ute to a false impression that they are hopelessly concrete (Simons and Keil, 1995). That difficulty in itself is an important developmental factor to consider in its own right.
Children can use many processes that can be thought of as the underpinnings of scientific reasoning (at least in certain contexts), certainly by the end of the preschool years and often well before. These include deductive reasoning, inductive reasoning, Bayesian reasoning and screening off, sensitivity to covariation, correlation, and contingency patterns in stimulus input, the ability to evaluate simple quantified and if-then rules, the ability to distinguish determinate and indeterminate evidence, and some general problem-solving heuristics and reasoning biases. In this sense, children are more competent than has been commonly supposed and bring a wealth of capacities to the learning process.
For example, reasoning about covariation and cause has been an active area of research on scientific reasoning in older children. Research has demonstrated that even preschool children are adept at using a variety of cues from the environment to identify the cause of an event from a set of potential candidates. Among these cues are temporal contiguity, spatial contiguity, consistent covariation between the candidate cause and the effect, and mechanism—that is, whether there is a plausible mechanism that would account for A causing B (Leslie, 1984; Shultz, 1982). There is mounting evidence that even very young children (ages 2, 3, and 4) are able to draw inferences about cause from viewing patterns of covariation of events.
In a series of investigations, Gopnik and her colleagues (Gopnik et al., 2004; Gopnik and Sobel, 2000; Gopnik, et al., 2001; Kushnir and Gopnik, 2005; Schulz and Gopnik, 2004) explored both how young children learn about new causal relations and whether these learning systems are domain specific or apply across different domains of knowledge, such as biological and physical systems. The strategy was to observe children as they went about learning a novel causal relation that they had not previously encountered or been taught.
In one series of studies, children were shown several small blocks and told that one or more of them were blickets. They were then introduced to the “blicket detector,” a machine that lights up and plays music when (and only when) “blickets” are placed on it. Children were asked to identify which of the blocks were the blickets, either by observing the response of the blicket detector as the researcher placed blocks on it, or, in some studies, by themselves placing blocks on the detector. Across trials within a study and across studies, the patterns of evidence that children used to make inferences became increasingly complex, ultimately including multiple causes and probabilistic relationships. In most cases, even the 2-year-olds made correct conclusions about causality by observing patterns of contingency, although younger children did not perform as well as older preschoolers on