amsterdam1

When in Rome…Norms as socio-mental processes not objects: 

When in Rome… Norms as socio-mental processes not objects Centre for Policy Modelling (CPM), Manchester Metropolitan University. Currently visiting: National Research Council (CNR), Institute of Psychology, Rome. David Hales Web: www.davidhales.com Email: dave@davidhales.com

Beliefs, goals and actions: 

Beliefs, goals and actions Beliefs are stored within agents Representations of believed states of affairs in the world Goals are desired states of affairs Actions are behaviours that an agent can perform in the world

Norms: 

Norms Norms are expressed as a socio-cognitive process encompassing beliefs, goals and actions The original sources of beliefs are also important in delineating a normative process Norms are not individually generated from base individual needs but are socially received. Norms are very often related to behaviour within social situations

Norms as process: 

Norms as process Beliefs which are socially received Concerning “proper” or “acceptable” goals in a given context (normative beliefs) Generate goals in those contexts Generate actions in those contexts Depending on existing beliefs, goals and strategies those actions may or may not be pursued

Norms as process: 

Norms as process A normative belief (b1) may produce goals in a given context. To achieve goals a set of actions need to be performed.

Current simulations…: 

Current simulations… The spread of normative reputation between and within agent groups Extending work done within an existing model Conte & Castelfranchi 1995 – Artificial Societies UCL Press. Castelfranchi, Conte & Paolucci 1998 – JASSS vol. 1, no. 3.

Quick simulation overview: 

Quick simulation overview 10x10 two dimensional grid world 50 agents, 25 food items Agents move around the grid looking for and consuming food items Agents use heterogeneous strategies Some respect a “possession” norm allowing agents initially near to food items to consume them Some do not respect the norm – attacking owners of food items to steal the food

Initial results with groups: 

Initial results with groups Certain forms of “group reputation” (ie stereotyping of a group of agents as either norm followers or norm breakers) are sufficient to give normative agents an overall advantage in food finding activities Additionally, groups of norm followers have a more egalitarian distribution of consumption.

The technical bit…: 

The technical bit… Agents are implemented as simple java objects Beliefs, actions and goals are all implicit – the behaviour is hard-coded. No BDI-type platform is used However, it would seem that all behavior is consistent with the outlined view of norms Does implementation matter? (opinions)

Questions from the armchair: 

Questions from the armchair Is this view of norms (as a socio-cognitive process involving beliefs, goals and actions) inconsistent with other frameworks? Is it too “thin” (adds nothing of value?) Is it too “wide” (covers too many phenomena) Is it a “so what” or “just so” theory?