Intuitive Physics

Definition

The innate human understanding of everyday physical interactions, enabling recognition of object permanence, continuity, and solidity without complex calculations.

Field of Study

Cognitive Science, Physics, Artificial Intelligence

Importance for AI

Critical for AI to understand and predict physical interactions, such as object motion, collisions, and gravity, akin to human common sense. Essential for robotic navigation, safe human-AI interaction, and inferring unstated knowledge.

Early Investigator

Paolo Bozzi

Research Approaches

Bayesian cognitive modeling, Information Integration Theory (IIT), self-supervised learning, phenomenological primitives (p-prims)

AI Acquisition Method

Can be acquired using a general learning principle, with factors like model size, pretraining data, and task influencing understanding.

Core Principles (Human)

Object permanence, continuity, solidity

1958-XX-XX

2010-XX-XX

## Understanding Intuitive Physics What Is Intuitive Physics? Intuitive physics is the innate human understanding of everyday physical interactions. It enables infants to recognize that a ball hidden behind a screen still exists (object permanence), that objects follow continuous paths (continuity), and that two objects cannot occupy the same space (solidity). These core principles allow humans to interact with the world without performing complex calculations consciously. [...] ## FAQ: FAQ: Intuitive Physics for AI: Teaching Machines How the World Works 1. What is intuitive physics in the context of AI? Intuitive physics refers to an AI's ability to understand and predict physical interactions, such as object motion, collisions, and gravity, akin to human common sense. This capability is critical for tasks like robotic navigation and safe human-AI interaction. 2. Why is intuitive physics important for AI systems? [...] DARPA’s MCS program prioritizes intuitive physics as foundational for AI to infer unstated knowledge (e.g., object permanence, gravity). It aims to bridge symbolic logic and human-like reasoning for safer, more adaptable AI systems. 7. How do self-supervised learning approaches improve intuitive physics in AI?

Intuitive physics has an obvious theoretical importance for understanding how people build cognitive and perceptual representations of the external world. However, the practical applications of intuitive physics are also worthy of discussion. A first important application of intuitive physics concerns the improvement of the teaching of physics. Converging evidence suggests that misconceptions about elementary physics are quite impervious to formal instruction and have a detrimental effect on [...] As discussed in Section 1, the field of intuitive physics is characterized by an open debate between the supporters of the hypothesis that intuitive physics is driven by sub-optimal heuristics, and the supporters of the hypothesis that intuitive physics is driven by the internalization of physical laws (Kubricht et al., 2017). The “sub-optimal heuristics” perspective actually includes a variety of specific hypotheses the discussion of which is beyond the scope of the present review. Most of [...] Intuitive physics has been explored from a variety of theoretical perspectives and using a variety of methodological approaches. The remainder of this review will focus on the application of Information Integration Theory (IIT) (Anderson, 1981, Anderson, 1982, Anderson, 2013, Weiss, 2006) to the field of intuitive physics. More specifically, this section provides a brief introduction to IIT (Section 2.1), showing with an example that IIT is a powerful and easily usable tool for the study of

The first investigations into what later on has been called intuitive physics were conducted by Paolo Bozzi, who examined the visual perception of pendulum motion and motion along inclined planes (for a recent reprint of a translation of Bozzi’s work, see Bressan and Gaudiano 2019). In his study on pendulum motion (Bozzi 1958), he discovered that the period of oscillation that appears most natural to observers is slower than the physically accurate period. This finding was partially supported [...] The 2010s marked the emergence of a new approach to studying intuitive physics, which was based on Bayesian cognitive modeling (Battaglia et al. 2013; Kubricht et al. 2017; Sanborn et al. 2013; Ullman et al. 2017). This approach involves comparing participants’ performance in intuitive physics tasks with predictions generated by Bayesian models. These models combine prior probability distributions based on Newtonian laws with current perceptual information that is characterized by stochastic [...] From a theoretical perspective, the development of heuristics in the field of intuitive physics involves a certain degree of knowledge systematization and generalization. The underlying idea is that perceptual experience is utilized to construct knowledge structures that encompass various physical phenomena. In his influential theoretical work, DiSessa (1993) termed these knowledge structures _phenomenological primitives_ or _p-prims_. Without delving into details, there exists a debate

behaviors. Under this account, the intuitive physics system would be both general in its capacity to operate over a broad range of physical scenarios and domain-specific for the processing of scene structure and behaviors that can be modeled with Newtonian mechanics. Our experiments here will test for both the flexibility of the mental processes underlying physical prediction and their specificity to the domain of intuitive physics. [...] on the table are softer than the walls surrounding the court. An intuitive physics system that models the underlying causal structure of a scene as its basis for prediction has the advantage of generalizing across a variety of physical scenarios even when the scenarios share few of their surface-level features. [...] within a shorter time frame, we constructed a condensed version of the physics task battery (the Test of Intuitive Physics; TIP). The TIP provides an efficient means of measuring general intuitive physics abilities in a variety of participant populations and experimental contexts going forward.