Ethics and Information Technology (2023) 25:46

https://doi.org/10.1007/s10676-023-09717-7

ORIGINAL PAPER

Cognitive warfare: an ethical analysis

Seumas Miller1,2,3

Accepted: 7 August 2023

© The Author(s) 2023

Abstract
This article characterises the nature of cognitive warfare and its use of disinformation and computational propaganda and
its political and military purposes in war and in conflict short of war. It discusses both defensive and offensive measures to
counter cognitive warfare and, in particular, measures that comply with relevant moral principles.

Keywords Cognitive warfare · Liberal democracy · Necessity · Proportionality · Freedom of communication ·

Disinformation · Computational propaganda

Characterising cognitive warfare neurophysiological techniques, such as transcranial direct

cranial stimulation (Bernal et al., 2020, p. 32; DeFranco
Cognitive warfare has been defined in various ways. Here et al., 2020), and its aim of destabilising institutions, espe-
are a couple of influential definitions to give the flavour of cially governments, albeit often indirectly by way of initially
what is meant by this term: “Cognitive Warfare is a strategy destabilising epistemic institutions, such as news media
that focuses on altering how a target population thinks and organisations and universities. Importantly, cognitive war-
through that how it acts” (Backes & Swab, 2019); “the weap- fare has been able to harness the new channels of public
onization of public opinion, by an external entity, for the communication, such as social media, upon which popula-
purpose of (1) influencing public and governmental policy tions have become increasingly reliant.1 Moreover, in some
and (2) destabilizing public institutions” (Bernal et al., 2020, contrast with traditional ideological contestation, e.g., the
p. 10). ideological conflict between the Soviet Union and the West
Accordingly, cognitive warfare is a recent development during the Cold War, in which each of the protagonists have
that has emerged from prior related non-kinetic forms of a system or quasi-system of ideas to try to ‘sell’, cognitive
warfare, such as PsyOps operations and Information War- warfare also has a very strong initial focus on sowing divi-
fare. In doing so it has relied heavily on new communica- sion and undermining cooperation in its target population by
tion and information technologies, notably AI. Key features emphasising existing differences and promoting polarising
of cognitive warfare include its targeting of entire popula- views, e.g., promoting both extreme left-wing and extreme
tions (as opposed to, for instance, merely military ones in right-wing views. In short, cognitive warfare makes heavy
wartime), its focus on changing a population’s behaviour by use of computational propaganda.
way of changing its way of thinking rather than merely by As is by now well-known, the advent of social media
the provision of discrete bits of false information in respect platforms and the associated cybertechnologies, such as
of specific issues (e.g., denying the extent of casualties in algorithms and automated software (e.g., bots that mimic
a kinetic war), its reliance on increasingly sophisticated real people), has brought with it an exponential increase in
psychological techniques of manipulation (and, potentially, the spread of disinformation, misinformation, conspiracy
theories, hate speech and propaganda on the part of a wide
array of actors (Cocking & van den Hoven, 2018), including
* Seumas Miller
1
[email protected] These developments have led some to conclude that the cognitive
domain is a new domain of operations in war. While information war-
1
Charles Sturt University, Canberra, ACT, Australia fare, Psyops and the like are hardly new, developments in technology
2 might, arguably, have greatly increased the importance of operations
University of Oxford, Oxford, UK
in the cognitive domain (Cao et al., 2021; MacDonald & Ratcliffe,
3
TU Delft, Delft, The Netherlands 2023).

13
Vol.:(0123456789)
 46 Page 2 of 10 S. Miller

individual citizens, single-issue pressure groups, right-wing ICT hardware itself). Fourth, there is institutional damage or
and left-wing extremist groups, terrorist groups, criminal harm3; that is, the undermining of institutional processes and
organisations, and, in some cases, such as Russia, govern- purposes, for example major breaches of confidentiality in a
ments. Following Woolley and Howard (2019, pp. 4–5), we security agency, loss of institutional control of territory. In
will refer to this latter phenomenon, in so far as it is under- this connection it should be noted that undermining specific
taken in the service of political agendas, as computational institutional process and purposes can be undertaken with a
propaganda. A particular feature of computational propa- view to undermining the institution itself, especially if the
ganda is its contribution to the generation of echo chambers beliefs and attitudes of the institutional actors themselves or
in which users are exposed to information that reinforces those they serve are targeted, e.g., if their trust in the institu-
their own point of view. Thus, social media algorithms tion is eroded as, for instance, occurred in the case of US
adjust the content that users are exposed to thereby creat- electoral institutions in the 2020 Presidential election. The
ing filter bubbles. As a result, the individual user is isolated main focus of cognitive warfare is on the first kind of harm,
from a wide spectrum of views and is exposed principally and more specifically psychological harm, and the fourth
to users with similar views to their own. This strengthens kind of harm, namely institutional harm or damage.
the user’s views at the expense of competing views and of In light of this are we to understand cognitive warfare as
information which might challenge the users’ view, thereby war, as a species of conflict short of war or as covert opera-
leading to an increase in entrenched ‘hard-shelled’ perspec- tions (or some combination thereof)?
tives that are not open to revision. The result is a weakening
of evidence-based discussions and a polarization of politi-
cal discourse that facilitates unevidenced extremist views Cognitive warfare: war, conflict short of war
(D’Alessio, 2021). and covert operations
We need to distinguish cognitive warfare from the (some-
times overlapping) categories of cyberwar, cyber conflict The first point to be made here is that these above-mentioned
short of war, cyber terrorism, cybercrime, cyber espionage two kinds of harm (psychological and institutional damage/
and what we refer to as covert cognitive warfare—a spe- harm) characteristic of cognitive warfare while not them-
cies of covert operations (Miller, 2016a; Miller & Bosso- selves typically thought of as definitive of war might, at least
maier, 2023). While the category of cybercrime is now well- in theory, have a threshold at which the term “war” might
established in law some of the other categories are not or, at be appropriately applied. Relatedly, these two kinds of harm
least, it is controversial whether they have been satisfactorily might have a threshold at which waging kinetic war might
worked out in detail.2 Specifically, there is a problem, or set be morally justified. Moreover, the threshold of psychologi-
of problems, in relation to the concept of war as it might, or cal or institutional harm/damage definitive of war might be
might not, apply to cyber-based conflict, including cogni- able to be attained even if the level of the other kinds of
tive warfare. harm/damage caused (i.e., the level of physical harm caused
In relation to the distinction between these different cat- to humans per se and the level of destruction of physical
egories we need to distinguish four kinds of harm or dam- property and the like) did not constitute war. Likewise, the
age. First, there is harm (physical or psychological) done to threshold of psychological or institutional harm/damage jus-
human beings per se. Here psychological harming is to be tifying war might be attained, even if the level of the other
understood broadly so as to include deceptive or manipu- kinds of harm/damage caused did not.4 Moreover, psycho-
lative inducement of false beliefs or unwarranted affective logical and institutional harm/damage might have thresholds
attitudes with a view to undermining self-mastery. Second, at which a seriously destructive or harmful response short of
there is damage done to buildings, ICT hardware and other war is morally, and perhaps legally, justified. Such responses
human artefacts (as well as to the natural environment in might include economic sanctions and the like; but they
so far as it supports individual and collective human life). might also include various forms of covert political action,
Third, there is, as Dipert (2010: 384) notes, cyber ‘harm’ notably covert cognitive warfare (of which more below).
(or rather ‘soft damage’ in our terminology), for example
damage to software and data (as opposed to the physical

2
The Tallinn Manual is a recent attempt to define cyberwar ade- Footnote 3 (continued)
quately (Schmitt, 2013). However, whether it has succeeded or not
on whether the human beings in question are harmed (and, if so, they
is controversial. See, for instance, Galliott (2019), Gross and Meisels
would be harmed qua members of the institutions in question).
(2017), Lucas (2017). 4
3 Or at least that cyber-harm and/or institutional harm could conceiv-
Since institutions are constituted by roles that are occupied by
ably reach such a threshold independently to some extent of the first
human beings they can be damaged or harmed (or both) depending
two kinds of harm.

13
Cognitive warfare: an ethical analysis Page 3 of 10 46

Some have claimed that cyberwar is a distinct new cat- nor has it resulted taking and holding territory. Moreover,
egory of war sitting alongside conventional war and nuclear cognitive warfare has not thus far resulted in the undermin-
war in particular. By analogy, it can be claimed that cogni- ing of institutions to the point at which the political order of
tive war is a distinct new category of war, albeit one that a nation state has been overthrown.5 So perhaps cognitive
evidently would overlap with cyber war given the nature of warfare (and cyber-based conflict more generally (Miller,
its cyber-based ‘attacks’. However, both claims are question- 2019; Miller & Bossomaier, 2023) is more appropriately
able. Roughly speaking, conventional war is held to neces- regarded as a species of conflict short of war (Galliott, 2019;
sarily involve ‘killing people and breaking things’ in the Gross & Meisels, 2017; May, 2017)—or as an ancillary
service of taking and holding territory (ultimately one’s own means of fighting a conventional war. Aside from its non-
territory in the case of a war of self-defence). However, nei- kinetic character, cognitive warfare often occurs in what are
ther cyber-conflict nor cognitive warfare necessarily involve acknowledged on all hands to be peacetime conditions, e.g.,
either of these things. But perhaps cyber war is a species Russian interference in the 2020 US Presidential election.
of cyber conflict involving organised groups engaged in an Moreover, many instances of cognitive warfare might be
ongoing series of cyberattacks in which there is massive appropriately regarded as species of covert operations. Let
destruction of critical infrastructure leading to large-scale us consider this suggestion.
loss of life, e.g., one of many cyberattacks destroys physical One problem in relation to cognitive warfare engaged in
components of an electricity power grid in the middle of by nation-states against other nation-states is the so-called
winter indirectly leading to numerous deaths. By analogy, problem of attribution; a problem also identified in relation
perhaps cognitive war is a species of conflict in cyberspace to cyber attacks, albeit developments in cyber forensics are
in which organised groups engage in an ongoing program evidently mitigating this problem (Lucas, 2013, p. 371;
of disinformation, propaganda and the use of manipulative Office of the Director of National Intelligence, 2018; Rowe,
techniques to control on-line discourse and discredit political 2013, p. 401). Unlike most attacks in conventional wars or,
opponents (including by destroying their reputations with for that matter, conventional crimes of assault or theft there
unfounded claims) and the profile-based, micro-targeting is a major epistemic problem in relation to such hostile activ-
of vulnerable groups (e.g., mentally-disturbed individuals) ity: the problem of reliably attributing responsibility and,
that undermines political institutions resulting, potentially, conversely, the credibility of denial of responsibility on the
in widespread violent insurrections and the collapse of the part of culpable aggressors (at least, if these attacks are not
existing political order. undertaken as part of a conventional war—since in the latter
In addition, of course, conventional war in contemporary case they might not be denied). Because actors in cyberspace
settings uses cyber weapons and, more generally, has an are densely interconnected by indirect pathways, it is often
important cyber dimension. Consider, for instance, the Feb- extremely difficult to pinpoint the source of such hostile
ruary 2022 Russian invasion of the Ukraine. It has involved cognitive activity or even to know that it is not simply the
a wide range of cyberattacks, including on Ukraine’s banks expression of ordinary citizens engaging in political com-
and government departments (Alazab, 2022). However, munication, albeit communication that is ill-informed and
arguably, the cyber dimension in an otherwise conventional permeated by ideology.
kinetic war would have to become the dominant dimension The existence of the ‘problem’ of attribution and, as a
in order for the war to be reasonably described as a cyber consequence, the credibility of denial, taken in conjunction
war. Moreover, conventional war in contemporary settings, with a commitment to freedom of communication on the
including the current war being waged by Russia against the part of liberal democratic states being targeted makes cog-
Ukraine, has an important cognitive warfare dimension. By nitive warfare an extremely useful strategy for authoritar-
parity of reasoning, arguably, the cognitive warfare dimen- ian nation-states seeking to undermine liberal democratic
sion in an otherwise conventional kinetic war would have to states while avoiding outright war (indeed, avoiding the use
become the dominant dimension in order for the war to be of lethal force or even coercive force). Nation-states respon-
reasonably described as a cognitive war. sible for cognitive warfare are typically engaged in the age-
However, arguably, cognitive warfare has not, at least thus old strategy of covert operations, sometimes referred to as
far, risen to the threshold of conflict reasonably characterised covert political operations (Johnson, 2021). Historically, the
as war; rather it has consisted in activity that is more aptly tactics deployed in covert political operations have included
characterised as conflict short of war (as opposed to force
short of war). Certainly, cognitive warfare has not in fact
resulted in large-scale ‘killing people and breaking things’
5
(even if it potentially could have done so, albeit indirectly); Although it has been argued that information warfare and related
earlier forms of ‘cognitive’ warfare have achieved this, such as in the
overthrowing of the Allende government in Chile. See Bernal et al.
(2020, p. 17).

13
 46 Page 4 of 10 S. Miller

assassination of the political leaders of such ‘enemy’ states, harm is grossly disproportionate to their culpability; and/or
targeted killing of terrorist leaders outside theatres of war, (3) violates a morally justified law.8 Paradigm instances of
the financing of coup d’etats and other insurrectionary dirty hands action are the torture of terrorist suspects to gain
movements, but also destabilizing ‘enemy’ states by spread- information and unlawful cyber-attacks on foreign govern-
ing disinformation and propaganda, deploying agent provo- ments’ suspected weapons installations in peacetime, such as
cateurs and so on (Perry, 2009). the Stuxnet attack on the Iranian nuclear facility. Notice that
Covert political operations are typically, but perhaps not in dirty hands scenarios the ‘dirty’ action might or might not
necessarily, unlawful, at least in the nation-state against be morally justified, all things considered. Either way, the
which they are directed, if not in international law. This ‘dirty’ action is pro tanto a legal9 or moral wrong and the
is one reason why they are not conducted openly albeit, person seriously harmed has been wronged, at least by virtue
arguably, not the main reason at least in the case of covert of having his or her legal rights violated.10 Indeed, this being
political operations conducted in peacetime. Covert political so, dirty hands actions are typically unlawful. This being so,
operations outside war, while they may involve killings and an important question arises as to how those who engage in
the destruction of property are typically designed to stop covert political action in a liberal democracy are to be held
short of war or, at least, short of kinetic war; the whole point accountable (Regan & Poole, 2021).
of such covert political operations is to weaken an enemy Here is it important to distinguish dirty hands actions
state, or defend oneself from being weakened, while plausi- from lawful and morally justifiable but, nevertheless, harm-
bly denying that one is doing so, thereby averting outright ful actions. Presumably, the lethal and other harmful actions
(kinetic) war. It is, therefore, no accident that during the of soldiers in wartime, in so far as they comply with Just
Cold War in the shadow of nuclear war, the covert political War Theory (both the jus ad bellum and the jus in bello) are
operation was a favoured tactic of both the Soviet Union and not instances of dirty hands actions.11 Nor are the harmful
the US or that it has been favoured by Russia in its aggres- actions of police officers, (e.g., the use of coercive force to
sive stance toward the US, e.g., as the recent interference in effect an arrest), instances of dirty hands in so far as they
US elections utilising Cambridge Analytica demonstrates. comply with legally enshrined, community accepted, objec-
The most appropriate moral category, or general descrip- tively correct, moral principles (Miller, 2016a, 2016c).
tion in the philosophical tradition, under which to file most6 If this is correct then covert political action and, there-
covert political actions and, therefore, many, if not most, fore, covert cognitive warfare poses particular challenges,
covert cognitive warfare is, we suggest, that of so-called both for the standard Law Enforcement model and for Just
dirty hands.7 Covert political action is typically a para- War Theory. On the one hand, covert cognitive warfare is
digm of dirty hands (although obviously many instances (more or less) by definition harmful action short of war; its
of dirty hands actions are not instances of covert political raison d’être is typically to harm an ‘enemy’ state without
action); doing what is pro tanto morally wrong (and, typi- triggering war and, especially, in the case of nuclear powers,
cally, unlawful) in order to achieve some putative greater to avoid triggering nuclear war. Moreover, its remit in terms
moral good and, in the case of covert political action, includ- of national security might be somewhat wider than that of
ing covert cognitive warfare, the greater moral good (it is national defence understood in terms of the territorial integ-
assumed) of the relevant nation-state. This greater moral rity and political independence of the nation-state. So the
good of the nation-state is presumably it’s nation security
(as opposed to, for instance, its national interest which
might in some instance not be a good, objectively speaking,
e.g., subjugation of a foreign country). The pro tanto moral
9
wrongness of a dirty hands action typically consists in the And the law in question is a law that ought to exist, e.g., the ‘dirty’
action is a violation of sovereignty and sovereignty is morally desir-
fact that the action either: (1) deliberately inflicts serious able.
harm on an innocent person or persons; or (2) deliberately 10
So this person did not consent to being harmed; nor is it a harm
inflicts serious harm on a culpable person or persons, but the of a kind and degree that the person could reasonably be expected to
suffer in order to realise the greater good to which it is an effective
and necessary means, e.g. as in the case of the use of coercive force
6
by police to arrest a suspect who later turns out to be innocent. More-
Albeit not all; not, for example, the 1981 US covert operation to over, the action was not in the person’s interest all things considered.
rescue the US diplomats and other US citizens held hostage by Iran— See Greenwald (2014) for an account of ‘dirty hands’ actions under-
its breach of Iranian sovereignty notwithstanding. taken, he alleges, by the western intelligence agencies.
7
For an influential treatment see Walzer (1973). 11
Arguably, combatants on both sides are governed by a particularist
8
Roughly speaking, a morally justified law is one that is promul- principle of reciprocity according to which each combatant of State
gated by a legitimate legislature in a procedurally correct manner and A is entitled to use lethal force against each combatant of State B,
is not morally unacceptable, e.g., by virtue of violating a fundamental on condition each combatant of State B is entitled to use lethal force
moral right. against each combatant of State A (Miller 2016a, 2016b, 2023).

13
Cognitive warfare: an ethical analysis Page 5 of 10 46

application of Just War Theory is somewhat inappropriate; which is explicitly or implicitly expressive of the politi-
it largely misses its mark. cal ideology of the communicator, (e.g., extremist jihad-
On the other hand, covert cognitive warfare is (more ist ideology communicated by members of Islamic State,
or less) by definition unlawful (at least in the nation-state right wing Russian nationalism communicated by Russian
against which it is directed). Accordingly, there is a strong state officials, the ideology of the Chinese Communist Party
moral presumption against its use. Yet, for reasons elabo- communicated by Chinese state officials), and, on the other
rated below, it does seem morally justified on some occa- hand, computational propaganda the content of which is not
sions and in some areas, for example the reciprocal targeting thus expressive, e.g., antivaxxer conspiracy theories or right
by liberal democratic security agencies of culpable authori- wing US nationalist quasi ideology communicated by Rus-
tarian state actors engaged in unjustified cognitive warfare sian state officials to US audiences to sow discord in the US.
(of which more in the following sections). Moreover (obvi- The challenges posed by the advent of cognitive warfare
ously) its raison d’etre is not the enforcement of the law, as are considerable, not the least for liberal democracies com-
in the case of police work conducted by law enforcement mitted to ethical or moral (we use these terms interchange-
agencies. So the application of the Law Enforcement model ably) values and principles, such as freedom of communica-
leaves the problem largely untouched; the problem being tion, democratic processes, the rule of law, evidence-based
the apparent moral justifiability of many instances of covert truth telling, and so on. Thus, while there is a need to curtail
political action and, therefore, of covert cognitive warfare, disinformation, nevertheless, there is a requirement that this
notwithstanding their unlawfulness and their inconsistency be done without undermining freedom of communication.
or, at least, incongruence with law enforcement activity. Again, there is a need to combat states engaged in cogni-
tive warfare, but it is problematic for a liberal democratic
state to do so by spreading its own self-serving disinforma-
Countering cognitive warfare tion or by seeking to manipulate citizens of authoritarian
states. A further issue pertains to responsibility. Given the
Cognitive warfare is likely to be more successful in the con- nature of cognitive warfare, there is a need for a variety
text of the already destabilising effects of war, economic of institutions, other than merely governments and security
depression, pandemics and other disasters or in a context of agencies, to shoulder responsibilities for combating cogni-
a pre-existing polarised society, e.g., the UK in the context tive warfare, e.g., to shoulder responsibilities for building
of Brexit, the US in the aftermath of the Global Financial resilience to disinformation, ideology and the use of manipu-
Crisis or the Middle East in the context of the Israel/Arab lative techniques. What precisely are these responsibilities
conflict. Hence Russia and China seized upon the opportu- and to which institutions ought they be allocated? Speaking
nity of the COVID pandemic to increase their operations in generally, we suggest that there is a collective responsibil-
cognitive warfare, e.g., to promote various conspiracy theo- ity (understood as joint responsibility (Miller, 2006, 2016b,
ries in the US population. Again, Russia infamously utilised Ch. 5)) on the part of multiple institutions (or, at least, the
Cambridge Analytica to sow discord in the US Presidential members thereof) including government, security agencies,
elections. Moreover, terrorist groups, such as Al Qaeda and media organisations and institutions of learning such as
Islamic State, have utilised cognitive warfare techniques to schools and universities.
recruit disaffected youths in various liberal democratic and Elsewhere we have proposed a raft of countermeasures
authoritarian states to their cause and, importantly, to sow to combat computational propaganda (Miller, 2020; Miller
discord by getting their ‘enemies’ to overreact, as in the case & Bossomaier, 2023). These included the following ones:
of the 9/11 bombing of the Twin Towers which proved to be
a spectacular success for Al Qaeda in terms of its visibility, • Government to enact legislation to hold mass social
prestige among disaffected Muslims and so on. media platforms, such as Facebook and Twitter, legally
It is important to understand that cognitive warfare is tak- liable for illegal content, such as incitement and hate
ing place in pre-existing social, institutional and technologi- speech, on their platforms.
cal contexts in which there have already been destabilising • Mandatory licensing of mass social media social plat-
effects arising from a proliferation on a massive scale of forms to be introduced with the licences to be held con-
disinformation, misinformation, conspiracy theories, propa- ditionally on the content on their platforms being com-
ganda, hate speech and so on, much of which has not been pliant with the minimum epistemic and moral standards
done in the service of an explicit political purpose (though determined and adjudicated by an independent statutory
it may have serviced such a purpose inadvertently). authority established by government, e.g., the Australian
We also need to distinguish between, on the one hand, Office of e-Safety Commissioner.
computational propaganda (e.g., disinformation, ideology/ • Lawful content which, nevertheless, fails to meet these
quasi-ideology/groupthink and hate speech) the content of minimum epistemic and moral standards, (e.g., by virtue

13
 46 Page 6 of 10 S. Miller

of being demonstrably false), and which is significantly Secondly, there is macro-level, profile-based, individu-
artificially (e.g., by means of bots) or otherwise illegit- ally targeted, speech to millions via mass media channels of
imately amplified, is to be liable to removal by social ostensibly private communication. This macro-level speech
media platforms, but only in accordance with the (pub- might involve the use of bots to send millions of emails to
licly transparent) adjudications of the above-mentioned selected individuals who are not necessarily aware that the
independent statutory authority. same communications are being sent to millions of recipi-
• Account holders with mass social media platforms are to ents and being sent (at least initially) from a single source).
be legally required to be registered with the independent This form of macro-level speech is favoured by computa-
statutory authority which will then issue a unique identi- tional propagandists, such as Cambridge Analytica.13
fier but only after verifying the identity of the account Clearly, as argued elsewhere (Miller, 2020; Miller & Bos-
holder, e.g., by means of his or her passport, driver’s somaier, 2023), there is no moral right to engage in macro-
licence and the like. level, profile-based, micro-targeted, speech to millions via
• Communicators of politically significant content (includ- mass media channels of ostensibly private communication.
ing, but not restricted to, content with national security Indeed, quite the reverse; there is a moral obligation on the
implications) on mass media channels of public commu- part of governments to combat such speech (including by
nication who have very large audiences, e.g., greater than recourse to the means we summarised above). However, it
100,000 followers, to be legally required to be publicly will also turn out that there is no moral right on the part of
identified (other things being equal). foreigners to engage in macro-level socially-directed speech
to the domestic citizenry and this has implications for ban-
These measures are all relevant to cognitive warfare. ning, for instance, Russian mass media channels, such as
However, they are not sufficient to combat a hostile state Russia Today. Accordingly, we are providing the justification
engaged in cognitive warfare (and, for that matter, prob- for a policy advocated by David Sloss; namely, the banning
ably not sufficient, absent some redesign of epistemic insti- of Russia Today and like mass media outlets (Sloss, 2022).
tutions, to combat computational propaganda in other set- Before doing so we need to get clearer on the notion of
tings). What more needs to be said about measures to be socially-directed speech (Miller, 1994, 2001, 2010); a form
implemented in liberal democracies to combat a hostile state of public communication.
engaged in cognitive warfare, such as in the case of Rus- Socially-directed speech is speech in which the speaker
sia’s computational propaganda campaign directed at the speaks to the rest-of-the-community qua member of that
Ukraine, and China’s directed at Taiwan? community (and does so publicly in our above-discussed
Here we need to distinguish micro-level interpersonal sense). Here the community is to be loosely understood
speech, (e.g., John Brown speaking to Mary Smith on a as a social group. So it could be a small local community
street corner) from macro-level speech utilising mass media or a large national, or even international, community; and
channels of communication. Here we also need to distin- it could be an academic, business or political community
guish two forms of such macro-level speech. Firstly, there is (to name but a few instances of social groups in our loose
macro-level socially-directed speech to a very large audience sense of that term). Examples of socially-directed speech
via mass media channels of public communication. Exam- include the UK Prime Minister making a national address,
ples of this would be CNN news broadcasts and former US Dr Anthony Fauci appearing on CNN to say to members of
President Donald Trump communications on Twitter. Such the US population that they ought to get vaccinated, and the
communications reach audiences numbered in the millions mother of a black man slain by local city police pleading for
and they emanate from a single known source known to the non-violent demonstrations in her city by way of response.
members of the audience. Moreover, importantly, these com- What of a supposed moral right to engage in socially-
munications are public in the sense that all of the above directed speech to millions via mass media channels
information is a matter of mutual knowledge12 to the com- of public communication, i.e. to engage in macro-level
municators and to the members of the audience. Thus, each
individual communicator and audience member knows who
13
the source is, what the communicative content is, and knows There are other, i.e., other than the two distinguished here, more
that everyone else in the audience knows this, and knows subtle forms of macro-level communication that utilize mass media
channels of public communication to communicate propaganda, such
that everyone else knows this, and so on. as the so-called content farms favoured by China. These can consist
of websites appealing to, for instance, a religious group known to
have a large following in China’s main propaganda target, Taiwan.
These sites offer a wealth of useful, factual information to the reli-
12
The concept of mutual or common knowledge has been analysed gious adherents in question. However, Chinese ideology and selected
extensively in the philosophical literature. See, for instance, Smith facts are always embedded in the content of these websites See Hung
(1982). and Hung (2020, p. 7).

13
Cognitive warfare: an ethical analysis Page 7 of 10 46

socially-directed speech? There is, at least in principle, a media channels of public communication. Thus, US citizens
moral right of citizen, A, qua member of A’s political com- have a joint right to listen to Russian state actors on Russia
munity to speak to the-rest-of A’s political community. This Today. Notice that being a joint right it would be jointly
is a liberty right in that if one person is exercising it at one exercised; that is, no single citizen acting alone has such a
time then others may not be able to and, indeed, it may be right. However, this joint right carries with it the joint right
that not everyone can exercise this right even over a reason- not to do so. Thus, US citizens have a joint moral right to
ably lengthy period of time; there are just too many citizens ban foreign state actors from using mass media channels
for this to be possible. More specifically, in modern mass of public communication, including social media, to pub-
societies the exercise of this liberty right requires access to licly communicate politically significant messages to them
mass media channels of public communication. But whereas i.e., to US citizens. As is the case with other joint rights of
mass media channels enable mass audiences and everyone members of the citizenry, this joint right can be exercised on
can be a member of a mass audience, they do not enable behalf of the citizenry by their democratically elected repre-
mass speakers to those mass audiences. It is not possible, sentatives. In short, a liberal democratic government, such
even in principle, for everyone, or even a majority of the as the US government, has a moral right to ban foreign state
population, to reach a mass audience. Only a few can be actor from using mass media channels of communication
mass communicators; there are too many citizens and too to publicly communicate politically significant messages to
few channels of public communication for everyone to be a the citizens of the liberal democracy in questions and may
mass communicator. Accordingly, here as elsewhere, there have a moral obligation to do so if, for instance, the commu-
is a need for a fair procedure to govern this liberty right; a nications in question consist in computational propaganda.
fair procedure that might be difficult to find. However, in Indeed, if the foreign state in question is engaged in cogni-
the case of a foreign state actor seeking to communicate tive warfare then there is a clear moral obligation to institute
to a domestic audience other than its own there is no need such bans. Accordingly, we agree with Sloss (2022, Ch. 6)
to identify such a fair procedure since such a foreign actor that Russian and China state actors’ accounts with Facebook,
does not possess the liberty right in question. Thus, Russian Twitter and other ‘big tech’ should be revoked, given that
state actors (and Russians citizens more generally), do not these actors have engaged in cognitive warfare with liberal
have a moral right (specifically, a liberty right) to engage in democratic states and, specifically, have engaged in compu-
macro-level communication on politically significant matters tational propaganda campaigns aimed at undermining key
to US citizens. Likewise, US state actors (and US citizens institutions in liberal democratic states, such as the US and
more generally) do not have such a liberty right to engage in Taiwan.
macro-level communication on politically significant matters It is important to note that this above-mentioned joint
to US citizens. moral right with respect to macro-level, socially-directed,
Naturally, foreign actors do not have a right to engage in politically significant speech is consistent with the micro-
socially-directed communications to members of a domestic level interpersonal right of each member of a community to
audience other than their own. After all, they cannot engage listen to foreign state actors via channels of communication
in socially-directed action as is it defined above, given they that are not mass media channels of public communication.
are not members of the relevant community. However, it Thus, the bans mentioned above would not apply to micro-
might be suggested that, nevertheless, foreign state actors level communications by Russian citizens based in Rus-
have a less stringent (less stringent than the right to engage sia to US citizens based in the US. On the other hand, this
in socially-directed communications to members of their micro-level interpersonal right is not an absolute right. As
own domestic audience) liberty moral right to use channels with most, if not all, moral rights it can be overridden under
of mass communication to publicly communicate to mem- certain conditions. However, it is essentially the fundamen-
bers of a domestic audience other than their own. The exer- tal natural moral right of human beings to engage in free
cise of such a macro-level moral right of foreign state actors speech and, as such, there is a strong presumption against
(e.g., Russian state actors), supposing it exists, would be infringing it; a presumption that can only be overridden by
dependent on members of the domestic audience in question specific weighty moral considerations and not, for instance,
(e.g., US citizens) being prepared to listen to the commu- by blanket appeals to national security.
nications in question; that is, the US citizens have no moral
obligation to listen. Here we need to invoke the concept of
a joint right once again. Cognitive warfare: offensive measures
Consistent with the above, let us assume that there is a
joint moral right of members of a political community qua Thus far we have concerned ourselves with defensive meas-
members of that community to listen to speakers who do ures against cognitive warfare. It is now time to turn to a
not have a right to socially-directed speech to them via mass consideration of offensive measures. Naturally, in an overall

13
 46 Page 8 of 10 S. Miller

context of self-defence, non-kinetic offensive measures influence and offer no material support. Moreover, in doing
against attackers are justified (supposing they are likely to so they might themselves be unknowing victims of the disin-
be effective) by a principle of reciprocity (Miller, 2016a, formation and manipulative propaganda of the authoritarian
2016b, 2016c; Miller & Bossomaier, 2023). state in question. Given that they are victims in this sense,
Let us assume that the offensive measures in question perhaps they are not really culpable. But, if so, how are they
that are non-kinetic. If so, and if these are directed at culpa- to be distinguished in practice from fellow citizens who dif-
ble attackers then it might be thought that there are few, if fer only in that they are fully aware of the techniques of
any, restrictions (other than the likelihood of effectiveness disinformation and manipulative propaganda being deployed
and, perhaps, of compliance with a principle of reciproc- by their security agencies and verbally endorse the use of
ity14). If certain members of an enemy state are spreading these techniques? Members of the latter group are culpable
disinformation, propaganda, ideology and hate speech and (or more culpable than members of the former group) but,
doing so by recourse to computational propaganda and other nevertheless, unable in practice to be distinguished from
manipulative means then the defender is morally entitled members of the former group.
to do likewise, at least if the target audience consists of the Let us distinguish cognitive warfare conducted in the con-
culpable members of the enemy state in question. Perhaps text of a kinetic war from cognitive warfare conducted in
so. However, two immediate problems arise at this point. ‘peacetime’, i.e., conducted in circumstances in which there
Firstly, these non-kinetic measures may have lethal or is no kinetic war. Thus, since the invasion of Ukraine by
other kinetic effects characteristic of kinetic wars. Con- Russia in February 2022, Ukraine and Russia are engaged in
sider, for instance, the dissemination of disinformation, a cognitive war in the context of a kinetic war. By contrast,
propaganda and hate speech designed with a view to incit- Russia has waged a cognitive war of sorts against the US,
ing violence. More generally, the use of cognitive warfare e.g., by virtue of its efforts to interfere in the US Presidential
techniques cannot be insulated from their kinetic effects, and elections, and sow discord more generally, but is not doing
certainly not from their intended kinetic effects. After all, the so in the context of a kinetic war being waged by Russia
whole point of engaging in cognitive warfare is ultimately against the US. Arguably, in the context of the latter kind
to change behaviour. of case, i.e., a morally justified (we assume) cognitive war
Secondly, many of these non-kinetic measures will not be being waged in ‘peacetime’ by a liberal democratic state,
effective if they only target culpable attackers. Consider, for it is not necessary, and may be counter-productive at least
instance, propaganda comprising (in part) in disinformation in the medium to long term, to resort to harmful offensive
that is aimed at weakening the enemy’s war effort (in the cognitive warfare measures that target non-culpable (or, at
overall context of a kinetic war); the obvious target is the least, much less culpable) members of the hostile state in
civilian population as a whole. Moreover, the application of question. Rather the following threefold combination of
the culpable/non-culpable distinction to cognitive warfare measures is likely to be sufficient: (1) essentially defensive
is problematic, and certainly does not mirror the relatively cognitive measures, e.g., implementing the measures men-
clear-cut combatant/non-combatant distinction relied upon tioned above to combat computational propaganda including
by Just War theorists and others in relation to the use of banning the hostile state’s propaganda on the channels of
lethal force in kinetic wars. public communication in the defending state; (2) develop-
The application of the culpable/non/culpable distinction ing counter-narratives to the hostile state’s disinformation,
in cognitive warfare is problematic since, for instance, many propaganda and use of manipulative but counter-narratives
civilian members of an authoritarian state the security forces that are not essentially false or manipulative and, therefore,
of which are engaging in cognitive warfare might support the not harmful offensive measures; and disseminating these
cognitive war in the weak sense that they verbally endorse counter-narratives in an ongoing, systematic manner to the
it to their friends and family but are otherwise without hostile state’s population; (3) deploying harmful offensive
measure that target culpable members of the ‘enemy’ state,
as appropriate, e.g., using profile-based, micro-targeting
14
It is unclear whether a third party state, C, has any obligation to techniques to disseminate disinformation or manipulative
use offensive cognitive warfare measures to intervene to defend mem- messages to culpable actors in the hostile state, e.g., mem-
bers of a state, A, being subjected to unjustified cognitive warfare by bers of security agencies.
members of a hostile state, B, by analogy with the obligation that C
might have to use lethal force against B if B was waging an unjust What of cognitive warfare undertaken in the context of a
kinetic war against A. There is, presumably, an expectation that an kinetic war (or perhaps the threat of a kinetic war)? Given
individual or state can stand up for themselves verbally (so to speak), that there is much more at stake in a kinetic war than in a
even if they cannot be expected to stand up for themselves physically. purely cognitive war and given what is at stake is in the here
On the other hand, there may be issues of great imbalances of com-
municative reach by virtue of, for instance, B’s possession of far more and now, a loosening of the restriction to avoid using harm-
sophisticated mass communication technologies. ful offensive measures against non-culpable members of the

13
Cognitive warfare: an ethical analysis Page 9 of 10 46

belligerent state is called for. (As above, we assume the per- of cognitive warfare in the context of waging a just kinetic
spective of a liberal democratic state determining its morally war. If these deaths or serious injuries were not intended
justified response to the morally unjustified use of cognitive then the use of the cognitive techniques in question might
warfare by a hostile state, albeit this time in the context of well be morally justified by recourse to the principles of
a kinetic war (being justly waged by the liberal democratic necessity and proportionality. Here there would be parity
state against the hostile, indeed belligerent, state.)) At this of reasoning with the morally justified, unintended killing
point, the general principles of necessity and proportional- of non-culpable citizens (or, at least, non-combatants) by
ity have a clear application. Moreover, in this context of combatants using lethal force in accordance with the prin-
a kinetic war the culpable/non-culpable distinction as it ciples of necessity, proportionality and discrimination. If,
applies to the use of the methods of cognitive warfare has on the other hand, the deaths of, or serious injuries to, the
much less purchase. In this respect it is akin to the closely non-culpable citizens were intended then they would likely
related moral and legal principle of discrimination, which violate the principle of discrimination. However, in these
has application to kinetic wars. According to the principle latter cases involving intended deaths or injuries there are
of discrimination, non-combatants cannot be intentionally likely to be moral complications arising from two factors.
targeted, although it is allowable for them to be uninten- Firstly, there is an indirect (causal) relationship between the
tionally killed in military operations if those operations are use of these cognitive techniques and the resulting deaths
compliant with the principle of military necessity and if the or serious injuries in question. Secondly, those who directly
numbers killed is not disproportionate by the lights of the cause the serious death or injuries must themselves bear
principle of proportionality. However, as we saw above, the some (and perhaps full) moral responsibility for these death
principle of discrimination (or related principles) has much or injuries, notwithstanding that they were acting on the
less purchase if the intended harm to non-combatants, or basis of beliefs and other attitudes to some extent induced in
innocent (i.e., non-culpable) civilians otherwise demarcated, them by those who targeted them with the cognitive warfare
is not death or serious physical injury, as it might well not techniques with the intention that their targets so act. Argu-
be in the case of the use of the techniques of cognitive war- ably, in these sorts of case there is joint moral responsibility
fare. Accordingly, intentionally harming non-culpable citi- (Miller, 2001, Ch. 8, 2006, 2016b, Ch. 5); the users of the
zens by disseminating disinformation, propaganda and/or techniques of cognitive warfare and their targets are jointly
hate speech to them, might be morally justified under some morally responsible for the resulting deaths or injuries to the
circumstances, e.g., if it did not directly or indirectly cause non-culpable citizens. The use of techniques of cognitive
death or serious physical injury (or it did not do so dispro- warfare successfully to incite violence against non-culpable
portionately—see below). citizens would be an example of this.
The justification in question would rely on the following
general considerations: (1) The nature of the harm done by
the use of the (inherently morally wrongful, let us assume) Conclusion
cognitive warfare technique in question, e.g., creating false
beliefs in non-culpable citizens (as well as culpable ones) In this article cognitive warfare has been characterised and
that results in the undermining of their well-founded (initial) found to be either be a non-kinetic dimension of kinetic war
confidence in the ability of their security forces to win a (as in the case of its use by Russians in their 2022 invasion
kinetic war; (2) The use of the cognitive warfare technique of Ukraine) or as a species of conflict short of war and, most
in question is effective, and there is no more effective, less importantly, of covert operations, namely, covert cognitive
harmful (all things considered), means available15 to achieve warfare (whether conducted in war or in peace-time). In
the moral weighty military or political end it serves; (3) The addition, an array of morally justifiable defensive measures
use of a morally wrongful means taken in conjunction with to combat cognitive warfare have been outlined and an argu-
the harm done by it was not disproportionate relative to the ment made in favour of restricted forms of offensive meas-
moral weight to be attached to the military or political end ures to combat cognitive warfare in light of the problem of
ultimately achieved by this means, e.g., the morally weighty targeting non-culpable members of a hostile state.
end of facilitating victory in the just kinetic war in question
greatly outweighed the harm done.
Funding Open Access funding enabled and organized by CAUL and
A final point pertains to deaths or serious injury to non- its Member Institutions.
culpable citizens that might result from the use of techniques
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
15
Or means that is as effective but less harmful or almost as effective as you give appropriate credit to the original author(s) and the source,
but much less harmful etc.

13
 46 Page 10 of 10 S. Miller

provide a link to the Creative Commons licence, and indicate if changes Miller, S. (1994). Social action. South African Journal of Philosophy,
were made. The images or other third party material in this article are 13(1), 9–17.
included in the article's Creative Commons licence, unless indicated Miller, S. (2001). Social action: A teleological account. Cambridge
otherwise in a credit line to the material. If material is not included in University Press.
the article's Creative Commons licence and your intended use is not Miller, S., (2006). Collective moral responsibility: An individualist
permitted by statutory regulation or exceeds the permitted use, you will account. In P. A. French (Ed.), Midwest studies in philosophy
need to obtain permission directly from the copyright holder. To view a (Vol. XXX, pp. 176–193).
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Miller, S. (2010). The moral foundations of social institutions. Cam-
bridge University Press.
Miller, S. (2016a). Cyber-attacks and ‘dirty hands’: Cyberwar, cyber-
crimes or covert political action? In F. Allfhoff, A. Henschke, &
References B. J. Strawser (Eds.), Binary bullets: The ethics of cyberwarfare
(pp. 228–250). Oxford University Press.
Alazab, M. (2022). Russia is using an onslaught of cyber attacks to Miller, S. (2016b). Shooting to kill: The ethics of police and military
undermine Ukraine’s defence capabilities. The Conversation. use of lethal force. Oxford University Press.
Retrieved February 24, 2022, from https://theconversation.com/ Miller, S. (2016c). Corruption and anti-corruption in policing: Philo-
russia-is-using-an-onslaught-of-cyber-attacks-to-undermine-ukrai sophical and ethical issues. Springer.
nes-defence-capabilities-177638 Miller, S. (2019). Jus ad Vim: The morality of military and police
Backes, A., & Swab, A. (2019). Cognitive warfare: The Russian threat use of force in armed conflicts short of war. In J. Galliott (Ed.),
to election integrity in the Baltic states. Belfer Center for Science Force short of war in modern conflict: Jus ad Vim. Edinburgh
and International Affairs, Harvard Kennedy School. University Press.
Bernal, A., Carter, C., Singh, I., Cao, K., & Madreperla, O. (2020). Miller, S. (2020). Freedom of political communication, propaganda and
Cognitive warfare. NATO Report, 10. the role of epistemic institutions in cyberspace. In M. Christen, B.
Cao, K., Cao, K, Glaister, S., Pena, A., Rhee, D., Rong, W., Rovalino, Gordjin, & M. Loi (Eds.), The ethics of cybersecurity. Springer.
A. (2021). Countering cognitive warfare. NATO Review. https:// Miller, S. (2023). War, reciprocity and the moral equality of com-
www.nato.int/docu/review/articles/2021/05/20/countering-cogni batants. Philosophia. https://link.springer.com/article/10.1007/
tive-warfare-awareness-and-resilience/index.html s11406-023-00678-1
Cocking, D., & van den Hoven, J. (2018). Evil on-line. Wiley Miller, S., & Bossomaier, T. (2023). Cybersecurity, ethics and collec-
Blackwell. tive responsibility. Oxford University Press.
D’Alessio, F. A. (2021). Computational propaganda: Challenges and Miller, S., Regan, M., & Walsh, P. F. (Eds.). (2021). National security
responses. Academia Letters. https://doi.org/10.20935/AL3468 intelligence and ethics. Routledge.
DeFranco, J., DiEuliis, D., & Giordano, J. (2020). Redefining neu- Office of the Director of National Intelligence. (2018). A guide to
roweapons. Prism, 8(3), 49–63. cyber attribution. https://www.dni.gov/files/CTIIC/documents/
Dipert, R. R. (2010). Ethics of cyberwarfare. Journal of Military Eth- ODNI_A_Guide_to_Cyber_Attribution.pdf
ics, 9, 384–410. Perry, D. L. (2009). Partly cloudy: Ethics in war, espionage, covert
Galliott, J. (Ed.). (2019). Force short of war in modern conflict: Jus ad action and interrogation. Scarecrow Press.
Vim. Edinburgh University Press. Regan, M., & Poole, M. (2021). Accountability of covert action in the
Greenwald, G. (2014). How covert agents infiltrate the internet to United States and the United Kingdom. In S. Miller, M. Regan,
manipulate, deceive and destroy reputations. The Intercept. & P. F. Walsh (Eds.), National security intelligence and ethics.
Retrieved February 24, 2014, from https://theintercept.com/2014/ Routledge.
02/24/jtrig-manipulation/ Rowe, N. C., et al. (2013). Perfidy in cyberwarfare. In F. Allhoff (Ed.),
Gross, M., & Meisels, T. (Eds.). (2017). Soft war: The ethics of Routledge handbook of ethics and war: Just war in the 21st cen-
unarmed conflict. Cambridge University Press. tury. Routledge.
Hung, T. C., & Hung, T. W. (2020). How China’s cognitive warfare Schmitt, M. M. (Ed.). (2013). Tallinn manual on the international law
works. Journal of Global Security Studies, 7, 4. applicable to cyberwar. Cambridge University Press.
Johnson, L. (2021). The “third option” in American foreign policy. Sloss, D. (2022). Tyrants on twitter. Stanford University Press.
In S. Miller, M. Regan, & P. F. Walsh (Eds.), National security Smith, N. V. (Ed.). (1982). Mutual knowledge. Academic Press.
intelligence and ethics. Routledge. Walzer, M. (1973). Political action: The problem of dirty hands. Phi-
Lucas, G. (2017). Ethics and cyber war. Oxford University Press. losophy and Public Affairs, 2, 160–180.
Lucas, G., (2013). Just in silico: Moral restrictions on the use of cyber- Woolley, S., & Howard, P. (Eds.). (2019). Computational propaganda.
warfare. In F. Allhoff (Ed.), Routledge handbook of ethics and Oxford University Press.
war: Just war in the 21st century. Routledge.
MacDonald, A., & Ratcliffe, I. (2023). Cognitive warfare: Manoeu- Publisher's Note Springer Nature remains neutral with regard to
vring in the human dimension. US Naval Institute. https://www. jurisdictional claims in published maps and institutional affiliations.
usni.org/magazines/proceedings/2023/april/cognitive-warfare-
maneuvering-human-dimension
May, L. (2017). The nature of war and the idea of “cyber war.” In M.
Gross & T. Meisels (Eds.), Soft war: The ethics of unarmed con-
flict. Cambridge University Press.

13