Purchase this article with an account.
Hayley Lagroix, Vincent Di Lollo, Thomas Spalek; Are accuracy and reaction time equivalent measures of the attentional blink?. Journal of Vision 2015;15(12):880. doi: 10.1167/15.12.880.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Perception of the second of two rapidly sequential targets (T1, T2) is impaired when presented soon after the first (attentional blink; AB). AB magnitude is indexed by the difference between performance at short and long inter-target lags. Conventionally, T2 performance is assessed using accuracy as the dependent measure. An inherent problem with this measure, often encountered in AB experiments, is the 100% response ceiling. For example, Visser (2007) reported greater AB magnitude with hard than with easy T1 tasks. That conclusion is questionable, however, because the two functions converged to the ceiling, thereby confounding the effect of T1 difficulty with the ceiling constraint. To avoid this problem, we used reaction time (RT) as the dependent measure and found AB magnitude to be invariant with T1 difficulty (Experiment 1). One interpretation of this result is that the invariance seen with RT would also obtain with accuracy but for the response ceiling. This implies equivalence of the two measures, which is not always the case (Santee & Egeth, 1982). In Experiment 2, we checked the equivalence of RT and accuracy measures of the AB using the phenomenon of lag-1 sparing, which refers to the finding that T2 performance is relatively unimpaired when T2 comes directly after T1 (Lag 1). Using accuracy, Visser et al. (1999) found lag-1 sparing only when T1 and T2 were presented in the same spatial location. Lag-1 deficit occurred otherwise. We replicated Visser et al.’s finding with accuracy; with RT as the dependent measure, however, lag-1 deficit occurred even when T1 and T2 were presented in the same location. This pattern of results suggests that RT and accuracy are not always equivalent measures of the underlying processes involved in the AB. Therefore, RT may not be a good way of avoiding the ceiling problem inherent in accuracy measures.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only