adults and peers, to learn with; thought-provoking tasks; tools that both boost and shape thinking; and activity structures that encapsulate learning-supportive norms and processes.

Indeed, observational and historical studies of working scientists reaf-firm the promise of looking closely at the ways in which environments support learning. These studies demonstrate that theory development and reasoning in science are components of an ensemble of activity that includes networks of participants and institutions (Latour, 1999); specialized ways of talking and writing (Bazerman, 1988); development of representations that render phenomena accessible, visualizable, and transportable (Gooding, 1989; Latour, 1990); and efforts to manage material contingency by making instruments, machines, and other contexts of observation (such as experimental apparatus). The alignment of instruments, measures, and theories is never entirely principled (e.g., Pickering, 1995), and, whether the scientists are professionals or school students, they wrestle with the relationships between these tools and the phenomena they are intended to capture.

A second major problem with assuming children’s learning will unfold without support is that what children are capable of doing without instruction may lag considerably behind what they are capable of doing with effective instruction. Further clouding the picture is that research on cognitive development may not be helpful in illuminating how instruction can advance children’s knowledge and skill. Often, studies in developmental psychology do not have an instructional component and therefore may be more informative about starting points than about children’s potential for developing scientific proficiency under effective instructional conditions.

For example, the idea that prior to middle school children are incapable of designing controlled experiments has been a ubiquitous assumption in the elementary school science community. This claim can be traced to Inhelder and Piaget’s (1958) influential study, The Growth of Logical Thinking from Childhood to Adolescence. Indeed the Benchmarks for Scientific Literacy (American Association for the Advancement of Science, 1993) included design of controlled experiments in their list of limitations of the scientific reasoning of third to fifth graders:

Consider the Benchmarks’ crucial—and unusual—caveat (p. 11):