آیا N1 / MMN، P3a و RON شکل دهنده زنجیره ای شدیدا همراه و منعکس کننده سه مرحله از حواس پرتی شنوایی هست؟
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38732||2008||9 صفحه PDF||سفارش دهید||7899 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Biological Psychology, Volume 79, Issue 2, October 2008, Pages 139–147
Abstract Distraction triggered by unexpected events is generally described in a serial model comprising (1) automatic detection of unexpected task-irrelevant events, (2) orienting towards the event, and (3) recovery from distraction. Processes taking place at the three stages are assumed to be reflected by the N1 and mismatch negativity (MMN); the P3a; and the reorienting negativity (RON) event-related potentials (ERPs), respectively. We investigated whether the processes indexed by these components form a strongly coupled chain, each co-varying with the preceding one. To this end, micro-sequence analysis of the ERPs elicited by unpredictable pitch-changes was conducted in an auditory duration discrimination task. Results indicated that the processes indexed by the above-mentioned ERPs are not strongly coupled. Pair-wise dissociations were found between the ERPs reflecting each processing stage: P3a can be elicited without concurrent N1-increase or MMN elicitation and without subsequent elicitation of the RON. Possible interpretations of P3a and RON are discussed.
1. Introduction In everyday life, optimal performance can often be achieved by focusing on the task-relevant aspects of the incoming stimuli while disregarding other information. However, a “perfect” selective attention set, in which resources are used only for processing task-relevant events, cannot be achieved: task-irrelevant sensory events occasionally draw away our attention, that is, we get distracted (passive attention, see James, 1890). Though involuntary shifts of attention lead to temporary deterioration in task performance, they may also bring potentially important information into the focus of attention, thereby allowing re-evaluation of the whole situation. In the optimal case, distraction and the maintenance of goal-directed behavior are well balanced. Distraction is usually triggered by unexpected events in the sensory environment. It is assumed that such events initiate a chain of processes starting with automatic detection of unexpected sensory information, followed by shifting of the focus of attention, and concluded by restoration of the task-optimal attention set (unless the distracting event requires immediate adaptive changes). In the present study, we investigated whether the above listed processes form a true causal chain, that is, whether or not each processing step is governed by the preceding one. Distraction (i.e., an involuntary change of the selective attention set) is usually investigated in the oddball paradigm, in which some stimuli of a regular sequence are replaced by irregular ones. Regular stimuli are termed standards, irregular ones deviants. The sequence of distraction-related processes can be described by a three-stage sequential model (for a detailed summary of the model, see Escera and Corral (2003) and Escera and Corral (2007); also Horváth et al., in press; different parts of this model were described by Escera et al. (2000), Friedman et al. (2001), Näätänen, 1990 and Näätänen, 1992, Polich (2007), Polich and Criado (2006), Schröger (1997) and Schröger and Wolff (1998b). Processes of the three stages are thought to be reflected by specific event-related potentials (ERPs). The first stage features sensory processes forming an adaptive filter, which minimizes the sensory-information load on capacity-limited processing resources. These processes do not require voluntary activation, although their outcome can be affected by some top-down processes (e.g., Hansen and Hillyard, 1988 and Sussman et al., 2002). Two of these processes are thought to be directly reflected in ERPs: (1) an onset and simple (first-order) change detection process eliciting the N1 wave (for a review, see Näätänen and Picton, 1987); and (2) a separate sensory memory-based deviance-detection process indexed by the mismatch negativity (MMN) ( Näätänen et al., 1978; for a review, see Näätänen et al., 2007). The modality-specific component of the N1 wave reflects differential activation of neural elements sensitive to various stimulus features: stimulus-repetition leads to decreased responses, whereas change restores the full activation. The memory-based comparison process indexed by the MMN component compares incoming stimuli to representations generated on the basis of temporal regularities extracted from the auditory input. When a stimulus proves to be incompatible with the current regularity-representations, MMN is elicited. Thus, N1 and MMN can be regarded as first- and second-order change detectors. Both N1 and MMN are modality-specific ( Czigler et al., 2002, Näätänen et al., 1978 and Shinozaki et al., 1998); N1 usually peaks between 80 and 120 ms, whereas MMN peaks between 100 and 200 ms from the onset of change/deviance, respectively. It is hypothesized that both processes can call for higher level, capacity-limited attentional processes, when their activation exceeds a variable threshold, the level of which is set by top-down control (e.g., stronger or weaker focus on the ongoing behavior; Schröger, 1997). Therefore, small changes or deviations may not reach the second processing stage (attention switching), whereas large changes or deviations usually trigger involuntary attention switching to the stimulus, which elicited the large N1/MMN (Escera et al., 1998, Näätänen, 1990, Rinne et al., 2006 and Schröger, 1997). Typically, slow gradual changes in the environment are automatically incorporated into an implicit representation of the sensory environment and thus may go unnoticed unless they are attended. In contrast, the same change occurring suddenly may capture one's attention. Some processes related to attention switching are thought to be reflected by the fronto-central P3a component, which usually peaks at about 300 ms from the onset of the change/deviance and is at least partly modality-independent (Escera et al., 2000, Friedman et al., 2001, Knight and Scabini, 1998 and Schröger, 1996). Note, that although P3a is often taken to reflect the process of involuntary attention switching itself, there is no general consensus on the precise role of P3a within the second processing stage (Dien et al., 2004). Finally, in the third processing stage, if the event that triggered an attention switch does not require immediate adaptive changes (i.e., the task priorities do not change), the task-optimal selective attention set is restored. This function is hypothesized to be reflected by the modality-independent, fronto-central reorienting negativity (RON) component, which peaks 400–600 ms after the onset of change/deviation ( Berti and Schröger, 2001, Schröger et al., 2000, Schröger and Wolff, 1998a and Schröger and Wolff, 1998b). Schröger and Wolff, 1998a and Schröger and Wolff, 1998b introduced a paradigm optimized for investigating all three stages of distraction by behavioral and ERP measures. In this auditory distraction paradigm, a sequence of stimuli varying in two features is presented. One auditory feature (e.g., sound duration) is task-relevant; the other (e.g., pitch) is task-irrelevant. The task-relevant feature varies equiprobably (e.g., 50% of the stimuli are short, the other 50% long), whereas levels of the task-irrelevant feature are unevenly distributed across the stimuli (e.g., 90% of the sounds have high, whereas 10% have low pitch). On each trial, participants perform a discrimination task based exclusively on the task-relevant feature. It has been shown that infrequent unexpected changes in the task-irrelevant feature (deviants) cause distraction: responses to deviants are usually slower than to standards; the N1 amplitude increases and MMN, P3a, and RON are elicited by deviants. Several studies using this paradigm provided evidence compatible with the three-stage model of distraction. It has been shown that larger deviations elicit larger N1/MMN, P3a and RON components, suggesting that activation of the corresponding processes is correlated with each other and with the magnitude of stimulus deviance (Berti et al., 2004, Escera et al., 2001 and Yago et al., 2001). Bendixen et al. (2007) found that all three components showed increasing amplitudes when the deviant was preceded by longer standard-stimulus sequences. On the other hand, it seems also likely that the processes indexed by P3a and RON are not fully governed by N1/MNN. A number of studies obtained evidence suggesting that P3a and RON can be modulated by factors not affecting N1/MMN. Winkler et al. (1998) found that whereas the MMN amplitude saturated at intermediate magnitudes of stimulus deviance, the P3a amplitude increased with further increases of the magnitude of deviance. Berti and Schröger (2003) found that some manipulations of working memory load did not affect the N1/MMN amplitude, but resulted in decreased P3a and RON amplitudes. Distraction-related processing can also be limited to the first stage of the model: Ritter et al. (1999) and Sussman et al. (2003) showed that it is possible to prevent the elicitation of P3a and RON by making the occurrence of deviants fully predictable through the use of visual cues. However these manipulations had no effect on the elicitation and amplitude of the N1/MMN. One model-conforming explanation of these findings is that it is possible to dynamically raise or lower the threshold through which bottom-up activation can reach the attention-switching mechanism (Schröger, 1997). On the other hand, it is also possible that knowledge about the occurrence of the deviant could only be utilized by the processes of the second and third stages of the distraction model, but not by those of the first stage. The auditory system rapidly adapts to changes in environment. A number of studies suggest that after one or two repetitions of a stimulus (i.e., two or three stimulus presentations), a subsequent change can elicit MMN (Bendixen et al., 2007, Horváth et al., 2001 and Winkler et al., 1996). Furthermore, increasing N1/MMN amplitudes are elicited by changes following homogeneous micro-sequences of increasing length (Giese-Davis et al., 1993 and Sams et al., 1983). If processes of the three-stage distraction model form a tightly coupled chain then the presumed ERP correlates of these processes should exhibit similar patterns of behavior as a function of the length of the homogeneous micro-sequence of sounds preceding a deviant stimulus. In the present study, we investigated distraction-related ERP responses elicited by sound change following repetitive micro-sequences. We term such stimulus changes as “local deviants”. We employed a variant of the auditory distraction paradigm introduced by Schröger and Wolff, 1998a and Schröger and Wolff, 1998b; described above), in which, instead of presenting different levels of the task-irrelevant stimulus feature with uneven probabilities, the two levels of this feature were delivered equiprobably (that is, there were no global deviant or standard stimuli within the sequences). Because the order of the sounds was randomized, repetitive micro-sequences of various lengths emerged incidentally within the sound sequences. Local deviants of the task-irrelevant stimulus feature (change in the task-irrelevant feature following a micro-sequence within which the task-irrelevant feature was constant) can be expected to cause distraction and thus elicit the distraction-related ERP components. In contrast, repetitions of the task-irrelevant stimulus feature (“local standards”) should not cause distraction. If the processes indexed by the sequence of N1/MMN, P3a, and RON form a causal chain, then the amplitudes of these components should exhibit similar behavior as a function of the length of the homogeneous micro-sequence preceding local deviants. Different patterns of behavior of these ERP components would suggest that the processes reflected by these components do not form a tightly coupled sequence, or that the functional interpretation of some of these components should be reconsidered.
نتیجه گیری انگلیسی
. Results 3.1. Behavioral measures Group-average reaction times and error rates are presented in Fig. 1. Group-average (N=12) reaction times and error rates (ER) for task-irrelevant ... Fig. 1. Group-average (N = 12) reaction times and error rates (ER) for task-irrelevant stimulus changes and repetitions as a function of the length of the preceding homogenous micro-sequence. Reaction times have been measured from tone onset, which proceeds by 200 ms the time when difference between the short and long tones commenced. Standard errors of mean are shown by whiskers (R stands for pitch-repetition and C for pitch-change). Figure options The Stimulus-type × Micro-sequence-length ANOVA of the reaction times showed a significant main effect of stimulus type: F(1,11) = 11.72, p < 0.01, η2 = 0.52; a significant main effect of the micro-sequence-length: F(3,33) = 4.87, p < 0.05, ɛ = 0.66, η2 = 0.31; and a significant interaction between the two factors: F(3,33) = 4.49, p < 0.05, ɛ = 0.62, η2 = 0.29. Post hoc Tukey-HSD tests showed that responses to changes were generally slower than to repetitions, except for the “CC” sub-sequence (at least p < 0.05 for all comparisons); and responses to “CRRRC” and “CRRC” were significantly slower than those to “CC” (p < 0.001 and p < 0.05, respectively). The Stimulus-type × Micro-sequence-length ANOVA of the error rates showed no significant effects. 3.2. ERP measures The group-average ERPs elicited by the “RRRRRRC” and the “RRRRRRR” stimuli and the “RRRRRRC”-minus-“RRRRRRR” difference waveform are presented in Fig. 2. The ERPs and the difference waveform were very similar to those elicited in oddball versions of the paradigm tested in previous studies (see e.g., Roeber et al., 2003). The tones elicited two successive fronto-central negative waves peaking at about 120 and 260 ms, respectively. The negative peaks were separated by a positive-going wave peaking at about 170 ms. The second negativity was followed by a positive-going peak at about 310 ms, a negativity peaking at about 410 ms, and a large parietal positivity peaking at about 600 ms. Because the earliest of these negativities (120 ms) showed polarity inversion at the mastoids, it was identified as N1, possibly also containing contribution from MMN in the responses elicited by the RRRRRRC micro-sequence. We shall thus refer to this peak as N1/MMN in the following. After the positive-going P2 (the trough peaking at 170 ms), the second negativity is probably an N2b (because it shows no polarity inversion at the mastoids), which is followed by a P3a response. The late parietal positivity was identified as P3b. The difference waveform showed the expected succession of the N1/MMN, P3a, and RON components. In the difference waveform, N1/MMN peaked at 120 ms, P3a at 288 ms, and RON at 516 ms (average latencies at frontal and central sites). All three components showed a fronto-centrally maximal scalp topography. In the following, these distraction-related ERP components are analyzed in detail. Group-average (N=12) ERPs elicited by the sequence-final stimuli of the ... Fig. 2. Group-average (N = 12) ERPs elicited by the sequence-final stimuli of the “RRRRRRC” and “RRRRRRR” micro-sequences together with the corresponding change-minus-repetition difference waveform. (R stands for pitch repetition and C for pitch change.) Stimulus onset is at the crossing of the two axes. Figure options 3.2.1. Elicitation of the distraction-related ERPs by change following a long homogeneous micro-sequence The ERP components elicited by the “RRRRRRC” and “RRRRRRR” micro-sequences were compared to determine whether the expected distraction-related ERPs were elicited in the paradigm (see Fig. 2). The Stimulus-type (“RRRRRRC” vs. “RRRRRRR”) × Electrode (F3, Fz, F4, C3, Cz, and C4) ANOVA showed a significant Stimulus-type main effect in the first measurement interval (N1/MMN: 100–140 ms): F(1,11) = 6.15, p < 0.05, η2 = 0.36, indicating that the ERP amplitude elicited by change was significantly more negative than that elicited by repetition in the N1/MMN interval (−3.86 ± 0.67 μV vs. −2.52 ± 0.67 μV; mean ± standard errors of mean for the change and repetition trials, respectively). Thus a significant N1-increase/MMN was elicited by changes following repetitive micro-sequences. The ANOVA of the same structure showed a significant stimulus type main effect for the second time interval (P3a: 268–308 ms): F(1,11) = 36.55, p < 0.001, η2 = 0.77; and a tendency for an interaction between the two factors: F(5,55) = 2.90, p < 0.1, ɛ = 0.53, η2 = 0.20. The stimulus main effect was caused by more positive ERP amplitude in response to change compared to that to repetition (2.39 ± 0.96 μV vs. −2.41 ± 0.63 μV, for the change and repetition trials, respectively). Thus change following a repetitive micro-sequence elicited P3a. The tendency for interaction suggests that the P3a scalp distribution (elicited by change trials) was different from that of the ERP activity occurring in the same time interval on repetition trials. For the third time window (RON: 496–536 ms) the same type of ANOVA showed a significant stimulus type main effect: F(1,11) = 5.94, p < 0.05, η2 = 0.35; and an electrode main effect: F(5,55) = 22.21, p < 0.001, ɛ = 0.45, η2 = 0.67. The stimulus main effect was caused by lower positive ERP amplitudes elicited by change compared to repetition (1.65 ± 0.89 μV vs. 3.79 ± 1.00 μV, for the change and repetition trials, respectively). The electrode main effect was brought about by a generally higher positive signal on the central than frontal electrodes (as shown by a Tukey-HSD test, p < 0.05 for all comparisons between central and frontal electrodes; the signal was also significantly lower at C4 than at Cz, p < 0.05). Thus RON was elicited on change trials following a repetitive micro-sequence. 3.2.2. Effects of micro-sequence-length in the N1/MMN time interval In the N1/MMN time interval (100–140 ms, see Fig. 3 and Fig. 4), the Stimulus-type (change, repetition) × Micro-sequence-length (2,3,4,5) × Electrode (F3, Fz, F4, C3, Cz, and C4) ANOVA showed a significant Stimulus-type main effect: F(1,11) = 39.82, p < 0.001, η2 = 0.78; a significant Stimulus-type × Electrode interaction: F(5,55) = 5.50, p < 0.01, ɛ = 0.54, η2 = 0.33; and a significant Stimulus-type × Micro-sequence-length interaction: F(3,33) = 9.79, p < 0.001, ɛ = 0.58, η2 = 0.47. Because the Stimulus-type was involved in both interactions, two further ANOVAs were carried out for changes and repetitions separately. For changes, the Micro-sequence-length (“CC”, “CRC”, “CRRC”, and “CRRRC”) × Electrode (F3, Fz, F4, C3, Cz, and C4) ANOVA on the N1/MMN amplitudes showed a significant main effect of Micro-sequence-length: F(3,33) = 7.99, p < 0.01, ɛ = 0.59, η2 = 0.42. Tukey HSD post hoc tests showed that the responses elicited by “CRRRC” micro-sequences were more negative than those elicited by “CC” (p < 0.001) and “CRC” micro-sequences (p < 0.1); also the “CRRC” responses were more negative than the “CC” responses (p < 0.01). For repetitions (see Fig. 3 and Fig. 4), the Electrode (F3, Fz, F4, C3, Cz, and C4) × Micro-sequence-length (“CR”, “CRR”, “CRRR”, and “CRRRR”) ANOVA showed no significant effects. That is, the significant Stimulus-type × Micro-sequence-length interaction was explained by the results showing that the N1/MMN amplitude increased with the length of the homogenous micro-sequence preceding a change, whereas the length of the homogeneous micro-sequence had no significant effect on the N1 elicited by repetitions. Group-average (N=12) ERPs to task-irrelevant pitch changes (first column) and ... Fig. 3. Group-average (N = 12) ERPs to task-irrelevant pitch changes (first column) and repetitions (second column) following pitch-homogenous micro-sequences of different lengths at the midline electrodes. The third and fourth columns show the corresponding change-minus-RRRRR and repetition-minus-RRRRR difference waveforms. (RRRRR is a task-irrelevant repetition preceded by at least four other repetitions.) Note that the calibration of the vertical axes is different for the third and fourth columns. Stimulus onset is at the crossing of the two axes. Figure options Group-average (N=12) ERP amplitudes and standard errors of means in the N1/MMN, ... Fig. 4. Group-average (N = 12) ERP amplitudes and standard errors of means in the N1/MMN, P3a, and RON time intervals averaged over measurements from frontal and central electrodes (F3, Fz, F4, C3, Cz, and C4) for task-irrelevant pitch changes and repetitions following pitch-homogenous micro-sequences of different lengths. Figure options 3.2.3. Effects of micro-sequence-length in the P3a time interval In the P3a time interval (268–308 ms, see Fig. 3 and Fig. 4), the Stimulus-type × Micro-sequence-length × Electrode ANOVA showed a significant Stimulus-type main effect: F(1,11) = 22.61, p < 0.001, η2 = 0.67; a significant Stimulus type × Micro-sequence-length interaction: F(3,33) = 16.48, p < 0.001, ɛ = 0.69, η2 = 0.60; and a tendency for a Stimulus type × Micro-sequence-length × Electrode interaction: F(15,165) = 2.29, p < 0.1, ɛ = 0.29, η2 = 0.17. To specify these effects, two separate ANOVAs were carried out for changes and repetitions. For changes, the Micro-sequence-length × Electrode ANOVA showed a significant main effect of Micro-sequence-length: F(3,33) = 15.73, p < 0.001, ɛ = 0.89, η2 = 0.59; and a significant interaction between the two factors: F(15,165) = 3.43, p < 0.05, ɛ = 0.31 η2 = 0.24. Because the Micro-sequence effect was present at all electrodes, a detailed analysis of the interaction was not relevant for the purpose of the present study. Post hoc Tukey HSD-tests for the Micro-sequence-length effect showed that the “CRRRC” responses were more positive than the “CC” (p < 0.001) and the “CRC” responses (p < 0.05); also the “CRRC” responses were more positive than “CC” responses (p < 0.001), and the “CRC” responses were more positive than the “CC” responses (p < 0.05). The same type of ANOVA for repetitions showed a Micro-sequence-length main effect: F(3,33) = 4.20, p < 0.05, ɛ = 0.70, η2 = 0.27. Tukey-HSD post hoc tests showed that the ERP response to the “CR” micro-sequence was more positive than that to the other three micro-sequences (at least p < 0.05 for all comparisons). In summary, the P3a amplitude increased with the length of homogenous sequences preceding a change; and, surprisingly, a P3a was elicited by the first tone repetition following a change, but not by further tone repetitions. 3.2.4. Effects of micro-sequence-length in the RON time interval In the RON time interval (496–536 ms, see Fig. 3 and Fig. 4), the Stimulus-type × Micro-sequence-length × Electrode ANOVA showed a significant electrode main effect: F(1,11) = 34.02, p < 0.001, η2 = 0.75; a tendency for a Micro-sequence-length main effect: F(3,33) = 2.61, p < 0.1, ɛ = 0.65, η2 = 0.19; and a tendency for a Stimulus-type × Electrode interaction: F(5,55) = 3.06, p < 0.1, ɛ = 0.49, η2 = 0.22. Post hoc Tukey HSD-tests showed that the signal at central electrodes was significantly more positive than at frontal electrodes (at least p < 0.01 for all comparisons); the signal was also more positive at Cz than at C4 (p < 0.01). 3.2.5. Comparison between the three ERP components as a function of micro-sequence-length The behavior of the three distraction-related ERP components as a function of the length of the homogeneous micro-sequence preceding a pitch repetition or change is shown in Fig. 4. Our first comparisons are for the ERP amplitudes elicited by pitch-change. The Component × Micro-sequence-length ANOVA on the z-transformed N1/MMN and P3a amplitudes (see Section 2) only showed a significant main effect of Micro-sequence-length: F(3,33) = 26.41, p < 0.001, ɛ = 0.71, η2 = 0.71. The same ANOVA for the N1/MMN and RON amplitudes also showed only a significant main effect of Micro-sequence-length: F(3,33) = 13.09, p < 0.001, ɛ = 0.70, η2 = 0.54. Finally, the same ANOVA for P3a and RON showed a significant main effect of Micro-sequence-length: F(3,33) = 23.67, p < 0.001, ɛ = 0.82, η2 = 0.68; and a significant interaction between the two factors: F(3,33) = 3.75, p < 0.05, ɛ = 0.80, η2 = 0.25. In summary, a difference between the amplitude-micro-sequence-length functions was indicated for P3a and RON (see Fig. 3 and Fig. 4), but not between the other pairs of components. Following up on the unexpected elicitation of P3a by “CR” micro-sequences (see Fig. 3), Component × Micro-sequence (“CR”, “CRR”) ANOVAs were conducted on the z-transformed ERP component amplitudes elicited by pitch repetitions. The ANOVA for N1/MMN and P3a showed a significant main effect of Micro-sequence: F(1,11) = 12.83, p < 0.01, η2 = 0.54; and a significant interaction between the two factors: F(1,11) = 14.73, p < 0.01, η2 = 0.57. For P3a and RON a significant main effect of Micro-sequence-length: F(1,11) = 6.34, p < 0.05, η2 = 0.36; and a significant interaction between the two factors were found: F(1,11) = 19.48, p < 0.01, η2 = 0.64. In summary, the P3a component was affected significantly differently by task-irrelevant pitch repetitions following a pitch change (“CR” vs. “CRR”) compared with either N1/MMN or RON. 3.2.6. Effects of micro-sequence-length in the P3b time interval P3b peaked at 600 ms (average across P3, Pz, and P4) in the group-averaged waveforms (see Fig. 3). The Stimulus-type × Micro-sequence-length × Electrode (P3, Pz, and P4) ANOVA showed a significant stimulus type main effect: F(1,11) = 8.90, p < 0.05, η2 = 0.45 (caused by the amplitude being higher on change than on repetition trials: 7,94 ± 0.93 μV vs. 7.25 ± 1.02 μV) and a significant electrode main effect: F(2,22) = 20.40, p < 0.001, ɛ = 0.95, η2 = 0.65. Post hoc Tukey HSD-tests showed that the signal at Pz (8.71 ± 1.12 μV) was higher than at the P3 (7.39 ± 0.89 μV) or the P4 (6.67 ± 0.94 μV) lead (at least p < 0.01 for both comparisons) and there was a tendency for the amplitude at P3 being higher than that at P4 (p < 0.1).