Advances in deep reinforcement learning enable better predictions of human behavior in time-continuous tasks

Abstract

Humans have to respond to everyday tasks with goal-directed actions in complex and time-continuous environments. However, modeling human behavior in such environments has been challenging. Deep Q-networks (DQNs), an application of deep learning used in reinforcement learning (RL), enable the investigation of how humans transform high-dimensional, time-continuous visual stimuli into appropriate motor responses. While recent advances in DQNs have led to significant performance improvements, it has remained unclear whether these advancements translate into improved modeling of human behavior. Here, we recorded motor responses in human participants (N=23) while playing three distinct arcade games. We used stimulus features generated by a DQN as predictors for human data by fitting the DQNs response probabilities to human motor responses using a linear model. We hypothesized that advancements in RL models would lead to better prediction of human motor responses. Therefore, we used features from two recently developed DQN models (Ape-X and SEED) and a third baseline DQN to compare prediction accuracy. Compared to the baseline DQN, Ape-X and SEED involved additional structures, such as dueling and double Q-learning, and a long short-term memory, which considerably improved their performances when playing arcade games. Since the experimental tasks were time-continuous, we also analyzed the effect of temporal resolution on prediction accuracy by smoothing the model and human data to varying degrees. We found that all three models predict human behavior significantly above chance level. SEED, the most complex model, outperformed the others in prediction accuracy of human behavior across all three games. These results suggest that advances in deep reinforcement learning can improve our capability to model human behavior in complex, time-continuous experimental tasks at a fine-grained temporal scale, thereby opening an interesting avenue for future research that complements the conventional experimental approach, characterized by its trial structure and use of low-dimensional stimuli. Author summaryIn our complex environment, we are constantly responding to outside influences. Traditional trial-based psychological experiments have limitations because they cannot account for the continuous nature of the world. In this study, we combine psychological questions with advanced technology to investigate how humans behave over time. Artificial neural networks have not only proven they can outperform humans in complex tasks but also have the ability to generate features that can be used to model human behavior. We investigated whether technological advancements in deep neural networks would lead to higher accuracy in predicting human motor responses recorded from subjects while playing three arcade games. We compared the predictive performance of features generated by three neural networks of varying complexity. Our results provide evidence that all three models can predict human behavior, with the most advanced model achieving the highest prediction accuracy. These results suggest that advanced neural networks are suitable models for studying human behavior in a time-continuous context and serve as a complement to the trial-based study designs.