Transitivity for height versus speed: To what extent do the under-7s really have a transitive capacity?

Abstract

Transitive inference underpins many human reasoning competencies. The dominant task (the “extensive training paradigm”) employs many items and large amounts of training, instilling an ordered series in the reasoner's mind. But findings from an alternative “three-term paradigm” suggest transitivity is not present until 7 + years. Interestingly, a second alternative paradigm (the “spatial task”), using simultaneously displayed height relationships to form premise pairs, can uphold the 4-year estimate. However, this paradigm risks cueing children and hence is problematic. We investigated whether a height-task variant might correspond to a more ecologically valid three-term task. A total of 222 4–6-year-olds either completed a modified height task, including an increased familiarisation phase, or a computer-animated task about cartoon characters running a race in pairs. Findings confirmed that both tasks were functionally identical. Crucially, 4-year-olds were at chance on both, whereas 6-year-olds performed competently. These findings contrast with estimates from all three paradigms considered. A theoretical evaluation of our tasks and procedures against previous ones, leads us to two conclusions. First, our estimate slightly amends the 7-year estimate offered by the three-term paradigm, with the difference explained in terms of its greater relevance to child experiences. Second, our estimate can coexist alongside the 4-year estimate from the extensive training paradigm. This is because, applying a recently developed “dual-process” conception of reasoning, anticipates that extensive training benefits a species-general associative system, while the spatial paradigm and three-term paradigm can potentially index a genuinely deductive system, which has always been the target of transitive research.

Keywords:

• Children's reasoning
• Dual-process theory
• Height task
• Spatial reasoning
• Transitive reasoning.