A closer look at working memory changing with adversity

The authors have addressed all of the concerns raised in an exceptionally thorough and satisfactory manner. Their point-by-point responses are well-reasoned and supported by additional clarifications and revisions to the manuscript. The changes made have significantly improved the quality and clarity of the work. I appreciate the authors' responsiveness and their commitment to strengthening the paper based on the feedback provided. With the revisions made, I believe the manuscript is now in excellent shape and ready for going to the next stage.

This registered report will examine whether adversity is related to lower or intact working memory (WM) performance. The authors will isolate variance in performance related to WM capacity from variance in performance related to updating ability as the two measures covary. They will combine existing and new data, and estimate participants' exposure to neighborhood threat, material deprivation, and unpredictability as measures of adversity. Structural equation modeling will be used to analyze the relationship between adversity and WM measures.

The relationship between working memory and the experience of adversity is a topic of great practical and clinical importance, especially in psychology and related fields of research. 

However, the significance of this as basic research is not clear enough, and there appear to be some major problems, as discussed below.

First, to the best of my knowledge, most previous research has supported that adversity is negatively associated with performance on WM tasks and executive function. It is true that some other research also showed no relationship between adversity and WM performance, but only a few studies reported the positive relationship between adversity and WM/executive function. In fact, the authors relied on research by Young and colleagues for this aspect. If the authors wish to test two hypotheses (deficit-based and adaptation-based models), they should cite more research on the adaptation-based models and justify that the research question in this study is worth investigating.

In addition, whether their argument can apply to other aspects of executive function is unclear. Previous research consistently showed that adversity can negatively affect children’ and adults’ executive function performance. Is WM special?

Second, and relatedly, although the authors compare the two hypotheses from the perspective of work memory capacity and updating, there are several other aspects that should be considered to explain different results in previous studies. For example, the two models may differ in the age of the participants and the population. Research supporting the adaptation model (i.e., Young and colleagues' research) has generally included children, whereas research supporting the deficit model has included both adults and children. Furthermore, the adaptation model may only be supported in a specific population (e.g., US), whereas negative effects on WM and executive function are observed globally (Western, Asian, and African).

This point is important because the authors will be using data from participants between the ages of 18 and 55. If their results support one model, we cannot determine whether the measures of working memory or the age of the participants/population are critical factors.

Third, the term "adversity" is ill-defined and somewhat vague. It encompasses a very broad range of indicators other than those used in this study. For example, some researchers use socioeconomic status to refer to it, while others refer to family criminal history or abuse, using the Adverse Childhood Experiences questionnaire. The authors must explain how the measures of adversity may affect the relationship between adversity and WM and justify why they chose their measure to address the research questions in this study. Also, for the analyses, they will be calculating composite scores without analyzing the relationship between the measures (e.g., Neighborhood Threat Composite). They should analyze the relationship between the measures, such as confirmatory factor analyses.

I recommend this manuscript to be accepted at Stage 1 with just a few clarifications.
 
The rationale is mapped out clearly with a strong theoretical and methodological basis for the study. Research questions and hypotheses are outlined in detail. What is the rationale for creating composite scores for adversity measures? Perceived and objective measures of crime, for example, might show different relationships to WM, and these nuances are lost in a composite score. Explanation of this choice would be useful.
 
In terms of the framework, how will you determine “lowered”, “intact”, and “enhanced”? Arguably you would need a repeated measures design with manipulations to determine whether WM was lowered or enhanced. Moreover, how would you define an intact score? Adjusting the language to reflect regression analyses would be more suitable, e.g., higher threat predicts lower WM capacity.
 
The methods are mostly outlined in detail. The authors describe two datasets collected via the LISS with multiple timepoints. Will the 800 participants be randomly selected? And will they be selected from one or more timepoints? What if a participant has data from all time points and another only has data for one timepoint? This should be explained more clearly. It is also not clear what the final sentence (point 4) of the “Data access” section is referring to – what do you mean by later timepoint? It might be helpful to label the timepoints at the start of the method section and use those labels throughout.
 
The number of trials is missing from the task descriptions; this is needed for future replication. For the updating task, 18 trials are referred to. Is this total number of trials or number of trials for the updating condition? Number of trials per condition and per task is needed. Can the authors explain why only 18 trials in this case? Is this enough trials to accurately capture performance?
 
The data analysis plan is well thought out with appropriate steps, control variables, and model fit checks.