitivity for small saccades, the mode tends to shift towards lower values. At the extreme, the range of 'regular' saccades smoothly merges with the range of eye-movement phenomena usually regarded as microsaccades, microtremor, and the like. That is, when only the range from approximately 1 deg to the maximal image extent is considered, the peak of saccades lies at the lower end of the distribution, with a continuously decreasing incidence towards longer saccades. This suggests that it might be reasonable to omit saccades below 1 deg for analyses in cognitive research. |
In diesem Absatz wird eine Schlussfolgerung für die methodische Durchführung von ähnlichen Experimenten herausgearbeitet. |
|
Our experimental results as well as the meta-analysis demonstrate that the size of stimulus images is the dominant factor influencing mean and median saccade amplitude during scene image viewing. It seems that other factors, such as stimulus properties, viewing instructions, measurement equipment specifications, and even object spacing, are only of subordinate importance. From a methodological standpoint, this underlines that stimuli for scene-viewing studies should be displayed in reasonable size (at least 20 deg) so as to improve the ratio between measured amplitude values and measurement error. However, in fixed-head experimental settings, horizontal image width should not exceed 35 to 40 deg, as it seems unnatural to visually explore such large scenes, and make accordingly long saccades, without concomitant head movements. |
Der letzte Abschnitt bildet das Fazit, der nochmals die Haupterkenntnis betont, und eine weitere methodische Schlussfolgerung zieht. |
