Abstract
In this paper we present an agent-based model (ABM) of scientific inquiry aimed at investigating how different social networks impact the efficiency of scientists in acquiring knowledge. The model is an improved variant of the ABM introduced in [3], which is based on abstract argumentation frameworks. The current model employs a more refined notion of social networks and a more realistic representation of knowledge acquisition than the previous variant. Moreover, it includes two criteria of success: a monist and a pluralist one, reflecting different desiderata of scientific inquiry. Our findings suggest that, given a reasonable ratio between research time and time spent on communication, increasing the degree of connectedness of the social network tends to improve the efficiency of scientists.
The research by AnneMarie Borg and Christian Straßer is supported by a Sofja Kovalevskaja award of the Alexander von Humboldt Foundation and by the German Ministry for Education and Research.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Accordingly, we can distinguish between models that provide how actually and how possibly explanations. While some have suggested that modeling a possibility is epistemically valuable [11], others have argued that if a model merely captures a possibility, it is epistemically and pragmatically idle [19]; instead, the presented possibility has to be understood within a specified context, which makes it relevant for real world phenomena.
- 2.
To this end, one of the tasks for future research is the empirical calibration of the parameters used in the model, as pointed out in [13].
- 3.
Our ABM is created in NetLogo [22]. The source code is available at: https://github.com/g4v4g4i/ArgABM/tree/LORI-VI-2017.
- 4.
Other ways of discovering arguments and attacks (via social networks) are discussed below.
- 5.
Given that theories in our model are conflict-free, our notion of admissibility is actually the same as the one introduced in [7]. In Dung’s terminology, our sets of defended arguments correspond to preferred extension (which are exactly the maximally admissible sets), except that we determine these sets relative to given theories.
- 6.
Such hunches are not considered when agents evaluate theories.
- 7.
In contrast to the current model, in AABMSI [3] network structures are generated probabilistically in specific time intervals.
- 8.
A representative agent is excluded from research for 1–4 rounds: she always pays the basic cost of information sharing which is 1 round, and in addition, for every 2 fully explored arguments she will pay an additional round. The cost of learning an attack is equivalent to learning one degree of exploration of an argument.
- 9.
Further parameters, with short explanations, are as follows. The move probability (set to 0.5) together with the degree of exploration of the argument an agent is situated at, determines the chance that she will move to another argument every 5 rounds (the move incentive is further decreased by \(\frac{1}{5}\) for time steps in between). The visibility probability (set to 0.5) is the probability with which a new attack is discovered when an agent further explores her argument. The research speed (set to 5) determines the number of time steps an agent has to work on an argument a before a reaches its next level of exploration. The strategy threshold (set to 0.9) concerns the fact that each theory with a degree of defensibility that is at least 90% of the degree of defensibility of the best theory is considered good enough to be researched by agents. The jump threshold (set to 10) concerns the number of evaluations an agent can remain on a theory that is not one of the subjectively best ones.
- 10.
While our criterion is moderately pluralist, a more radical version would make plurality a necessary condition of success (i.e. populations would be punished for converging on one theory). We leave this consideration for future research.
- 11.
The plots concern the landscape consisting of three theories. The results were similar in case of two theories in all the discussed respects, except that the agents were comparatively more efficient.
- 12.
One example of an ABM that studies deception in science is [12], which examines the effects of a deceptive agent in a community of epistemically pure agents. The authors show that in general a higher degree of connectivity helps against deceptive information. While our model doesn’t examine the case of mixed (reliable and deceiving) agents, our results are, generally speaking, in line with their results.
- 13.
Interestingly, comparing the results of our model that employs the heuristic behavior (HB) and the results produced when HB is removed, shows that HB has hardly any impact on the success of agents, and in some cases it even slightly lowers their success. This seems to suggest that HB, by making agents stay on an undefended argument, waiting to find how to defend it, shields not only the best theory but also the worse ones, leading to an overall less successful inquiry. Examining this issue in more detail remains a task for future research.
- 14.
References
Bench-Capon, T.J.M.: Value based argumentation frameworks. arXiv:cs.AI/0207059 (2002)
Betz, G.: Debate Dynamics: How Controversy Improves Our Beliefs, vol. 357. Springer, Dordrecht (2012)
Borg, A., Frey, D., Šešelja, D., Straßer, C.: An Argumentative Agent-Based Model of Scientific Inquiry, pp. 507–510. Springer, Cham (2017). http://dx.doi.org/10.1007/978-3-319-60042-0_56
Caminada, M.: Truth, lies and bullshit: distinguishing classes of dishonesty. In: Social Simulation Workshop at the International Joint Conference on Artificial Intelligence (IJCAI) (2009)
Chang, H.: Is Water H2O? Evidence, Pluralism and Realism. Springer, Dordrecht (2012)
Douven, I.: Simulating peer disagreements. Stud. Hist. Philos. Sci. Part A 41(2), 148–157 (2010)
Dung, P.M.: On the acceptability of arguments and its fundamental role in nonmonotonic reasoning, logic programming and n-person games. Artif. Intell. 77, 321–358 (1995)
Gabbriellini, S., Torroni, P.: A new framework for ABMs based on argumentative reasoning. In: Kamiński, B., Koloch, G. (eds.) Advances in Social Simulation. AISC, vol. 229, pp. 25–36. Springer, Heidelberg (2014). doi:10.1007/978-3-642-39829-2_3
Grim, P.: Threshold phenomena in epistemic networks. In: AAAI Fall Symposium: Complex Adaptive Systems and the Threshold Effect, pp. 53–60 (2009)
Grim, P., Singer, D.J., Fisher, S., Bramson, A., Berger, W.J., Reade, C., Flocken, C., Sales, A.: Scientific networks on data landscapes: question difficulty, epistemic success, and convergence. Episteme 10(04), 441–464 (2013)
Hartmann, S., Reutlinger, A., Hangleiter, D.: Understanding (with) toy models. Br. J. Philos. Sci. (2016)
Holman, B., Bruner, J.P.: The problem of intransigently biased agents. Philos. Sci. 82(5), 956–968 (2015)
Martini, C., Pinto, M.F.: Modeling the social organization of science. Eur. J. Philos. Sci. 2(7), 1–18 (2016)
Muldoon, R., Weisberg, M.: Robustness and idealization in models of cognitive labor. Synthese 183(2), 161–174 (2011)
Rosenstock, S., O’Connor, C., Bruner, J.: In epistemic networks, is less really more? Philos. Sci. (2016)
Šešelja, D., Straßer, C.: Abstract argumentation and explanation applied to scientific debates. Synthese 190, 2195–2217 (2013)
Thimm, M.: A probabilistic semantics for abstract argumentation. In: ECAI, pp. 750–755 (2012)
Thoma, J.: The epistemic division of labor revisited. Philos. Sci. 82(3), 454–472 (2015)
Van Riel, R.: The content of model-based information. Synthese 192(12), 3839–3858 (2015)
Walton, D., Krabbe, E.C.: Commitment in Dialogue: Basic Concepts of Interpersonal Reasoning. SUNY Press, Albany (1995)
Weisberg, M., Muldoon, R.: Epistemic landscapes and the division of cognitive labor. Philos. Sci. 76(2), 225–252 (2009)
Wilensky, U.: Netlogo. Center for Connected Learning and Computer Based Modeling. Northwestern University (1999). http://ccl.northwestern.edu/netlogo/
Zollman, K.J.S.: The communication structure of epistemic communities. Philos. Sci. 74(5), 574–587 (2007)
Zollman, K.J.S.: The epistemic benefit of transient diversity. Erkenntnis 72(1), 17–35 (2010)
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer-Verlag GmbH Germany
About this paper
Cite this paper
Borg, A., Frey, D., Šešelja, D., Straßer, C. (2017). Examining Network Effects in an Argumentative Agent-Based Model of Scientific Inquiry. In: Baltag, A., Seligman, J., Yamada, T. (eds) Logic, Rationality, and Interaction. LORI 2017. Lecture Notes in Computer Science(), vol 10455. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-55665-8_27
Download citation
DOI: https://doi.org/10.1007/978-3-662-55665-8_27
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-662-55664-1
Online ISBN: 978-3-662-55665-8
eBook Packages: Computer ScienceComputer Science (R0)