We did research that examined the relationship between attention and word learning. The procedure involved using an eyetracker to play videos to infants. In these videos, two novel objects would appear on the screen, one object would have a black square flash over it while a novel word was said aloud. We wanted to know if this sort of artificial attention capture would create an association between the word and the object. Another point of interest was how that association compares to associations made by other attention captures (like longer and shorted pointing gestures).
During a pilot study with adults, the average of target looks seems to be significantly greater than competing looks across conditions. Adult participants looked at the target object and average of 68.6% of the time and spend only 5% of their time looking at the competing object. As a general trend, participants in the attention capture condition tend to look longer at the target object than in long and short point conditions. Adults looked at the target object in the attention capture condition 80.5% of the time and at the competitor object 8.6% of the time. In the long point condition adults looked at the target object 59.6% of the time and at the competitor object 3.7% of the time. Adults looked at the target object in the short point condition 64.4% of the time and at the competitor object 3.0% of the time.
In both learning and test trials, competitor looking time is consistently higher for attention capture conditions than for long and short point conditions. Long point looking times seem to have the lowest target looking times among each of the three conditions. A possible explanation for this difference could be that adults are curious about the unexplained flash and spend more time looking at the objects on the screen after the first flash, while in other trials they would have spent that time looking at other parts of the screen. Short point might be performing better than long point for a similar reason. The shorter grasp of attention might be arousing more curiosity from participants, whereas, in a long point condition, nothing is ambiguous and participants might lose interest more quickly. In the graph of learning conditions, attention to the target object declines more quickly than in short point and attention capture conditions.