Endogenous Uncertainty and the Dynamics of Constraints

An agent- based model of social dynamics is introduced using a deformable fitness landscape, and it is shown that in certain clearly specifiable situations, strategies that are different from utility maximization outperform utility maximizers. Simulation results are presented and intuitive interpretations of the results provided. The situations considered occur when individuals' actions affect the outcomes for other agents and endogenous effects are dominant. The Tragedy of the Commons is merely a special case of this. Arguments are given that constraints are to be encouraged in some circumstances. The appropriate role of constraints in various types of society is assessed and their use justified in identifiable types of situations.

Analytic versus Synthetic Understanding

It is argued that understanding is an epistemically more important concept than explanation. Two kinds of understanding are discussed, that based on analysis of the explanandum and a kind that is based on a synthetic method. The distinction between primary and secondary understanding shows that who is asking the why question is of central importance. A problem for causation is introduced that revolves around the degree of isolation that is needed to identify a causal factor and to distinguish causes from their carriers. It is shown that the analytic and synthetic approach can deal with partial understanding, that there can be an increase in understanding without an explanation, and that laws are not necessary for understanding.

Aleatory Explanations

The question of what role negatively relevant factors should play in explanations that refer to probabilistic causes is addressed. A distinction between contributing and counteracting causes is drawn. Arguments are given that citing the probability value as part of the explanation is unnecessary, that restricting the explanatory factors to only the positively relevant factors is misleading, and that listing some, but not all, of the contributing and counteracting causes allows for partial but true explanations to be given. The canonical explanatory form for probabilistic causal explanations "X because Y despite Z" is introduced and linguistic variants of the canonical form are given.

Emergence

Retrospective reflections are provided on the papers “How Properties Emerge,” “Emergence, Not Supervenience,” “Synchronic and Diachronic Emergence,” and “Computational and Conceptual Emergence” by Paul Humphreys. Remarks are made on the respective merits of diachronic and synchronic approaches to emergence and it is noted that fusion emergence is a special case of the author’s more recent transformational emergence. An argument is given that holism should be imposed as a requirement for emergence only on synchronic accounts, and a reevaluation of the six conditions for emergence given in the author’s “Emergence, Not Supervenience” paper is provided. Elaborations of why features are conceptually emergent or inferentially emergent are given and brief remarks on theoretical approaches to emergence are provided. A clarification of the author’s position that introducing levels of properties is in many cases a methodological convenience rather than a recognition of ontological reality is given.

Emergence, Not Supervenience

It is argued that supervenience is an inadequate device for representing relations between different levels of phenomena and that emergence is a better vehicle for interlevel relations. Various interpretations of supervenience relations are examined and they are found to be excessively thin compared to explainable emergence relations. A question is raised about the differences between `vertical’ and `horizontal’ determination relations. Some possible examples of emergence are given and six criteria are provided that emergent phenomena seem to satisfy. Using examples drawn from macroscopic physics, it is suggested that such emergent features may well be quite common in the physical realm.

Computer Simulations

The need to solve analytically intractable models has led to the rise of a new kind of science, computational science, of which computer simulations are a special case. It is noted that the development of novel mathematical techniques often drives scientific progress and that even relatively simple models require numerical treatments. A working definition of a computer simulation is given and the relation of simulations to numerical methods is explored. Examples where computational methods are unavoidable are provided. Some epistemological consequences for philosophy of science are suggested and the need to take into account what is possible in practice is emphasized.

Explanation, Understanding, Ontology, and Social Dynamics

Retrospective reflections are provided on the papers “Aleatory Explanations,” “Analytic versus Synthetic Understanding,” “Scientific and Speculative Ontology,” and “Endogenous Uncertainty and the Dynamics of Constraints” by Paul Humphreys. Arguments are given for restricting probabilistic causation to situations in which single case propensities are present and for keeping the pragmatics of explanation as a secondary, rather than a primary, feature of explanations. A finite representation schema for ceteris paribus conditions is provided and some suggestions for how to deal with the pessimistic induction are floated. Some reasons why traditional empiricism should not place constraints on the ontologies we use are given. Motivations for a previously unpublished paper that identifies situations in which individual utility maximization strategies perform worse at the society level than do some common individual strategies are provided and the appropriate role for encouraging certain types of constraints on self-interest is discussed.

Probability and Propensities

Retrospective reflections are provided on the papers “Why Propensities Cannot Be Probabilities,” “Some Considerations on Conditional Chances,” and “Probability Theory and Its Models” by Paul Humphreys. A discussion of whether probability theory is a mathematical or an empirical theory is provided and the point made that mathematical theories are not revised but replaced when used as models of empirical phenomena. Probability theory qua formal theory has a mathematical interpretation but any empirical interpretation, contra Quine, is completely detachable. A replacement for Quine’s web metaphor is suggested. The author assesses Donald Gillies’ response to Humphreys’ Paradox, and reasons not to abandon the single case propensity interpretation of probabilities are given. Responses to the paradox by Mauricio Suárez, Isabelle Drouet, Leslie Ballentine, and David Miller are discussed, and an argument given that the temporal evolution approach is primary for absolute propensities.

Introduction

During a summer break in college, I set out to solo hike the Pennine Way, a 267-mile footpath snaking across the moors of northern England. One day, my first goal was Stoodley Pike, a hill topped by a 121-foot stone folly. That morning, the Pike was covered in a dense fog, with visibility down to about ten yards. I lost the trail, and for about an hour, I had no idea where I was or whether I was hiking toward or away from the monument. I seemed destined to spend the rest of my holiday on the Pike, but suddenly the mist parted and right in front of me was the folly. When doing philosophy I am often reminded of that morning....

Why Propensities Cannot Be Probabilities

Propensities are often taken to provide an interpretation for probability theory that is congenial to single-case probabilities. An argument is provided to show that propensities cannot be correctly represented by conditional probabilities because they do not in general satisfy Bayes’ theorem or the multiplication principle. The first section of the chapter provides an informal version of the argument, indicating how the causal nature of propensities cannot be adequately represented by standard probability theory. In the second section, a full version of the argument is given so that the assumptions underlying the informal account can be precisely identified. The third section examines those assumptions and deals with objections that could be raised against the argument and its conclusion. The fourth and final section draws out some rather more general consequences of accepting the main argument.