@tdietterich @yudapearl That makes a lot of sense to me. Still there seems to be a lack of clarity around how variables should be chosen, which choices are valid, which are useful. See e.g. https://t.co/OWSaOYFZTq https://t.co/IoOdprp6qs For CRL one woul
..and probabilistic models https://t.co/0wGOtX6YfL If variables are not properly defined/constructed (cf. above), we may run into problems of ambiguous manipulations https://t.co/GUUVzvAsA9 and green and grue variables https://t.co/Hc8GcoOcD5 2/
@atypical_me @KordingLab @neuroccino That is an excellent question :-) Some works at UAI 2017/18 addressing this from a conceptual methodological perspective by Beckers and Halpern https://t.co/3FFNcvIHHG based on this https://t.co/iWaFzMYnJD; more here ht
@atypical_me @GunnarBlohm @KordingLab @dan_marinazzo This great article by Frederick Eberhardt may be a good starting point, instructive, illustrative, and well written https://t.co/Hc8GcoOcD5 "observations of some variables Xi, let's infer the causal str