Abstract representations emerge naturally in neural networks trained to perform multiple tasks

Humans and other animals demonstrate a remarkable ability to generalize knowledge across distinct contexts and objects during natural behavior. We posit that this ability to generalize arises from a specific representational geometry, that we call abstract and that is referred to as disentangled in machine learning. These abstract representations have been observed in recent neurophysiological studies. However, it is unknown how they emerge. Here, using feedforward neural networks, we demonstrate that the learning of multiple tasks causes abstract representations to emerge, using both supervised and reinforcement learning. We show that these abstract representations enable few-sample learning and reliable generalization on novel tasks. We conclude that abstract representations of sensory and cognitive variables may emerge from the multiple behaviors that animals exhibit in the natural world, and, as a consequence, could be pervasive in high-level brain regions. We also make several specific predictions about which variables will be represented abstractly.

Automated summary

Humans and animals can generalize knowledge across different contexts and objects during natural behavior. This ability arises from abstract representations, observed in recent neurophysiological studies, that emerge through the learning of multiple tasks using neural networks. These abstract representations enable few-sample learning and reliable generalization on novel tasks, and may be pervasive in high-level brain regions. Specific predictions are made about which variables will be represented abstractly.