Effect of Task Difficulty on the Division of Attention between Nonverbal Signals: Independence or Interaction?

Models of divided attention are grouped into two classes according to whether they assume that efficiency on one task is independent of or interacts with the level of difficulty on a second simultaneously performed task. On the basis of contradictory evidence from previous studies, four necessary requirements are proposed for an empirical test between the two classes of model. Two experiments successfully embodying these requirements were run, in which subjects performed a two-choice recognition task, involving frequency and intensity signals presented together and alone. The independent variable was the level of difficulty on one dimension; the dependent variable was the accuracy of performance on a second dimension presented at the same time. The results from both experiments showed an interactive trend, performance becoming worse as the difficulty of the accompanying task was increased. In neither case was the effect significant. When pooled over experiments the effect appeared to be reliable, although small. The outcome is interpreted as supporting interactive models of divided attention rather than independent models. Possible reasons for the smallness of the effect are considered. The most likely reason appears to be the competing demands between two of the test requirements adopted initially. An alternative to one of the requirements is proposed.