Perspectivism in Science: The Good, the Bad, and the Ugly
There are ways in which science is undeniably perspectival, as when different contexts of investigation – for example, the use of different apparatuses and experimental setups, the application of different models to specific target systems, and the analysis of systems at different scales or “levels” of organization – generate data and descriptions of scientific phenomena that can only be produced in idiosyncratic ways. The routine pursuit of contextual investigations of these sorts and the genuine forms of knowledge they may produce are, of course, good things. The notion of perspectivism in science becomes a bad thing, however, when the general slogan that scientific knowledge is perspectival is bandied without careful attention to the epistemological consequences of perspectival investigation in different sorts of cases. So long as putatively perspectival knowledge concerns different entities, properties, and processes in the world, or so long as it concerns the same things but in ways that are unifiable in prospect or in principle, there is no threat of contradiction and perspectivism remains a good thing vis-à-vis knowledge. In cases where putatively perspectival knowledge comes in the form of direct and apparently irreducible contradictions regarding the properties of one and the same thing, however, this is bad for anything other than an instrumentalist view of the relevant science. Indeed, this kind of situation rapidly becomes ugly for the scientific realist in the absence of a clear specification of how such inconsistencies can be understood. I argue that to the extent that one can make sense of combining perspectivism with realism, the combination is a good thing methodologically speaking, but epistemologically unremarkable. On the other hand, where inconsistency threatens, the combination of perspectivism and realism is either unstable or incoherent regarding knowledge, which is bad and ugly, respectively. I explore some options for how a realist might embrace the good and avoid the bad and the ugly.
Models and Methodology: More Problems for Perspectival Realism
One of the goals of scientific perspectivism is to address problems that arise as a result of the use of inconsistent or incompatible models. In Morrison (2015) I argued that perspectivism is unsuccessful in this respect because for difficult cases it collapses into a form of instrumentalism. And, where it is a successful methodology, as in cases of scaling, it is already implicit in the practice and so philosophically redundant. However, as a general philosophical thesis or methodology for modelling, perspectivism faces further problems in that there are a variety of modelling practices for which perspectivism is difficult, if not impossible, to motivate. I discuss some of these cases and practices, in particular multi-scale modelling, toy models, and the turbulence modelling with an eye to showing why in each case perspectivism is unhelpful in providing philosophical insights on these challenging cases. While perspectivism may have interesting implications as a global epistemic position, it seems to offer little hope of underwriting any form of scientific realism.
Cognitive and formal bases of perspectival models
Much of the work on “perspectivalism” about scientific theories and objects has focused on ways that a scientific model could be “perspectival” in some interesting sense, and the impacts on scientific practice or interpretation. In this talk, I instead explore whether it might be, in some sense, inevitable that we have perspectival scientific models. That is, I focus on the causes (rather than effects) of such models, though not contingent causal factors such as specific technologies or power relations within particular scientific communities. In contrast, I am principally concerned with cognitive and formal features (of human scientists investigating scientific problems) that are largely universal or ubiquitous in scientific inquiry, and that (nearly) inevitably lead to perspectival models. My talk considers multiple such factors of each type; in this abstract, I give just one example of each.
On the cognitive side, consider the cognitive necessity of conceptualization. Humans simply cannot function unless they understand the world in terms of concepts and categories that group together individuals with shared features, structures, or roles. The use of concepts to understand our world is not a contingent matter; we cannot do otherwise. Unsurprisingly, our scientific models are also based on our scientific concepts. Interestingly, however, a number of recent cognitive science experiments have revealed that our concepts—both everyday and scientific—do not simply “mirror” the structure or natural kinds of the world. Instead, our concepts are heavily shaped by our goals: the reason why we need some conceptual scheme influences both the structure and content of the resulting concepts. For example, the plant-centric concepts of a landscape designer/architect are importantly different from those of an evolutionary biologist. And since our scientific concepts are goal-relative (because all of our concepts are goal-relative in this way), then the resulting scientific models will have a measure of goal-relativity; they will be perspectival.
On the formal side, one standard view is that we have strong (though perhaps defeasible) reason to be realists about the objects of our best scientific theories. The notion of ‘best’ here is, however, relative to an evaluation standard: different theories will be superior depending on the metric by which we judge them. This goal-relativity of theories does not necessarily imply that they are perspectival, however, as we might be able to reconcile them in some way. However, one can prove that, for a wide range of formal conditions, the objects of different “best” theories for some domain will be incompatible (i.e., the objects that cannot be inter-translated). In these cases, we have no reason to privilege one of the evaluation standards, and so the formal constraints of the scientific learning problem lead directly to perspectival models.
Other cognitive and formal factors lead to the same overarching conclusion: our scientific models should almost certainly be perspectival, in the sense that the scientists’ goals and desired functions for the models matter. More generally, these considerations provide a normative justification for perspectival pluralism.
Cancer Modelling and the Advantages and Limitations of Multiple Perspectives
Cancer is a paradigmatic case of a complex causal process; causes of cancer operate at a variety of temporal and spatial scales, and the temporal order and interactive effects between cancers’ causes can have significant effects on how cancer progresses. Because of this complexity, models of cancer often involve deliberate choices to focus on one time scale, one causal pathway, or one aspect of cancer’s dynamics. Indeed, as in most of biology, modelling cancer involves simplification and idealization. Yet, for this very reason, critics of some models of cancer progression argue that these models and the theoretical framework associated with them are “degenerating research programs.” This paper will discuss when and why it is permissible to ignore some features of cancer’s causes in the context of modelling cancer’s dynamics. Part of this will involve explicating an account of modelling in the sciences; part of this will involve defending multiple perspectives on cancer causation. While there are certainly limitations to simple models, the seriousness of these limitations will depend upon our views of the aims and scope of theoretical modelling. I hope to bring this example to bear on debates among philosophers of science over perspectivism and realism, as well pluralism about the aims and scope of scientific theory. I take my view to be in the spirit of Massimi’s epistemic perspectivism, drawing in part also on Chang’s argument for the epistemic benefits of pluralism.
Mary S. Morgan
Measuring Instruments for Multiple Perspectives
The most interesting and intractable measurement problems in economics arise in developing measuring instruments for economic concepts that refer to complex kinds: non-simple aggregates of various sorts (the economy, the price level) or situations/states that are multidimensional in aspect (poverty, development, class). For these cases, some kind of a perspectivalism in measurement is helpful as it enables the separate and different aspects to be taken into account. But the danger in such measurement systems, in trying to capture all the different aspects of that complexity, is not to create the subjective ‘view from nowhere’ (one person’s perspective) but rather the ‘view from everywhere’ (where all perspectives are taken into account), making the project meaningless. The most effective of these measuring instruments rely on strong principles that both allow for multiple perspectives and provide the rules to integrate them to produce single measurements.
Two Senses of Integration
Scientists and philosophers often assume that integration is good in and of itself. This talk will consider integration from a pragmatic viewpoint. Integration will be viewed as a heuristic, useful for some purposes in some contexts, but possibly counterproductive for other purposes or contexts. I will examine integration in the the sciences of human behavior by analyzing the debate between James Tabery, who argues for integration, and Helen Longino, who argues that the perspectives of the different sciences cannot and should not be integrated. My analysis will distinguish between two senses of integration. Explanatory integration, perhaps the dominant sense in the contemporary literature, points towards connecting perspectives. The other sense, activity integration, does not. I will argue that Tabery is correct in thinking that integration is important for the advancement of human behavioral sciences. But this integration involves the investigative activities of one science informing the investigative activities of another, not the connecting of explanations or perspectives. I will argue that Longino is correct in thinking that the explanations of different behavioral sciences will remain separate. I will conclude that a plurality of perspectives is valuable, and seeking explanatory integration could be detrimental for the purposes of investigating human behavior.
Concept dynamics and the realism question
There is no knowledge without concepts, and there might be no scientific knowledge without perspectives. My talk takes those observations as starting point to explore the dynamics of scientific concepts and its relation to perspectives. I shall treat various developments in the history of electricity and magnetism with this goal in mind. Particular attention will be given to the varying roles of realistic commitments in those developments, and further questions for history and philosophy of science be raised.
Perspectival realism about what? Tracking the electron across shifting theoretical perspectives
The history of scientific objects is rife with diverse perspectives on them, associated with different concepts, theories, models, and experimental devices. This diversity prima facie undermines the realist aspirations of perspectivism: A plurality of perspectives on an object, especially a hidden (‘theoretical’ or ‘unobservable’) one, throws doubt upon its very identity, leaving us wondering whether those perspectives refer to a single, same object. In my talk I will address this worry with reference to the early history of the electron. Throughout that history, from the 1890s to the 1920s, the electron was considered from multiple, often incompatible, theoretical perspectives. I will make a case for understanding those multiple perspectives in a realist manner, that is, as perspectives on the same ‘thing’. To that aim, I will classify perspectives on the electron into three distinct levels: high level theoretical frameworks specifying its ‘nature’ and the laws that it obeys; intermediate level models representing its behavior in specific physical systems; and low level knowledge of its properties derived from particular experimental situations. I will then argue that low level, experimentally obtained knowledge is sufficient for tracking the electron across shifting higher level perspectives.