Causal Learning: Understanding the World

Abstract

This chapter focuses heavily on empirical research on whether causal learning is evident very early in life as an associative or as a primitive inferential, abstract fashion. The current predominant view is that it is Bayesian, statistical, probabilistic, computational, and so on, and not governed by either innate preformed abstraction-ready modules or associative, nonrepresentational mechanisms. The Bayesian point of view in this chapter is complemented by the interventionist and causal mapping one. In working in this area, the traditional Piagetian perspective on mental schemas still appears useful, and it is much cited. However, others dismiss its utility. In my compromise position, I show how a modified, integrative Neo-Piagetian view can be informative. The associative point of view is promoted by theorists who argue that too much is read into studies of very young infants in terms of their early abstractive abilities. Rather than being little logicians, young children are intuitive statisticians. A view that accommodates to the opposition of the fast minimal nativist and slow constructivist points of view on early causal learning concerns the middle-of-the-road one of rational constructivism. Early cognitive structures in the associationist camp have been referred to as intuitive and nontheoretical, with motor resonance involved. Yet the field also encounters contrary concepts, such as infants possessing an abstract framework and the blessing of abstraction. In a nativist-friendly approach, neonates might even understand physical causation/Michottian launching events. Yet, in the contrary view, only older children might develop a full theory of mind, or a ``theory'' theory. Aside from innate factors, the chapter refers to natural pedagogy, and observational causal learning/interventionist, causality-informative behavior. For some of the intriguing methods used in the research, they include ``blicket'' detectors, sticky mittens, everted rabbits, and win-stay/lose-shift strategies. Other concepts in the chapter include causal, higher-order relational cognition and the quantum probability model of causal reasoning.