Resisting hostility generated by terror: An agent-based study

Abstract

We propose an agent-based model leading to a decrease or an increase of hostility between agents after a major cultural threat such as a terrorist attack. The model is inspired from the Terror Management Theory and the Social Judgement Theory. An agent has a cultural identity defined through its acceptance segments about each of three different cultural worldviews (i.e., Atheist, Muslim, Christian) of the considered society. An agent's acceptance segment is composed from its acceptable positions toward a cultural worldview, including its most acceptable position. An agent forms an attitude about another agent depending on the similarity between their cultural identities. When a terrorist attack is perpetrated in the name of an extreme cultural identity, the negatively perceived agents from this extreme cultural identity point of view tend to decrease the width of their acceptance segments in order to differentiate themselves more from the threatening cultural identity. We generated a set of populations with cultural identities compatible with data from a survey on attitudes among a large sample representative of the population of France; we then simulated the reaction of these agents facing a terrorist attack from Muslim extremists. For most populations, the average attitude toward Muslims becomes more negative. However, for some specific populations, we noticed the opposite effect as the average attitude of the population toward Muslims becomes less negative. In these populations, the Muslim agents strongly differentiate themselves from the terrorists' extreme cultural identity, and the other agents are aware of these changes. These reactions are due to particular properties of their cultural identities that are identified in this paper.

Fig 1

Graphs of the attitude values (on the y-axis) ${\mathit{\omega }}_{\mathit{k}}^{\mathit{i}}\left(\mathit{p}\right)$ of agent i in orange, and ${\mathit{\omega }}_{\mathit{k}}^{\mathit{j}}\left(\mathit{p}\right)$ of agent j in green for possible positions p on a worldview (on the x-axis). The values ${\omega }_k^i\left(p\right)$ depends on the acceptable positions of i. and the one of ${\omega }_k^j\left(p\right)$ depends on the acceptable positions of j. For both agents, the acceptable positions are recognizable in the figure as going from -0.85 to -0.15 on the x-axis since these positions are valued 1 on the y-axis Agents i and j have a similar acceptance segment but a different m.a.position (see the orange and the green squares respectively). Thus they have a different margin(h) (outlined by the orange and green arrows). The green agent, having a smaller margin(h) decreases more quickly the attitude given to positions outside its acceptance segment. A similar position 0.2 is valued -0.46 by the green agent while it is valued -0.24 by the orange agent.

Fig 2. Attitudes of each group—Muslims (M), Christians (C), and areligious (A)—toward each other (averages and standard deviations in parentheses for each relation computed from the representative survey of the French population).

Negative values represent the “unfavorable” attitudes. Positive values represent the “favorable” attitudes. The percentages in the middle of each hexagon is the percentage of people each group represents in the survey–it is close from their distribution in the French population: 3.6¨% for Muslims (M), 35.8% for areligious (A); 60.6% for Christians (C).

Fig 3. Distribution in percentages of types of evolutions of agent’s attitude toward M agents (increase in black, decrease or do not change in grey) for all initial 120 populations and times.

Fig 4. Evolution of the attitude of non-M agent (in orange) for the M agent (in blue): Dotted line, before the terrorist attack; plain line, after the agression.

(a) On the left, the attitude does not change despite the change of M margin(h) of acceptance, it remains -1: the maximum change of M margin(h) is limited by ${\mathrm{a}}_{\mathrm{M}}^{\mathrm{j}}$ which is valued -1 by i in any cases. Thus the attitude of the non-M agent for the M agent can’t change; (b) On the right, the attitude decreases (from 0.23 to 0.14) due to the decreasing of the margin(h) of acceptance of the non-M agent and despite the part of the overlap over the M acceptance segment has increased. Indeed, with a smaller margin(h), the non-M agent sees the part of the inclusive M’s segment external to it strongly more negatively than before.

Fig 5. Evolution of the attitude of the non-M agent about the M agent (in blue): Dotted line, before the terrorist attack; plain line, after the aggression.

(a) On the left, the attitude increases (from 0.01 to 0.17) due to the very small decreasing of the margin(h) of acceptance of the non-M agent, and the part of the overlap over the M acceptance segment has increased; (b) On the right, the attitude increases (from -0.03 to 0.00) due to the very small decreasing of the margin(h) of acceptance of the non-M agent, and the part of the overlap over the M acceptance segment has increased.