摘要:Four normally developing 2-year-old children were exposed to a story in which six novel characters appeared and were named. To assess learning of the character name -> picture relations, all six pictures were presented together as comparisons in matching-tosample test trials and the children were asked to make selections as each of the six names were dictated as samples. No child learned any of the target relations, but 2 made unreinforced conditional selections (URCSs) to some of the sample names (i.e., made their own one-to-one assignments of sample names to comparison pictures). This was the first time URCS had been demonstrated on an auditory-visual matching task, on any matching task with more than three comparisons, and with children so young. The long-term retention of a specific pattern of URCS was also shown for the first time, with 1 child maintaining his sample-comparison assignments across a 43-day gap in testing. The findings highlight the potential threat of URCS-based false-positives on tests of young children's emergent matching-to-sample (e.g., in tests of stimulus equivalence or ostensive learning).