Evolving Minds: Logic-Informed Inference from Temporal Action Patterns

Understanding human mental states—such as intentions and desires—is crucial for natural AI-human collaboration. However, this is challenging because human actions occur irregularly over time, and the underlying mental states that drive these actions are unobserved. To tackle this, we propose a novel framework that combines a logic-informed temporal point process (TPP) with amortized variational Expectation-Maximization (EM). Our key innovation is integrating logic rules as priors to guide the TPP’s intensity function, allowing the model to capture the interplay between actions and mental events while reducing dependence on large datasets. To handle the intractability of mental state inference, we introduce a discrete-time renewal process to approximate the posterior. By jointly optimizing model parameters, logic rules, and inference networks, our approach infers entire mental event sequences and adaptively predicts future actions. Experiments on both synthetic and real-world datasets show that our method outperforms existing approaches in accurately inferring mental states and predicting actions, demonstrating its effectiveness in modeling human cognitive processes.

To work well with humans, AI needs to understand what people are thinking—like their goals, intentions, or desires—but this is difficult because we can’t directly observe people’s thoughts, and their actions don’t always happen in predictable ways. Our research tackles this problem by developing a new method that helps AI make better guesses about what a person might be thinking based on how they behave over time. We combined a type of mathematical model that tracks when events happen with logic-based rules about human behavior, helping the AI make smarter predictions even with limited data. These logic rules act like common-sense guidelines, shaping the model’s understanding of how actions and thoughts are connected. Because it’s very hard to figure out people’s mental states directly, we also created an efficient way to approximate them in a way computers can manage. Our method learns to improve itself over time, using both observed actions and inferred mental states to make more accurate predictions about future behavior. This research brings us closer to AI systems that can truly understand and respond to human behavior in natural, intelligent ways.