Perceiving the arrow of time in autoregressive motion
Authors: Kristof Meding, Dominik Janzing, Bernhard Schölkopf, Felix A. Wichmann
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Here we show that human observers can indeed discriminate forward and backward autoregressive motion with non-Gaussian additive independent noise, i.e. they appear sensitive to subtle asymmetries between the time directions. We employ a so-called frozen noise paradigm enabling us to compare human performance with four different algorithms on a trial-by-trial basis: |
| Researcher Affiliation | Collaboration | Kristof Meding University of Tübingen Neural Information Processing Group Tübingen, Germany kristof.meding@uni-tuebingen.de Dominik Janzing Amazon Research Tübingen Tübingen, Germany janzind@amazon.com Bernhard Schölkopf* Max-Planck-Institute for Intelligent Systems Empirical Inference Department Tübingen, Germany bs@tuebingen.mpg.de Felix A. Wichmann* University of Tübingen Neural Information Processing Group Tübingen, Germany felix.wichmann@uni-tuebingen.de |
| Pseudocode | No | The paper describes algorithms but does not include structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper states 'We refer to the supplementary material for detailed explanations and all information needed to allow all experiments to be reproduced.' but does not explicitly state that the methodology's source code is open-source or provide a direct link to it. |
| Open Datasets | No | The paper describes generating synthetic time series data for experiments and training ('We constructed time series from a generative additive noise model', 'For each noise distribution the network was trained with 30000 time series'), but does not provide a link or specific citation to a publicly available dataset. |
| Dataset Splits | No | The paper mentions training time series for the neural network ('For each noise distribution the network was trained with 30000 time series') and screening for human participants, but does not provide explicit train/validation/test split percentages or sample counts for the data used in the experiments. |
| Hardware Specification | No | No specific hardware details (like GPU/CPU models or memory) used for running experiments are mentioned in the paper. |
| Software Dependencies | Yes | The following psychometric functions and Bayesian Credible Intervals (CI) were calculated with the Beta Binomial Model in Psignifit 4 [54]... Res Dep relies on the ARMA method in MATLAB; |
| Experiment Setup | Yes | For each noise distribution the network was trained with 30000 time series. We used the Adam optimizer with an initial learning rate of 0.01. The network was trained for a maximum of 30 epochs. |