Evolutionary accounts of feelings, and in particular of negative affect and of pain, assume that creatures that feel and care about the outcomes of their behavior outperform those that do not in terms of their evolutionary fitness. Such accounts, however, can only work if feelings can be shown to contribute to fitness-influencing outcomes. However, simply assuming that a learner that feels and cares about outcomes is more strongly motivated than one that doesn’t is not enough — if only because motivation can be tied directly to outcomes through an appropriate reward function, without leaving any apparent role to feelings (as it is done in state-of-the-art engineered systems based on reinforcement learning). Here, we propose a possible mechanism whereby pain contributes to fitness: an actor-critic functional architecture for reinforcement learning, in which pain reflects the costs imposed on actors in their bidding for control, so as to promote honest signaling and ultimately help the system optimize learning and future behavior.

In a new animation, scientists map the planet’s plate tectonics over the last 40 percent of its history. It’s the longest such reconstruction yet.

Google partners with GASA and DNS RF to launch Global Signal Exchange, enhancing real-time scam detection.

Mozilla urges users to update Firefox after critical CVE-2024–9680 vulnerability is actively exploited.

In recent research on Large Language Models (LLMs), there’s been a focus on improving how these models handle complex reasoning tasks.

It’s not a tool, but an alien agent, says Yuval Noah Harari in an interview.