Oscillatory correlates of controlled speed-accuracy tradeoff in a response-conflict task

Abstract

In making decisions, people have to balance between the competing demands of speed and accuracy, a balance generally referred to as the speed-accuracy tradeoff (SAT). In this study, we investigated the role of controlled SAT in a two-choice task in which manual responses were either validly or invalidly cued. Examining electrophysiological measurements of oscillatory brain activity, theta power in the anterior cingulate cortex (ACC), alpha power in the occipital cortex, and beta power in the motor cortex were found to be related to SAT. Because oscillatory effects of SAT were found to emanate from the SAT baseline interval preceding the two-choice task, the results indicate that SAT is modulated by a change of visuo-motor baseline activities rather than a change of response threshold. Moreover, in the two-choice task, conflict-induced theta power in the ACC was found to be more pronounced in speed than in accuracy trials, whereas priming-related beta power dynamics in the motor cortex were unaffected by SAT. These results indicate that conflict processing, but not response priming, depends on SAT.

Figure 1

Depiction of the procedure. In the two‐choice task, subjects were instructed to respond to a red cross which was presented in one of the two vertically aligned square boxes with the index finger of their left hand (targets above fixation) and right hand (targets below fixation) while keeping fixation on the centre of the screen. Targets were validly or invalidly cued by brightening of one of the squares or uncued. Prior to the two‐choice task, subjects were instructed to focus on either speed or accuracy in the current trial. In speed trials, subjects saw the message “too slow” whenever they exceeded the response time criterion of 450 ms. In accuracy trials, they saw the message “incorrect” whenever they made an incorrect response. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 2

Behavioral data. Main effect of SAT instruction: compared with accuracy trials, in speed trials, responses were faster but more error‐prone. Main effect of response cuing: compared with uncued trials, responses were faster in validly cued trials and more error‐prone in invalidly cued trials. An interaction between SAT and response cuing was observed for response errors, but not for RT. All error bars: standard errors. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 3

Topography and source localization of differences in oscillatory brain activity between speed and accuracy trials in the SAT baseline interval. A: Differences in mean theta power were localized to the ACC, (B) differences in mean alpha power to the occipital cortex, and (C) differences in mean beta power to the motor cortex. In topographical maps, warm color coding indicates event‐related increases of power and cold color coding indicates event‐related decreases of power. Difference maps between conditions are also expressed by topographical plotting of the P‐levels obtained by nonparametric Wilcoxon sign‐rank tests. Randomization tests (see Methods) revealed that differences in all three frequency bands were reliable (all Pcorr's < 0.001). LORETA significance probability maps are based on P‐values (P's < 0.01, one‐tailed). Coordinates in Talairach space of the maximum‐value voxels are shown. Left: sagittal slices, seen from the left at X; middle: coronal slices, seen from the rear at Y; right: axial slices, seen from above at Z. X, Y, and Z are Talairach coordinates: X from left (“L”) to right (“R”); Y from posterior (“P”) to anterior (“A”); Z from basal to dorsal. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 4

Time courses of oscillatory activities in the SAT baseline interval. Time‐frequency spectrograms and time courses of SAT‐cue‐induced (A) mean theta power combined for all electrode sites, (B) mean alpha power combined for central and parietal sites, and (C) mean beta power combined for central electrode sites. Combination of sites for each frequency band was based on significance of the interaction between SAT and ROIs (see Results). In the right side, point‐to‐point Wilcoxon sign‐rank tests (50 ms bins, P's < 0.05) showed that differences in oscillatory power between SAT conditions sustained throughout the SAT interval in all three frequency bands. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 5

Topography and source localization of differences in oscillatory brain activities in the 300‐ms interval following target onset in the two‐choice task. A: Differences in mean theta power between invalidly and validly cued trials were localized to the ACC and (B) differences in mean beta power between uncued and validly cued trials were localized to the motor cortex. Warm color coding indicates event‐related increases of power and cold color coding indicates event‐related decreases of power. Difference maps between conditions are also expressed by topographical plotting of the P‐levels obtained by nonparametric Wilcoxon sign‐rank tests. Randomization tests (see Methods) revealed that differences in both frequency bands were reliable (all Pcorr's < 0.001). LORETA significance probability maps are based on P‐values (P's < 0.01, one‐tailed). Coordinates in Talairach space of the maximum‐value voxels are shown. Left: sagittal slices, seen from the left at X; middle: coronal slices, seen from the rear at Y; right: axial slices, seen from above at Z. X, Y, and Z are Talairach coordinates: X from left (“L”) to right (“R”); Y from posterior (“P”) to anterior (“A”); Z from basal to dorsal. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 6

Time courses of oscillatory activities in the two‐choice task. Time‐frequency spectrograms and time courses of (A) mean theta power and (B) mean beta power, both combined for mid‐central electrode sites, triggered to target onset. Combination of sites for each frequency band was based on significance of the interaction between SAT and ROIs (see Results). In the right side, point‐to‐point Wilcoxon sign‐rank tests (50 ms bins, P's < 0.05) showed (i) a significant theta power increase in the speed condition compared with the accuracy condition in (invalidly) cued trials and (ii) sustained SAT‐induced differences in beta power lasting until about 300 ms after target onset in all three response‐cuing conditions. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]