Normal = Normative? The role of intelligent agents in norm innovation

Abstract

Abstract. In this paper the results of several agent-based simulations, aiming to test the role of normative beliefs in the emergence and innovation of social norms, are presented and discussed. Rather than mere behavioral regularities, norms are here seen as behaviors spreading to the extent that and because the corresponding commands and beliefs do spread as well. On the grounds of such a view, the present work will endeavour to show that a sudden external constraint (e.g. a barrier preventing agents from moving among social settings) facilitates norm innovation: under such a condition, agents provided with a module for telling what a norm is can generate new (social) norms by forming new normative beliefs, irrespective of the most frequent actions.

A coherence maximisation process for solving normative inconsistencies

“…Specifically, we focus on how an agent resolves the inconsistencies among the instances that affect it. To make such decisions, we propose that the agent considers: the salience of norms (which has been defined in [5] as the degree of activity and importance of a norm within a social group and a given context) that have given rise to the instances, the ease of compliance of these instances (i.e., how difficult it is for an agent to comply with these instances), the agent environment (i.e., the certainty of the beliefs about the agent's current circumstances) and the impact of instances (i.e., the importance of the desires hindered or favoured by the instances). In order to capture these different parameters, and to allow our agents to deal with dynamic and non-deterministic environments, we use the Normative Graded BDI architecture (known as n-BDI) [11], which extends the Graded BDI architecture proposed by Casali et al in [6] with an explicit representation of norms and instances.…”

“…These sets are composed of M (γ, ρ) expressions, where: M ∈ {belief, desire, intention} is a graded modality used for representing graded beliefs, desires or intentions, respectively; γ is a grounded formula of L; and ρ ∈ [0, 1] represents the degree associated with this mental formula. ρ represents a certainty degree in case of belief, a desirability degree in case of desires 4 , and an intentionality degree in case of intentions 5 . -N is a set formed by norm(n, ρs) expressions, where n is a norm, and ρs ∈ [0, 1] is a real value that assigns a salience to this norm.…”

“…-A set of literals L is created such that for each atom in P we have a positive literal and a negative literal: i.e., ∀p ∈ P : p ∈ L ∧ ¬p ∈ L. -A set of roles R is created such that the size of R is randomly determined to be in the interval [1,5]. -A set N of x norm expressions is created.…”

Reasoning about norms under uncertainty in dynamic environments

“…The term norm salience was defined by Campennì et al in [11] as "the degree of activity and importance of a norm within a social group and a given context". As psychological [15,5] and behavioural economics [6,50] studies have pointed out, norm acceptance is strongly influenced by the norm salience.…”

“…To model this reasoning process we define the norm relevance bridge rule as follows: If an agent considers that a norm -represented by the expression D, C, T, A, E, S, R -is currently active -i.e., there is a substitution σ such as the expression (B σ(A), ρ σ(A) ) is deduced in BC; where σ(A) denotes the result of applying σ to A, and ρ σ(A) is a real number within the [0, 1] interval representing the certainty about this belief -and the agent knows that there is an agentrepresented by the expression AgentID -that it is under the influence of the norm -i.e., the expression (B play(AgentID, T ), ρ T ) is deduced in BC; where play(AgentID, T ) denotes fact that AgentID is playing role T , and ρ T is a real number within the [0, 1] interval representing the certainty about this belief-, then a new instance is generated 11 .…”