حواس پرتی در یک کار تشخیص مستمر تحریکی

Abstract Event-related potential (ERP) correlates of distraction are usually investigated in the oddball paradigm following a discrete, trial-by-trial stimulation protocol. In this design, participants perform a discrimination task while oddball stimuli deviate in a task-irrelevant stimulus feature. In our experiment, participants detected gaps in a continuous tone while infrequent frequency glides served as distracting events. Glides preceding a gap by 150 ms delayed the response to the gap and elicited the ERP sequence of N1, probably MMN, P3a, and reorienting negativity, suggesting that these responses reflect distraction-related processes which are neither task- nor stimulation-specific. When participants watched a silent movie and the auditory stimulation was task-irrelevant, glides preceding a gap by 150 ms enhanced the amplitude of the gap-elicited N1. However, when the auditory stimulation was task-relevant, the gap-elicited N1 was attenuated. These results show that the glides drew attention away from the ongoing task, both from watching the silent movie and from detecting gaps.

1. Introduction Distraction is an involuntary attentional change triggered by events which are irrelevant with respect to the current behavior. Whereas distraction has often adverse effects on the immediate task performance, it may be crucial in many situations where behavioral goals have to be changed to adaptively follow situational changes. For example, breaking your phone conversation to jump out of the way of a honking car might serve you better than continuing the conversation without noticing the danger. An experimental model for the study of distraction is the oddball paradigm. Here, stimuli of a regular, predictable stimulus sequence (standards) are occasionally replaced by irregular stimuli (deviants) in an unpredictable manner. Some of the differences observed in behavioral and event-related potential (ERP) responses to deviants and standards have been interpreted as being directly related to distraction. One variant of the oddball paradigm, the auditory distraction paradigm, introduced by Schröger and Wolff (1998a), was found to be especially suited for the investigation of distraction. In this paradigm, on each trial, participants perform a two-alternative forced choice (2AFC) discrimination-task based on one (task-relevant) property of the stimuli (for example, they have to press one button for short tones and another one for long tones). The two types of stimuli differing in the task-relevant property are presented equiprobably. A different (task-irrelevant) property of the stimuli (e.g., tone pitch) changes on some (deviant) trials, usually unpredictably. Because the same task has to be performed on both deviant and standard trials, some of the behavioral and ERP response differences between deviants and standards can be attributed to distraction-related processing. Whereas the distraction paradigm has proven to be a useful research tool, it is important to assess the generality of the phenomena observed in this paradigm, because distraction obviously occurs in many different situations. One characteristic that is common to all variants of the auditory distraction paradigm is stimulus discrimination and categorization. More and more evidence suggests that some of the distraction-related responses may reflect working memory and task-related processes. For example, Escera et al. (2001) found that the distraction-related ERP waveform contained at least one subcomponent time-locked to the task-relevant aspect of the stimulation. Roeber et al. (2005) found that response change/repetition modulated distraction-related behavioral and ERP responses (see also Roeber et al., 2009). Thus task-related processes interact with distraction. Furthermore, Berti (2008a) suggested that some of the distraction-related ERP components actually reflect working memory operations, such as switching between objects manipulated in working memory (Berti, 2008b). In light of these studies, a different, non-discrimination-based task may help to distinguish task-specific from genuine distraction-related effects. The aim of the present study was to test the effects of distraction in a detection task and compare them with the results usually obtained in the prototypical auditory distraction paradigm (Schröger and Wolff, 1998a). To this end we developed a novel continuous-stimulation distraction paradigm. The typical results obtained in the original distraction paradigm and their interpretations can be described in a three-stage model of distraction (for recent summaries, see Escera and Corral, 2007 and Horváth et al., 2008c). Processes in the first stage form a sensory information filter, which reduces the processing load on capacity-limited processing resources by absorbing predictable sensory events. Two of these processes are reflected in the ERPs: a first-order change detection process, which is reflected by the modulation of the modality-specific N1 component (peaking around 100 ms after change onset; for a review, see Näätänen and Picton, 1987); and a deviance-detection process working on the basis of regularities extracted from the previous stimulation, indexed by the mismatch negativity (MMN, peaking 100–200 ms after deviance onset; Näätänen et al., 1978; for a recent summary on the interpretation of the MMN component see Winkler, 2007). The processes reflected by N1 and MMN can trigger an involuntary attention switch (e.g., a shift towards the unpredictable change; Näätänen, 1990), thus initiating the second stage of the distraction model. Some processes playing a role in this attention-change are thought to be reflected by the modality-independent, fronto-central P3a (or novelty P3, see Friedman et al., 2001), which probably takes both sensory and cognitive aspects of the information carried by the given stimulus into account (relation to goals, conceptual context, subjective importance, etc., see, e.g., Barcelo et al., 2006 and Friedman et al., 2003). However, the link between the processes underlying the ERP components of the first stage (N1 and MMN) and those generating the P3a have been recently questioned (Horváth et al., 2008c and Rinne et al., 2006). When various complex sounds (often termed as novel sounds) are occasionally presented within a sequence predominantly made up of a regularly repeating simple tone, an early P3a subcomponent can be observed in the novel-minus-standard ERP difference waveform (Escera et al., 2000). The early P3a is probably generated in auditory cortex (Alho et al., 1998). It is, however, not known, whether this subcomponent is related to attention switching, or it reflects acoustic deviation. In active versions of the auditory oddball paradigm (i.e., when participants perform a task related to the stimuli), a central negativity, the N2b (Näätänen and Gaillard, 1983) is also often elicited by both target and non-target deviants. N2b is generally thought to reflect the controlled registration of the infrequent task-relevant event (Ritter et al., 1992). Processes taking place after the attention switch constitute the third stage of the model. The functions of these processes seem to be manifold, and there is no consensus yet regarding the correspondence between cognitive functions and the observable ERP components. If task priorities do not change as a result of the evaluation of the distracting event (i.e., the task should be carried on), the task-optimal attention has to be restored: this function is thought to be reflected, at least in part, by the modality-independent, reorienting negativity (RON), which peaks 400–600 ms after the onset of change/deviation (Berti and Schröger, 2001, Schröger et al., 2000 and Schröger and Wolff, 1998b). RON usually shows a frontal (e.g., Berti and Schröger, 2001), fronto-central (e.g., Roeber et al., 2003), or central (e.g., Horváth et al., 2008c) scalp maximum. Performing the task optimally after distraction may also require adjustments to response- and decision-related aspects of task-related processing (Berti, 2008a, Escera et al., 2001 and Horváth et al., 2008a), which may also be partly reflected by RON, or by the modulation of the parietal P3b component, which is typically elicited by targets requiring a response. The P3b probably also reflects the maintenance of the task-related stimulus context information in working memory, or decision-related processes regarding stimulus–response associations (Donchin and Coles, 1988 and Polich, 2007; but see Verleger, 1988 and Verleger, 2008). Several studies employing the 2AFC auditory distraction paradigm found the succession of N1/MMN, N2b, P3a, RON components on deviant trials (e.g., Jankowiak and Berti, 2007, Roeber et al., 2003 and Wetzel et al., 2006). These studies also found slower behavioral responses and usually more errors on deviant than on standard trials. The go/no-go variant of the auditory distraction paradigm produced the same pattern of ERPs as the 2AFC task (see, e.g., Sussman et al., 2003), although the P3a and RON component amplitudes were found to differ between short (go) and long (no-go) trials (Horváth et al., 2009). In the original version of the auditory distraction paradigm (tone duration discrimination task with pitch as task-irrelevant property, see Fig. 1), there are two moments within each trial where important stimulus information can be discovered: the onset of a tone, and the latency at which the short tone ends. Stimulus onset informs the participant about the beginning of the trial (probably facilitating preparation for the occurrence of the task-relevant information) and about the (task-irrelevant) standard/deviant status of the stimulus. At the latency at which the short tone ends the information governing the discriminative response becomes available in the form of the presence or absence of an offset of the tone. In the present study we reduced (see below) cueing information by presenting a continuous tone instead of discrete tones (continuous-stimulation paradigm, see Fig. 1). The target event was a short gap occasionally inserted into the continuous tone. Participants were required to respond to the gaps (the detection condition). The task-irrelevant distractor was an infrequent frequency glide, which is known to elicit N1 (Noda et al., 1999) and MMN whether embedded in a continuous tone (Lavikainen et al., 1995) or in short complex tones (Sams and Näätänen, 1991 and Winkler et al., 1998). The present paradigm was not designed to separate these two ERP responses. However, they both are indices of a call for attention (Näätänen, 1990) and, based on the above mentioned studies both can be expected to be elicited by the rare glides. Half of the glides were followed by a gap. The delay from the glide to the gap (when it immediately followed a glide) was constant. Note that on deviant trials, the glide could have served as a somewhat informative cue of the occurrence of gap (i.e., the glide could reduce the temporal uncertainty regarding the occurrence of the gap, although it did not fully predict its occurrence; if participants made use of this information, it might have made possible for them to temporally focus their attention on the latency of the potential gap). The effects resulting from this possibility will be taken into account when interpreting the results. Participants of the current experiment also performed a go/no-go variant of the original auditory distraction paradigm (the discrimination condition). The go/no-go task was employed to match the temporal stimulation, task, and response characteristics of the two conditions. Finally, in order to distinguish task-related from stimulus-related ERP effects, stimuli of the detection condition were also presented in a “passive” Experiment in which participants watched a silent, subtitled movie. Schematic diagram of the stimulation in the discrimination condition (a go/no-go ... Fig. 1. Schematic diagram of the stimulation in the discrimination condition (a go/no-go variant of the prototypical auditory distraction paradigm) and the detection condition (the new continuous-stimulation distraction paradigm). Time is represented on the vertical axis. Stimuli are shown as black horizontal black lines with their length representing tone duration. The vertical position of the stripes represents tone pitch, separately for the two conditions. Times at which distracting (tone/glide onset) and task-relevant information (the offset of the short tone/gap) can be discovered are marked with grey vertical lines. Figure options Differences between the pattern of ERP responses recorded in the detection and discrimination conditions (with the exception of possible N1 and MMN differences, see above) may reflect task-specific effects. Such differences, especially at the late ERP components (RON and P3b), should affect the distraction-related interpretation of the ERP results obtained in the auditory distraction paradigm.

3. Results 3.1. Behavioral data In the discrimination condition, responses to short (go) deviants were significantly slower than to short standards: t(11) = 3.05, p < 0.05 (reaction times of 361 ± 10 ms vs. 335 ± 10 ms; mean ± standard error of mean; RT's were calculated from start of the falling ramp of the tones). Sensitivity scores (d′) were significantly lower for deviants than for standards: t(11) = 5.53, p < .01 (4.23 ± 0.18 vs. 4.99 ± 0.16). In the detection condition, responses in “glide and gap” trials were significantly slower than those in “gap only” trials: t(11) = 3.76, p < .01 (363 ± 22 ms vs. 343 ± 20 ms; RT's were calculated from the start of the falling ramp of the gap). Hit rates tended to be lower for “glide and gap” than for “gap only” trials: t(11) = 1.88, p < .1 (95.6 ± 1.7% vs. 99.2 ± 0.1%). False alarm rates in the “glide only” trials in the detection condition were 13 ± 3%, whereas for long deviant trials in the discrimination condition the false alarm rate was 4 ± 1%. 3.2. Event-related potentials 3.2.1. The influence of the immediately preceding glide on the processing of gaps Group-average ERPs obtained in the passive experiment are shown in the left column of Fig. 2, separately for “glide only”, “gap only” and “glide and gap” trials. In the right column, the ERPs elicited by “glide and gap” trials are compared with the sum of the ERPs elicited by the “glide only” and “gap only” trials. Note that the zero time point on the horizontal axis is where a glide onset could occur; gaps could occur at 147.5 ms. The infrequent glides (both in the “glide only” and “glide and gap” trials) elicited an N1 peaking fronto-centrally at 128 ms, and a P2 peaking with a central maximum at 214 ms. For “glide only” trials, a relatively small fronto-central positivity peaking at 344 ms could be also distinguished (t(11) = 2.43, p < .05, at Fz). This component was identified as P3a. The gap occurring in the frequent “gap only” trials elicited an N1 peaking at 288 ms (i.e., 140 ms after the gap onset) followed by a P2 peaking at 362 ms (i.e., 214 ms after the gap onset). ERP responses elicited in the passive experiments. On the left, the ERPs ... Fig. 2. ERP responses elicited in the passive experiments. On the left, the ERPs elicited in the “gap only”, “glide only” and “glide and gap” trials are overplotted separately at the Fz, Cz, Pz electrodes and for the average of the signals recorded from the two mastoids (CM). The relevant ERP components have been labeled. On the right, the sum of the ERP waveforms elicited in the “glide only” and “gap only” trials is compared with the waveforms elicited in the “glide and gap” trials. Time zero marks where the infrequent glide (if present) would commence. The arrow on the ruler marks the latency where a gap could occur. Figure options We investigated how the gap-related ERP responses elicited in the passive experiment were affected by the presence of an immediately preceding glide by comparing the ERP amplitudes elicited in the “glide and gap” trials with the sum of the amplitudes measured for the “glide only” and “gap only” trials (see Fig. 2, right panel). Statistical results are summarized in Table 1. The divergence of the two waveforms commenced at the gap-elicited N1 wave: the trial [“glide and gap” vs. “gap only” + “glide only”] × electrode [Fz, Cz, Pz, CM] ANOVA of the amplitudes showed a significant electrode main effect and an interaction between the trial and electrode factors. The interaction was followed-up by t-tests between the two trial levels at the electrodes, which showed that the interaction was brought about by higher-amplitude negativity at Fz to “glide and gap” trials compared with the sum of the responses elicited in the “gap only” and “glide only” trials: (t(11) = 2.51, p < .05); and a higher-amplitude positivity at the mastoids in the same type of comparison (t(11) = 2.40, p < .05). That is, the interaction was caused by higher N1 amplitude for “glide and gap” trials than the sum of the responses elicited in the “gap only” and “glide only” trials. The analysis of the normalized amplitudes (an ANOVA of the same structure, see Section 2) did not yield a significant interaction. For the gap-elicited P2 amplitudes, we found significant trial and electrode main effects as well as an interaction between the two factors. The analysis of the normalized amplitudes yielded a significant interaction. Table 1. Results of the statistical analyses comparing the gap-elicited ERP amplitudes measured for “glide and gap” trials with the sum of the amplitudes measured for “gap only” and “glide only” trials, separately in the passive experiment (left column) and the detection condition of the active experiment (right column). When the interaction between the trial [“glide and gap” vs. “gap only” + “glide only”] and electrode [Fz, Cz, Pz, CM] factors was significant, an ANOVA of the same structure was performed on normalized amplitude values (see Section 2). Only the interaction is checked in these analyses (see rows labeled as “Normalized”). Latencies are measured from the onset of the gap. Passive experiment Active experiment N1 (124–156 ms) Trial ns. F(1,11) = 6.07, ηp = .35, p < .05 Electrode F(3,33) = 14.18, ɛ = .41, ηp = .56, p < .01 F(3,33) = 13.82, ɛ = .59, ηp = .56, p < .01 Interaction F(3,33) = 8.06, ɛ = .63, ηp = .42, p < .01 ns. Normalized ns. N/A P2 (198–230 ms) Trial F(1,11) = 18.38, ηp = .62, p < .01 F(1,11) = 28.07, ηp = .72, p < .01 Electrode F(3,33) = 20.60, ɛ = .37, ηp = .65, p < .01 F(3,33) = 32.64, ɛ = .74, ηp = .75, p < .01 Interaction F(3,33) = 18.96, ɛ = .59, ηp = .63, p < .01 F(3,33) = 28.79, ɛ = .50, ηp = .72, p < .01 Normalized F(3,33) = 9.69, ɛ = .58, ηp = .47, p < .01 F(3,33) = 24.23, ɛ = .73, ηp = .69, p < .01 P3b (352–432 ms) Trial N/A ns. Electrode N/A F(3,33) = 31.22, ɛ = .64, ηp = .74, p < .01 Interaction N/A ns. Normalized N/A N/A Table options Group-average ERPs obtained in the detection condition of the active experiment are shown in Fig. 3, left column. In the right column, the ERPs elicited by “glide and gap” trials are compared with the sum of the ERPs elicited by the “glide only” and “gap only” trials. The structure of the ERP responses elicited in the detection condition of the active experiment is similar to that in the passive experiment, except that P50 became discernible and P3b was elicited in the active experiment. The glide-elicited N1 and P2 showed similar topographies and peaked at or close to the latencies of the corresponding ERPs obtained in the passive experiment (128 and 210 ms, respectively). In “gap only” trials P50 peaked at 236 ms (88 ms following the gap onset); N1 and P2 peaked at the same latency as in the passive experiment; the parietal P3b peaked at 540 ms (392 ms from gap onset). ERP responses elicited in the detection condition of the active experiment. On ... Fig. 3. ERP responses elicited in the detection condition of the active experiment. On the left, the ERPs elicited in the “glide only”, “gap only”, and “glide and gap” trials are overplotted separately at the Fz, Cz, Pz electrodes and for the average of the signals recorded from the two mastoids (CM). The relevant ERP components have been labeled. On the right, the sum of the ERP waveforms elicited in the “glide only” and “gap only” trials is contrasted with the waveforms elicited in the “glide and gap” trials. Time zero marks where the infrequent glide (if present) would commence. The arrow on the ruler marks the latency where a gap could occur. Figure options Testing the effects of the preceding glide on the gap-elicited N1 the trial × electrode ANOVA yielded significant trial and electrode main effects. For P2, we found significant trial and electrode main effects, as well as an interaction between the two factors. The ANOVA of the normalized amplitudes also yielded a significant interaction. Finally, only the electrode factor yielded a significant effect for P3b. In summary, in the passive experiment, the infrequent glide preceding the gap increased the amplitude of the N1, and reduced the amplitude of the P2 elicited by the gap. The P2-reduction did not affect the response recorded at the mastoid leads suggesting that generators outside auditory cortex were affected by the processing of the preceding glide. In the detection condition of the active experiment, the glide immediately preceding the gap decreased the amplitude of the gap-elicited N1 and P2. The decrease of the P2 was larger at central and posterior areas than frontally and no effect was observed at the mastoid leads. The presence of the glide did not influence the P3b amplitude. 3.2.2. The effects of attention to the sound on the processing of gaps Fig. 4 shows the ERP responses elicited by the “gap only” trials of the passive and active experiments. Gaps were task-relevant in the active experiment allowing us to test the effects of task-relevance on the ERP responses by comparing them between the two experiments. Statistical results are summarized in Table 2. ERP responses elicited in “gap only” trials in the active and passive ... Fig. 4. ERP responses elicited in “gap only” trials in the active and passive experiments at the Fz, Cz, Pz electrodes and for the average of the signals recorded from the two mastoids (CM). Time zero marks where the infrequent glide (if present) would commence. The relevant ERP components have been labeled. The arrow on the ruler marks the latency where a gap could occur. Figure options Table 2. Results of the statistical analyses comparing the ERP amplitudes measured for “gap only” trials, between the passive experiment and the detection condition of the active experiment. When the interaction between the trial [passive vs. active experiment] and electrode [Fz, Cz, Pz, CM] factors was significant, an ANOVA of the same structure was performed on normalized amplitude values (see Section 2). Only the interaction is checked in these analyses (see rows labeled as “Normalized”). Latencies are measured from the onset of the gap. P50 (72–104 ms) Trial F(1,22) = 8.28, ηp = .27, p < .01 Electrode F(3,66) = 10.79, ɛ = .57, ηp = .33, p < .01 Interaction F(3,66) = 4.12, ɛ = .57, ηp = .16, p < .05 Normalized N/A N1 (124–156 ms) Trial F(1,22) = 23.94, ηp = .52, p < .01 Electrode F(3,66) = 39.56, ɛ = .49, ηp = .64, p < .01 Interaction F(3,66) = 5.42, ɛ = .49, ηp = .20, p < .05 Normalized ns. P2 (198–230 ms) Trial F(1,22) = 8.55, ηp = .28, p < .01 Electrode F(3,66) = 66.94, ɛ = .62, ηp = .75, p < .01 Interaction F(3,66) = 10.68, ɛ = .62, ηp = .33, p < .01 Normalized F(3,66) = 4.55, ɛ = .59, ηp = .17, p < .05 P3b (352–432 ms) Trial F(1,22) = 18.96, ηp = .46, p < .01 Electrode F(3,66) = 5.06, ɛ = .75, ηp = .19, p < .05 Interaction F(3,66) = 13.30, ɛ = .75, ηp = .38, p < .01 Normalized N/A Table options The trial [passive vs. active experiment] × electrode [Fz, Cz, Pz, CM] ANOVA of the average amplitudes in the P50 interval yielded significant trial and electrode main effects, as well as an interaction between the two factors. Because no P50 was observable in the passive experiment, topographical differences were not analyzed. For N1, we found significant trial and electrode main effects, as well as an interaction between the two factors. The ANOVA of the normalized amplitudes did not yield a significant interaction. For P2, we found significant trial and electrode main effects, as well as an interaction between the two factors. The ANOVA of the normalized amplitudes also showed a significant interaction. Finally, for P3b, we found significant trial and electrode main effects, as well as an interaction between the two factors. Because no P3b was observable in the passive experiment, topographical differences were not analyzed. In summary, compared to the passive experiment, gaps elicited a P50, a highly enhanced (more than twice as large) N1, as well as a P3b. The P2 amplitude was also enhanced in a way which suggested a difference in the generator structures of the deflections. 3.2.3. The effects of attention to the sound on the sensory processing of infrequent glides Although glides were largely task-irrelevant in the active experiment (noting their possible temporal cueing role), the continuous sound was attended in the active but not necessarily in the passive experiment, in which participants watched a movie. In Fig. 5, the group-average ERPs elicited in the “glide only” trials are compared between the active and passive experiments. The N1 and the P2 elicited by the glide in the two experiments did not significantly differ from each other (the trial [active vs. passive experiment] × electrode ANOVAs of the average N1 and P2 amplitudes showed only electrode main effects: F(3,66) = 94.30, ɛ = .47, ηp = .81, p < .01 and F(3,66) = 114.26, ɛ = .60, ηp = .84, p < .01, for N1 and P2, respectively). ERP responses elicited in the “glide only” trials in the active and the passive ... Fig. 5. ERP responses elicited in the “glide only” trials in the active and the passive experiments at the Fz, Cz, Pz electrodes and for the average of the signals recorded from the two mastoids (CM). Time zero marks where the infrequent glide (if present) would commence. The relevant ERP components have been labeled. The arrow on the ruler marks the latency where a gap could occur. Figure options 3.2.4. Distraction-related ERP components in the detection condition of the active experiment Because the presence of the glide affected the ERPs elicited by the gap in the “glide and gap” (“deviant”) trials compared with the “gap only” (“standard”) trials even in the passive experiment, the best estimate of the effects of distraction in the detection condition of the active experiment can be obtained by assessing the elicitation of the late distraction-related ERP components in the “glide only” trials (in which a low-amplitude P3a was elicited, and RON was not visible at all; see Fig. 5). “Glide only” trials elicited the P3a peaking at 360 ms (t(11) = 5.42, p < .01 at Fz), RON peaking at 448 ms (t(11) = 1.91; p < .05 at Fz), and P3b peaking at 536 ms (t(11) = 5.83, p < .01 at Pz). 3.2.5. ERP components in the discrimination condition Group-average ERPs and deviant-minus-standard difference waveforms for short (go) and long (no-go) stimuli in the discrimination condition are shown in Fig. 6. Both standard- and deviant-stimulus ERPs exhibited an N1 peaking at about 130 ms, and a P3b peaking between 500 and 600 ms from stimulus onset. ERPs for standards showed a slow frontal negativity in the 200–400 ms range, whereas ERPs for deviants showed a centrally maximal N2 with no polarity reversal at the mastoids, which peaked at about 180 ms from stimulus onset and a centrally maximal positivity peaking at about 320 ms from stimulus onset. Judged by their latency and scalp distribution, these waveforms were identified as N2b and P3a, respectively. The deviant-minus-standard difference waveforms showed an N1-increment/MMN peaking at 130 ms from stimulus onset, followed by an N2b at 188 ms and a P3a peaking at 314 ms. For the short (go) stimuli, an RON was observable peaking at 424 ms, which was followed by a P3b peaking at 620 ms. ERP responses elicited in the discrimination condition of the active experiment. ... Fig. 6. ERP responses elicited in the discrimination condition of the active experiment. The left column displays the ERPs elicited by standard and deviant short stimuli together with the corresponding deviant-minus-standard difference waveforms, separately at the Fz, Cz, Pz electrodes and for the average of the signals recorded from the two mastoids (CM). The right column displays the ERPs elicited by standard and deviant long stimuli together with the corresponding deviant-minus-standard difference waveforms. The relevant ERP components have been labeled on the difference waveforms. The arrows on the rulers mark the latency when task-relevant information could be discovered (i.e., the offset of the short tone). Figure options ERPs elicited by short and long tones were separately analyzed by ANOVAs of trial [deviant vs. standard] × electrode [Fz, Cz, Pz, CM] structure, because in a previous study employing a similar stimulus paradigm (Horváth et al., 2009), we found that P3a and RON amplitudes were lower for long (no-go) than for short (go) stimuli. Difference waveforms were then compared between long and short tones by ANOVAs of duration [long vs. short] × electrode [Fz, Cz, Pz, CM] structure. The results of the statistical analyses are summarized in Table 3. Note that direct comparison between the detection and discrimination conditions (between the temporally and functionally corresponding “glide only” and “long deviant tone” trials) is not meaningful, because of the different nature of the transients (onset vs. glide and offset vs. gap). A qualitative comparison will be given in Section 4. Table 3. Results of the statistical analyses of the ERP amplitudes measured in the discrimination condition of the active experiment, separately for short (left column) and long tones (right column). When the interaction between the trial [standard vs. deviant] and electrode [Fz, Cz, Pz, CM] factors was significant, an ANOVA of the same structure was performed on normalized amplitude values (see Section 2). Only the interaction is checked in these analyses (see rows labeled as “Normalized”). Results of the comparison of the deviant-minus-standard difference amplitudes between short and long tones duration [short vs. long tones] × electrode [Fz, Cz, Pz, CM] is shown in the central column. Short tones Comparison Long tones N1/MMN (114–146 ms) Trial/duration F(1,11) = 21.89, ηp = .66, p < .01 ns. F(1,11) = 6.52, ηp = .37, p < .05 Electrode F(3,33) = 111.18, ɛ = .60, ηp = .91, p < .01 F(3,33) = 51.93, ɛ = .58, ηp = .53, p < .01 F(3,33) = 89.22, ɛ = .66, ηp = .89, p < .01 Interaction F(3,33) = 43.04, ɛ = .68, ηp = .79, p < .01 ns. F(3,33) = 42.69, ɛ = .49, ηp = .79, p < .01 Normalized ns. N/A ns. N2b (166–194 ms) Trial/duration F(1,11) = 9.94, ηp = .47, p < .01 ns. F(1,11) = 6.52, ηp = .37, p < .05 Electrode F(3,33) = 10.39, ɛ = .48, ηp = .49, p < .01 F(3,33) = 12.07, ɛ = .48, ηp = .52, p < .01 F(3,33) = 12.4004, ɛ = .48, ηp = .53, p < .01 Interaction F(3,33) = 3.56, ɛ = .62, ηp = .24, p < .05 ns. ns. Normalized ns. N/A N/A P3a (306–334 ms) Trial/duration F(1,11) = 130.79, ηp = .92, p < .01 ns. F(1,11) = 35.16, ηp = .76, p < .05 Electrode F(3,33) = 15.06, ɛ = .76, ηp = .91, p < .01 F(3,33) = 29.05, ɛ = .54, ηp = .72, p < .01 F(3,33) = 16.49, ɛ = .68, ηp = .60, p < .01 Interaction F(3,33) = 59.18, ɛ = .81, ηp = .84, p < .01 ns. F(3,33) = 14.61, ɛ = .60, ηp = .57, p < .01 Normalized F(3,33) = 51.34, ɛ = .83, ηp = .82, p < .01 N/A F(3,33) = 14.45, ɛ = .61, ηp = .57, p < .01 RON (408–440 ms) Trial/duration F(1,11) = 9.61, ηp = .47, p < .05 F(1,11) = 13.43, ηp = .55, p < .05 ns. Electrode F(3,33) = 15.43, ɛ = .68, ηp = .58, p < .01 F(3,33) = 3.41, ɛ = .58, ηp = .24, p < .01 F(3,33) = 16.26, ɛ = .67, ηp = .60, p < .01 Interaction F(3,33) = 8.23, ɛ = .61, ηp = .43, p < .01 F(3,33) = 10.01, ɛ = .70, ηp = .48, p < .01 ns. Normalized F(3,33) = 7.62, ɛ = .62, ηp = .41, p < .01 N/A N/A P3b (580–660 ms) Trial/duration F(1,11) = 12.24, ηp = .53, p < .01 F(1,11) = 6.45, ηp = .37, p < .05 ns. Electrode F(3,33) = 46.10, ɛ = .87, ηp = .81, p < .01 F(3,33) = 10.80, ɛ = .69, ηp = .49, p < .01 F(3,33) = 23.06, ɛ = .73, ηp = .68, p < .01 Interaction F(3,33) = 7.70, ɛ = .63, ηp = .41, p < .01 ns. F(3,33) = 5.82, ɛ = .72, ηp = .35, p < .01 Normalized ns. N/A ns. Table options N1/MMN. For short tones, the ANOVA of the N1/MMN amplitudes yielded significant trial [deviant vs. standard] and electrode [Fz, Cz, Pz, CM] main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes did not yield a significant interaction. For long tones, we found significant trial and electrode main effects, as well as an interaction between the two factors. Again, the analysis of the normalized amplitudes did not yield a significant interaction. The ANOVA comparing the deviant-minus-standard differences between short and long tones only yielded a significant Electrode main effect. Thus similar N1's and N1-increment/MMN responses were found at both tone durations. N2b. For short tones, the ANOVA of the amplitudes in the N2b latency range yielded significant trial and electrode main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes did not yield a significant interaction. For long tones, we found significant trial and electrode main effects. The ANOVA comparing the deviant-minus-standard differences only yielded a significant electrode main effect. Thus similar N2b responses were elicited by deviants at both durations. P3a. For short tones, the ANOVA of the amplitudes in the P3a latency range yielded significant trial and electrode main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes showed a significant interaction. For long tones, we found significant trial and electrode main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes showed a significant interaction. The ANOVA comparing the deviant-minus-standard differences between short and long tones only yielded an electrode main effect. Thus similar P3a responses were elicited by deviants at both durations. RON. For short tones, the ANOVA of the amplitudes in the RON latency range yielded significant trial and electrode main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes showed a significant interaction. For long tones, we only found a significant electrode main effect. The ANOVA comparing the deviant-minus-standard differences between short and long tones yielded significant duration [short vs. long tones] and electrode main effects, as well as an interaction between the two factors. Thus a significant RON response was only found for the short (go) stimuli. P3b. For short tones, the ANOVA of the P3b amplitudes yielded significant trial and electrode main effects, as well as an interaction between the two factors. The analysis of the normalized amplitudes did not yield a significant interaction. For long tones, we found a significant electrode main effect and an interaction between the trial and electrode factors. The analysis of the normalized amplitudes did not yield a significant interaction. Using Student's dependent two-sided t-tests, separately for each electrode, it appeared that the difference was significant only at Pz: t(11) = 3.24, p < .01. The ANOVA comparing the deviant-minus-standard differences between short and long tones yielded significant duration and electrode main effects. Thus the deviation-related amplitude increment of the P3b was larger for the short (go) than for the long (no-go) stimuli.