آگاهی از ساختار توالی مانع از حواس پرتی شنوایی: مطالعه ERP

Abstract Infrequent, salient stimuli often capture attention despite their task-irrelevancy, and disrupt on-going goal-directed behavior. A number of studies show that presenting cues signaling forthcoming deviants reduces distraction, which may be a “by-product” of cue-processing interference or the result of direct preparatory processes for the forthcoming distracter. In the present study, instead of “bursts” of cue information, information on the temporal structure of the stimulus sequence was provided. Young adults performed a spatial discrimination task where complex tones moving left or right were presented. In the predictable condition, every 7th tone was a pitch-deviant, while in the random condition the position of deviants was random with a probability of 1/7. Whereas the early event-related potential correlates of deviance-processing (N1 and MMN) were unaffected by predictability, P3a amplitude was significantly reduced in the predictable condition, indicating that prevention of distraction was based on the knowledge about the temporal structure of the stimulus sequence.

. Introduction Many tasks in our everyday life require the filtering of task-relevant and task irrelevant sensory events: Task-relevant events have to be processed as fast as possible, while task-irrelevant events should not consume processing resources at all. Such a “perfect” selective attention set, however, cannot be established: Unpredictable, rare stimuli easily capture our attention and disrupt the ongoing task-related behavior, that is, we get distracted. A number of studies show that the sensory system automatically responds to unpredictable, rare stimulus events (for a summary, see Escera et al., 2000), which may lead to involuntary allocation of attention to such events ( Schröger, 1997). Recent studies show that when forthcoming, potentially distracting events are preceded by informative cues, the effects of distraction are reduced or eliminated ( Sussman et al., 2003, Horváth et al., 2011, Horváth and Bendixen, 2012, Wetzel and Schröger, 2007, Wetzel et al., 2009 and Wetzel et al., 2012). The goal of the present study was to investigate whether the prevention of distraction was also possible by providing information on forthcoming distracters without relying on explicit cues. Cognitive processing related to distraction is usually investigated in oddball-paradigms, in which the presentation of frequent standard stimuli is interrupted by infrequent deviants. A variant of the oddball paradigm developed by Schröger and Wolff (1998b) allows unique insights into distraction-related processing. In this paradigm, long and short tones are presented equiprobably, and participants perform a duration discrimination task. Occasionally, randomly, the task-irrelevant tone pitch is changed (in about 10% of the trials). For such deviants, prolonged response times, reduced hit rates and more false alarms were found than for standards. Distraction effects can be found at the electrophysiological level as well: After deviance onset, a characteristic waveform can be observed in the deviant-minus-standard event-related potential (ERP) difference, starting with an enhanced N1 and mismatch negativity (MMN) at 100–250 ms, followed by a positivity at around 300 ms (P3a), and finally a negative deflection occurs peaking around 500 ms (reorienting negativity — RON). The N1-effect and MMN reflect the activity of sensory change detection processes ( Näätanen, 1982). P3a is generally assumed to reflect involuntary attention switching ( Friedman et al., 2001 and Polich, 2007), while RON is theorized to reflect the reorientation of attention to the original task ( Schröger and Wolff, 1998a and Sussman et al., 2003). Similar results were found in auditory–visual paradigms in which targets were visual stimuli (e. g. odd or even numbers) and the distractors were sounds ( Escera et al., 1998, Escera et al., 2000 and Escera et al., 2001). Although the early studies using either auditory ( Berti and Schröger, 2003, Schröger and Wolff, 1998a and Schröger and Wolff, 1998b) or auditory–visual ( Escera et al., 1998, Escera et al., 2000 and Escera et al., 2001) paradigms consistently found prolonged response times (RTs) and decreased accuracy, recent studies found abolished or even reversed behavioral effects ( Li et al., 2013, Parmentier et al., 2010, SanMiguel et al., 2010a, SanMiguel et al., 2010b and Wetzel et al., 2012). These studies suggest that alerting and fore-period effects differ between standards and deviants, and these differences influence the behavioral results. Interestingly, the paradigm can be also utilized to assess whether distraction can be prevented or reduced. Sussman et al. (2003) utilized the paradigm developed by Schröger and Wolff (1998b) but they presented visual cues before each tone. In the predictable condition, cues indicated whether the forthcoming tone was a standard or a pitch-deviant. In the unpredictable condition, the cues did not allow predicting whether the forthcoming tone was a standard or a deviant. In the unpredictable condition, the expected distraction effects were found: (delayed RTs to deviants in comparison to standards, and the elicitation of N1/MMN, P3a, and RON). In the predictable condition, however, the RT-delay, P3a and RON were abolished (predictability had no effect on the N1/MMN). These results were replicated in several studies using different experimental designs and manipulations of presentation (Horváth et al., 2011, Horváth and Bendixen, 2012, Wetzel and Schröger, 2007 and Wetzel et al., 2009). These studies showed that cues providing different degrees of predictability allow the reduction of distraction, but the mechanism behind the cuing effect is not fully understood yet. Although the prevalent interpretation of the cuing effect is that cues allow one to prepare for, and prevent distraction caused by deviants (“preparation”-hypothesis), other interpretations are also possible. The main alternative interpretation is that distraction-prevention is a “byproduct” of cue-processing: Because cues deliver information commensurate to that of the forthcoming deviant (i.e. their presentation frequencies are necessary the same, therefore deviant cues are deviants themselves within the cue sequence), processing this sudden “burst” of information may temporarily deplete processing resources, which in turn, may lead to reduced distraction effects. Direct evidence against the “byproduct”-hypothesis is scarce. There is only one study, conducted by Parmentier and Hebrero (2013), which showed that cues allowing the prediction of forthcoming deviants reduced distraction-related response-time delays even if the cues preceded the deviants by as much as 2250 ms (i.e. the reduction of RT-delay did not differ from that at 250 ms cue-tone separation). Because it seems unlikely that cue-related processing would block further processing for such a long time, this result supports the “preparation” account of the cuing effect. The goal of the present study was to investigate distraction-prevention using the method of ERPs in a setting in which information on forthcoming distracters was not delivered in “bursts”, but was available continuously. Investigating whether distraction can be reduced in this setting is important, because such an arrangement would allow the comparison of distraction-prevention ability between groups potentially differing in their ability to process and utilize “bursts” of information. That is, the continuous availability of cue-information would eliminate confounds due to potential between-group cue-processing abilities. For example, if processing “burst”-like cues required 300 ms on average in one group, but required 500 ms in another, then cues appearing 400 ms before distracters would allow one group to fully prepare for the forthcoming distracters, while leaving the other groups prone to their distracting effects. In this example, one would measure between-group differences in the efficiency in distraction-prevention, but these differences would not reflect the ability to prevent distraction, rather, they would reflect a difference in cue information processing speed. Furthermore, even if the cue-distracter separation allowed both groups to process cue information in time, the utilization of this information depends on the willingness of participants to do so. The amount of effort needed to process cue information in the short time available may reduce the participants' motivation to utilize cue information at all (Horváth, 2013). We administered an auditory distraction paradigm in which the presentation order of tones was either predictable (every 7th tone was pitch-deviant) or random (with 1:6 deviant:standard ratio). The tones virtually moved either to the left or to the right and participants responded to the direction of the movement, ignoring sound frequency. As in previous studies, deviants in the predictable condition should be less distracting than those in the random condition because of the availability of information on forthcoming deviants. This arrangement, however, still provides a challenge: participants have to keep the current position within the sequence in mind to be able to prepare for forthcoming deviants. In order to minimize the effort needed, a visual counter showing the sequence position was presented as a constant reminder, which made information on forthcoming tones continuously available throughout the experimental blocks of the predictable condition. We hypothesized that knowledge about the stimulus sequence would reduce or abolish behavioral and ERP effects of distraction.

. Results 3.1. Behavioral performance Neither the analyses of d′ nor that of response times showed significant effects. The group-mean response time in the predictable condition was 576 ms in standard (standard deviation, SD = 50 ms) and 579 ms in deviant trials (SD = 55 ms), while in the random condition 577 ms was the average speed on standards (SD = 49 ms) and 578 ms on deviants (SD = 59 ms). These response times are referred to the onset of the tones (and not the time point the virtual movement started). Neither the main effect of Condition (F[1,13] < .001, p = .99, MSE < 0.001, η2G < .001), nor the main effect of Stimulus: standard or deviant (F[1,13] = .015, p = 0.7, MSE < .001, η2G < .001) was significant; and the Condition × Stimulus interaction did not show any significant effect either: (F[1,13] = .133, p = .133, MSE < .001, η2G < .001). Regarding sensitivity, the mean of d′-s in the predictable condition was 2.91 for standards (SD = .65) and 2.8 for deviants (SD = .58). In the random condition, the mean of d′-s was 2.98 for standards (SD = .79) and 2.85 for deviants (SD = .73). No significant effect was found (Condition main effect: F[1,13] = .22, p = .65, MSE = .183, η2G = .002, Stimulus main effect: F[1,13] = .72, p = .408, MSE = .287, η2G = .008, Condition × Stimulus interaction: F[1,13] = .011, p = .916, MSE = .034, η2G < .001). 3.2. ERPs After excluding artifact-contaminated epochs, individual ERPs were averaged for 88 deviants in the predictable condition (SD: 13.6); for 65 deviants in the random condition (SD: 13.44); 81.5 standards in the predictable condition (SD: 12.19) and 64.6 standards in the random condition (SD: 14.36). The group-average ERPs elicited at midline electrodes in the two types of trials and conditions, and corresponding deviant-minus-standard waveforms are presented in Fig. 1. Group-average (N=14) ERPs to deviants, and standards preceding them in the ... Fig. 1. Group-average (N = 14) ERPs to deviants, and standards preceding them in the random (left) and the predictable (right) conditions, and the corresponding deviant-minus-standard difference waveforms (middle column) at selected midline and averaged mastoid leads. Figure options The ERP waveforms at FCz showed a negative-going trend before tone onset suggesting preparatory activity for the forthcoming tone. Tones elicited an N1 and a P2, which was followed by a negativity between 200 and 300 ms and a negative sustained activity of duration comparable to that of the tone. For deviants, the second negativity was overlapped by a positive waveform, and the sustained negativity persisted longer than for standards. This suggests that participants probably kept their attention slightly longer on deviants than on standards. The deviant-minus-standard difference waveform in the random condition showed an early negative difference (N1-effect/MMN/N2b) with two negative peaks at 100 ms and 162 ms, and a fronto-central P3a peaking at 286 ms. In parallel with the differential fronto-central negativity resulting from the persistence of the sustained negativity for deviants, the difference waveform also showed a slow positive activity after 500 ms, peaking at 634 ms on the POz lead in the random condition (identifiable as a P3b). The ANOVA of the amplitudes at the first peak of the early negativity showed a significant Stimulus main effect: F[1,13] = 39.766, p < .001, MSE = 1.667, η2G = .102, indicating larger (more negative) N1 (and possibly MMN) amplitudes. Neither the Condition main effect (F[1,13] = 2.14, p = .16, MSE = 1.69, η2G = .006) nor the Stimulus × Condition interaction (F[1,13] = .019, p = .89, MSE = 2.629, η2G < .001) was significant. For the second peak only a marginal Stimulus main effect was found: F[1,13] = 3.75, p = .075, MSE = 6.53, η2G = .034. Neither the main effect of Condition: (F[1,13] = .75, p = .4, MSE = 5.53, η2G = .006), nor the interaction of Stimulus × Condition (F[1,13] = 1.02, p = .32, MSE = 3.048, η2G = .004) reached statistical significance. The ANOVA of the amplitudes in the P3a latency-range showed a significant Stimulus main effect: F[1,13] = 25.05, p < .001, MSE = 22.15, η2G = .35 and a Condition × Stimulus interaction: F[1,13] = 8.20, p = .013, MSE = 2.04, η2G = .016, showing that P3a amplitude was smaller in the predictable than in the random condition. A significant Condition main effect was not found: F[1,13] = .095, p = .76, MSE = 4.52, η2G < .001. The topography of the P3a in the two conditions and the modulatory P3a-effect (the difference of the deviant-minus-standard differences) are presented in Fig. 2. The ANOVA of the P3b activity on POz lead showed significant stimulus effect: F[1,13] = 30.366, p < .001, MSE = 3.837, η2G = .053, indicating that deviants evoked larger positive responses than standards. Neither the main effect of Condition (F[1,13] = .008, p = .92, MSE = 8.813, η2G < .001) nor the Stimulus × Condition interaction was significant (F[1,13] = 1.64 p = .22, MSE = 8.033, η2G = .014). The ANOVA of the negative difference on AFz electrode showed a stimulus main effect: F[1,13] = 4.80, p = .047, MSE = 8.495, η2G = .028, indicating that amplitudes for deviant tones were more negative than for standards. No significance was present regarding the Condition main effect (F[1,13] = .69, p = .42, MSE = 8.654, η2G = .028) and the Stimulus × Condition interaction (F[1,13] = .92, p = .76, MSE = 6.37, η2G < .001). Group-average (N=14) topographical distribution of the P3a in the random (left ... Fig. 2. Group-average (N = 14) topographical distribution of the P3a in the random (left panel) and in the predictable condition (middle panel). The P3a-effect (right panel) is calculated as the between-condition difference of the deviant-minus-standard ERP difference.