Recovering Quantitative Models of Human Information Processing with Differentiable Architecture Search
This work addresses the challenge for cognitive scientists who accumulate large datasets but lack resources to manually integrate them into theories, though it appears incremental as it applies existing machine learning techniques to this domain.
The authors tackled the problem of automating the construction of quantitative models of human information processing by developing a pipeline using neural architecture search and automatic differentiation, and found it capable of recovering basic quantitative motifs from synthetic data in areas like psychophysics, learning, and decision making.
The integration of behavioral phenomena into mechanistic models of cognitive function is a fundamental staple of cognitive science. Yet, researchers are beginning to accumulate increasing amounts of data without having the temporal or monetary resources to integrate these data into scientific theories. We seek to overcome these limitations by incorporating existing machine learning techniques into an open-source pipeline for the automated construction of quantitative models. This pipeline leverages the use of neural architecture search to automate the discovery of interpretable model architectures, and automatic differentiation to automate the fitting of model parameters to data. We evaluate the utility of these methods based on their ability to recover quantitative models of human information processing from synthetic data. We find that these methods are capable of recovering basic quantitative motifs from models of psychophysics, learning and decision making. We also highlight weaknesses of this framework and discuss future directions for their mitigation.