Using the Blind Spot to Investigate Trans-Saccadic Perception

Round 1

Reviewer 1 Report

I appreciate restructuring the manuscript as a concept paper and as such allowing a rapid communication of the mere idea. No further comments or suggestions.

Reviewer 2 Report

Thank you for revising this contribution. It isn't a regular research contribution but as a short communication it has some value for those working in this area. 

Reviewer 3 Report

No further comments

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This paper describes a technique for presenting targets within the blind spot, so that they become visible only after a saccade.

 

The idea seems perfectly fine but .. the paper seems to be missing its Results section? The Methods describes a sensible experiment, but I was expecting Results figures. First, I was expecting a figure corresponding to Fig A (right hand side), showing % of saccades made to the different locations. Presumably this would be 100% for the visible targets and 0% for the ones in the middle of the blind spot, with a more graded response for targets near the edge. 

 

Similarly Fig C is a nice diagram for illustrating the concept, but where are the actual results, showing target locations and mean +/- SEM of saccade landing positions? 

 

Presumably also you have to monitor eye position and reject trials where eye movements could have revealed the target prematurely. Data on the visibility of targets within the blind spot would be critical here, and it's exactly these sorts of implementation details that I felt were missing.

 

Ideally the results of this novel technique would also be compared with the results obtained with a conventional double-step experiment, perhaps showing a timing difference that could be attributed to the latency. 

 

Without quantitative results, the paper says "we *demonstrate* that using the blind spot, a double-step paradigm (as well as many other trans-saccadic detection paradigms) can be generated without the need for online saccade detection", but frankly it reads more like an assertion than a demonstration. I do personally buy their assertion, but is that enough for even a brief report? If they have collected these data from one or more participants, why not analyse and display them?

 

Finally, the paper says "using the blind spot, a double-step paradigm (as well as many other trans-saccadic detection paradigms) can be generated". Can they give examples of these other trans-saccadic tasks? Are there any tasks that could NOT be generated using this technique?

 

 

Reviewer 2 Report

This idea is brilliant. However, the manuscript obviously just describes the idea and does not seem to be an empirical study in which a hypothesis was tested and quantitative results were obtained.

 

Major issues

- Results are missing. The methods section sounds like the size of the blind spot of a single participant was measured, but no results are reported. Was anything measured in the second test/double-step paradigm? Were there any differences in gaze behaviour for forward versus backward step conditions? Was there any comparison of blind-spot to a non-blind-spot/traditional double-step paradigm?

 

 

Minor points

 

- lines 48-50, the description of location and size seem to be about the blind spot, i.e. area of the visual field, and not about the optic disk, area of the retina. Please make that more explicit to avoid confusion. In addition, the numbers specifying its extent seem to be very specific. I prefer to see ranges or some estimates of individual variability instead for all values.

 

- Figure C, the second “left panel” is supposed to be “right panel”? Please label the axes in the figure and specify units.

 

- lines 108-111 references for these two techniques and the associated shortcoming are missing.

 

- line 113 “or any other intra-saccadic detection paradigms”. That’s not true anymore if we consider trans-saccadic manipulations more generally. There are limitations to the proposed ingenious method., because it can only change visual input from pre-saccadically “invisible”/filled-in at a certain eccentricity and size (so that it fits the size of the blind spot) to a certain post-saccadic input. With gaze-contingent methods, both pre- and post-saccadic visual input can be varied arbitrarily. Seems like a trade-off between timing requirements and required stimulus material.

 

- Instead of making stimuli suddenly appear in a gaze-contingent way this method could in theory also be used to make things disappear. Although perhaps less useful, the authors might want to discuss this possibility.

Reviewer 3 Report

An interesting idea. It is very brief description though and it would be nice to have some more detailed methodology. For example, will you make code available to replicate? Do you have some practical experimental results/outcomes? A study with a small subject group?  

Reviewer 4 Report

This brief report proposes a novel method to present stimuli for the double-step saccade with no latency. The idea is that the second stimulus, which has to be invisible before a saccade and visible after the saccade initiation, is located within the blind spot before the saccade is made. Since any stimuli within the blind spot are invisible by definition and escape from it whenever the eye moves by a distance larger than the blind-spot size, the use of the blind spot ideally meets the specifications. A feasibility test has been successfully completed.

I believe that sharing this nice idea will absolutely benefit eye-movement researchers, and I find every sentence is very clearly written step by step. I have only four minor things for clarification.

(1) The second target has to be located within the blind spot, so there is an inevitable correlation between the first target's absolute position and the second one's relative position in the shown example. This correlation should be often avoided when stimulus predictability is a serious matter. Some discussion about this issue will enhance the impact of the paper. Of course researchers could create workaround plans that would effectively dissipate this potential issue. For example, place the second target to the left, right, above, or below the first target, and only one of these locations corresponds to the natural blind spot (so only the data for this location are relevant) whereas the other locations are displayed by a conventional gaze-contingent algorithm (so you may throw away all data for these latency-prone sham locations).

(2) There is no description about the offset of the first target in the blind-spot experiment. In the explanation of the double-step target paradigm in general (Figure B), there is an explicit mention ("the first target is removed") so it would be better to add a similar mention in the blind-spot experiment as well.

(3) Figure A: The depiction of the blind spot looks OK in the lefthand panel but slightly atypical (too high) in the righthand one. If this oval shows a schematic rather than an actually measured one, lower it by 1 deg or so and the location will become more consistent with the one in the left panel.

(4) Figure C legend has subsections (A) and (B), so the two panels should be titled "(A) Feedforward Step" and "(B) Feedback Step".