Two research movements—cognitive science and cognitive robotics—pursue different agendas in studying the minimal self. Cognitive science creates abstract models that predict human experimental data, while cognitive robotics builds embodied learning machines that develop a self from scratch, like human infants. Although both approaches can produce causal mechanistic models, building a minimal self in a robot is not the same as understanding the human minimal self. Conclusions drawn from robotic models about humans, or vice versa, require caution. Incorporating constraints from multiple levels of analysis is essential for models that predict, generate, and causally explain real-world behavior.
A neuro-computational model explains how the brain gates information into consciousness, specifically in the distractor-induced blindness task. In this task, participants view rapid streams of colored dots and report intervals of coherent motion preceded by a target color. When distractors (motion intervals before the target color) appear, detection of the target motion declines, especially when the delay between target color and target motion is short. The model proposes that conscious perception requires reverberating activity in cortico-subcortical loops, with basal ganglia pathways either allowing or inhibiting this reverberation. Distractors trigger suppression via the inhibitory hyperdirect pathway, and when a target follows closely, lingering suppression prevents its identification. The model reproduces experimental data on how delay length affects detection probability.