British Academy: The UK's National Academy for the Humanities and Social Sciences
Enquiry, Evidence and Facts: An Interdisciplinary Conference
Evidence, Inference and Action: Towards a New Philosophy of Evidence
Dr Grant Fisher
Department of Science & Technology Studies, University College London,
Gower Street, London, WC1E 6BT
An abstract presented to the conference
‘Enquiry, Evidence and Facts: An Interdisciplinary conference’
at the British Academy, London, on 14 December 2007
Biography
Grant Fisher obtained a PhD in Philosophy from the University of Leeds in 2004. He has taught philosophy of science at the University of Leeds and history of science at Durham University. He is currently a Research Fellow in the Department of Science & Technology Studies at University College London. He works on the Evidence in the Natural Sciences project within the Evidence, Inference & Enquiry programme. His interests are in philosophy of science, especially confirmation, theories and models, scientific representation and history and philosophy of chemistry.
Abstract
This paper is concerned with outlining a new philosophy of evidence, a theory of epistemic action, illustrating how it might be applied to some of the fundamental problems in philosophy of science. Within traditional philosophy of science, particular emphasis is placed on expressing the confirmation relation, the relationship between evidence and hypotheses or theories, as a logical relationship between statements. Adopting this perspective has proved to be enormously productive, resulting in intuitively plausible qualitative analyses of confirmation and applications of various interpretations of probability to the logic of induction. However developments in confirmation theory have been conjoined by a number of puzzles and paradoxes that have had a strong impact in not only philosophy of science but also epistemology more broadly. We argue that in order to confront these problems and to effectively capture the practices of evidence collection, interpretation, assessment, deployment and evidential reasoning in the natural sciences, an analysis is required in which confirmation is understood as consisting in more than merely the logical relationships between statements.
Admitting extra-logical “contextual factors” into the study of evidence in philosophy of science has prima facie plausibility. In recent years, some contributions to the study of scientific explanation have shifted from the idea that explanation issues in the logical relationship between statements describing events or regularities to be explained and the laws and initial conditions that explain them. What constitutes an effective explanation is not a matter of logical subsumption of explanadum by explanans; it depends on pragmatic and contextual factors related to who requests an explanation, who provides it and for what reasons. Yet while such contextual factors might be admitted some significance in philosophy of science, they are rarely attributed a central function in the justification of beliefs in the natural sciences. By looking to the practices of evidence gathering, interpretation and use, we claim that confirmation is intrinsically contextual. This claim amounts to more than the idea that contextual factors are somehow “relevant” to confirmation in some loosely specified sense. Rather, the relationship between hypotheses and evidence is intrinsically contextual in a sense that is robustly epistemic.
A philosophy of evidence so conceived is born out of an analysis of scientific practices, and hence human agency takes centre-stage. Evidence is employed by agents in a variety of contexts, for example in testing hypotheses, generating new theories, or in providing explanations. Testing, theory-building and explaining each constitute different evidential activities executed within different contexts of scientific practice. To hypothesis testing itself, one might ascribe practices such as evidence seeking, gathering, interpretation and presentation. The manner in which these practices are conducted has an impact upon the justification of scientific beliefs because hypothesis testing is an epistemic activity: an activity in which evidence is used in accordance with certain constraints and conditions of coherence. Hypothetico-deductive confirmation, Popperian falsification, and Bayesian confirmation are all different forms of epistemic activities in the sense that they are distinctive testing activities. They are characterised by different actions directed to, for example, the collection of evidence, different choices regarding what constitutes admissible evidence and how that evidence relates to an hypothesis.
Human action is normally a topic associated with philosophy of mind and ethics, but it is rarely engaged within philosophy of science and epistemology. Yet we show that the study of evidence in the context of epistemic action not only offers a unique perspective on evidence broadly, but can also yield fresh perspectives on some old problems. For example, reappraising Hempel’s famous paradox of confirmation (the ‘ravens’ paradox) has been crucial to developing our ideas on evidence and has resulted in an alternative response to the problem. We reformulate the paradox, claiming that the genuinely paradoxical point is not the evidential relevance of seemingly irrelevant objects, but the variable evidential value of one and the same object with respect to a given hypothesis. The central thrust of our claim concerning the ravens paradox, and our view of evidence more broadly, is that observations do not have an intrinsic evidential value. Observations of events or phenomena do not constitute “evidence” in themselves or in isolation; they come to have evidential value as part of an epistemic activity, that is, a coherent testing activity. That a single observation can have a variable evidential value depends on what sort of testing activity is employed by agents in order to test an hypothesis. Seemingly irrelevant evidence, such as contrapositive instances, can have confirmatory value with respect to a hypothesis and that is dependent on which testing activity one is engaged with.
The idea that a single observation can have a variable evidential value may seem to be purely a psychological matter, devoid of epistemological import. The way in which observations are made, and their dependency on a particular testing activity, makes no difference to the justification of hypotheses, or so it might be claimed. However, we illustrate how treatments of scientific observation within the philosophy of science have suffered from what we call the “shoebox illusion”. Scientific observation is often assumed to be as simple a matter as making an observation of a shoe in a box, which is permanently there, static in all its detail, and we can open the box at any time we please in order to observe any aspects of the shoe we care to note. Yet this view fails to take account of the complexities of scientific observations and the nature of human agency. For example, scientific observations can be fleeting and partial, the temporal order of noticing an object’s properties can influence whether or a not an item of evidence constitutes relevant evidence with respect to a given hypothesis, and sometimes we simply cannot, as cognitive agents, record all of the properties an object possesses because our representation of some properties can preclude the representation of others. Hence the idea that observations can have variable evidential values is not “merely” psychological, but is epistemic and integral to the new philosophy of evidence outlined here.