Purchase this article with an account.
Thomas U. Otto, Pascal Mamassian; Divided Attention and Sensory Integration: The Return of the Race Model. Journal of Vision 2010;10(7):863. doi: 10.1167/10.7.863.
Download citation file:
© 2017 Association for Research in Vision and Ophthalmology.
The sensory brain processes different features of incoming stimuli in parallel. The combination/integration of features within or across sensory modalities can often improve perception and cognition as expected from probability summation. Interestingly, the literature on divided attention abounds with reports of integration effects that exceed probability summation. In the so-called redundant target paradigm, participants respond as quickly as possible to targets that are defined, for example, by their color (red vs. green) and/or orientation (vertical vs. horizontal). Typically, reaction times to redundant targets defined by both features are faster than to targets defined by a single feature only. In analogy to a higher probability for a “small number” when playing with two rather than one dice, probability summation or race-models can in principle account for a speeding up of reaction times. However, according to the influential race-model-test (Miller, 1982, Divided attention: evidence for coactivation with redundant signals, Cognitive Psychology 14, 247–279), reaction times are even faster and performance is even better than predicted by probability summation. Consequently, race models are rejected and a benefit due to sensory integration is assumed. Here, we critically evaluate the race-model-test and its interpretation. Fifteen observers participated in the experiment described above and we determined cumulative reaction time distributions based on a total of 3000 trials per condition. We show that not only fastest reaction times are faster but also that slowest reaction times are slower than predicted – a finding that is neglected by Miller's test and also in the literature. Importantly, in terms of variance, this result indicates that performance is not better but in fact worse. Hence, race models cannot be rejected. We hypothesize that the increased variance is related to capacity limited decision processes and provide new interpretations for a large variety of studies using the redundant target paradigm.
This PDF is available to Subscribers Only