A new robot controller model called an ASM-network, built from adaptive sensorimotor maps, enables a robot to learn object discrimination without explicit representations or external rewards. The model combines a mechanism that generates continuous motor activity from past sensorimotor trajectories with an evaluative mechanism that reinforces trajectories supporting higher-order sensorimotor coordinations. In a minimal cognition task, a single robot learned through random exploration and repetition of supportive trajectories. The results demonstrate that recognizable learning behavior can emerge from enactive principles, adapting based on the internal requirements of the action-generating mechanism.
A workshop at the Artificial Life XV conference revisited enactivism's contributions to biology, aiming to ground the concept of autonomy in quantitative definitions based on observable phenomena. Discussions covered identifying emergent individuals from environmental backgrounds, the roles of autonomy and normativity in biological theory, the spontaneous emergence of autonomous agents at life's origins, and scientific approaches to subjective experience.
The work argues that enactive robotics, which models how organisms interact with their environment through sensorimotor contingencies, can help understand how autonomous habits emerge beyond simple problem-solving frameworks. It proposes that the constraints imposed by sensorimotor contingencies shape the forms these habits take, leading to behaviors that are not merely goal-directed but arise from the dynamic interaction between agent and environment. This perspective challenges traditional cognitive science views that focus solely on internal representations and problem-solving, offering a broader account of autonomous agency and learning.