Evidence for early recognition of inconclusive data in children’s evaluation of evidence

• In Event: 1-029 - The emergence of scientific reasoning skills in early childhood

Integrative Statement

In the current study, we explored early markers of scientific reasoning. Although preschoolers intuitively form accurate inferences from observed evidence (e.g., Gopnik, 2012), they often fail to explicitly recognize the effects of indeterminacy of knowledge, and do not distinguish inferences that are “warranted” from those that are not (e.g., Legare et al., 2013; Piekney et al., 2014). Here we aimed to scaffold children’s ability to evaluate and interpret evidence by presenting side-by-side experiments, conducted by adult experimenters. Although both experiments produced the same outcome and yielded the same conclusion, they presented a critical contrast between a conclusive and inconclusive test. We hypothesized that although young children often struggle to generate a controlled experiment that successfully isolates a variable of interest, they may be able to evaluate which of two alternatives yield an interpretable outcome, and selectively draw inferences from the conclusive result. In addition, we examined whether these observations might scaffold children’s ability to isolate a variable of interest on their own, in a subsequent task.

A total of 56 preschoolers (28 4-year-olds and 28 5-year-olds) were introduced to a type of causal machine that ‘activates’ (plays music) when blocks with a certain feature (button or antenna) are placed on top. Children then watched side-by-side videos (presented on an iPad) in which two experimenters (A and B) each produced a machine and a set of three blocks with different combinations of these features (i.e., both features, button only, neither feature). Both experimenters tried the block with both features first, activating their machine. Then, experimenter A tried the block with a button only (controlled), while experimenter B tried the block with neither feature (confounded). In both cases, their machine failed to activate, and the experimenter concluded, “Antennas make it play!” At the conclusion of the video, both of the completed experiments remained visible, along with memory cards indicating the outcome (i.e., music or no music) for each block (see Figure).

In the generalization task, children were asked to use a novel block that they had not previously observed (antenna only) to activate one of the two machines. Although 4-year-olds responded at chance (54%, p=.43), 5-year-olds correctly selected the machine used by experimenter A (71%, p=.02). Afterwards, in the transfer task, children in both age groups successfully applied this skill to a subsequent experiment in which they were asked to determine which of two new features (i.e., color or shape) would activate a novel machine (C). After observing the first outcome (i.e., a purple cube activated the toy), children correctly selected between two additional blocks (i.e., a red cube [controlled] vs. a red sphere [confounded]), conducting a controlled experiment in a forced choice context (82% and 75%, ps<.01). Results demonstrate that preschoolers can evaluate the quality of evidence, use it to draw novel inferences, and transfer these skills to support their own experiments. These findings suggest that the evaluation of evidence may provide a more accessible platform, allowing children to more readily demonstrate their early recognition of control of variables.

Authors

• Caren Walker, University of California, San Diego
  Presenting Author

• Disha Goel, University of California, San Diego
  Non-Presenting Author

• Angela Nyhout, University of Toronto
  Non-Presenting Author

• Patricia Ganea, University of Toronto
  Non-Presenting Author