جلوگیری از حواس پرتی با سرنخ های احتمالی

Abstract Involuntary attention switches triggered by infrequent, unpredictably occurring sensory events (distraction) can be prevented when participants are made aware of the forthcoming distractor. Previous studies exploring this phenomenon presented visual cues before each stimulus in an auditory oddball sequence. In one condition, cues were completely reliable in predicting the forthcoming distractor or standard sound, in another, separate condition, they were completely unreliable. These studies found that in the condition with reliable cues, distraction was reduced compared to that with unreliable cues, as signaled by decreased reaction time delay as well as reduced P3a and reorienting negativity event-related potentials. Whereas these results are generally interpreted as the results of preparatory processes initiated by the cues, it could be argued that the preventive effect is a byproduct of increased information processing load in the condition with informative cues compared to that in the condition with uninformative ones. In the present study, using 80% reliable visual cues preceding tones in an oddball sequence, it was demonstrated that distraction can be prevented when the trials with valid and invalid cues were presented within a single experimental condition, as shown by reduced reaction time delay and P3a amplitude. These results are compatible with the notion that the distraction is prevented by means of preparatory processes initiated by the cues.

1. Introduction Many tasks in everyday life demand the maintenance of a selective attention set: to perform efficiently, we have to monitor task-relevant sources of information while disregarding others. Our efforts, however, are often unsuccessful: unpredictably occurring, rare sensory events capture our attention despite being task-irrelevant; in other words, we get distracted. Under normal circumstances, processes leading to distraction and those supporting the maintenance of a focused attention set are well-balanced: we can perform the task at hand without many interruptions, but occasional episodes of distraction allow us to re-evaluate our goals and priorities (e.g. the distraction caused by the fire-alarm allows us to change our behavior adaptively). Recent studies show that this balance can be dynamically adjusted, that is, we can prevent or counteract distraction when we are made aware of forthcoming, potentially distracting events. A cue preceding such an event allows one to reduce distraction as measured by behavioral and event-related potential (ERP) indices ( Sussman et al., 2003, Wetzel and Schröger, 2007, Wetzel et al., 2009 and Horváth et al., 2011). The goal of the present study was to investigate the mechanism behind this cueing effect. Distraction-related processing is usually investigated in studies presenting oddball stimulus sequences to participants. In such sequences, most stimuli (termed standards) conform to a regularity, which is occasionally violated by unpredictably occurring stimuli (deviants). A paradigm using such an oddball sequence introduced by Schröger and Wolff (1998a) proved to be highly useful for investigating distraction. In the prototypical paradigm, a sequence of short and long tones is presented with 50–50% probability, and participants perform a two-alternative forced choice duration discrimination task. Occasionally (typically with 5–20% probability), a task-irrelevant feature of the given tone (e.g. its frequency) is changed (deviants). Because participants perform the same duration discrimination task for deviants as for standards, it is assumed that differential responses to deviants and standards reflect processes solely related to distraction. For deviants, reaction times are typically delayed, error rates may increase, and a characteristic sequence of ERPs can be observed ( Escera and Corral, 2007): deviants elicit an enhanced N1 (around 100 ms post stimulus onset; Näätänen and Picton, 1987) and the mismatch negativity (MMN, peaking between 100 and 250 ms post deviance onset; see Näätänen et al., 1978; for recent summaries see Kujala et al., 2007 and Winkler, 2007). These are followed by the P3a or novelty-P3 (peaking around 300 ms after deviance onset; Friedman et al., 2001 and Polich, 2007), which is usually interpreted as the correlate of attention switching ( Escera et al., 2000; but see Horváth et al., 2008). Finally, because the deviant information is irrelevant for the participant in this experimental setting, the task-optimal attention set has to be restored, which is assumed to be reflected by the reorienting negativity (RON, peaking between 400 and 600 ms after deviance onset; Schröger and Wolff, 1998b). Sussman et al. (2003) modified the auditory distraction paradigm summarized above by presenting visual cues before each tone. In one condition, the cues were informative regarding the task-irrelevant dimension of the forthcoming tone (i.e. its frequency), in the other they were not. Informative cues allowed participants to reduce the effects of distraction, as signaled by decreased reaction time delay as well as reduced P3a and RON amplitudes compared to the condition with uninformative cues (see also Wetzel and Schröger, 2007, Wetzel et al., 2009 and Horváth et al., 2011). This result was interpreted as the reflection of preparatory activity for the forthcoming distractor. In the following, we offer an alternative explanation for these results, which is based on the fact that the studies cited above manipulated cue reliability in separate conditions. In one condition, the cue was completely reliable, that is, the correlation between the appearance of the cue and the distractor was 100%; in the other condition, the cue was completely unreliable, that is, the correlation was 0. Since participants knew about the reliability of the cues within each experimental block, one could argue that in each condition (and each block) participants adopted different strategies for processing the cues: When the cue was completely unreliable, participants probably refrained from making an effort to “figure out” the meaning of the cue, whereas they were engaged in processing the meaning of the cue when the cues were reliable. These hypothetical cue-evaluation strategies would result in different information processing loads, which may lead to the observed distraction-preventive effects: Distractors may be less efficient when cues are completely reliable because information processing resources are engaged by cue processing. That is, the distraction-preventive effect may not be due to direct preparatory activities, but rather, it may be an indirect (but useful) byproduct of a difference in information processing load. That a higher information processing load may result in lower distractibility as indexed by the P3a and RON amplitudes has been shown in the context of the prototypical distraction paradigm by Berti and Schröger (2003). That information load may be different between the two conditions is supported by previous studies: in the study by Wetzel and Schröger (2007) infrequent visual cues did not elicit a P3b at all in the uninformative condition. Similarly, in the Sussman et al. (2003) and Horváth et al. (2011) studies, P3b amplitudes elicited by infrequent visual cues were higher when cues were informative than when they were not. Since the P3b amplitude correlates with the task-relevancy of a stimulus (see e.g. Donchin et al., 1978), these results indicate that obviously uninformative cues may simply be disregarded by the participants. The goal of the present study was to investigate whether distraction could be prevented by cueing when the adoption of different cue-processing strategies was not feasible. To this end, we used a single setting with 80% reliable cues instead of two experimental conditions with completely reliable in one and unreliable cues in the other. Using such probabilistic cues has proven to be a robust experimental manipulation of the allocation of attention (see, e.g. Posner, 1980 and Mondor and Bregman, 1994), therefore, it seems to be safe to assume that cues with 80% reliability would still encourage participants to utilize this information, and it would allow the presentation of standard and potentially distracting events preceded by valid and invalid cues within the same experimental setting.

. Results 3.1. Behavioral performance Group-average reaction times and sensitivity scores (d′) to tone duration are presented in Fig. 1. The ANOVA of the reaction times showed a stimulus main effect: F(1,11) = 21.99, p < .001, a cue main effect: F(1,11) = 40.98, p < .001; and a Stimulus × Cue interaction: F(1,11) = 13.26, p < .01. That is, responses to deviants were slower than those to standards; also responses to tones preceded by invalid cues were slower than those preceded by valid ones; moreover, the reaction time delay for deviants compared to standards was longer for trials with invalid than those with valid cues. The ANOVA of the d′ scores showed a significant stimulus main effect only: F(1,11) = 25.67, p < .001, showing a lower sensitivity to duration in deviant than in standard trials. Group-average reaction times and d′ values (with standard errors of means). Fig. 1. Group-average reaction times and d′ values (with standard errors of means). Figure options 3.2. ERPs After the rejection of single-sweep epochs potentially contaminated by artifacts as described above, individual ERPs for the deviants preceded by an invalid cue were averaged from 60 trials on average (standard deviance, SD: 15); for the deviants preceded by valid cues this was 242 (SD: 61); for the standards preceded by an invalid cue 396 (SD: 93), and for the standards preceded by valid cues it was 1559 (SD: 385). The group-average ERPs elicited in the four types of trials and corresponding deviant-minus-standard difference waveforms are presented in Fig. 2. The visual cues elicited two positivities peaking occipitally at 135 and 233 ms following cue onset (− 211 and − 113 ms with respect to tone onset, respectively) and a fronto-central negativity at 151 ms following cue onset (− 195 ms with respect to tone onset). The ERP curves elicited by the two (deviant vs. standard) cue types start to bifurcate at around 150 ms after cue onset (− 194 ms with respect to tone onset), with the ERP curve corresponding to the standard cues displaced in the negative direction. As in previous studies (Sussman et al., 2003 and Horváth et al., 2011) this bifurcation disappears before the tone-elicited P3a, but may overlap the tone-elicited N1 and N2 (see below). Group-average event-related potentials and deviant-minus-standard difference ... Fig. 2. Group-average event-related potentials and deviant-minus-standard difference waveforms at AFz, FCz, CPz, POz, and the average mastoid signal (CM). The onset of the visual cue is at the gray vertical line, the onset of the tone is at the crossing of the axes. Figure options Tones elicited an N1 (peaking at 99 ms with respect to tone onset). For standard tones, this was followed by a second negativity between 200 and 300 ms, but in the same interval a positivity (P3a) was present for deviants. This was followed by a parietal positivity between 400 and 600 ms (identified as a P3b). The difference waveforms showed an N1-effect/MMN (peaking at 99 ms post tone onset — not analyzed), an N2b (194 ms post tone onset — not analyzed), a P3a (283 ms post tone onset) and a fronto-central negativity identified as RON between 366 and 466 ms. The ANOVA of the P3a amplitudes at FCz showed a stimulus main effect: F(1,11) = 39.24, p < .001; a cue main effect: F(1,11) = 11.81, p < .01, and a Stimulus × Cue interaction: F(1,11) = 12.16, p < .01. Follow-up Student's t-tests showed that the deviants preceded by an invalid cue elicited more positive amplitudes than those preceded by valid cues (t[11] = 3.70, p < .01), whereas there was no significant difference between validly and invalidly cued standards. Fig. 3 (left panel) shows the topographical distribution of the cueing-related P3a-effect measured as the difference between ERPs to deviants preceded by an invalid cue and deviants preceded by a valid cue. To ascertain that the topography of the effect corresponded to that reported in the literature, an ANOVA over the amplitudes measured at the midline with Cue and Anterior–Posterior (AP) factors was conducted. This analysis showed a significant Cue × AP interaction: F(3,33) = 3.71, ε = .58; p < .05. The interaction was followed-up by pair-wise Student's t-tests, which showed that the amplitude at FCz was significantly higher than at AFz or POz, and at CPz being higher than at POz (all t values > 2.24; p values < .05). The Cue × AP × Laterality ANOVA over the amplitudes measured at the 3- and 4-lines showed no significant interactions involving both Cue and Laterality factors. Group-average topographical distributions of the cue-related ERP effects in the ... Fig. 3. Group-average topographical distributions of the cue-related ERP effects in the P3a (left panel) and the RON (middle and right panels) time ranges. The P3a effect (left) is calculated as the difference of the ERPs elicited by deviants preceded by invalid and valid cues. The RON effects are calculated as the difference of the ERPs elicited by standards preceded by invalid and valid cues (middle); and as the difference of the deviants and standards preceded by valid cues (right). For the P3a (left), isopotential lines are separated by 0.5 μV. For RON (middle and right), they are separated by 0.25 μV. Figure options The ANOVA of the amplitudes in the RON latency-range at FCz showed a cue main effect: F(1,11) = 6.15, p < .05; and a Stimulus × Cue interaction: F(1,11) = 6.01, p < .05. Follow-up tests showed that standards preceded by invalid cues elicited a more negative signal than those with valid cues (t[11] = 4.72, p < .001), while there was no significant difference between deviants. Fig. 3 (middle and right panels) shows the topographical distributions of the cueing- and stimulus-related RON-effects measured as the difference between ERPs to standards preceded by an invalid cue and standards preceded by a valid cue; and the difference between ERPs to deviants preceded by a valid cue and standards preceded by a valid cue, respectively. To ascertain that the topography of the cue-related effect corresponded to that reported for RON in the literature, an ANOVA over the amplitudes measured at the midline with Cue and Anterior–Posterior (AP) factors was conducted. This analysis showed a significant Cue × AP interaction: F(3,33) = 13.42, ε = .49; p < .001. The interaction was followed-up by pair-wise Student's t-tests, which showed that the difference amplitude at FCz was significantly higher (more negative) than at any of the other electrodes, there was no significant difference between AFz and CPz, while difference amplitudes measured at both AFz and CPz were higher (more negative) than that at POz (all t values > 2.40; p values < .05). The Cue × AP × Laterality ANOVA over the amplitudes measured at the 3- and 4-lines showed a Cue × Laterality interaction: F(1,11) = 6.04, p < .05, indicating higher (more negative) difference amplitudes on the left side. For the deviance-related RON effect, the ANOVA over the amplitudes measured at the midline with Stimulus and Anterior–Posterior (AP) factors showed a significant Stimulus × AP interaction: F(3,33) = 7.16, ε = .70; p < .01. The interaction was followed-up by pair-wise Student's t-tests, which showed that the difference amplitude at FCz was significantly higher (more negative) than at any of the other electrodes, and it was higher (more negative) at CPz, than at POz (all t values > 2.40; p values < .05). The Stimulus × AP × Laterality ANOVA over the amplitudes measured at the 3- and 4-lines showed no significant interactions including both Stimulus and Laterality factors.