Epistemic Bias and Pluralism in Science

Project coordinator: Marco Viola

Risultati immagini per cognitive bias

On a daily basis, researchers are called to act as judges for the papers, projects, and even scientific careers of their peers. When they do so, their decision is of tantamount importance, as they affect the acceptance/rejection of the paper, the funding of some project, and whether to hire some colleague, respectively. Thus, it is important to study the factors that bias such decisions in given directions. While a good share of attention has been devoted to sociological phenomena linked to nepotism, racial or gender biases, few analyses have dealt with epistemic factors linked to the theoretical commitments. However, when a discipline is fragmented into different schools of thought, we can expect that a scholar who is closest to a given school of thought will judge works that she feels closest to her approach differently than those she feels more distant. Is there actual evidence of such phenomenon? Does it dampen or promote the pluralism in scientific communities in desirable ways? Also, given the evasiveness of concepts such as ‘school of thought’, how can it be investigated and measured? And what is impact of different institutional assets on this bias?

These concerns become especially relevant in the light of the institutional context of contemporary science. The social organization of science is undergoing some major shifts in recent decades. Following the mainstream of new public management, scientific institutions worldwide have been reformed.  In the name of accountability, audit culture has been imposed. Formal evaluation of research have flourished, whose results significantly affect the funding of scientific communities.

This project aims at studying various notions of epistemic biases with either theoretical and quantitative approaches. The former include conceptual and historical analyses of the very notions at play (e.g. specifying the notion of ‘school of thought’ and reflecting upon how it can be measured). The latter include agent-based-simulation and network analyses, and are meant to test and quantify the phenomena and theories previously framed in conceptual analysis.