Talk:Eye movements

The idea that this can be reviewed like a journal article is a bit naïve and I’m not sure how much of this goes immediately into the public domain but here are some first thoughts from John Findlay.

Eye movements are a behavioral measurement

This describes the focus of the current article but it fails to address the interest in eye movements that can come from neuroscience, physiology, optometry, ophthalmology etc. I would suggest that a title ‘Eye movement studies of cognitive processes’ would be more appropriate with several other articles being solicited to address the other aspects

Basic characteristics of eye movements

Again the title implies a much broader content than occurs. Saccades are only one type of eye movement and even for them, the ‘basic characteristics’ (stereotypy, ballistic nature) are scarcely described.

>Once initiated, it is difficult (though not impossible) to change their trajectory.

This might be misleading. I’d suggest “The end-point is pre-selected before the beginning of the movement”.

>While it has generally been assumed that the two eyes move in synchrony and that >they fixate the same point in space, recent research clearly demonstrates that this is >not the case and the two eyes are frequently deviated from each other.

The general assumption here only relates to reading research. For more general purposes, something needs to be said about vergence, saccadic disconjugacy etc.

Table 1 At the moment this table presents a range as a ‘Mean’, Maybe say ‘Typical mean fixation duration’.

Brief History of Eye Movements

Again this is narrowly contextualised. Wade and Tatler have written a book on the topic.

Eye Movements and Visual Search

What about how saccade destinations are selected – see later comment on saliency ?

Eye Movements and Scene Perception

As well as scene concept, scene layout seems to form part of gist information (Sanocki and Epstein, 1997 etc)

The concept of saliency maps has also been considerably used in visual search, although the usage is somewhat different (not simply bottom-up saliency)

‘about 2.6 degrees’ - mixture of precision

Interestingly, viewers are rather insensitive to large changes … I suggest this needs qualifying, particularly since the changes in Figure 9 are all quite small ones. I suggest “viewers may be surprisingly insensitive to quite large changes”

References

I’d like to put in a pleas that Active Vision (Findlay and Gilchrist) might get a plug here since it’s about the level that potential readers may find useful.

Review of Rayner and Castelhano "Eye Movements" by Simon Liversedge

This short review article is written in an informal way, presumably with a non-expert audience in mind. As such, it does a good job of providing basic information from which further research can be undertaken. I've no major criticisms, but there were a few points that I thought could do with reivision - I'll list these as I cam upon them as I read the paper.

Striclty speaking, the opening sentence is not correct (but I understand what the authors mean I think). I'd suggest changing it to something like "Eye movements are a behavior that can be measured and their measurement provides a sensitive means of learning about visual processing".

Insert "psychological" between "about processes" in the next sentence.

I may have missed it, but I couldn't find Figure 1.

In the text starting "During the actual..." the digit "1" appeared after the word suppressed. Should this be deleted?

For the bullet point starting "While it has..." maybe cite a reference? E.g., Liversedge, S.P., Rayner, K., White, S.J., Findlay, J.M., & McSorley, E. (2006). Binocular coordination of the eyes during reading. Current Biology, 16, 1726-1729.

For the section on eye movements and reading, delete "in reading" from the opening sentence (it appears twice).

For the Figure 2 legend, delete "a lot", add in "one" between "than" and "fixation", and change the phrase "some fixations" to "some saccades".

For the bullet point starting "A very important issue..." in the first sentence move the word "is" two words downstream. Also, in the next sentence, should the comma come one word later?

I didn't understand the reference to Eye-contingent experimental paradigms. Also, the digit "2" appeared in the next section - should this be deleted?

The final two bullet points in the section on Visual Search were not at all clear - I'd recommend these be changed. The text talks about letters, but in principle, visual search could use distractors of any type. The final point was related and did not make sense either...

Figures 6 & 7 were a little confusing to me. I wasn't entirely clear what Fig 7 showed that Fig 6 didn't also show. Perhaps one of these could be deleted, or instead it could be made clearer what the difference between the two is.

For the point about salience maps, I'd recommend that Findlay and Walker (1999) be cited (the authors may argue that this model was not a model of eye movements in scenes, but in my view, it was put forward as a general model of eye movement control with a salience map as a central component, and it could easily, therefore, be taken as a model of eye movements in scenes). Also, I'd change the wprd "prominent" to "salient". I'm not sure the two mean the same thing.

Figure 8 legend "on the right a different person is..."

Figure 10 legend "which ones did you miss".

Simon P. Liversedge

--

Review by John M. Henderson

My own bias would be to have separate entries for Eye Movements in Reading and Eye Movements in Scene Perception, and perhaps even a third in Eye Movements in Visual Search, particularly given that this topic is under Visual Cognition. Although there certainly are some commonalities in eye movements across these situations there are some very important differences as well, both at the level of the empirical phenomena and the level of theory. In fact, at this point there is very little overlap in the literatures. It also is not at all clear that models (including computational models) developed in one domain are going to generalize to the other domains, and in fact there are currently independent sets of such models in reading and in scene perception, and yet another set developing in search. For all of these reasons, it is a little odd to lump reading, search, and scene perception together.

Having said that, I think the authors have done a very nice job of highlighting some of the key findings and theoretical issues in each area.

Here are some minor points and suggestions.

Basic Characteristics of Eye Movements

Second Point. I’m not at all sure that it makes much sense to define a functional parafoveal area, particularly in scene perception. As the acuity figure shows, there really is a continuous acuity function.

Third Point. Do you really want to say our eyes move on average 250-350 ms? You say later that in reading the average is 225-250, so don’t you want to use say 225 ms as the lower bound for the average? At a later point you say that the lower bound for reading is 200 ms. Same for the next point where the same values are used.

Table 1, I think the saccade amplitude average given for scene perception is misleading, since the value varies hugely depending on the size of the scene (as has been shown in several recent papers), and varies yet again for the real world itself. Perhaps it would be fair to say that this is the value you would get for picture perception on an average sized computer screen at an average viewing distance, but even that is debatable. I think the same is true for visual search: It really doesn’t make all that much sense to give a saccade amplitude average like this given that it varies depending on the nature and size of the array, the nature of the search targets and distractors, and so on. Unlike reading, there really isn’t a standard type of search display. You sort of end up making this point yourselves later in the Eye Movements and Visual Search section, as well as in the General Comments section, so perhaps the best option is just deleting Table 1.

Eye Movements in Reading

Sixth point. You give yet a different range of fixation durations (200-250). Given this one, you might want to drop your range in the earlier point to 200-350.

Eighth point. I think it was a paper by Pollatsek and Rayner in Canadian Journal of Experimental Psychology that showed that information is not acquired from below the current line in reading, not Rayner (1998).

Eye Movements and Scene Perception

Point 10. You might want to make clear that Torralba et al presented a *computational* model.

Point 11. Henderson and Hollingworth (1999, Psych Science) along with other related papers (e.g., Hollingworth and Henderson, 2002) demonstrated that the window for detecting object changes in photorealistic scenes (rather than the line drawings used by Loftus) is closer to 2-3 degrees. This is more relevant that the Nodine et al study.

Point 12. I think the issue of the perception span for objects in scenes is actually pretty well established by the papers indicated above and other similar object-processing papers. See the discussion in the cited Henderson & Ferreira (2004) chapter for a review. What we don’t know yet in the field is what the perceptual span is for other sorts of information in scenes.

Point 13. I think it's misleading to just point to the original change blindness demonstrations without citing the papers that show why the theoretical interpretations (and even some of the basic phenomena) in those early papers were wrong. At least one critical paper (e.g., Hollingworth and Henderson, 2002) on the role of eye movements and particularly fixation should be cited.

Table 8. This scene figure has already been published so there may be copyright issues with using it again.

One other thought: You might want to cite the Henderson (2007) paper Regarding Scenes, in Current Directions in Psychological Science, as an introduction to eye movements in scene viewing for the non-expert.

John M. Henderson