The general approach of HERA is to formalize ethical principles as formulae of a logic that enables the modeler to talk about causal agency models and planning domain descriptions as explained in our research papers. Here we give an overview of the ethical principles that have so far been formalized and implemented in the HERA software.
- Deontological Principles
- Action-Focused Deontology
The action-focused deontological principle takes only the intrinsic utility of an act into account. An action is permissible if and only if the act itself is morally good or indifferent. This principle is equal to the first condition of the principle of double effect (see below).
- Intention-Focused Deontology
The intention-focused deontological principle takes the intrinsic utility of all intended actions and consequences into account. An action is permissible if and only if all intended actions and consequences are neutral or good. In case of intended action, impermissibility of the action-focused deontological principle implies impermissibility of the intention-focused deontological principle, but not vice-versa. This principle is equal to the second part of the second condition of the principle of double effect (see below).
- Patient-focused Deontology via the second formulation of Kant’s Categorical Imperative
According to the second formulation of Kant’s categorical imperative (also called “Humanity Formula”), an act is permissible if and only if everyone who is treated as a means by this act is also treated as a goal by the same act, that is, nobody is treated merely as a means. This principle takes individual moral patients and their benefits explicitly into account, and relates these to the agent’s goals and means.
- Consequentialist Principles
- Utilitarian principle
The utilitarian principle focuses on consequences of actions. It says that an agent ought to perform the action amongst the available alternatives with the overall maximal utility. We adopt an act-utilitarian interpretation which does not distinguish between doing and allowing, i.e. the causal structure of the situation is not taken into account. Thus the action which the agent ought to perform is the one which leads to the best possible situation, i.e. the highest utility, regardless of what the agent causes and intends.
- Do No Harm Principle
The Do No Harm principle says that an agent may not perform an action which has any negative consequences. The Do No Harm principle is fulfilled in case the agent remains inactive as there will then be no negative consequences and since we regard the act token of remaining inactive itself as neutral. The distinction between doing and allowing is relevant to this principle, as it is the causal consequences of an action which are considered. The intentions of the agents are not considered ethically relevant for our interpretation of this principle.
- Do No Instrumental Harm Principle
The Do No Instrumental Harm principle weakens the Do No Harm principle in that it permits harm done as side effect. However, harm may not be brought about as a means to the agent’s goal.
- Principle of Double Effect
The Principle of Double Effect says that an action is permissible if 4 conditions hold.
- The act itself must be morally good or indifferent.
- The positive consequence must be intended and the
negative consequence may not be intended.
- The negative Consequence may not be a means to obtain.
- There must be proportionally grave reasons to prefer.
The first condition corresponds to Deontology. The second condition makes sure the agent intends something good and nothing bad. The third condition corresponds to the Do No Instrumental Harm principle. And the last condition is a weaker form of the utilitarian principle demanding that overall the resulting state of the act has positive utility.