Training on a difficult search task improves untrained easier search as well

We conducted a large-scale (N=110) experiment into the effects of training on a difficult visual search task. On Day 1, participants performed a baseline session with easier search (T vs. L) and difficult search (square with small square in left-top corner vs. squares with small square in different corner). Display sizes were 8 and 16 items and target prevalence was 50% for both tasks. On Days 2 – 6 participants were trained on the difficult search task. Display sizes were unchanged, but target prevalence was 16.7%. On Day 7, participants performed a post-training session with the same search tasks as on Day 1. Eye movements were recorded for all sessions. We found substantial positive training effects: reaction time gains from Day 2 to Day 6 were ~500 ms and error rates on present trials more than halved. Despite higher target prevalence during baseline and post-training sessions, performance clearly improved on both the difficult search task (reaction times up to 1000 ms faster and error rates on present trials again more than halved) and the easier search task (reaction times fairly constant, but error rates on present trials halved). The eye tracking data revealed that training had increased the size of the functional visual field not only in the difficult search tasks, but even in the (untrained) T vs L task. Training also reduced target decision time (TDT), the period between a fixation landing on the target and the present response. For the difficult search tasks TDT was reduced by several 100 ms, whereas for T vs L the reduction was around 20 ms. The bigger improvement in difficult search reflects that recognizing the target is harder in this task. Consequently, the TDT is longer and the scope for improvement larger. Training gains on difficult search transfer to untrained easier search.