Title: Reasoning from Data in Science: Rigor, Context, Faith, and Progress
Abstract:
In statistical research and teaching we rightly focus on technical methods but, in my view, especially in teaching, the underlying attitudes and principles are often given too little attention. I will use two examples drawn from neuroscience to illustrate four general themes that help explain many great advances in reasoning from data. The first part of the lecture will summarize analyses of data from neural spike trains and local field potentials, based on a pair of papers. In the second part I will go over the themes, explaining what those four labels are supposed to mean, and why they are so important.
Background and additional details can be found from my website. The two examples come from Chen et al 2022, J. Neurophysiology; Klein et al, 2020, Ann. Applied Stats. General background on computational neuroscience is covered in a 2018 article I wrote with 24 others for the Annual Reviews (and I invite people to watch the 10-minute video, What is computational neuroscience? https://youtube.com/watch?v=wm-wxS1CsY4) Many specific methods are summarized (without math, aiming at experimentalists) in a 2023 review I wrote with 4 of my graduate trainees for the Journal of Neurophysiology. Some of the philosophical perspective comes from my 2011 article, "Statistical Inference: The Big Picture" and my 2021 commentary, "The Two Cultures: Statistics and Machine Learning in Science."
