Pharmacogenetic manipulation of Locus coeruleus activity in monkeys demonstrates its role in cognitive effort
Curation statements for this article:-
Curated by eLife
eLife Assessment
This valuable study uses chemogenetic inactivation techniques in behaving macaque monkeys to ask whether the locus coeruleus (LC), an area that is involved in the control of arousal, is also involved in regulating access to mental resources for a task that requires working memory. The data show that moderate locus coeruleus inhibition impairs foraging efficiency without affecting response times, total rewards accrued, or overall motivation. However, the evidence supporting the more specific claim that the locus coeruleus regulates cognitive and/or working memory function is currently incomplete, as the experimental design is confounded by elapsed time, lacks formal statistical interaction tests, and relies on behavioral measures that cannot cleanly separate effort allocation from working memory capacity or strategy shifts. These limitations may be ameliorated with further analyses.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
The noradrenergic nucleus locus coeruleus (LC) is involved in numerous cognitive functions. Its activation often enhances sensory and motor performance and its activity correlates with arousal, which altogether suggest a general role in the mobilization of resources for cognition and action. We recently showed its strong and specific implication in physical effort and indirect evidence suggest that it might also be involved in cognitive effort. To address this question directly, we used a pharmacogenetic approach in rhesus macaques to selectively and reversibly inhibit LC neurons in a cognitively challenging task. Two monkeys were injected with viral vectors expressing inhibitory DREADDs (hM4Di) specifically in noradrenergic LC neurons, allowing reversible suppression of LC activity via systemic administration of deschloroclozapine (DCZ, 0.1 mg/kg). A third monkey served as a control and only received DCZ injections. Monkeys performed a simple hole-board task in which they searched for food rewards (raisins) hidden in a 5×5 grid of wells. In the transparent condition, rewards were visible, requiring minimal cognitive effort. In the opaque condition, rewards were hidden, such that monkey had to rely upon working memory to avoid revisiting empty wells. Thus, performance in opaque condition required more cognitive effort. Behavioral analysis showed that LC inhibition had no effect on performance in the transparent condition. However, in the opaque condition, it significantly impaired performance by increasing errors (revisits), without affecting the total number of rewards obtained, response times, or overall motivation. Monkeys compensated the decrease in success rate by performing more trials, indicating reduced efficiency rather than disengagement. These findings demonstrate that the LC plays a critical causal role in mobilizing cognitive resources for a demanding task. This is in line with the idea that the noradrenergic system contributes broadly to cognitive control and effort, extending previous findings on its involvement in physical effort. Overall, the study provides strong evidence linking LC activity to cognitive effort regulation, thereby complementing non-invasive studies in humans.
Article activity feed
-
eLife Assessment
This valuable study uses chemogenetic inactivation techniques in behaving macaque monkeys to ask whether the locus coeruleus (LC), an area that is involved in the control of arousal, is also involved in regulating access to mental resources for a task that requires working memory. The data show that moderate locus coeruleus inhibition impairs foraging efficiency without affecting response times, total rewards accrued, or overall motivation. However, the evidence supporting the more specific claim that the locus coeruleus regulates cognitive and/or working memory function is currently incomplete, as the experimental design is confounded by elapsed time, lacks formal statistical interaction tests, and relies on behavioral measures that cannot cleanly separate effort allocation from working memory capacity or strategy …
eLife Assessment
This valuable study uses chemogenetic inactivation techniques in behaving macaque monkeys to ask whether the locus coeruleus (LC), an area that is involved in the control of arousal, is also involved in regulating access to mental resources for a task that requires working memory. The data show that moderate locus coeruleus inhibition impairs foraging efficiency without affecting response times, total rewards accrued, or overall motivation. However, the evidence supporting the more specific claim that the locus coeruleus regulates cognitive and/or working memory function is currently incomplete, as the experimental design is confounded by elapsed time, lacks formal statistical interaction tests, and relies on behavioral measures that cannot cleanly separate effort allocation from working memory capacity or strategy shifts. These limitations may be ameliorated with further analyses.
-
Reviewer #1 (Public review):
Summary:
This study has a very well-posed question and interesting proposed finding. It proposes that because LC activity has been shown to correlate with arousal, perhaps its activity is related to mobilizing resources for cognition and action. They use chemogenetics to inactivate LC neurons while monkeys perform a foraging task with homogeneous reward options and a working memory component. They use two task conditions, one where the grid of reward wells is visible and one where they are opaque, and the monkey has to open them to retrieve the food inside and remember where he has been. The authors claim that inactivation of LC had a selective effect on the opaque well task by increasing the number of revisits to the same wells without affecting total rewards, response times, or overall motivation, which …
Reviewer #1 (Public review):
Summary:
This study has a very well-posed question and interesting proposed finding. It proposes that because LC activity has been shown to correlate with arousal, perhaps its activity is related to mobilizing resources for cognition and action. They use chemogenetics to inactivate LC neurons while monkeys perform a foraging task with homogeneous reward options and a working memory component. They use two task conditions, one where the grid of reward wells is visible and one where they are opaque, and the monkey has to open them to retrieve the food inside and remember where he has been. The authors claim that inactivation of LC had a selective effect on the opaque well task by increasing the number of revisits to the same wells without affecting total rewards, response times, or overall motivation, which they interpreted as reduced efficiency rather than disengagement. This is a particularly appealing outcome because it implies that LC suppression, which should lower arousal, does not cause disengagement, and rather affects working memory in some fashion. However, there are many issues with this study that make the abstract misleading, in a concerning way. The abstract and introduction are written well, but the data do not support the overall claims, and there are fundamental flaws that aren't easily resolvable.
Strengths:
The bigger question is very interesting and has broad appeal. The writing of the abstract and introduction sets the paper up clearly.
Weaknesses:
(1) The title and abstract claim effort; the paper repeatedly discusses working memory. It does not appear that any metric of effort was measured or manipulated in this study (no pupillometry, no effort-cost model, no offered choice between effort levels).
(2) The authors cite their previous paper as strong evidence of chemogenetic manipulation of LC neurons. They use that paper to justify the dose they use of DCZ, the activator of the chemogenetic construct. That study recorded only 5 neurons at that dose, and 3 of them had a reduction in rate. The statistical significance of that result is questionable (given that I think the authors averaged % change across 3 neurons and it was 15%). The authors treat that as a robust result in multiple places, for example:
"...induced a reliable 15% decrease in the firing of LC neurons..." The only significance test reported in the source paper for that data is t(5.8) = -3.6, p = 0.01, and that test compares low-dose against high-dose neurons. It establishes dose-dependence. It does not establish that 0.1 mg/kg produced a reliable decrease in firing.
"...the dose of DCZ was chosen based on its known influence on LC activity in monkey Jf, i.e. a 15% decrease in firing lasting over 30 minutes...".
This is misleading to readers with regard to the effectiveness of chemogenetics in LC and for creating a foundation for this paper.
(3) The authors seem to undermine their main argument that the opaque task requires working memory at all, and I am not convinced the monkeys were not just operating like a random agent. The random agent model seems to sample uniformly with replacement and an arbitrary termination rule, which is a straw man argument for a model. It does not allow for testing memoryless vs. memory. A monkey executing a systematic scan with short, coherent moves will avoid revisits without using working memory at all. The correct control is an agent matched to the animal's own movement statistics (step-length and turn-coherence distributions) with no memory of visited wells and with memory of visited wells. There's currently no evidence that the monkeys are using working memory in the opaque well task.
(4) The models used are not described sufficiently for interpretation or replication. The methods are brief, and the models seem pulled out of thin air with no reference to other work in foraging or any mathematical construct. If the authors have used these models before, then they should cite them, but the statements "...performance of a random agent is harder to evaluate intuitively...", "random agents have no reason to stop generating actions...", "...the number of trials was set arbitrarily...", and the qualitative comparison of performance "was intermediate between that of the optimal and average random agent..." raised concern as to whether the models were principled at all. The distance model was strange, although fine, but why not use geometry? It is also not clear why a reach would have this kind of spatial relationship when the monkey eats food between visiting locations and does not go from one box to the other without a reach to his mouth and a reach back to the board in between.
(5) The figures in this paper need work; axes are not labeled properly, or have labels that are undefined. Legends are not labeled clearly or described properly in the figure legends to interpret the figures. In multiple cases, the box and whisker plots have too little data, and the edges of the boxes collapse on themselves because IQR doesn't make sense for only 6-7 data points.
(6) The statistics seem questionable; although they used parametric and non-parametric tests, they report results of tests that assume normality for small sample sizes <10. Other choices regarding the trial number to stop counting data (e.g., on the 20th or 23rd) are suspicious and seem arbitrary, not data-driven. Some choices regarding analysis are stated in the text as arbitrary when they could be derived from hypotheses or rigor. The extremely small p-values for the ANOVA result testing performance made me suspect that this was combining all trials across sessions (given there were only 6 or 7 sessions per animal) and there was no repeated measures or nesting to account for repeated sampling within an animal. Sessions within an animal under matched conditions are the natural paired unit. So a paired or mixed-model treatment would be both more powerful and more appropriate.
(7) Lastly, there is no evidence that chemogenetics actually worked in this experiment in a way that tests their hypothesis. In Figure 7, they show their only effects that passed statistical significance (p=0.03 and p=0.028) for two monkeys, comparing across saline (D) and DCZ (D) sessions, and show a small effect on mean performance. I don't think there were corrections for multiple tests across all their monkeys and performances. But that aside, the monkeys' performances were not different from those of their random agents on mean performance, as shown in Figure 5(A). So if they are performing randomly, what would inactivation be affecting?
(8) There seems to be an ordering confound of weeks 1 and 2 being saline and weeks 3 and 4 being entirely DCZ, and only the last 2 weeks are randomized, so for 20 of the 28 sessions per animal, treatment is confounded with time.
(9) There does not appear to be any group x treatment interaction test for any analyses. Claims rest on statements like "significant in monkey 1 but not in monkey 3," etc.
(10) There are duplicated statistics between pages: "The effect reached significance for monkey Jf (T-tests, Jf: t(10.79) = -2.31, p = 0.042), but only approached significance for monkey Js (t(8.89) = 2.04, p = 0.072)." on page 12, "The number of trials performed after the 23rd reward increased under DCZ for monkey Js (T-test, t(10.79)=-2.31, p=0.042) and Jf only showed a non-significant trend (T-test, t(8.89)=-2.04, p=0.072)." on page 13 for a separate section.
-
Reviewer #2 (Public review):
Summary:
The authors investigate the causal role of the LC in mobilizing cognitive resources. Two rhesus macaques expressing TH-dependent inhibitory (hM4Di) DREADDs in the LC, alongside one non-transduced control, performed a spatial foraging task under visible (transparent) and hidden (opaque) reward conditions. The authors report that systemic DCZ (0.1 mg/kg) administration selectively impaired performance in the memory-guided opaque condition in the DREADD animals, concluding that LC activity is causally required for allocating cognitive effort
Strengths:
(1) Achieving cell-type-specific, reversible chemogenetic inhibition in the macaque LC is a substantial technical achievement that successfully bypasses the limitations of systemic noradrenergic pharmacology.
(2) The experimental setup elegantly isolates …
Reviewer #2 (Public review):
Summary:
The authors investigate the causal role of the LC in mobilizing cognitive resources. Two rhesus macaques expressing TH-dependent inhibitory (hM4Di) DREADDs in the LC, alongside one non-transduced control, performed a spatial foraging task under visible (transparent) and hidden (opaque) reward conditions. The authors report that systemic DCZ (0.1 mg/kg) administration selectively impaired performance in the memory-guided opaque condition in the DREADD animals, concluding that LC activity is causally required for allocating cognitive effort
Strengths:
(1) Achieving cell-type-specific, reversible chemogenetic inhibition in the macaque LC is a substantial technical achievement that successfully bypasses the limitations of systemic noradrenergic pharmacology.
(2) The experimental setup elegantly isolates memory-guided action selection from global arousal, sedation, or motivational deficits, providing a useful quantitative constraint for LC functional models.
(3) The introduction and discussion clearly identify the limitations of using indirect pupillometry to assess locus coeruleus activity. The authors thoughtfully frame their targeted manipulation as a critical complement to existing literature without overstating their claims.
(4) Demonstrating that the DREADD animals navigated the board similarly regardless of reward visibility is a valuable null finding. This specific result provides crucial evidence for addressing the memory-versus-effort debate. Elevating this insight from a subordinate clause to a more prominent position will further maximize the paper's impact.
Weaknesses:
(1) The current behavioral metrics-specifically mean inter-choice distance and success rates bounded by a "random agent" baseline-cannot distinguish between a working memory deficit and a fundamental shift in search strategy. Because performance is a many-to-one function of memory quality and search policy, animals could achieve the observed outcomes either by executing a tight local sweep with "leaky" memory for visited wells, or by abandoning tight spatial locality for a dispersed, near-random walk with intact memory. Without a detailed characterization of the animals' actual step-by-step visit sequences and spatial consistency across conditions, the data cannot adjudicate whether LC inhibition degrades an existing habitual policy (a memory/maintenance deficit) or impairs the adoption of a structured policy (an effort-mobilization/task-set deficit). Clarifying this is critical, as the two interpretations engage distinctly different LC-prefrontal mechanisms.
(2) The analyses of within-session behaviors rely on arbitrary reward-inclusion thresholds (e.g., 20 or 23 rewards) and censoring rules that obscure the temporal dynamics of the task. Because wells are depleted without replacement, the difficulty and working memory burden rise steeply as the board empties. A constant, time-invariant memory deficit could readily produce the observed early vs. late performance pattern, as the detectable cost of forgetting accrues heavily after the 20th reward. Furthermore, early-session failures to maintain an effective policy naturally lead to continued responding after depletion. However, the imposed time caps compress variance exactly where this within-session causal chain would be most evident, leaving the claimed independence of the early and late deficits unsupported.
(3) The experimental timeline heavily confounds chemogenetic manipulation with elapsed time. Test weeks 1-2 were exclusively saline, and weeks 3-4 were exclusively DCZ, with treatment alternation only occurring in weeks 5-6. Consequently, the critical contrast between opaque saline and DCZ conditions is largely a between-block comparison across a month of testing. Plausible time-varying factors-such as habituation, declining novelty of a daily home-cage task, and accumulating satiety-all directionally align with the reported DCZ impairment, undermining the isolation of the chemogenetic effect.
-
-
-
Pharmacogenetic manipulation of Locus coeruleus activity in monkeys demonstrates its role in cognitive effort https://www.biorxiv.org/content/10.64898/2026.07.23.740243v1
-
Pharmacogenetic manipulation of Locus coeruleus activity in monkeys demonstrates its role in cognitive effort https://www.biorxiv.org/content/10.64898/2026.07.23.740243v1
-
