Inhibitory Control, Shifting, and Working Memory Updating Domains form Cognitive Phenotypes in Non-human Primates
Curation statements for this article:-
Curated by eLife
eLife Assessment
The primary aim of this study is to measure the different dimensions of executive function in non-human primates and the classification of animals into distinct cognitive phenotypes. To achieve this, it uses a multi-task touchscreen battery for six rhesus macaques with repeated measurements to examine how performance across inhibitory control, shifting/learning, working memory, and updating covaries within and across individuals. Overall, this is an important finding with relevance for anyone interested in behavioural assessment of cognitive functions in non-human primates or dimensionality of executive functions. The strength of evidence is overall convincing, with some minor methodological weaknesses that should be addressable in future versions.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Executive functions comprise at least four major subdomains: Inhibitory Control, Updating, Shifting, and Working Memory. Cognitive abilities in these subdomains are partially separable and partially unified in a common cognitive control factor in humans, but how these functions are organized in the nonhuman primate (NHP) is largely unknown. Here, we used a multi-task assessment approach and found that NHPs show within single sessions reliable cognitive markers of Inhibitory Control in an antisaccade task, ( ii ) Updating abilities in a multidimensional continuous updating task, ( iii ) Working Memory in a delayed matching task, and ( iv ) Shifting abilities in a feature-based rule learning task. First, we found that subjects’ performance fell into three separable cognitive phenotypes with unique strengths and weaknesses across cognitive subdomains. Second, beyond individual tasks cognitive metrics gave rise to four latent cognitive factors separating abilities for shifting/learning and working memory updating, and distinguishing abilities in inhibitory control of exogeneous versus endogenous interference. These results document four separable latent cognitive factors underlying executive control performance in this NHP sample, with inter-subject differences of these factors forming cognitive phenotypes.
Article activity feed
-
eLife Assessment
The primary aim of this study is to measure the different dimensions of executive function in non-human primates and the classification of animals into distinct cognitive phenotypes. To achieve this, it uses a multi-task touchscreen battery for six rhesus macaques with repeated measurements to examine how performance across inhibitory control, shifting/learning, working memory, and updating covaries within and across individuals. Overall, this is an important finding with relevance for anyone interested in behavioural assessment of cognitive functions in non-human primates or dimensionality of executive functions. The strength of evidence is overall convincing, with some minor methodological weaknesses that should be addressable in future versions.
-
Reviewer #1 (Public review):
Summary:
The authors aim to characterize how executive function (EF) subdomains (shifting/learning, inhibitory control, working memory maintenance, and working memory updating) are organized in NHPs. They train six macaques on four touchscreen tasks per session and derive a battery of behavioral metrics. They use reliability screening, correlation analyses, unsupervised clustering, and latent factor models to argue for separable EF factors and consistent cross-session cognitive phenotypes.
Strengths:
(1) Clear behavioral effects within tasks.
(2) Battery approach with repeated sampling.
(3) Scalable.
Weaknesses:
(1) Fixed task order may be a confound for between-task correlations. Curious about the potential impact of motivational changes, reward satiety, fatigue, etc., especially if later tasks have lower …
Reviewer #1 (Public review):
Summary:
The authors aim to characterize how executive function (EF) subdomains (shifting/learning, inhibitory control, working memory maintenance, and working memory updating) are organized in NHPs. They train six macaques on four touchscreen tasks per session and derive a battery of behavioral metrics. They use reliability screening, correlation analyses, unsupervised clustering, and latent factor models to argue for separable EF factors and consistent cross-session cognitive phenotypes.
Strengths:
(1) Clear behavioral effects within tasks.
(2) Battery approach with repeated sampling.
(3) Scalable.
Weaknesses:
(1) Fixed task order may be a confound for between-task correlations. Curious about the potential impact of motivational changes, reward satiety, fatigue, etc., especially if later tasks have lower accuracy or higher variance?
(2) The claim about the common EF factor is limited with only 6 animals.
(3) The claim about phenotypes and clustering is limited as one cluster is a single animal (subject F).
(4) Framing of the AS task as such is inappropriate, as it is a manual response task and is not interchangeable with classic oculomotor AS paradigms.
The authors demonstrate with solid evidence that a multi-task battery can yield repeatable performance metrics across multiple EF-relevant tasks in macaques. The data support the idea that multiple separable dimensions can summarize performance in this cohort and that individuals differ consistently across sessions. However, the claims about the three cognitive phenotypes, four-factor EF architecture, and lack of common EF factor in NHPs are only partially supported given the fixed task order, small sample size, and ambiguity about whether factors reflect stable traits or covariance between sessions. These claims would be strengthened if the authors analyze effects of the fixed task order and time since session start.
-
Reviewer #2 (Public review):
Summary:
The authors set out to characterize how four studied components of executive function (Inhibitory Control, Shifting, Updating, and Working Memory) are organized in rhesus monkeys, using a battery of four touchscreen tasks performed repeatedly across many sessions. They report that individual monkeys fall into distinct "cognitive phenotypes," and that statistical modeling of the task scores reveals four separable latent factors, which they interpret as evidence against a single shared "general" executive function factor in this species. The four latent factors (Shifting/Learning, Exogenous Inhibitory Control, Endogenous Inhibitory Control, and Working Memory Updating) are slightly different from those mentioned in the introduction, which is of interest to refine the definition and assessment of …
Reviewer #2 (Public review):
Summary:
The authors set out to characterize how four studied components of executive function (Inhibitory Control, Shifting, Updating, and Working Memory) are organized in rhesus monkeys, using a battery of four touchscreen tasks performed repeatedly across many sessions. They report that individual monkeys fall into distinct "cognitive phenotypes," and that statistical modeling of the task scores reveals four separable latent factors, which they interpret as evidence against a single shared "general" executive function factor in this species. The four latent factors (Shifting/Learning, Exogenous Inhibitory Control, Endogenous Inhibitory Control, and Working Memory Updating) are slightly different from those mentioned in the introduction, which is of interest to refine the definition and assessment of executive functions in non-human primates.
Strengths:
(1) The multi-task, repeated-measures design is a substantial undertaking and yields a large, richly characterized behavioral dataset from the same animals across many sessions. The use of 4 different cognitive tasks is uncommon in this field and a real asset to the study.
(2) The overall approach of testing the reliability of each behavioral measure before including it in further analysis is a sound and appreciated practice, backed up with overall appropriate data analysis strategies.
(3) The attempt to move beyond single-task designs, which dominate the non-human primate literature, toward a broader profiling of executive sub-functions is conceptually valuable and timely.
Weaknesses:
(1) The four factors introduced early in the paper are not the same four factors that emerge from the final statistical analysis, and this shift is not clearly flagged or explained to the reader, which is likely to cause confusion about what the paper is actually demonstrating. In addition, working memory vs. updating are not clearly defined or distinguished. The introduction and discussion sometimes treat "updating" and "working memory" as separate executive functions, but the literature the authors themselves cite (e.g., Friedman & Miyake, 2017) treats updating as a component nested within working memory rather than a fully independent dimension. This conceptual ambiguity carries through the paper and affects how the reported four-factor structure should be interpreted.
(2) The grouping of very different types of behavioral measures (accuracy-based executive function metrics, reaction-time/processing-speed metrics, and reward/loss motivational-effect metrics) under a single "cognitive metric" label in Figure 2 is potentially misleading, since these are not conceptually equivalent, and the paper later treats them separately in supplementary analyses. This could give readers the impression that all reported metrics are equally informative tests of the four-domain hypothesis, when only a subset actually is.
(3) The rationale for how motivational & valence effects (gains/losses) were analyzed is not well supported. Effects of loss versus gain may reflect risk attitudes or loss-aversion documented in the primate literature, rather than motivation per se, and this alternative explanation is not discussed.
(4) The Discussion would benefit from acknowledging other factors that could influence performance and phenotype membership across sessions (such as age, sex, social rank, or stress) and from clarifying whether the identified cognitive phenotypes are expected to be stable traits or could shift with further training. Additionally, some interpretive statements appear in the Results section rather than the Discussion, which is inconsistent with how results and interpretation are normally separated, and could give undue weight to preliminary interpretations.
(5) Concerns about the reliability of some response-time-based measures. In the Continuous Updating task, reaction times appear to drop unexpectedly at the highest difficulty levels for each animal, which may reflect animals disengaging from the task (e.g., random choice heuristic) rather than a genuine cognitive effect, and could also reflect thinner, unevenly sampled data at these levels rather than a true average. If so, the slope measures derived from these data may need to be recalculated after appropriately filtering the trials. Related to this, individual differences in hand-resting position in front of the screen are acknowledged by the authors as a source of noise in reaction times across tasks, but no correction or normalization for this is applied.
(6) The Supplementary Methods referenced in the text (e.g., for the variance decomposition and additional factor analyses) do not appear to be included anywhere in the submitted materials, only Supplementary Results, making some analyses difficult to fully evaluate. There is no information about the subject age, sex, species, etc.
Achievement of aims and impact:
The authors largely achieve their stated aim of demonstrating that monkeys' executive function performance is reliable and structured in different dimensions rather than guided by a single factor, and the discovery of stable individual "cognitive phenotypes" across sessions is a genuinely interesting and useful contribution. However, the conceptual framework of the study needs to be better clarified (i.e. are working memory and updating truly separable?) and enriched with relevant inputs from the fields of comparative cognition and cognitive ethology in non-human primates and beyond. The conclusions about Response Times (RT) may change after proper or further analysis of these metrics; additional relevant dimensions may then be disentangled by considering the sources of variability in these RT data (i.e., motivation, emotional state). With clarification of these points, this work has the potential to become a useful reference and methodological resource for the field of comparative cognition and non-human primate cognitive neuroscience.
-
