Cognitive control networks in human and macaque

Reviewer #2 (Public review):

Summary:

Mione et al. aim to resolve a long-standing question in comparative neuroscience: whether the macaque brain contains a functional analogue to the distributed human multiple-demand (MD) network. To address this, they employ a direct cross-species fMRI comparison using a multi-step saccadic maze task in humans and a simplified two-step version in macaques. By contrasting goal-directed navigation against a control condition that requires similar motor responses but no strategic planning, the study isolates the neural signatures of cognitive control across species.

Strengths:

The most compelling aspect of this work is its methodological alignment. Previous attempts to compare these systems often relied on comparisons of human BOLD signals and macaque single-unit recordings. By running parallel fMRI protocols, the authors establish a shared measurement basis that allows for a more direct comparison. The resulting activation maps demonstrate conserved network topology in dorsolateral and dorsomedial frontal cortex and insula. Individual-level data are presented alongside a conjunction map across the two animals, with the threshold used to define common activations stated explicitly, and supplementary analyses restricted to correctly completed trials or using a smaller smoothing kernel leave the main features intact. This strengthens confidence in the frontal correspondences that appear robust.

Weaknesses:

A design asymmetry between the human and monkey tasks remains. In humans, 2-, 4- and 6-step problems were intermixed within blocks, whereas macaques solved only two-step trials. Because participants never know when a harder problem will occur, they may stay in a state of proactive cognitive control (Braver, 2012) throughout the block, which could inflate MD activation even during simpler trials and complicate direct comparison with the monkey data. The authors acknowledge this limitation but do not quantify or correct for the effect.

Individual-level consistency for some claimed MD regions is limited. The two monkeys differ considerably in parietal cortex, and a conjunction analysis shows no common activation there even at a permissive threshold of t > 1.5. Strong dorsal/medial parietal and ventrolateral prefrontal activations were each seen in only one animal, and the abstract presents them accordingly. This is an honest presentation, but it leaves the evidence for these regions resting on one subject, so whether they can be regarded as part of a macaque MD network remains uncertain.

In the Discussion, the authors conclude that results from single animals "tell against" genuine species differences in these regions. The implied reasoning appears to be that if macaques truly lacked parietal and ventrolateral prefrontal nodes, neither animal would have shown activation there. Since one monkey did show it, they take this as evidence that the regions are not genuinely different between species, attributing the other animal's absence to limited power or insufficient task demands. This inference is difficult to justify with only two animals and no common activation across them at any meaningful threshold. A single positive result does not establish that a region reliably belongs to the macaque MD network.

Overall, this manuscript provides solid evidence for functional correspondence between human and macaque cognitive control networks in dorsomedial frontal cortex and insula. Individual-level data are presented transparently, with conjunction thresholds stated explicitly and limitations acknowledged throughout. What remains is largely interpretive. Given only two animals and no cross-animal convergence in parietal or ventrolateral prefrontal regions, claims that these form part of a macaque MD network should be read as provisional.

Author response:

The following is the authors’ response to the previous reviews.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The "multiple-demand" (MD) system is a well-known finding of human brain imaging and is thought to play a central role in cognitive control. To directly compare the MD system in humans and monkeys, Mione et al. used functional magnetic resonance imaging to measure whole brain activation in a multi-step saccadic maze task. In humans, the authors found a distributed pattern of brain activity close match to the canonical MD network and extending to adjacent regions of dorsal attention and other networks. While there was good correspondence between monkey and human data, differences were also notable in lateral frontal cortex, dorsal parietal cortex, and sensorimotor cortex.

Strengths:

Though previous data hint at a corresponding network in the macaque, there has been no direct comparison to human data. This study provides a direct cross-species comparison with whole-brain data of fMRI, and the findings suggest an extended and strongly interconnected brain network recruited by increased cognitive challenge.

Weaknesses:

In previous human imaging, the MD system is defined by overlapping activation for many kinds of cognitive demand. In the present work, however, the authors used just a single task. Although there is some overlap between putative monkey MD network and canonical MD network identified in human imaging, it should be cautious to link current findings to MD system based on limited task events.

We agree that present conclusions are limited by data from a single task. As we note, however, our human data show that this contrast works well for identifying the usual components of the MD network, suggesting it is a suitable candidate for use in the monkey.

Reviewer #2 (Public review):

Summary:

Mione et al. aim to resolve a long-standing question in comparative neuroscience: whether the macaque brain contains a functional analogue to the distributed human multiple-demand (MD) network. To address this, the authors employ a direct cross-species fMRI comparison using a multi-step saccadic maze task in humans and a simplified two-step version in macaques. By contrasting goal-directed navigation against a control condition that requires similar motor responses but no strategic planning, the study isolates the neural signatures of cognitive control across species.

Strengths:

The most compelling aspect of this work is its methodological alignment. Previous attempts to compare these systems often relied on comparisons of human BOLD signals and macaque single-unit recordings. By running parallel fMRI protocols, the authors establish a shared measurement basis that allows for a more direct comparison. The resulting activation maps demonstrate conserved network topology in dorsolateral and dorsomedial frontal cortex and insula.

Weaknesses:

However, there is concerning inter-individual variability in the macaque data, along with a notable design asymmetry between the human and monkey tasks. In the human experiment, 2-, 4-, and 6-step trials were mixed within the same blocks, whereas macaques performed only 2-step problems. This design difference likely places human participants in a state of sustained proactive cognitive control (Braver, 2012), as they must remain prepared for more demanding trials at any moment. This elevated baseline arousal may potentially inflate MD network activation even during the simpler 2-step trials in humans, making direct comparisons with the macaque data difficult.

The authors acknowledge this concern and note that similarities between species survive despite this procedural difference. However, their own individual-level data raise doubts about how robust these "similarities" really are. A defining feature of the MD system is consistent recruitment across individuals performing the same task. In this study, however, the two monkeys show markedly different activation maps in parietal cortex. The conjunction map (Supplementary Figure 6) does not support the claim that lateral and medial parietal regions are commonly activated across both animals at any meaningful statistical threshold, even when using a permissive threshold of |t| > 1.5, corresponding to approximately p ≈ 0.13 - 0.14 (two-tailed).

This suggests that the group-level parietal activations are largely driven by a single animal, raising questions about whether these regions can reliably be considered as part of the monkey MD network. Notably, in Premereur et al. (2018), the only other direct macaque fMRI study of a cognitively demanding task, their conjunction map between the two animals also failed to show common parietal activation during task switching. Taken together, these findings suggest that either the monkey MD network is not as similar to its human counterpart as claimed, or the monkey version of the maze task was simply not sufficiently challenging to fully engage parietal MD nodes.

Because of the design asymmetry described above, it remains difficult to determine which of these possibilities is correct (a genuine species difference versus insufficient task demands in macaques). The authors' current claims treat the group-level activations, which appear largely driven by one animal, as reliable evidence for homology, but without consistent individual-level support, such conclusions remain tentative.

In summary, the authors demonstrate functional correspondence between human and macaque cognitive control networks in dorsolateral and dorsomedial frontal regions. However, assertions that this extends to lateral and medial parietal cortex are not consistently supported at the individual level. While the overall conclusions are plausible, they would be significantly strengthened by demonstrating consistent activation across both animals in all claimed MD regions, ideally with a properly thresholded conjunction map.

These are all good points, and in the revision, we have placed greater emphasis on differences as well as similarities between animals. We have also pointed out the limitations of the conjunction map. In particular, two strong components of the human MD network (vlPFC and parietal cortex) are each seen in only one of the two animals. As we note, limited task difficulty, long training, and limited statistical power may all contribute to negative results. For these regions we agree that between-animal differences limit conclusions, though provisionally, results for single animals tell against genuine species differences.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) Although the whole-brain data obtained by fMRI can provide a direct comparison between species, a single cognitive task might be insufficient to link the findings with MD system. A cognitively challenging task likely activates multiple regions of the MD system; however, it may also recruit some task-specific regions, which do not belong to the canonical MD network.

Though this is a very reasonable concern, the results suggest little influence on the present findings. In the human, strong MD activation was accompanied by activation of the DA network, perhaps reflecting the spatial content of the task. In the monkey, however, there was little evidence of DA activation, likely reflecting limited statistical power.

(2) Based on the current work, the title seems to be too general.

For the above reason, we think the title is reasonable – acknowledging limitations of data from single animals (see below), the cortical and subcortical distribution of monkey activations indeed suggests correspondence to the human MD network.

Reviewer #2 (Recommendations for the authors):

(1) Please clearly state the statistical threshold used for your conjunction analysis in Supplementary Figure 6. The current display appears to rely on |t| > 0, which is not a meaningful criterion: it effectively includes any voxel with positive activation regardless of significance. Even at a more permissive but still unjustified threshold of t = 1.5 (corresponding to roughly p ≈ 0.13-0.14, two-tailed), lateral and posterior medial parietal regions do not appear in the conjunction map. Without a properly justified threshold, it is difficult to accept the claim that these regions are commonly activated across both animals.

The threshold and limitations of the conjunction map are now described (p. 13), and differences between animals given stronger emphasis.

(2) Given that lateral and medial parietal activations do not survive even a low uncorrected threshold in the conjunction analysis and are therefore largely driven by one animal at the individual level, I recommend explicitly acknowledging this limitation in the Discussion rather than treating these regions as reliable components of the macaque MD network. Alternatively, acquiring data from a third monkey would help determine whether this reflects limited statistical power or genuine inter-individual variability.

We have reworded our conclusions concerning parietal activation throughout (beginning with Abstract), emphasising their dependence on data from a single animal.

(3) The design asymmetry between humans and monkeys makes it difficult to interpret cross-species differences in activation. For parietal and ventral prefrontal regions specifically, the monkey version of the task may simply have been insufficiently demanding to fully engage these MD nodes, which might require higher cognitive load before they are reliably recruited. Alternatively, this could reflect a genuine species difference. I recommend discussing this possibility explicitly when interpreting missing or inconsistent activations in these regions.

These possibilities are now considered in a new paragraph of the Discussion (p. 18).

(4) Please specify exactly how Figure 7 was thresholded and state whether FDR correction was applied separately for cortical and subcortical analyses. The colorbar suggests t > 0 as a starting point, but large empty regions within the slices imply that some higher masking or threshold was applied.

See revised figure legend.