Early research in the field of intuitive physics provided extensive evidence that humans succumb to common misconceptions and biases when predicting, judging, and explaining activity in the physical world. Recent work has demonstrated that, across a diverse range of situations, some biases can be explained by the application of normative physical principles to noisy perceptual inputs. However, it remains unclear how knowledge of physical principles is learned, represented, and applied to novel situations. In this review we discuss theoretical advances from heuristic models to knowledge-based, probabilistic simulation models, as well as recent deep-learning models. We also consider how recent work may be reconciled with earlier findings that favored heuristic models.

中文翻译：

---

直观物理学：当前的研究和争议

直觉物理学领域的早期研究提供了大量证据，表明人类在预测、判断和解释物理世界中的活动时会屈服于常见的误解和偏见。最近的工作表明，在各种情况下，一些偏见可以通过将规范的物理原理应用于嘈杂的感知输入来解释。然而，目前还不清楚物理原理的知识是如何学习、表示和应用于新情况的。在这篇综述中，我们讨论了从启发式模型到基于知识的概率模拟模型以及最近的深度学习模型的理论进展。我们还考虑了如何将最近的工作与支持启发式模型的早期发现相协调。