The Study Of Causal Power Models

Causal power is a new representative approach to the study of causation,whose core idea is to introduce the concept of power in natural science to explain causation.Firstly,this thesis explains the meaning of causal power by examining different definitions of power and analyzes its main characteristics,namely intrinsicality,directedness,reality and independent.Then I study two representative models,the power probabilistic contrast model and vector model of power.The power probabilistic contrast model depicts causal powers in terms of the covariation-a core concept of probability theory,including the characterization of simple powers,contextual powers and conjunctive powers.The basic idea is to represent unobservable powers in terms of observable frequencies(i.e.,covariation).The vector model of power is based on the directedness of power,in which vector is used to represent causal power.This model can depict how causal powers interact with each other accurately and intuitively.On this basis,I discuss the latest research progress of these two representative paths,and point out the trend of causal power research.The main work of this paper is as follows:First,by examining different analyses of power,the essential meaning of power is revealed,and basic on this,how causal power acts in causation is discussed.This paper also examines two strategies for explaining causal power,concludes two models for understanding causality by virtue of causal power:stimulus-response model and mutual manifestation model,and explores the similarities and differences between them.Secondly,through the realistic analysis of causal power,the power probabilistic contrast model lays a philosophical foundation,which makes it more reliable.Since the power probabilistic contrast model has some obvious defects,this paper adds the Glymour's scheme and Korb's causal information scheme.Based on these discussions,I combine the vector model with them to give the trend of causal power research in the future.Third,this paper expands the concept of'threshold'in the vector model,which ignored by Stephen Mumford,but I think it is necessary to do more in-depth analysis.The threshold is also used to solve two basic problems of causality in this thesis,which shows that causality has limited necessity and no transitivity.This proves the modification can further improve the problem solving ability of vector model.