Your Paperpal-edited Manuscript

John Paperpal, Jane Cactus

Corresponding Author: example@example.com

Abstract

Estimation of learning curves is ubiquitously based on the proportion of correct responses within moving trial windows. ThereforeThereby, it is assumed that the learning performance remainsstaysconstant within the moving windows, which may oftennot be the case. In the present study, we demonstrate that this assumption's violations lead to systematic errors in the analysis of learning curves, and. We explore the dependency of these errors on window size, different statistical models, and the learning phase. To reduceFor reducing these errors in the analysis of single subject data, we propose adequate statistical methods for the estimation of learning curves and construction of confidence intervals in a trial-by-trialtrial by trial manner. Applied to data from an avoidance learning experiment with rodents, these methods revealed performance changes occurringhappening at multiple time scales across training sessions. Our work also shows that the proper assessment of the behavioral dynamics of learning at high temporal resolution can highlight specific learning processes, thus allowing the refinement ofallows to refine existing learning concepts. It further disambiguates the interpretation of neurophysiological signal changes recorded during training in relation to learning.

Introduction

Learning, the acquisition of knowledge through experience, manifests as behavioral changes duringin the course of training. Learning behavior reliesdoes rely on a multitude of neural and cognitive processes that act on different spatial and temporal scales [1–3]; however, many of these processes are not experimentally accessible. Therefore, any particular learning experiment is influenced by numerous uncontrolled variables. This entails a certain degree of unaccountable variability of behavior across time within a subject, as well as between subjects [4]. As a consequence, single behavioral responses of individual subjects are difficult to interpret with respect to learning.