Miriam Spering

Yanchao Bi

eLife assessment
This study, based on a large-scale data set and established behavioural tasks, has the potential to provide a valuable contribution to the literature if the authors could combine and correlate their behavioural evaluations with neural data and/or clinical assessments. As a standalone dataset, however, the current study appears incomplete because it does not go beyond merely replicating existing findings in a large cohort of children. In order to elevate their study, the authors are encouraged to publish their full dataset and explore the relationship between behavioural and neural or clinical data.

anonymous

Reviewer #1 (Public Review):
The authors provide a large-scale study of 18-month-olds, tested on a battery of tests (7 tasks designed to study attention, 1 to study working memory). Most of these tasks are already well-established, and the authors provide an additional replication. 
They further show that the variability in toddler's behavior (in terms of accuracy and reaction time) can be best and most parsimoniously accounted by two variable, one would correspond to attention, and the second one to social attention (i.e., the well-known interest for faces across the lifespan). Additionally, the authors find no evidence for a distinction between endogenous and exogenous attention. One may however argue that it is unclear whether any of the tasks actually tap endogenous attention. More detailed discussion of the cognitive functions involved in each of the tasks would enrich the paper.
Arguably, the working memory task is the task that is most likely to involve endogenous attention, as the behavior being tested is a spontaneous interest for a hidden object, hence attention triggered by an internal mental representation. Unfortunately, this task yielded null results and did not replicate previous findings.
Altogether, these findings provide an interesting method. Its value will be assessed when these behavioral evaluations will be combined with neural data and clinical assessment, as teased by the authors towards the end of the paper.

Reviewer #2 (Public Review):
The manuscript describes a large scale study of 8 eye tracking tasks in a large cohort of 18 month old children. The dataset is impressive and allows a comparison across children in different tasks that assess social, endogenous, and exogenous attention tasks. As such, it provides a benchmark for future studies that examine eye movements within different cohorts of children and across development and offers exciting possibilities to correlate these measures with behavior, other measures of motor and neural development, and to compare these measures with children diagnosed with neurodevelopmental disorders.
It does seem like additional insights can be gained from the study that could potentially address important topics in development, attention, and eye movements. Which components of attention are similar and in what way? The distinction between social vs non social is interesting but not ground breaking (e.g., the preference of toddlers to attend to faces); maybe looking at specific sub-tasks and clusters of participants the study can reveal new insights about the differences and similarities across tasks. The manuscript describes the importance of characterizing profiles of attention and individual differences, what kind of profiles are found in the study? Are there different profiles among this large cohort? 
Moreover, to allow comparison across analysis methods, ages, and neurodevelopmental disorders, it is important that the full dataset will be available online (i.e., all eye tracking data not just the metrics) as well as the software to run tasks that should also be made available to encourage using the battery across different research communities.

Reviewer #3 (Public Review):
Braithwaite et al. present data from a comprehensive large-scale study of 18-month-old's visual attention. The authors leverage a battery of well-known visual attention tasks to replicate canonical effects found in the literature and assess the latent structure of these tasks. They find that, while controlling for eye tracking precision and accuracy, two factors best fit the data - attention to social and non-social stimuli.
Strengths: 
The current study represents what amounts to years of hard work collecting data from a population that is challenging to work with - young children. The authors have diligently attended to data cleaning and sample size throughout the manuscript. Not only do they provide a large-scale replication of several well-known tasks, but they use advanced statistical modeling to discover the structure of visual attention in these 18-month-olds. Overall, this is a valuable contribution to the literature and provides a useful framework for studying visual attention development.
Weaknesses: 
While the study is clearly a valuable addition to the extant literature, I have several concerns that might be addressed to improve the manuscript. These primarily center around clarity and conciseness. First, the introduction seems to lack clarity at times. For example, the first paragraph seems to introduce several ideas (e.g., brain and cognitive development, direct and indirect measures of cognition, eyetracking, etc) that make it hard to understand where the paper is going. The authors might consider homing in on 2 main points to motivate eye tracking as a tool. Second, there are many different eye tracking measures may make it difficult for the reader to track which measures were used for each task and which were relevant for the larger model. This may be remedied by adding a section to the methods that briefly describes how each measure was calculated and perhaps a table that lists each task, the measure, and how it was calculated. Third, the results are exciting but hard to visualize in the supplementary figures. I commend them on using raincloud plots to visualize the individual data, but I would strongly encourage the authors to rethink how they display the data. For example, I find the supplementary images hard to see and as a result the effects reported are hard to discern in the image. Fourth, I believe the current data warrant a deeper discussion of what these findings mean. For example, given the developmental nature of the current study, it would be valuable for the authors to discuss how the structure visual attention might change or stay the same across development. For example, do the authors believe the current two factor model would replicate in older children, or would exogenous and endogenous attention emerge as separable components? How do these predictions relate to the extensive research in the adult literature?

Objective assessment of visual attention in toddlerhood

Figures and data

Be the first to read new articles from eLife