As part of the AISB 2017 conference, we are hosting a one-day workshop, which is concerned with emergence and immergence as cognitive phenomena.
Distributed approaches to language, reasoning, and decision-making suggest that cognition is grounded in three fundamental domains: internal, bodily mechanisms, external interactions with the choice environment and other people, multi-temporal timescales that influence the possibility for cognition. While cultural norms may emerge across time, simulated agents are both reactive and anticipatory and are capable of learning from their immersion in a norm saturated environment. ABM enables agents to produce changing futures as they self-configure. That is, the immergent emergence of cognitive function and culture occurs in interactive relations and across different timescales (ranging from a few milliseconds to evolutionary millennia, which suggests the need for models that explore the emergence and evolution of lived cognition in vivo).
While burgeoning evidence supports the interactive element of cognition, the multi-scalar temporal hypothesis is harder to test in that not only do large timescales affect the evolution of culture and cognition, but the process is also affected by a myriad of other phenomena. Therefore, research has typically lacked a controlled method for testing the ontogenesis of cultural phenomena as well as their direct influence on cognitive functionality.
In the past decade, researchers have increasingly made use of Agent-Based Models to simulate complex interactions between agents and across timescales. This allows for highly controlled simulations of cognitive models in computationally intractable environments. As such, Agent-Based Models provide a powerful tool in exploring the emergence of inter-personal phenomena such as language, decision-making, and the evolution of culture more generally. This symposium draws together researchers who have contributed to the field of distributed cognition and how social norms can enrich Agent-Based simulations.
Language Distributed accounts of language argue that 1st-order usage of language (at a specific time and place) draws on 2nd-order languages that have evolved across history. In a simulated Agent-Based Model, contributors show that accents can indeed develop across time given interactions and mobility. Their models track the development of accentual differences that result from these interactions, despite their agents having identical cognitive design.
Decision-making Traditional models of decision-making isolate expected utilities and calculate optimal (or just beneficial) strategies given the constraints to the specific situation. These approaches, however, neglect to discuss the dynamic actions of others, the fluidity of utilities, and the development of decision-making across time and within social groups. Taking a point of departure in dynamic environments where other agents engage with the same environment, we show that decision-making evolves across time, both for each individual agent, but also for aggregates of agents. Given time and interactions, each agent will adapt and engage in significantly different decision strategies.
Culture On a broad level, the models presented in the symposium show the importance of the development of cultural practices and norms. Although these develop slowly in simulated societies, they have a profound effect on decision-making. This suggests that culture, which can only emerge across multiple timescales, is profoundly influential on language and decision-making, meaning that models of either necessarily require cultural, and thereby temporal, components. In this way, Agent-Based Models are used as proof of concept to test the importance of timescales and interactions for language and decision-making.
In addition to the contribution to and testing of models of language and decision-making, the symposium focuses on a discussion of Agent-Based Models as a novel methodology in cognitive sciences. We believe this method will become increasingly useful in years to come, and we hope the symposium will provide participants with a good understanding of the method, its requirements, its potential, and its limitations. As such, the symposium reaches beyond the contributions of each contributor in isolation and points to a general methodological discussion.
11.00-11.30: Introduction (Jens Koed Madsen) + Stephen Cowley & Jens Koed Madsen: Immergence: how the evolution of culture influences human forms of life, language and decision-making
11.30-12.00: Bruce Edmonds: Co-developing beliefs and social influence networks
12.00-12.30: Matthew Harvey: What changes in language change? A deflationary model of the emergence of intrusive and linking /r/ in nonrhotic varieties of English
13.30-14.00: Davide Secchi & Rafaello Seri: Power analysis for agent-based modelling determining the appropriate number of runs
14.00-14.30: Joanna Bryson: Semantics derived automatically from language corpora necessarily contain human biases
14.30-15.00: Martin Neumann: Attitudes and ideologies: how do political issues map on a left-right scale?
15.30-16.00: Davide Secchi: The social dependence of cognition: An agent-based simulation of the perception of scientific value as a function of impact factor
16.00-17.00: Panel Discussion: Equation-based and Agent-based Models in Social Sciences
Stephen Cowley, Professor, University of Southern Denmarkcowley@sdu.dk
Jens Madsen, Post-doctoral researcher, University of Oxfordovermuren@gmail.com
Mario Paolucci, Researcher, CNR, Romemario.firstname.lastname@example.org
Davide Secchi, Associate Professor, University of Southern Denmarksecchi.email@example.com
Copyright © All Rights Reserved