Experimental argument analysis: Reasoning with stereotypes
PI Eugen Fischer, Reader (Senior Associate Professor) in Philosophy, University of East Anglia
Co-I Paul E. Engelhardt, Associate Professor of Psychology, University of East Anglia
Post-doc Dimitra Lazaridou-Chatzigoga, Affiliated Lecturer in Linguistics, University of Cambridge
Verbal reasoning is shaped by automatic inferences that have received relatively little attention in the philosophical literature: Arguments couched in natural language are strongly shaped by automatic inferences that continuously occur in language comprehension and production.
This interdisciplinary project develops the new research program of experimental argument analysis (EAA). EAA represents the first sustained effort to examine the roots of verbal reasoning in language processing. As developed by members of the project team, EAA examines how automatic comprehension inferences shape thinkers’ moves from premises to conclusions – and investigates when and why automatic language processes lead even competent thinkers to go along with inferences they ought to reject. To do so, EAA innovatively combines traditional philosophical approaches with cutting-edge empirical methods, and repurposes methods, findings and theoretical frameworks from
psycholinguistics; it adapts tools previously used to examine utterance comprehension and production to examine and evaluate verbal reasoning – for a start, in philosophical arguments. Given their fundamental role in language comprehension and their wider social relevance, the project focuses on stereotypical inferences (everyday example: the ‘secretary’
will be female) and studies the circumstances under which such inferences influence further judgment and reasoning even in contexts that defeat them (‘the secretary was a Mr Brown’). While competent language users are generally good at contextualising, specific comprehension biases hinder contextualisation and lead to over-reliance on stereotypes. EAA experimentally documents and explains inferences of interest by reference to specific comprehension biases, and deploys findings to expose poor reasoning and cultivate sound reasoning.
This new approach is much needed and highly productive: Reasoners have no introspective access to automatic inferences; these need to be examined experimentally. Moreover, the interpretation of philosophical and other reasoning is governed by hermeneutic principles of charity that allow us to attribute fallacies to competent thinkers only when we can explain when and why even competent reasoners commit fallacies of this kind. Psychological explanations that consider reasoning, as a cognitive process, can therefore facilitate decisive philosophical criticism. Equally important, empirically informed explanations improve our self-understanding as human reasoners, let us critically re-assess our own reasoning, and help us formulate arguments in ways not liable to trigger problematic inferences. The findings will help us expose unnoticed fallacies, understand opaque reasoning, develop improved conceptual tools, and formulate fresh solutions to longstanding philosophical problems.
The PI and Co-I created new tools and starting-points for EAA, in interdisciplinary work that pioneered the adaptation of psycholinguistic methods for philosophical purposes and experimentally documented a previously unrecognized but crucial comprehension bias: The linguistic salience bias affects comprehension inferences from polysemous words, i.e., words with several distinct, but related senses (e.g., ‘see’ her hat vs. ‘see’ her point). Polysemes account for at least 40% of words in languages including English, and play a key role in philosophical discourse, where philosophers often give new uses to familiar words. The bias leads to fallacies of equivocation: Under circumstances we predicted, we observed that subordinate uses of polysemous words trigger stereotypical inferences supported only by the words’ dominant sense. The bias is liable to arise where people give special uses to familiar words, e.g., to conceptualise new phenomena, or to advance socio-political agendas.
On this basis, our present project with a multidisciplinary team (philosophy, psychology, linguistics) will break new ground by developing the first complete paradigm (worked example) of EAA. This paradigm will anchor diverse innovative applications and forge new connections between experimental work and the philosophical tradition. The project will combine eight cutting-edge psycholinguistic experiments with philosophical exegesis and analysis. At its heart will be a case study that examines how and why contextually defeated stereotypical inferences drive classical philosophical arguments. The case study targets stereotypical inferences from verbs, which play a central role in language
comprehension; it examines how an important comprehension bias, namely, the linguistic salience bias, leads to contextually inappropriate inferences; it studies how these fallacious inferences shape one historically influential family of philosophical arguments, namely, arguments ‘from illusion’ and arguments ‘from hallucination’ that jointly engender the philosophical ‘problem of perception’. This longstanding problem, situated at the strategic intersection of three key areas of theoretical philosophy (epistemology, metaphysics, and the philosophy of mind), is a linchpin of Western philosophy since the 18th century and has become a renewed focus of debate. We explore how our findings help resolve this central
problem and how our approach can address other philosophical problems with a similar structure.
Together with the first worked example of EAA, the project will develop new methods. In a spirit of methodological pluralism, the project breaks new ground in combining eye-tracking methods from psycholinguistics with hermeneutic and analytic methods from philosophy. Our prior work has adapted psycholinguistic methods, initially devised to examine sentence comprehension, for the new use of studying verbal reasoning. The present project will optimise these methods for the new use and trial a sophisticated new paradigm for examining inferences from polysemous words. In addition, the project will advance the use of individual differences measures (like verbal IQ and inhibition) to support
nomothetic theory construction and to examine the susceptibility of different demographic groups to the targeted bias.
Findings of this foundational research project will unlock innovative applications across philosophy, with relevance beyond philosophy. Among other things, they will provide powerful new tools for critical thinking, launch a novel approach that combines empirical and hermeneutic methods and findings to interpret otherwise opaque texts, and provide empirical foundations for ‘conceptual engineering’, i.e., prominent efforts to develop new conceptual tools for purposes ranging from improving scientific theorising to changing socio-political attitudes in pursuit of social justice. EAA thus unlocks philosophically, scientifically, and societally relevant applications.