In a normative environment an agent's actions are not only directed by its goals, but also by the norms activated by its actions and those of other actors. The potential for conflict between agent goals and norms makes decision-making challenging, in that it requires looking-ahead to consider the longer term consequences of which goal to satisfy or which norm to comply with in face of conflict. We therefore seek to determine the actions an agent should select at each point in time taking

more »

... of its temporal goals, norms and their conflicts. We propose a solution in which a normative planning problem is the basis for practical reasoning based on argumentation. Various types of conflict within goals, within norms and between goals and norms are identified based on temporal properties of these entities. The properties of the best plan(s) with respect to goal achievement and norm compliance are mapped to arguments, followed by mapping their conflicts to attack between arguments, all of which are used to identify why a plan is justified. Shams, Z. et al the regulatory norm, by advising an agent what it is obliged to do (i.e., obligation norm), or forbidden from doing (i.e., prohibition norm), either of which it may violate. Our focus is on action-based regulatory norms [Meneguzzi et al. 2015; , while state-based regulatory norms (i.e.; norms that oblige or forbid certain state of affairs) have also been discussed in the literature [De Vos et al. 2013; Kollingbaum 2005] . Norm enforcement approaches (e.g., [Pitt et al. 2013; y López et al. 2005] ) often associate punishment with norm violation, so that avoiding punishment incentivises agents to comply with norms. However, agents might not wish or be able to comply with all norms imposed on them, due either to norm conflicts 2 , or to their wishing to achieve a goal that outweighs the cost of non-compliance. In such cases, the agent must identify which norm(s) it might violate, and accept the punishment or lack of reward for so doing, or which goals it might not satisfy. In identifying actions, long-term impacts must be considered. For instance, a goal satisfied at the cost of violating a norm might hinder or prevent more important goals from being achievable. One way of reasoning about such long-term impacts of compliance and violation requires the agent to consider the conflicts in the context of plans available to it. In so doing, an agent can take into account the benefit of goal achievement and norm compliance against the cost of goal failure and norm violation in different plans, and hence determine which plans are best to follow in presence of conflict. Norms have been considered in both planning and plan selection in the past [Belchior et al. 2018; Broersen et al. 2002; Kollingbaum 2005] . In order to generate conflict-free plans these works aim at resolving all normative conflicts. Since norms cannot be violated, if conflict resolution is not possible, the planning fails. While generating conflict-free plans, due to the possibility of violation, we allow more freedom in action selection, where an agent can take an action that may cause a normative conflict. Alternative ways of resolving conflict then give rise to different plans. Our departure point in doing this is [Shams et al. 2016] , which analyses norm-goal and goal-goal conflicts statically, considering whether they require different states of affairs to hold, but not the times at which the norms and goals are active. Here, we extend that model by (i) considering temporal properties for goals, and (ii) proposing temporal solutions to goal-goal and goal-norm conflicts as well as norm-norm conflicts. Apart from the above distinction, earlier work mentioned [Belchior et al. 2018; Broersen et al. 2002; Kollingbaum 2005; Panagiotidi et al. 2012 ] focus on the reasoning processes of a fully autonomous system, without considering the explainability of the system. In contrast, we consider domains where humans may need to understand why some action or plan was selected for execution (such as human-agent teams, or where a developer is debugging agent behaviour). This requires a transparent reasoning mechanism, rather than the numerical utilities of [Broersen et al. 2001; Kollingbaum and Norman 2003] , that can serve as the basis for the justification of agent behaviour. We utilise formal argumentation to derive such a reasoning process. Argumentation serves as an effective computational tool for various agent activities including agent reasoning [Amgoud 2003; Bench-Capon et al. 2009; Dung 1995; Gaertner and Toni 2007; Oren et al. 2007 ]. It supports the derivation of consistent conclusions from conflicting, inconsistent and incomplete information as generic arguments (or argument schemes [Walton 1996]). Such schemes also allow us to capture conflicts between arguments through the use of critical questions, representing the context in which an argument is invalid. We can then use argumentation semantics [Dung 1995 ] to determine plan justifiability with respect to goal satisfaction and norm compliance/violation. In addition, we describe how to discriminate between justified plans and identify a most preferred set of plans. Finally, we investigate the formal properties of what our approach considers the best plan(s). 2 This may well be due to the fact that norms come from different authorities aiming at regulating different aspects of agent behaviour.