We propose an agent-based model where a fixed finite population of tagged agents play iteratively the Nash demand game in a regular lattice. The model extends the bargaining model by Axtell, Epstein and Young.

This model has been developed together with the publication ‘Modelling Value Change - An Exploratory Approach’

Value change and moral change have increasingly become topics of interest in the philosophical literature. Several theoretical accounts have been proposed. Such accounts are usually based on certain theoretical and conceptual assumptions and their strengths and weaknesses are often hard to determine and compare, also because they are based on limited empirical evidence.

We propose that a step forward can be made with the help of agent-based modelling (ABM). ABM can be used to investigate whether a simulation model based on a specific account of value change can reproduce relevant phenomena. To illustrate this approach, we built a model based on the pragmatist account of value change proposed in van de Poel and Kudina (2022). We show that this model can reproduce four relevant phenomena, namely 1) the inevitability and stability of values, 2) how different societies may react differently to external shocks, 3) moral revolutions, and 4) lock-in.

The modeling includes citizens, bounded into families; firms and governments; all of them interacting in markets for goods, labor and real estate. The model is spatial and dynamic.

A model for simulating the evolution of individual’s preferences, incliding adaptive agents “falsifying” -as public opinions- their own preferences. It was builded to describe, explore, experiment and understand how simple heuristics can modulate global opinion dynamics. So far two mechanisms are implemented: a version of Festiguer’s reduction of cognitive disonance, and a version of Goffman’s impression management. In certain social contexts -minority, social rank presure- some models agents can “fake” its public opinion while keeping internally the oposite preference, but after a number of rounds following this falsifying behaviour pattern, a coherence principle can change the real or internal preferences close to that expressed in public.

An agent-based model for the diffusion of innovations with multiple characteristics and price-premiums

This is a re-implementation of a the NetLogo model Maze (ROOP, 2006).

This re-implementation makes use of the Q-Learning NetLogo Extension to implement the Q-Learning, which is done only with NetLogo native code in the original implementation.

This model implements a classic scenario used in Reinforcement Learning problem, the “Cliff Walking Problem”. Consider the gridworld shown below (SUTTON; BARTO, 2018). This is a standard undiscounted, episodic task, with start and goal states, and the usual actions causing movement up, down, right, and left. Reward is -1 on all transitions except those into the region marked “The Cliff.” Stepping into this region incurs a reward of -100 and sends the agent instantly back to the start (SUTTON; BARTO, 2018).

The problem is solved in this model using the Q-Learning algorithm. The algorithm is implemented with the support of the NetLogo Q-Learning Extension

CINCH1 (Covid-19 INfection Control in Hospitals), is a prototype model of physical distancing for infection control among staff in University College London Hospital during the Covid-19 pandemic, developed at the University of Leeds, School of Geography. It models the movement of collections of agents in simple spaces under conflicting motivations of reaching their destination, maintaining physical distance from each other, and walking together with a companion. The model incorporates aspects of the Capability, Opportunity and Motivation of Behaviour (COM-B) Behaviour Change Framework developed at University College London Centre for Behaviour Change, and is aimed at informing decisions about behavioural interventions in hospital and other workplace settings during this and possible future outbreaks of highly contagious diseases. CINCH1 was developed as part of the SAFER (SARS-CoV-2 Acquisition in Frontline Health Care Workers – Evaluation to Inform Response) project
(https://www.ucl.ac.uk/behaviour-change/research/safer-sars-cov-2-acquisition-frontline-health-care-workers-evaluation-inform-response), funded by the UK Medical Research Council. It is written in Python 3.8, and built upon Mesa version 0.8.7 (copyright 2020 Project Mesa Team).

This is a simulation of an insurance market where the premium moves according to the balance between supply and demand. In this model, insurers set their supply with the aim of maximising their expected utility gain while operating under imperfect information about both customer demand and underlying risk distributions.

There are seven types of insurer strategies. One type follows a rational strategy within the bounds of imperfect information. The other six types also seek to maximise their utility gain, but base their market expectations on a chartist strategy. Under this strategy, market premium is extrapolated from trends based on past insurance prices. This is subdivided according to whether the insurer is trend following or a contrarian (counter-trend), and further depending on whether the trend is estimated from short-term, medium-term, or long-term data.

Customers are modelled as a whole and allocated between insurers according to available supply. Customer demand is calculated according to a logit choice model based on the expected utility gain of purchasing insurance for an average customer versus the expected utility gain of non-purchase.

How do bots influence beliefs on social media? Why do beliefs propagated by social bots spread far and wide, yet does their direct influence appear to be limited?

This model extends Axelrod’s model for the dissemination of culture (1997), with a social bot agent–an agent who only sends information and cannot be influenced themselves. The basic network is a ring network with N agents connected to k nearest neighbors. The agents have a cultural profile with F features and Q traits per feature. When two agents interact, the sending agent sends the trait of a randomly chosen feature to the receiving agent, who adopts this trait with a probability equal to their similarity. To this network, we add a bot agents who is given a unique trait on the first feature and is connected to a proportion of the agents in the model equal to ‘bot-connectedness’. At each timestep, the bot is chosen to spread one of its traits to its neighbors with a probility equal to ‘bot-activity’.

The main finding in this model is that, generally, bot activity and bot connectedness are both negatively related to the success of the bot in spreading its unique message, in equilibrium. The mechanism is that very active and well connected bots quickly influence their direct contacts, who then grow too dissimilar from the bot’s indirect contacts to quickly, preventing indirect influence. A less active and less connected bot leaves more space for indirect influence to occur, and is therefore more successful in the long run.