Cognitive Science Speaker Series talk examines ‘The Proxy Problem’ in machine learning

On March 24, the next Cognitive Science Speaker Series talk will discuss the ethical issues arising from the use of proxies in machine learning algorithms. “Proxies Aren’t Intentional, They’re Intentional” features guest speaker Gabrielle Johnson, an assistant professor in the Philosophy Department at Claremont McKenna College.

Gabrielle Johnson
Gabrielle Johnson. Photo: Claremont McKenna College

The Cognitive Science Speaker series is presented by York University’s Department of Philosophy in the Faculty of Liberal Arts & Professional Studies. All talks take place on Wednesdays from 12:30 to 2:30 p.m. via Zoom. Prior to each talk, the Zoom link will be emailed to all students and faculty from Cognitive Science and Philosophy. If you’re affiliated with York but not in one of those groups, and want to receive the Zoom links, email from your York email address.

Johnson works in philosophy of psychology, philosophy of cognitive science, philosophy of science, and philosophy of technology. She specializes in computational models in vision and social cognition, and her projects in these areas explore the nature and structure of social bias. Currently, her research extends theories of bias beyond individual agents to biases that manifest in machine learning algorithms and larger structural systems of injustice, and investigates the role of human values in scientific explanations involving burgeoning technology.

This talk concerns ‘The Proxy Problem’: often machine learning programs utilize seemingly innocuous features as proxies for social sensitive attributes, posing various challenges for the creation of ethical algorithms. Johnson argues that to address this problem, we must first settle a prior question of what it means for an algorithm that only has access to seemingly neutral features to be using those features as ‘proxies’ for, and so to be making decisions on the basis of, protected class features. She argues against theories of proxy discrimination in law and political theory that rely on overly intellectual views of the intentions of the agents involved or on overly deflationary views that reduce proxy use to mere statistical correlation. Instead, using insights from philosophy of language and mind, she adopts an anti-individualist account of representational content to argue for a constitutive account of ‘contentful proxy use.’ On this view, proxies represent socially sensitive features when and only when they constitutively depend on discriminatory practices against members of marginalized groups.