Presentation on theme: "1 Mechanistic Causality: Dispositions vs. Structures Lorenzo Casini"— Presentation transcript:
1 Mechanistic Causality: Dispositions vs. Structures Lorenzo Casini
2 Outline Mechanisms & complex systems Glennans account & latest views Dispositionalist interpretation The case of asset pricing Between dispositions and structures
3 Complex systems Systems whose behaviour result from (rule-based) interactions of many (different) components and exchange (of e.g. energy, mass, information) with environment Can display one or more among: –Nonlinearities –Sensitivity to initial conditions –Self-organisation –Adaptivity
4 Mechanistic causality The given: –Complex systems sciences study mechanisms (cf. Bechtel & Richardson, B&A, Kuhlmann, etc.) –In C S S, talk of causal relations and of mechanisms often go together Working hypothesis: –causal relations in complex systems have to do with mechanisms Desiderata for account: –informative about truth conditions –provide explanation of phenomena
5 Glennans account (C) Event A causes event B iff there is a mechanism (M) which connects them (1996: 49, 56, 64) All sorts of mechanisms between any two events. How to select the right one? (M) A mechanism for a behavior is a complex system that produces that behavior by the interaction of a number of parts, where the interactions between parts can be characterized by direct, invariant, change-relating generalizations (2002: S344) i.Interactions are only so characterised – what are they? ii.Events are related by mechanism (=complex system); complex system = object; whence, events mediated by an object – not a process ? Relation between object and process?
6 Latest views Causal claim relates events (property instances) and has the form: Event c causes event e (in background conditions B) in virtue of properties P (of c, e, or B) (...) For instance, Bob's coughing (c) caused Carol to wake up (e) in virtue of cough's loudness (P). (2010a: 364) productionrelevance relates eventsrelates properties singulargeneral non-counterfactualcounterfactual truth conditionsexplanation production provides truth conditions : to say that one event produced another is to say that in fact the causative event is connected to the effect via a continuous chain of causal processes (2010a:365-6)
7 (i) relation process – object Mechanism is both a system and a process – which are so related: Mechanism" is used to describe two distinct but related sorts of structures. First, mechanisms are systems consisting of a collection of parts that interact with each other in order to produce some behavior. (…) Second, mechanisms are temporally extended processes in which sequences of activities produce some outcome of the mechanisms operation. (…) There is a natural relationships between processes and systems, for the operations of systems give rise to processes. (2008: 376) (Couldnt it be other way round ? )
8 (ii) nature of interactions Interactions are only characterised as.. – what are they? an interaction is an occasion on which a change in a property of one part brings about a change in a property of another part. For instance, a change in the position of one gear within a clock mechanism may bring about the change in the position of an interlocking gear. Interaction is a causal notion that must be understood in terms of the truth of certain counterfactuals. (2002: S344) What makes counterfactual assertion true? singular determination, i.e. exercise of power When a change in a produces a change in b, it follows (with the usual caveats about overdetermination, etc.) that if a had not changed, b would not have changed. But the counterfactual locution should be understood not as a claim about non-actual worlds, but a claim about the determining power of a in this world. (2010b, sec.5)
9 Ambiguity remains What is the truth maker of a causes b ? 1.determining power of a or 2.continuous chain of causal processes between a and b Other sources of ambiguity: Glennan also talks of causal rel between events as if it is relevance rel the set of events causally sufficient to bring about an effect are typically large, so that when we speak of the cause of an event, we are using pragmatic criteria to single out a certain event as especially salient. (2010a: 364-5) Powers are usually ascribed to objects not events – can be OK but we need a (dispositionalist?) story here..
10 Chakravartty (2007) : A causal property is a property conferring to particulars that have it dispositions to behave in certain ways when in the presence or absence of other particulars with causal properties of their own (p.108) causation is a relation of de re necessity between properties, or property instances account of de re necessity follows from account of causal properties identity (pp ) (DIT): what makes a causal property the property that it is are the dispositions it confers to the objects that have it (p 129) causal phenomena are the result of continuous processes of interaction among particulars with causal properties A dispositionalist interpretation
11 talk of events as relata is convenient but elliptical for description ofaspects of such processes identity of particulars (objects, events, processes) is derivative from identity of causal properties. Position entails holism, or ontological circularity: –All laws (general relations between properties) and all causal properties are fixed at once given a set of properties and their distribution What is the truth maker of a causal claim? (...) it is a consequence of DIT that networks of causal properties have a holistic nature. This furnishes a more radical solution to the problem of truthmaking than it is generally appreciated. The existence of any one causal property is a sufficient truthmaker for counterfactuals about all possible relations applicable to the world in which that property is found (p. 146)
12 All this seems in line with complex systems scientists views: Causal relations are not something extra added to predefined noncausal objects. They appear simultaneously with objects in a world that becomes, as a result of systematic individuation, a complex causal network of things and events. Causal relations obtain among states of things in static conditions and among events in dynamic conditions. An example of a static causal relation is the suspension of the Golden Gate Bridge by steel cables. Two states or events are causally relatable if they are connectible by a process, which can be stationary, related if they are so connected. If the connecting process is the change of a thing, then the thing is the agent of interaction. (…) (Auyang 1998, p. 260) Weed (2005): Auyangs conception of a state space, prior to analysis is [that of a reality] composed of actually indefinite strings of activity.
13 For Chakravartty, if the account of de re necessity is viable, then 1.it gives criterion to distinguish causal / accidental regularities; and 2.this criterion is explanatory (p. 130) How does Glennans revised account fare wrt (1) and (2) in complex systems? Are dispositions and de re necessity helpful tools? –First, whilst guaranteeing existence of sufficient truth maker, holism (obviously) doesnt help determine minimally sufficient (local) truth conditions. But these may nonetheless exist, whenever system is sufficiently isolated / carefully described. –Let us consider an example then..
14 (Apoptosis) Weinberg (2007), p. 354
15 Asset pricing. Stylised facts Prob that tomorrows price goes up equals goes down given available evidence (conditional distribution is approx Gaussian). Yet: big (/little) price changes follow big (/little) price changes: changes not uniformly distributed (volatility clustering); asset returns at different t show a dependency (volatility persistence): autocorrelation (correlation between values at different t, as function of t difference) of squared returns decays slowly; distributions of unconditional returns at frequencies of less than one month are fat-tailed: too many observations near the mean, too few in mid range and too many in the tails to be normally distributed. Mechanistic account needs to answer, e.g. What causes crash/bubble? What explains time series? Gaussian and other distributions
16 Asset pricing. Time series Lux &Marchesi (1999), p. 397
17 Asset pricing. Model Lux & Marchesi (1999) – analogy with phase transition phenomena in physics Summary from Kuhlmann (2009)
18 Structural vs. mechanistic account What are the truth conditions for, e.g., Switch of fundamentalists into chartists caused the bubble? What explains, e.g., specific event (crash) or general pattern (fat tails, volatility clustering/persistence)? Smith (1998): no ontological commitment is needed –fit between geometrical structure of model and model of data is sufficient for approximate truth –explanation of behaviour just is a geometrical feature of dynamical model – property of representation of a concrete structure (cf. Goldstein, 1996; Huneman, forthcoming) no appeal to causal notions
19 For Glennan, instead, more is needed: It is possible to formulate a mechanical model using a state space representation but not all state space models are mechanical models. The requirements for a model being a description of a mechanism place substantive constraints on the choice of state variables (such as the fact that state variables should refer to properties of parts), parameters, and laws of succession and coexistence. The satisfaction of these additional constraints is what accounts for the explanatory power of mechanical models. (2005: ) The point is whether these additional constraints can be met in complex systems..
20 Kuhlmann (2011) is halfway between Smith and Glennan: although doesnt reify structure/geometry, contrasts compositionally complex mechanisms (MDC, B&A) and dynamically complex mechanisms (e.g. nonlinear systems, chaotic systems, CA) Similarity: essential to explanation of both –reference to interactions of systems parts (e.g. agents, comparison of profits) –behaviour of the whole system must show some degree of robustness (high volatility for wide rage of parameter values, thresholds for transitions) Difference: these features must be filled in differently behaviour of c.c.m. –ontological details are important –parts maintain identity and function throughout process behaviour of d.c.m. (e.g. apoptosis, asset pricing) –ontological details are less important /irrelevant (e.g. whether and what specific agent buys or sells) –parts can change identity and function (fundamentalists become chartists, optimistic become pessimistic)
21 Result: entities dispositions in complex systems explain less than we hoped: explanation depends largely on structural features of the arrangement. Depending on this: –A has the capacity to produce B if not interfered –A has the capacity to prevent B if not interfered –And vice versa (B has the capacity to produce/prevent A if not interfered) Analogously, for apoptosis: –Depending on geometry, XIAP, by binding to Casp3, which would normally prevent apoptosis, can also promote it, due to XIAPs inability to inhibit Casp9, which is then left free to trigger Casp3 –Caspases are synthetised as inactive and become active by proteolitic cleavage – one can say they are disposed to become active, but mechanists (seem to) view procaspases and caspases as different things with different functions
22 Summary Glennans account and his latest views have ambiguities as regards the nature of truth makers and explanation His account can be made coherent by employing a dispositionalist metaphysics (Chakravartty) In complex systems, talk of mechanisms and causality as involving properties and (dynamical) relations is legitimate However, reference to specific parts with stable identities and functions to explain behaviour and provide truth makers is less appropriate vis-à-vis structural features of the arrangement
23 References Auyang, S. (1998). Foundations of complex-system theories. Cambridge: CUP Bechtel, W., and Abrahamsen, A. (2005): Explanation: A mechanist alternative. Studies in History and Philosophy of Biological and Biomedical Sciences, 36: Bechtel, W., and Abrahamsen, A. (forthcoming): Complex biological mechanisms: Cyclic, oscillatory, and autonomous. In Collier, J., and Hooker, C.A. (eds.): Handbook of the Philosophy of Science, Vol. 10: Philosophy of Complex Systems. New York: Elsevier. Glennan, S. (1996). Mechanisms and the nature of causation. Erkenntnis, 44, 49–71. Glennan, S. (2002). Rethinking mechanistic explanation. Philosophy of Science, 69, S342–S353. Glennan, S. (2005). Modeling mechanisms. Stud. Hist. Phil. Biol. & Biomed. Sci., 36, 443–464 Glennan, S. (2008). Mechanisms. In Psillos, S. and Curd, M. (eds): The Routledge Companion to the Philosophy of Science, 376–384 Glennan, S. (2010). Mechanisms, causes, and the layered model of the world. Philosophy and Phenomenological Research, 81(2): Glennan, S. (2011). Singular and general causal relations: a mechanist perspective. In Illari, P., Russo, F. and Williamson, J. (eds.): Causality in the Sciences. Oxford: OUP. Goldstein, J. (1996). Causality and Emergence in Chaos and Complexity Theories. In Sulis, W. H. and Combs, A. (eds.). Nonlinear Dynamics in Human Behavior, pp World Scientific Publishing. Huneman, P. (forthcoming). Topological explanations and robustness in biological sciences. Synthese Kuhlmann, M. (2011). Mechanisms in dynamically complex systems. In Illari, P., Russo, F. and Williamson, J. (eds.): Causality in the Sciences. Oxford: OUP. Lux, T., and M. Marchesi (1999): Scaling and criticality in a stochastic multi-agent model of a financial market, Nature 397: Machamer, P., Darden, L., and C. Craver (2000): Thinking about mechanisms, Philosophy of Science, 67: Smith, P. (1998). Explaining chaos. Cambridge: CUP Weed, L. E. (2005). Sunny Auyang on complex systems theory: a field being perspective in philosophy of science. Weinberg, R. A. (2007). The biology of cancer. Garland Science, Taylor & Francis Group, New York.