Mimicry in Civil War

Workshop with Diego Gambetta

Please note: This page refers to an event that has already taken place.

Time: 07 Dec 2007 - 08 Dec 2007

Deceptive signalling and identity mimicry

A term derived from biology, mimicry – in the deceptive variety that interests us here – consists of passing off either as a different individual or as a member of a group to which one does not belong. Each mimicry episode has three protagonists: a mimic, a model and a dupe. In order to improve either his aggressive or defensive payoffs the mimic aims to make the dupe believe that he is the model. For an overall view of the phenomenon, which has not been systematically studied, we refer to a paper by Gambetta (2005), which we attach.

Here it is important to set mimicry in the broader context of deceptive signalling. Humans can feign certain dispositions – curiosity, excitement – or physical and emotional states – headaches, grief, or orgasm – by imitating the looks, postures, words, or level and type of perceivable activity directly associated with these states. They can deceitfully pretend to possess an unobservable but desirable property k, such as benevolence or honesty, by adopting a sign m that is associated with k, e.g., looking people in the eyes to persuade them of one's honesty. The model for the mimic in these cases is not a specific agent or group of agents, but a generic state or property and its manifestations. This is not, however, the type of deceptive mimicry that concerns us here.

A great deal of human signalling takes place indirectly by signalling one's identity, both as a specific individual, or as a member of a group or category. After we encounter an individual or group member and experience dealing with them, we form and retain an idea of whether this person or group has or lacks the k property that interests us – it creates a “reputation”, whether positive or negative. Identity signalling enables the signaller to exploit a positive reputation. In identity signalling, instead of using a two-layered inferential structure (m → k), we use a three-layered structure, g → i →k, where, i denotes identity and g denotes a sign of identity, or signature. If persons or group members are re-identifiable by some signature, the next time we meet them we infer the presence or absence of k via their identity i.

The re-identification of a signature, however, can be problematic. This is because frequently the fact that someone has a certain reputation is an unobservable property of that person. For example, Armani has a reputation for selling well-designed clothes, but to exploit this reputation, the seller must convince customers that he is Armani. Islamic jihad has a reputation for carrying out its threats against hostages, but to exploit this reputation a group of kidnappers must convince governments that they really belong to Islamic jihad. When a model signals his qualities via his identity, the threat of mimicry of k through m is replaced by the threat of mimicry of i through g. Much of human deceptive mimicry exploits signalling via identity, and our interest is mostly in this case.

Note also that while the mimic and the dupe are always ultimately one or more individuals – one can only intentionally fool or be fooled as an individual - the model may be both an individual and the features of the members of a group or category (or, in the case of camouflage, the features of an environment).[1] In short while the mimic and the dupe always require a mind, the model does not necessarily require one. Depending on the mimicry ‘system’, the model may be inert or indifferent, but may also either fight against or collude with mimic.

Types of mimicry in civil wars

‘Fight and flight’ are the fundamental tactics in human conflict, but both are often supported by the strategic use of identity signalling. Perhaps the simplest way to convey the range of identity mimicry episodes is by giving a set of examples, in which the typical protagonists of a civil war can take any of the three mimicry roles –we have insurgents, regular armies and civilians all using deceptive mimicry (the latter group for defensive purposes only); we have each of them serving as models to the mimicking agents, and finally each of them can be the dupe, the intended victim of a mimicry act.

To keep matters simple we organise the sample simply by whether mimic and model are groups or individuals. 

Individual mimicry of specific individuals

In this case the model is a specific person. When organisations operate in secrecy, it is often difficult to establish whether a video or recorded message from, say, ‘bin Laden’ is really a message from bin Laden, rather than from someone who pretends to be bin Laden – intelligence agencies regularly engage in matching operations, trying to, say, match the voice pitch or the facial features with models they have of the real bin Laden. There could be at least two types of mimics of bin Laden. First, his friends – as Kurosawa’s film “Kagemusha” suggests, there are instances in which subjects pretend their dead chiefs are still alive (some cases have been found in the mafia too). Next, his enemies who want to make ‘him’ say things which discredit him, or confuse and misinform his followers.

Another case of individual mimicry in conflict situations is the use of doubles – in this case the model himself recruits the mimic to avoid exposing himself to dangerous situations, and letting the mimic do that instead (as the following joke conveys: during the Iraqi war a general of Saddam summoned his doubles and told them: the good news is that Saddam is alive, the bad news is that he lost an arm.)

Individual mimicry of group membership

In this case the model is the set of signatures of belonging to a group or category, more precisely the set of signatures that identify a group in the perception of the dupe. Under this category examples abound whether for aggressive or defensive goals. Think of suicide bombers disguising themselves as orthodox Jews or pretending to be pregnant; of insurgents passing off as civilians or as members of the enemy fighters; of Jews passing off as Poles to escape Nazi persecution; of men passing off as women to escape detection, but also of women passing of as men to be able to fight, as in the American Civil War.

Group mimicry of other groups

Some of the most interesting cases are found in the mimicry of groups which pretend to be other groups, a phenomenon also known as ‘false flags’, and there is a whole variety of cases. The model group can be real or invented.

In Colombia there are reported instances of both criminal kidnappers passing off as insurgents to increase the credibility of their blackmail, and, conversely, insurgents passing off as criminals when kidnapping or robbing a bank in order not to tarnish their political image.

Much of terrorist bombings in Italy 1969-1980 had right wing extremists or rogue state agents ‘signing’ their acts with left wing signatures. Another case of mimicry aimed at directing the public blame on opponents is found in Algeria where “The techniques of violence used in the Algerian civil war have consisted among other things of disguising the militants’ identities at the “roadblocks” mounted by the armed fundamentalists who pass themselves off as members of the Popular National Army. In many cases, the manipulation of violence has been used by the Algerian regime as well as by the fundamentalists” (Selma Belaala).

Mimicry can also be used for competitive purposes. Very often terrorist acts have multiple claims of paternity from groups on the same side of the conflict – when only one group is responsible, the mimics do it in order to gain at the lower mimicry cost from the reputation deriving from committing such an act.

Finally, insurgents have set up local NGOs in order to operate publicly and legally behind a neutral ‘face’.  

In Iraq, in recent times there were some episodes of double mimicry: when carrying out violent acts real rogue policemen kept their police uniforms counting on the fact that people would believe they were insurgents disguised as policemen. 


For the workshop we are looking for three kinds of papers

  • Theory papers on identity mimicry
  • Case-based papers on specific instances of mimicry that cut across mechanisms
  • Mechanism-based papers, which tackle in theory and with empirical examples specific questions that cut across cases.

We have identified some of the general questions:

  • Under which conditions are there more or less opportunities for identity mimicry to occur in civil wars? For instance, what happens in conflicts between co-ethnics and non co-ethnics? What happens, in other words, where friends and foes looks alike, such as Bosnia or Northern Ireland, as opposed to conflicts in which the members of the fighting parties can be easily distinguished from one another by simple observation? Or what are the differences between territorial and non territorial insurgents?
  • Which part does mimicry have in conflicts in which the insurgents are selective in the type of actions they carry out as opposed to those in which insurgents hit everyone and everything? For instance, the IRA wanted to keep a strict control on where and when and whom to hit and with what level of force, thus its energetic attempt to avoid mimicry was essential to maintain control; by contrast, al-Qaeda is far less selective – any Western target anywhere in the world can do. So anyone who does something bad and pretends to be al-Qaeda de facto works for al-Qaeda (short of harming children and bombing an elderly people’s home) – mimicry dissolves into useful proliferation of clones. So while mimicking al-Qaeda is easier it is also pointless. On the other hand al-Qaeda has reached such a monopoly of terror brands, that we could have the reverse case, in which some other terrorist group does some big act of violence and then truthfully claims it, and no one believes them, as everyone will be knee-jerkily blame al-Qaeda, who can then effortlessly and fraudulently claim it. As with the blackout that struck the US east coast in August 2003, al-Qaeda can even claim responsibility for accidents.
  • Which part does mimicry have in conflicts in which insurgents are fragmented in several groups as opposed to those in which one group monopolises action? Does group mimicry have a more relevant part say in Palestine than Sri Lanka?
  • Which strategies are open to groups to protect themselves from mimicry from other groups, which are the mimic-proof signatures? The IRA agreed a code with the police which was used in warning calls. Other groups call before hand for only the culprit can know that something is going to happen. Other use video recorded evidence of their intentions.
  • Which are the strategies and opportunities for infiltration, and which the screening counter strategies?

[1] Camouflage can be conceptualised as a case of negative mimicry. There are often signs that are likely to be interpreted by the signal receiver, rightly or wrongly, as indicating the absence of a property k of the signaller. Both an honest signaller with k who expects to be unjustly perceived if he displays such a sign m, and an opportunist non-k who is afraid of being detected if he does, have a reason to camouflage. That is, they take steps not to show m. Camouflaging can be considered as a special case of mimicking, since the strategy of camouflaging non-k-ness by suppressing m is just that of mimicking k through displaying the notional sign ‘no m’ (Gambetta 2005).