1. A Conditional View of Causality Friedel Weinert University of Bradford

2. Friedel Weinert, University of Bradford (UK) 2 Motivation for Conditional Model  The primary motivation for a conditional model of causation: a) Causal explanations in the sciences seem to be concerned with actual obtaining conditions, rather than counterfactual scenarios (‘What-if-things-had-been-different?’). b) Causal interpretations of quantum mechanical (Franck Hertz experiment) and social events (Weber) force us to abandon the link between the notions of conditionality and regularity. Mackie’s INUS account still retains this link. c) Counterfactual analyses of causation as they appear in interventionist models of causation (Woodward) presuppose knowledge of lawful regularities in the physical world in order to impose constraints on the counterfactuals, which appear in hypothetical inferences. In the absence of such regularities, counterfactuals are much more difficult to evaluate.

3. Friedel Weinert, University of Bradford (UK) 3 Counterfactual Evaluation  Counterfactual Evaluation and the Role of Laws o C 1 : “If the Franck-Hertz experiment had been conducted in Copenhagen, it would have established the discreteness of quantum energy levels in atomic systems.” o C 2 : “If Hitler had been hit by the bullet, which killed his neighbour in a march November 9, 1924, the Second World war would not have happened.” Analysis Analysis: C 1 is easy to evaluate because it is grounded in lawful regularities and the symmetry of laws with respect to space&time reversal C 2 is difficult to evaluate precisely, without additional evidential support, because of the lack of lawful regularities in the social world and its openness.

4. Friedel Weinert, University of Bradford (UK) 4 Mackie’s INUS condition and the conditional view In order to apply Mackie’s account to quantum mechanical and social events, the idea of succession in the Humean sense must be abandoned. We need to think of causality as conditional dependence. ‘Causation’, Mackie holds, ‘is not something between events in a spatio-temporal sense, but is rather the way in which they follow one another.’ Causation is analysed in terms of clusters of conditions. For Mackie, an INUS condition is a (partial) cause, an Insufficient but Non-redundant part of an Unnecessary but Sufficient condition for some E. Thus there is a cluster of factors, making up the cause C, which bring about the effect E. Unlike Mill, however, who took the cause to be the sum total of the conditions, Mackie makes a distinction between necessary and sufficient conditions. Mackie also introduces the notion of a causal field: A causal field comprises the background conditions, which make the normal running of things possible.

5. Friedel Weinert, University of Bradford (UK) 5 Applications of condition model: Franck Hertz experiment  Applications of Conditional Model to 1. Franck-Hertz experiment (1914) This experiment provided experimental evidence of the discreteness of energy levels in atomic systems, which had been proposed in Bohr’s atomic model of the hydrogen atom (1913) In the experiment electrons are ejected from a cathode, C, into a chamber filled with mercury atoms. The energy of the electrons is controllable through the accelerating potential V a. The electrons will make inelastic collisions with the mercury atoms. In the process they will transfer quantized amounts of energy to the atoms, which will be excited by this process to higher energy levels. The following diagrams show the experimental set-up and the measured result.

6. Friedel Weinert, University of Bradford (UK) 6 Franck-Hertz experiment: experimental set-up

7. Friedel Weinert, University of Bradford (UK) 7 Franck-Hertz experiment: measured results

8. Friedel Weinert, University of Bradford (UK) 8 Franck-Hertz experiment: measured results  Some further remarks on the graph:  The peaks and troughs in the graphs are explained causally; if electrons near the grid, G, have the necessary amount of energy to excite the atoms to higher energy levels, they will not have sufficient energy to overcome the retarding potential, V r, to reach the anode, A, which explains the drop in I a.  The electronic energy input and the quantum state of the atoms in the gas can be considered as necessary conditions, since in their absence we would not expect the results (on the Bohr model).  The type of atom used in the chamber is a sufficient condition, since various kinds of atomic systems can be used to run the experiment.  The experiment confirmed the loss of the electronic energy at discrete levels and confirmed that the higher states of energy of the atoms corresponded to the Bohr postulate of discrete energy levels.  In order to interpret this experiment causally we need a probabilistic notion of causality, since we are not in a position to predict which electron will collide with which mercury atom and how much energy it will transfer. There is no causal trace from the antecedent to the consequent conditions.

9. Friedel Weinert, University of Bradford (UK) 9 Applications of condition model: Weber’s ‘adequate causation’  Applications of Conditional Model to 2. Weber’s notion of ‘adequate causation’: If C 1, C 2, C 3 represent a cluster of antecedent causal conditions and E represents an effect in the social world, then generally Pr 1 ( E|C 1, C 2, C 3 ; B ≠ 0) > Pr 2 (E|C 3, C 4, C 5 ; B ≠ 0). The expression ‘Pr 1 > Pr 2 ’ captures Weber’s notion of adequate causation. B ≠ 0 means that background conditions cannot be set to zero; social systems always remain open and are characterized by symbolic dimensions. In terms of a conditional model of causation, Weber clearly argues that some clusters of conditions are more adequate explanations of a social event than others. The ‘adequacy’ must be established by evidence, which favours one causal interpretation over alternatives. (E.g. some causal accounts of the outbreak of WWII are more adequate, in view of the evidence, than others.)

10. Friedel Weinert, University of Bradford (UK) 10 Background Conditions and Trends  ‘Adequate’ causation in the social sciences differs from causation in the physical sciences in at least two ways: a) the background conditions, even in laboratory experiments with human subjects, cannot be set to zero (B ≠ 0); the background effects are not calculable; this is due to the open-endedness of social systems and their symbolic dimensions; the epistemic status of social regularities is also unclear; b) it is questionable whether lawlike regularities exist in the social sciences, although trends clearly do exist in the social world. This claim can be supported by two arguments:

11. Friedel Weinert, University of Bradford (UK) 11 Claim: there are no social laws justified  Trends depend on initial conditions; they are inductive generalizations over initial conditions; they do not impose structural constraints, like laws, but only practical constraints on the social world.  Trends are clearly social regularities. But they differ markedly from lawlike regularities in the physical world. Lawlike regularities, either of a deterministic or statistical nature, cannot be reversed. They constitute structural constraints on the natural systems, which obey them. But social trends can be reversed, say through changing social practices. Moreover, the very knowledge of trends can be used to reverse the very trend that has been observed. (Example: alcohol consumption and car accidents.) Therefore trends impose at most practical constraints on social agents.

12. Friedel Weinert, University of Bradford (UK) 12 Conclusion  Conclusion: A philosophical model of causation should be workable in the practice of the natural and social sciences. It should also capture causal relations in the physical and social world, where according to the evidence they seem to exist. The openness of social systems and the lack of lawlike generalities in the social world make it unlikely that either causal-mechanical or counterfactual-interventionist models of causation can be successfully employed by the social scientist. Although there are significant differences in the sets of antecedent and consequent conditions, depending on whether the conditional model discusses a physical or social event, the conditional model is sufficiently general to be applied to both the natural and the social sciences.