Abstract
In recent studies, children aged below 10 years did not integrate estimates from different modalities to reduce response variance (Nardini et al., 2008, Gori et al., 2008). We asked whether there is earlier variance reduction for two cues within a modality. Adults and 6 – 11-year-olds judged which of two simultaneously presented slanted planes, defined by disparity, texture, or both, was shallowest. Adults improved their thresholds given both cues rather than either one, and weighted the texture cue most for planes approaching the horizontal, in which texture provides the most reliable slant information. In children, this adaptive pattern of weighting was first seen at 10 years, and a reduction in variance at 11 years. Therefore we found no earlier variance reduction within a modality than was previously found between modalities.
Whether judging single or multiple cues, it is necessary to decide when to stop collecting information and respond. In a second study we asked whether changes in this decision rule might underlie the development of variance reduction. We analysed the latency and accuracy improvements of adults, 8- and 10-year-olds, given both disparity and texture cues compared with one alone. For difficult stimuli close to threshold, adults exploited a second cue to improve accuracy. For easy stimuli already accurate unimodally, they improved latency. This implies that adults respond once they judge the probability of being correct to have reached some threshold. As before, children did not improve accuracy given a second cue. However older children exploited the second cue to improve latency, even with difficult near-threshold stimuli for which adults instead improve accuracy. We propose that one reason for children's failure to reduce variance given multiple cues is that they use a decision rule that does not take into account the quality of sensory evidence, i.e. the probability of being correct.
Supported by ESRC grant RES-062-23-0819.