Abstract
Successful recognition of an object often requires binding of features of the object by attention. Withdrawing attention to a very demanding visual task at fixation can impair discrimination performance of even very simple stimuli such as colored bisection disks presented in the periphery (Fei-Fei, VanRullen, Koch, Perona, 2002). A number of studies reported that the impaired performance under dual-task conditions cannot be compensated even after training for more than 10 hours across multiple days (e.g., Fei-Fei, VanRullen, Koch, Perona, 2005; Reddy, Reddy, Koch, 2006). These results suggest that the underlying processes of feature binding is rigid and cannot be trained. Contrary to this view, we report significant training effects on discrimination performance under a dual-task condition. In a single central task, participants were asked to report whether all the letters shown on the fovea are identical or not. In a single peripheral task, they discriminated a red/green disk from a green/red disk in the periphery. Performances for the single-task condition were kept at 80% accuracy by QUEST during the entire experiment (Matthews et al., 2017). In the dual-task, participants were asked to conduct both tasks simultaneously, with emphasis on the central task. Consistent with the previous studies, this dual-task condition resulted in chance performance in the peripheral task presumably due to exhaustion of attention by the central task. However, after 10 days of training on the dual-task with trial-by-trial feedback, performance on the peripheral task significantly improved. The training effect was not specific to the trained location, suggesting improvements in readout of sensory representations by the training. We also found the training effects are long-lasting and can be observed even after a month. These results suggest that feature binding process becomes less attention-dependent as a result of extensive training.
Acknowledgement: This work is supported by JSPS Kakenhi Grant Number 17H04789 to KS, ARC Discovery Projects (DP180104128, DP180100396) and Templeton World Charity Foundation “Intelligence and consciousness in flies” to NT.