Transitive Inference as a probabilistic preference learning task, using one-parameter Mallows models.
This repository contains the code implementation for the paper "Transitive Inference as Probabilistic Preference Learning" by Francesco Mannella and Giovanni Pezzulo, affiliated with the Institute of Cognitive Sciences and Technologies at the National Research Council in Rome, Italy.
TransitiveInference.ipynb contains the python code for the first simulation which explores the decision dynamics of a well-trained Mallows ranking model, replicating key features of TI observed in human and animal studies and the second simulation demonstrates the learning dynamics of the model when confronted with a novel TI task, showcasing its ability to connect previously separate rankings.
MallowsNN.ipynb contains the python code reguadsing the third simulation provides insights into the neural-like codes harnessed by the model to perform TI.