Learning Intuitive Physics and One-Shot Imitation Using State-Action-Prediction Self-Organizing Maps

Human learning and intelligence work differently from the supervised pattern recognition approach adopted in most deep learning architectures. Humans seem to learn rich representations by exploration and imitation, build causal models of the world, and use both to flexibly solve new tasks. We suggest a simple but effective unsupervised model which develops such characteristics. The agent learns to represent the dynamical physical properties of its environment by intrinsically motivated exploration and performs inference on this representation to reach goals. For this, a set of self-organizing maps which represent state-action pairs is combined with a causal model for sequence prediction. The proposed system is evaluated in the cartpole environment. After an initial phase of playful exploration, the agent can execute kinematic simulations of the environment’s future and use those for action planning. We demonstrate its performance on a set of several related, but different one-shot imitation tasks, which the agent flexibly solves in an active inference style.

Scalar Induced Gravitational Waves Review

We provide a review on the state-of-the-art of gravitational waves induced by primordial fluctuations, so-called induced gravitational waves. We present the intuitive physics behind induced gravitational waves and we revisit and unify the general analytical formulation. We then present general formulas in a compact form, ready to be applied. This review places emphasis on the open possibility that the primordial universe experienced a different expansion history than the often assumed radiation dominated cosmology. We hope that anyone interested in the topic will become aware of current advances in the cosmology of induced gravitational waves, as well as becoming familiar with the calculations behind.

Stable individual differences in infants’ responses to violations of intuitive physics

Infants look longer at impossible or unlikely events than at possible events. While these responses to expectancy violations have been critical for understanding early cognition, interpreting them is challenging because infants’ responses are highly variable. This variability has been treated as an unavoidable nuisance inherent to infant research. Here we asked whether the variability contains signal in addition to noise: namely, whether some infants show consistently stronger responses to expectancy violations than others. Infants watched two unrelated physical events 6 mo apart; these events culminated in either an impossible or an expected outcome. We found that infants who exhibited the strongest looking response to an impossible event at 11 mo also exhibited the strongest response to an entirely different impossible event at 17 mo. Furthermore, violation-of-expectation responses in infancy predicted children’s explanation-based curiosity at 3 y old. In contrast, there was no longitudinal relation between infants’ responses to events with expected outcomes at 11 and 17 mo, nor any link with later curiosity; hence, infants’ responses do not merely reflect individual differences in attention but are specific to expectancy violations. Some children are better than others at detecting prediction errors—a trait that may be linked to later cognitive abilities.