Giordano and DiEuliis’s case for treating cognitive engagement as a decisive domain of irregular warfare is not wrong. The underlying diagnosis — that human prediction, heuristics, and trust are exploitable terrain, and that peer competitors have said so in their own doctrinal language — is sound, and the five recommendations that follow it are the recommendations a serious institution would be expected to produce. Below are the well-meant observations on some points mentioned in the article, the clarification of which would add value to the field.
WHAT THE PAPER ESTABLISHES
Three claims in the piece are genuinely load-bearing, and worth stating plainly before any critique, because the recommendations that follow are only as strong as this foundation. Irregular warfare does exploit predictive cognition — this is well-documented across the surprise and deception literature, not a novel claim. China’s Three Warfares and Russia’s reflexive-control doctrine are real, published, and cited correctly. And the cross-domain point — that cognitive operations do not stop at a battlefield’s edge — is empirically the least contestable sentence in the paper. None of what follows disputes any of this.
WHERE THE PAPER OVERSTATES
Doctrine treated as capability. The paper moves from “China and Russia have articulated doctrines” to “adversaries… are becoming increasingly adept at its exploitation” within the space of one paragraph, without a citation marking the seam. A published doctrine is a declared intention. Adeptness is a demonstrated outcome. Those are different categories of claim, and the paper’s argument for urgency depends on collapsing the distance between them. A reader is never told whether the evidence for adversary effectiveness is an intercepted operational result, an open-source doctrine document, or an inference from the doctrine document alone.
Technological determinism. AI, big-data targeting, and neurotechnology are presented as capabilities moving in one direction — toward adversary advantage — without engaging the empirical literature on how effective influence operations have actually been. Independent analyses of the Internet Research Agency’s campaigns, for instance, have generally found measurable persuasion effects to be modest and short-lived relative to the resources committed. A threat model is stronger, not weaker, for citing that literature rather than omitting it; a paper that only cites capability and never cites measured effect risks mistaking investment for outcome.
Mirrored coherence. The paper ascribes unified, well-coordinated execution to Chinese and Russian cognitive-warfare institutions while simultaneously recommending the United States build a “unified doctrine” because none yet exists here. Both claims may be true, but the paper never considers that adversary bureaucracies may face the same coordination friction it assumes only the U.S. side suffers from. Assuming the other side has already solved the problem you are being asked to fund a solution for is a specific and common failure mode in threat assessment, and this paper does not avoid it.
WHAT THE PAPER OVERLOOKS
Attribution. Every recommendation in the paper — doctrine, resilience training, NEURINT, intelligence integration — presumes that a cognitive operation can be reliably distinguished from organic information-ecosystem noise, domestic political actors, or ordinary human bias. The paper never states how. This is arguably the hardest unsolved problem in the entire field, and it is not a detail; it is the precondition for every downstream recommendation working as intended. Metrics of “perception, sentiment, and behavioral intent,” collected without an attribution standard, do not reduce uncertainty — they generate a larger surface for false positives, at exactly the moment decision-makers are told to trust the metric.
Governance of neurocognitive capability. Recommendation three calls for both defensive and offensive neurocognitive capability, including real-time neuromodulatory intervention on warfighters, with the entire ethical apparatus handled in a single subordinate clause. That is a checkbox, not a design constraint. There is no proposed oversight body, no consent framework for a warfighter subject to “optimization,” no sunset provision, and no acknowledgment of the domain’s own precedent — a serious proposal to build state neurocognitive-modulation capability would be expected to reckon openly with that history rather than pass over it in silence.
Civilian resilience. The paper states that cognitive operations extend to civilian populations, political institutions, and allied networks — then addresses every recommendation to the Department of War alone. By the paper’s own stated threat model, a DoW-only response leaves the larger part of the attack surface it just described uncovered.
Allied legal friction. “Strengthen interagency and allied coordination” does not engage the fact that NEURINT-style behavioral and biometric collection runs directly into allied privacy and human-rights law. Coordination without a harmonization mechanism is an aspiration, at best.
ADDITIONS TO THE RECOMMENDATIONS
- An evidentiary standard for attribution. Before any resilience or response measure is triggered, require a formal confidence tiering — comparable to existing intelligence confidence levels — that separates observed effect, inferred adversary involvement, and confirmed attribution. This should be a precondition on Recommendation 4 (intelligence integration), not an afterthought to it.
- Independent oversight and sunset provisions for neurocognitive programs. A review mechanism analogous to existing SIGINT congressional oversight, an explicit prohibition on domestic scope creep, and a named, revocable consent framework for any warfighter-facing neuromodulation. This should be written into Recommendation 3, not left to a general reference to “ethical and legal considerations.”
- A civilian resilience track outside DoW ownership. Since the paper’s own threat model includes civilian populations and allied networks, resilience education and platform coordination for that population should run through non-military channels, coordinated with but not subordinate to the Department of War.
- A standingred-teamrequirement. Independent, adversarial review of the paper’s own working assumption — that adversary doctrine equals adversary capability — before that assumption is used to justify budget or authority. Doctrine documents should be treated as one input to a capability assessment, not as the assessment.
- A declaratory framework for offensive cognitive engagement. Cyber and nuclear deterrence both operate with escalation thresholds and signaling conventions designed to reduce miscalculation. Offensive cognitive engagement currently has no comparable structure. Recommendation 3’s call for offensive capability should not proceed without one, given how easily cognitive operations could be misread as something more severe than intended — or the reverse.
CLOSING
None of this argues against the paper’s central claim. The cognitive domain is a real and under-institutionalized front, and the recommendation to build doctrine, train for it, and resource it is the correct instinct. But a doctrine built on an uninterrogated assumption about adversary capability, with an attribution problem left unsolved and a governance framework reduced to a clause, is a doctrine that could commit real institutional weight — and, in the neurocognitive case, real human subjects — to a threat model that has not yet earned the confidence the paper asks readers to extend it. The paper does the harder, more valuable work of naming the domain. What it still owes the reader is the discipline of marking, unambiguously, where its evidence ends and where its recommendation begins.
