NIAS Thesis Proposal Presentation: Theoretical and Experimental Investigations into Causality, its Measures and Applications
National Institute of Advanced Studies
Indian Institute of Science Campus
Bengalure-560012
Thesis Proposal Presentation
Title:
Theoretical and Experimental Investigations into Causality, its Measures and Applications
Date: Friday, 11 August 2017
Speaker:
Ms. Aditi Kathpalia
Time: 2:00 pm
Venue: Lecture Hall, NIAS
Abstract:
Testing of Causality between any two given systems is of enormous interest in a wide variety of disciplines including economics, network dynamics, epidemiology, physics, earth sciences, neuroscience and also consciousness research. Several model-based and model-free measures for estimating causality between two time series exist, however they suffer from some severe limitations as we shall demonstrate. In this thesis, we propose, implement and compare the performance of a new measure of causality and its variations with the existing measures on simulated and real datasets. This new measure is based on the idea of compression-complexity and draws upon learning and modelling implicit in lossless data compression algorithms. As a second step, we would like to understand what aspect of causality each of the existing measures is trying to capture – is the causal agent in each case, the predictor/ controller/ chooser of the caused variable or is there simply an information flow from one to the other?; and investigate if the function of all these causality measures can be understood under a single mathematical framework. In the last part, we wish to model retrocausal processes (in which the future could cause or influence the present)with possible implications to explain phenomena such as the experience of dreams during sleep, existence of free will and the logic of the mind. For this, we would proceed by questioning some of the fundamental assumptions like one dimensional linear arrow of time on classical scales, and draw inspiration from self-referential nature of paradoxes as well as coupled nonlinear chaotic dynamical systems exhibiting the phenomenon of ‘anticipating synchronization’ where the causality trend appears to be reversed.
All are invited to attend