Selective attention to pitch amid conflicting auditory information: context-coding and filtering strategies

A variety of rules can generate conflicting differences in the current paradigm; the present rule excludes conflicting differences in the Standard pair presented in the first observation interval. Displaying identical conflicting differences in all three observation-intervals would remove the within-trial conflicting differences, and result instead in a roving-standard task with between-trial stimulus variability. Presenting identical conflicting differences in both Comparisons precludes correlation between task-relevant and conflicting differences; for example, displaying conflicting differences consistently in the Comparison without the task-relevant difference would create correlation, and the “conflicting differences” would be useful to discriminate the relevant difference in target frequency. Trial-to-trial randomization of the Comparison conveying conflicting differences introduces an additional source of stimulus variability beyond the scope of this study.

The Spearman-rank correlation between FDTs and amount of training was computed using individual FDT estimates rather than averages of 5–6 FDT estimates.

Owing to the small participant samples and the large inter-individual differences, statistical significance employed non-parametric tests computed separately for each participant’s FDT data.

In some cases, the final conflict-laden FDTs were not significantly higher than the no-conflict FDTs; however, further verification seemed justified because of the small participant samples, and the relatively low power of the Wilcoxon rank-sum test.

The presentation order and time proximity (ISI) effects suggest that the trailing flanker (with or without conflicting differences) could interfere retroactively with the sensory memory of task-relevant differences in a manner resembling backward recognition/discrimination masking (BRM). It resembles the decline in pitch discrimination and recognition that occurs when a masker (without any conflicting difference) follows the target tone with ISIs shorter than 300 ms (e.g., Leshowitz & Cudahy, 1973; Massaro, 1975; Ronken, 1972; Turner, et al., 1994). To account for BRM, one hypothesis is that the masker interferes with the readout of information stored in the target sensory-memory trace (Massaro, 1975). However, it is unlikely that BRM alone could account for the FDT elevations induced by conflicting differences in the frequency or level of trailing flankers. First, the 40- or 80-ms targets used here are much longer than the 10–25-ms targets typically used in BRM tasks (e.g., Massaro, 1975). Well-trained participants show little or no BRM with targets longer than 20 ms (Sparks, 1976; Ronken, 1972; Leshowitz & Cudahy, 1973), and much less would occur in the present task. Second, prior to presenting conflicting differences in Experiment 1, training in the no-conflict target–flanker condition proceeded until FDTs approached closely those for no-conflict targets in isolation, i.e., training flattened the FDT elevations due merely to the presence of a flanker. With a trailing flanker, it might be difficult to read out or attend to the task-relevant information held in sensory memory, but the information itself is not lost; otherwise, training could not equate the two no-conflict FDTs. In addition to evincing after equating the no-conflict FDTs, the effects of conflicting differences in flanker frequency or level were much stronger and persistent than those due to the mere presence of a flanker. This occurred in spite of the fact that, in cumulative terms, participants had far more training when tested in conflict-laden than in no-conflict conditions. Moreover, conflicting information effects depend strongly on the conflicting-difference dimension: they are strong with frequency or level, but not with duration. Finally, in Experiment 2, a weak BRM effect could occur in the 60-ms ISI condition, but in addition to being significantly higher with than without conflicting frequency differences, the FDT elevations were significant also in the 350-ms ISI condition, in which no BRM would be expected (e.g., Massaro, 1975). Given the above, it is unlikely that sensory-memory interference could account for the effects of conflicting differences in the frequency or level of trailing flankers. However, a trailing flanker may produce effects resembling BRM, and in doing so facilitate conflicting-information interference. For example, the flanker could function as a distracter and make it difficult either to attend and access the information held in the target sensory memory trace or to ignore the conflicting information held in the flanker sensory memory trace, or both (e.g., Näätänen, 1988, 1992; Turner et al., 1994); this could account for presentation order effects.