Signal-related contributions to stopping-interference effects in selective response inhibition

The probability of stop-signal trials employed in selective-stopping tasks requiring stopping one of the two hands is usually higher than that in typical stop-all tasks (e.g. Aron and Verbruggen 2008). This is because in the selective-stopping tasks, stopping has to be signalled for either the left or the right hand, and that increases the frequency of the stop-signal trial. Regarding the probability of the controlled-go trials in the present study, we thought that SI effects obtained by comparing signal-inhibit trials and no-signal go trials have been clear, and our main purpose of this study was to test whether SI effects remain when the extra signal in signal-inhibit trials is controlled by introducing a similar extra signal in the controlled-go trials. For that reason, the probability of the controlled-go trials was set to be much higher than that of the no-signal go trials. Although the low probability of no-signal-go trials might cause the participant to adapt a delay strategy in those trials, this possibility does not affect our main concern of testing SI effects by comparing signal-inhibit and controlled-go trials.

This pattern of result was exactly the same when the comparison was made between signal-inhibit trials and their RT-matched no-signal go trials: there was no difference between the nontarget hand RTs of the two types of trials (648.85 and 648.70 ms; F(1,19) = 1.90, p > .1). The nontarget hand PFs were larger in signal-inhibit trials than in their RT-matched no-signal go trials (337 vs. 274 cN; F(1,19) = 21.56, p < .001).

Regarding typical two-choice RT tasks requiring a single-hand response to the imperative stimulus, it is true that the LRP is more closely associated with the response (i.e. the RT) to the imperative stimulus. However, in our selective-stopping task, a single-hand response (as it is in signal-inhibit trials) is prepared and executed after the presentation of the X, rather than the imperative stimulus (two filled-in squares). In other words, any LRP in our selective-stopping task is in response to the X, and therefore, the presentation of the X is the critical event for time-locking in our analyses. Nonetheless, when RT-aligned averages are considered, all patterns of results and their implications remain the same. Specifically, a clear LRP in the signal-inhibit trials and a small LRP in the controlled-go trials before the response were found. Mean LRPs in the 200-ms interval right before the response were calculated. The mean LRP in signal-inhibit trials was −1.98 muV, which was significantly different from zero [F(1,15) = 134.15, p < .0001], as was expected. The mean LRP in controlled-go trials was −.44 muV, which was also significantly different from zero [F(1,15) = 9.36, p < .01], indicating the extra signal has its effects on the motor stage of the information processing.

In Fig. 2, a 10-point moving-average filter was used as an off-line low-pass filter to attenuate high-frequency fluctuations in LRPs. To have a consistency between descriptions about statistical results and Fig. 2, the results reported here are those after the low-pass filtering. Importantly, all patterns of results were the same even if the low-pass filtering was not used: the mean LRP in signal-inhibit trials was −2.39 muV, which was significantly different from zero [F(1,15) = 82.47, p < .0001]. The mean LRP in controlled-go trials was −.67 muV, which was also significantly different from zero [F(1,15) = 14.94, p < .01].

We thank Sander Los for initiating this discussion.