|
My primary areas of research are philosophy of psychology (including perception and social cognition), philosophy of mind, and philosophy of language. My work in these areas explores the nature and structure of bias and purported cases of implicit content---e.g., transformation principles within the visual-perceptual system, computational processes in socio-cognitive systems, and predictive models in artificial intelligence programs---as well as the implications the possibility of implicit content has for computational theories of mind more generally.
|
From Newell's Unified Theories of Cognition, 1994 |
Current Papers:
(drafts available upon request)
The Structure of Bias
(drafts available upon request)
The Structure of Bias
The aim of this paper is to attain conceptual clarity on the notion of bias in social cognitive theories of mind. In it, I argue that focusing only on the conscious accessibility of a bias is insufficient for recognizing various forms bias can take. Crucially, I maintain that extant accounts of implicit bias either fail to consider or mischaracterize one important possibility for the form of a bias: one that influences an individual’s beliefs about and actions toward other people, but that is, nevertheless, nowhere represented in that individual’s cognitive repertoire. Toward a positive proposal, I sketch a functional account of bias—which encompasses so-called “explicit” and “implicit” biases—that defines it in terms of propositional inputs and outputs. I then argue that this definition aptly allows for unrepresented social cognitive biases, a heretofore unexplored version of implicit biases. Finally, I end with some general reflections about what this possibility entails for representational and dispositional theories of mind more generally.
Algorithmic Bias: on the Implicit Biases of Social Technology
Often machine learning programs inherit social patterns reflected in their training data without any directed effort by programmers to include such biases. Computer scientists call this algorithmic bias. This paper explores the relationship between machine bias and human cognitive bias. In it, I argue that they are of the same basic kind and that by adopting a functional model that extends to both cases, we gain two advantages over extant models of bias. First, adopting a functional model captures a heretofore neglected possibility of human cognitive bias: those that influence an individual’s beliefs about and actions toward other people, but are, nevertheless, nowhere represented in that individual’s cognitive repertoire. Second, adopting a functional account allows for robust predictive and explanatory exchange between the machine and cognitive domains. I end by demonstrating this in the case of mitigation techniques, explaining one reason human implicit biases resist revision: cognitive biases, like machine biases, can rely on proxy attributes.
Against Phenomenal Priority: Implicit bias as a counterexample to the view that central cognition is consciously accessible
Roughly, the Phenomenal Priority Thesis holds that a mental state’s being possibly phenomenally conscious is a necessary condition for that state’s being intentional. A purported counterexample to this general view is provided by vision science and its need to postulate low-level, intentional perceptual states in explanations of processes within an individual’s visual system. In light of this case, proponents of Phenomenal Priority have responded by restricting the main tenet of the view to just those states within the central cognitive capacities of an agent. In this paper, I argue that implicit bias is a counterexample to this refined version of Priority. To do this, I establish first, that implicit biases are within the domain of central cognition; second, that implicit biases are unconscious; and third, that some implicit biases are intentional. Having argued that the existence of implicit bias is a reason to reject Priority wholesale, I briefly explore the implications this counterexample has for a proposed core social cognitive mechanism as well as the organization of the mind more generally.
The Psychology of Bias (forthcoming in An Introduction to Implicit Bias: Knowledge, Justice, and the Social Mind, edited by Erin Beeghly and Alex Madva)
The purpose of this chapter is to provide a comprehensive review of the empirical data and psychological theories surrounding the existence of implicit biases. In it, I present reviews of the empirical evidence surrounding implicit biases as well as extant models within philosophy and psychology. Although the existence of implicit biases is, at this point, beyond reasonable doubt, the questions of how to regard the nature, structure, and processes governing these biases are still open. This survey highlights the benefits and short-comings of some of the most prominent theories attempting to answer these questions such as the Associative-Propositional Evaluative (APE) model, the Motivation and Opportunity as Determinants (MODE) model, and various propositional models.
Works in Progress:
Inside the Black Box
What's going on inside the head of a person with an implicit social bias? Popular dispositional accounts of bias claim that such mental constructs are underlain by combinations of states and processes. In this paper, I explore plausible candidate states and processes by relating them to prominent views of the origins of human social categorization, inductions based on social kind membership, and generics. Next, I demonstrate that although definitive answers to which states and processes are involved in the operation of bias remain beyond reach, contemporary empirical work on the relationship between implicit bias and psychological essentialized kinds suggests that some relationship between various core social cognitive mechanisms exists. Ultimately, I argue that such explanations that unify these mechanisms are, all things equal, better.
Why we probably don't *see* social traits and why it doesn't matter
Contemporary theories of visual content disagree about the attributes that make it into the content of visual perceptions. So-called “sparse” views of perceptual content limit the representational capacities of the visual system to low-level attributives such as shape, size, and color. So-called “rich” views of perceptual content maintain that these capacities are more expansive and include high-level features of distal stimuli such as emotional and mental states, artifact categories, and social kinds. Toward reconciliation, I argue that on many useful carvings of the perception-cognition boarder in cognitive science, the visual perceptual system likely doesn’t represent high-level features; however, we needn’t suppose that it does in order to achieve many of the predictive and explanatory aims high-level theories are often invoked to serve.
