People will often fight not for individual or collective material gain, but because of their commitment to abstract moral & sacred ideas; decisions to support or oppose war are descriptively deontological & are relatively insensitive to material costs or benefits

The Moral Logic of Political Violence. Jeremy Ginges. Trends in Cognitive Sciences, https://doi.org/10.1016/j.tics.2018.11.001

Abstract: There is a moral logic to reasoning about political violence. People will often fight not for individual or collective material gain, but because of their commitment to abstract moral and sacred ideas. Moreover, decisions to support or oppose war are descriptively deontological and are relatively insensitive to material costs or benefits.

---
For example, in one set of experiments carried out with Israelis and Palestinians in the West Bank and Gaza, participants were randomly presented with one of two compromises over a disputed issue such as Jerusalem: a straight compromise for peace, or the same compromise for peace plus a material incentive such as the promise of a life free of violence, or billions of dollars to the collective.  When participants were moral absolutists with respect to the issues in conflict (regarding them as sacred values) adding a material incentive backfired, ironically ‘increasing’ support for violent opposition to the deal. The existence of culturally specific sacred values does not itself impede tolerant interactions across cultures (Box 1). However, aggression may occur when one group acts to demean or threaten the second group’s sacred values.

Decisions about War Are Relatively Insensitive to Consequences
Rather than being a product of a breakdown in morality, war is often regarded as a moral necessity if not a moral good [2,7,13]. While moral reasoning is important to collective action in general [5], people tend to make decisions about war in a deontological way such that violence is either seen as prohibited or mandated.  This leads to decisions that are relatively insensitive to material consequences.  In anonymous surveys of Jewish Israelis living in the West Bank and Gaza, participation in non-violent protest (aggressive and non-aggressive) was related to both the perceived effectiveness of such behavior and its perceived ‘righteousness’. However, participation in violent attacks was predicted by righteousness and unrelated to effectiveness [3].

Cross-cultural experiments show similar effects [2]. In one experimental paradigm, participants were randomly assigned to consider their support for either a non-violent response (negotiation) or a violent response (armed attack) to the kidnapping and imminent murder of 100 innocent civilians.  In prior tests, participants thought both options were equally appropriate and desirable. However, when asked to indicate how many hostages they required to be rescued to support the response they were considering, participants in the negotiation condition demanded between 80 and 100 hostages to be rescued, while those in the armed attack condition option required only one hostage to be rescued.  Participants in themilitary conditions would often give strategic reasons for their responses, typically by arguing that violence will deter future attacks. Yet, a subsequent experiment showed that support for military options was similarly insensitive to its deterrent capability [2]. Thus, people reason differently about violent and nonviolent option in intergroup conflicts, using the logic of instrumental rationality for nonviolence, but deontological reasoning when making choices about political violence.  This can lead to systematic inconsistency of preferences for military action (Box 2).