Accident Vulnerability and Vision for Action: A Pilot Investigation

Round 1

Reviewer 1 Report

This is a straightforward correlational study, showing a relationship between performance on a visual and action attentional cueing paradigm and a self report measure of accidents.

Main comments

I was initially a bit sceptical about the premise that vision-for-perception would not be associated with accident proneness. While some accidents involve rapid actions (associated with dorsal processing), others must surely involve conscious recognition and decision making. Failing to see a red light, or searching for a safe path and failing to see a hole would be two examples. Perhaps the introduction could elaborate on this a little to explain whether there are perhaps different types of accident which might be associated with different abilities. This also ties in with critiques of the two streams view that they must, to some degree, interact in real behaviour. There is no getting away from the fact that the sample size is low to be making firm conclusions. The discussion mentions this regarding stage two, but I also think more caution is necessary for stage one. I don't know what the guidelines are for stats like cronbach's alpha, but it would be better to validate this questionnaire on a larger sample (hundreds). For readers not familiar with the VAT, I think a bit more detail and justification could be provided. How is this task different from a more conventional exogenous cueing task? I was a bit surprised to see that a trial was scored as correct if either the saccade OR the button response was correct (line 267). Doesn't that mean that participants could actually ignore the target and just ensure they are saccading towards the correct cue shape? It is also not clear whether participants are actually told which cue is the correct shape, or whether they have to learn this. Is it surprising that the APQ and SPPB are not associated? Again (see point 1), surely some accidents are not caused by vision per se and we would expect them to be associated with problems of movement.

Minors

5. interpreting the direction of the correlations in stage 2 is not always straightforward because it is not specified whether a high score on the different tests indicates better performance/acuity etc. For example there is a correlation of -.10 on line 282, but I'm not sure if a low score on the SPPB is actually good.

6. It is also worth noting, and this is probably a sample size issue, that some of the assumptions are not confirmed in stage one. We might assume that older adults would have more accidents but that does not seem to have been the case, in any of the three measures (not just non-significant but in the wrong direction).

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This research paper aims to explain increase accident risk with failures of the “vision for action pathway”. The authors correlate results from a “vision for action test”, which relies on subjects detecting and identifying a target in the periphery of their visual fields, with scores from a self-reported proneness to accidents questionnaire, while controlling for visual acuity (Freiburg Visual Acuity test) and physical performance. 

First of all, I would like to salute the authors endeavour to predict risk of accidents with indicators of vision for action, as this could indeed be an area of intervention. However, the method for assessing “vision for action” is either, flawed, unclear, or both.

 

The authors mention that they adjusted the SOA using a staircase procedure. Please give details about the procedure (Steps, number of reversals). Also, please clarify what the threshold measured at 75% performance corresponds to.

Furthermore, the authors mention: ”If accuracy at the initial SOA value did not exceed 75%, the threshold value for that participant was recorded at 1000 ms.” (line 263-264). Again, what is the threshold measured here if the SOA isn’t varied? What is the percentage of subjects that had their thresholds measured at 75% performance? How could have this percentage contributed to the final result?

Figure 1 shows VAT thresholds ranging -400 to 600, and APQ scores from -5.0 to 7.5. It is not clear what the thresholds represent, and given that average APQ scores in table 1 is 20.79, the numbers in Figure 1 don’t make any sense, what transformation was applied?

The authors mention: “A trial was scored as correct if a participant launched a saccade of 4 or more towards the landmark shape and target, or they discriminated the identity of the target correctly, by pressing the appropriate arrow key.” (line 267-269). Was the goal to test the ability to detect and initiate a saccade to a target in the periphery or identify a target in the periphery? Detection of target onset and identification of target would engage different but overlapping networks, although with very different processing times.

Furthermore, it has been recurrently demonstrated that performance in attentional tasks depends on the visual field (right vs left) as it is influenced by language direction among other things (e.g Kermani et al. 2018). Was the threshold calculated by taking into account left and right visual field performances together? Authors are advised to take this into consideration.

 

Maybe a more direct measure of the ability of participants to react to the sudden onset of a target at predicted location, would be median saccadic reaction times (SRT) to target onset for all SOAs, which I suspect authors must have for each subject.  Correlating median individual SRT to APQ scores, might provide a more straightforward indicator than thresholds. Authors might even consider correlating, misses or false-alarms to APQ scores, if those are available.

It is to be noted also that visually guided saccades, have been shown to be disrupted after impairment of the superior-colliculus ventral pulvinar pathway in monkeys with V1 lesions (Kinoshita et al. 2019). This entails that visually guided saccades, such as the ones tested in this study, may not be mediated by the dorsal vision for action pathway, but by subcortical circuitry.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The present study introduces a novel method to measure „vision for action“ and shows that this construct is predictive of accident vulnerability even after controlling for age and other variables. I agree with the authors that this is a very promising idea and the results appear to support the conclusions. However, I have strong concerns regarding the small sample size as discussed below.

Major comments

My main concern is the sample size of the present study. Investigating interindividual differences with such a small sample (<20 in Stage 2) is clearly inappropriate. In my eyes, there is only one way how this study can be published: First, the authors should make explicit that this study is a pilot study with a small, preliminary sample - this should be highlighted in the abstract and in the title (e.g., by adding: „- a pilot study.“). I leave it to the Editor to decide whether such a study is eligible for this journal. Second, the statistical methods must be adjusted to this small sample size. Either assumptions underlying parametric statistical tests must be checked (e.g., normality of variables), or non-parametric methods (where available) must be applied.

Lines 144-147: Please provide mean and standard deviation of age for the two groups.

Lines 210-212: It’s unclear what scoring high and scoring low means. Please provide all test statistics for the subgroups used in stage 2 (tests and mean/SD of age). Was there an a-priori criterion for selecting these participants?

Lines 267-269: „A trial was scored as correct if a participant launched a saccade of 4 o or more towards the landmark shape and target, or they discriminated the identity of the target correctly, by pressing the appropriate arrow key.“ Does that imply that a trial could be correct if there was a valid saccade but an incorrect response (and vice versa)? This is an arbitrary criterion and doesn’t make sense.

Minor comments

There are many errors in citations. Brackets are incorrectly used.

A figure illustrating the procedure of a trial in the VAT would be very helpful.


 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors have handled all of my concerns.

Reviewer 3 Report

The authors have addressed all my comments and I have no further concerns.