Breaching the Bottleneck: Evolutionary Transition from Reward-Driven Learning to Reward-Agnostic Domain-Adapted Learning in Neuromodulated Neural Nets

TL;DR

This paper explores how biological evolution enables neural networks to transition from reward-based learning to reward-agnostic domain-adapted learning, significantly improving learning efficiency and flexibility.

Contribution

It introduces a biologically plausible evolutionary pathway for neural networks to develop domain-adapted learning by integrating non-reward information through neuromodulation.

Findings

DAL agents learn 300 times faster than pure RL agents.

Evolution eliminates reliance on reward signals, enabling learning solely from non-reward information.

Neuromodulatory mechanisms facilitate local weight updates based on non-reward stimuli.

Abstract

Advanced biological intelligence learns efficiently from an information-rich stream of stimulus information, even when feedback on behaviour quality is sparse or absent. Such learning exploits implicit assumptions about task domains. We refer to such learning as Domain-Adapted Learning (DAL). In contrast, AI learning algorithms rely on explicit externally provided measures of behaviour quality to acquire fit behaviour. This imposes an information bottleneck that precludes learning from diverse non-reward stimulus information, limiting learning efficiency. We consider the question of how biological evolution circumvents this bottleneck to produce DAL. We propose that species first evolve the ability to learn from reward signals, providing inefficient (bottlenecked) but broad adaptivity. From there, integration of non-reward information into the learning process can proceed via gradual accumulation of biases induced by such information on specific task domains. This scenario provides a biologically plausible pathway towards bottleneck-free, domain-adapted learning. Focusing on the second phase of this scenario, we set up a population of NNs with reward-driven learning modelled as Reinforcement Learning (A2C), and allow evolution to improve learning efficiency by integrating non-reward information into the learning process using a neuromodulatory update mechanism. On a navigation task in continuous 2D space, evolved DAL agents show a 300-fold increase in learning speed compared to pure RL agents. Evolution is found to eliminate reliance on reward information altogether, allowing DAL agents to learn from non-reward information exclusively, using local neuromodulation-based connection weight updates only. Code available at github.com/aislab/dal.