Abstract

Exposure to rhythmic stimulation results in facilitated responses to events that appear in-phase with the rhythm and modulation of anticipatory and target-evoked brain activity, presumably reflecting “exogenous,” unintentional temporal expectations. However, the extent to which this effect is independent from intentional processes is not clear. In two EEG experiments, we isolated the unintentional component of this effect from high-level, intentional factors. Visual targets were presented either in-phase or out-of-phase with regularly flickering colored stimuli. In different blocks, the rhythm could be predictive (i.e., high probability for in-phase target) or not, and the color could be predictive (i.e., validly cue the interval to the target) or not. Exposure to nonpredictive rhythms resulted in faster responses for in-phase targets, even when the color predicted specific out-of-phase target times. Also, the contingent negative variation, an EEG component reflecting temporal anticipation, followed the interval of the nonpredictive rhythm and not that of the predictive color. Thus, rhythmic stimulation unintentionally induced expectations, even when this was detrimental. Intentional usage of predictive rhythms to form expectations resulted in a stronger behavioral effect, and only predictive cues modulated the latency of the target-evoked P3, presumably reflecting stimulus evaluation. These findings establish the existence of unintentional temporal expectations in rhythmic contexts, dissociate them from intentional expectations, and highlight the need to distinguish between the source of expectation (exogenous–endogenous) and the level of voluntary control involved in it (unintentional–intentional).

INTRODUCTION

The contingencies between events in the world are often not random, allowing the cognitive system to predict properties of future events and facilitate or inhibit their processing (Correa, Rao, & Nobre, 2008; Sterzer, Frith, & Petrovic, 2008; Posner, Snyder, & Davidson, 1980). One such property of upcoming events is the time at which they occur. Several lines of research found facilitated motor behavior (Sanabria, Capizzi, & Correa, 2011; Praamstra, Kourtis, Kwok, & Oostenveld, 2006) and improved discrimination ability (Rohenkohl, Carvo, Wyart, & Nobre, 2012) in response to temporally anticipated events. Applying the taxonomy from spatial orienting (Yantis & Jonides, 1990; Posner et al., 1980), it has been suggested that temporal expectations can be created exogenously, implying a low-level, data-driven, unintentional process (in contrast to endogenous, high-level, top–down, intentional process; Coull & Nobre, 2008; Nobre, Correa, & Coull, 2007; see Shiffrin & Schneider, 1977). The current study was designed to establish the existence of unintentional exogenous expectations and differentiate them from intentional ones.

Temporal expectations are created exogenously when the input dynamics have a nonrandom temporal structure (e.g., music, speech or motion, especially biological; Jones, 2010; Nobre et al., 2007). The simplest case is when the stimuli sequence has a fixed interonset interval (IOI). Evidence for formation of expectations comes from findings that detection of threshold stimuli (Mathewson, Fabiani, Gratton, Beck, & Lleras, 2010), perceptual discriminations (Rohenkohl et al., 2012), and choice RT performance (Martin et al., 2005) were improved for targets that appeared at the regular IOI (in-phase) compared with when the target appeared early or late (out-of-phase; Jones, Moynihan-Johnston, & Puente, 2006; Jones, Moynihan, MacKenzie, & Puente, 2002) or in a non rhythmic condition (Rohenkohl et al., 2012).

In addition to behavioral evidence, electrophysiological measurements of brain activity revealed several neural correlates of temporal expectations created in rhythmic contexts. One phenomenon is the contingent negative variation (CNV), a widespread slow increase in scalp negativity, which is observed when expecting an impending stimulus (Walter, Cooper, Aldridge, McCallum, & Winter, 1964). It has been shown that the time course of the CNV reflects the fixed interval of the rhythm such that negativity increases faster as the interval becomes shorter (Praamstra et al., 2006; see also Pfeuty, Ragot, & Pouthas, 2005; Macar & Vidal, 2004, for a similar phenomenon in nonrhythmic context). Other studies found reduced amplitude (desynchronization) of alpha-band activity in occipital electrodes just before the expected occurrence of a stimulus in a rhythmic sequence, possibly reflecting bias in cortical excitability according to temporal expectations (Rohenkohl & Nobre, 2011; Praamstra et al., 2006). In posttarget activity, it was found that when targets appeared in-phase with a predictive rhythmic stimulation, there was modulation of several ERPs related to both early sensory processing (attenuation of N1 component) and later activity related to stimulus evaluation (larger N2 and earlier P3; Correa & Nobre, 2008; Doherty, Rao, Mesulam, & Nobre, 2005).

Exogenous temporal expectations are conceived as nonintentional because the expected interval is already encoded implicitly in the temporal pattern of activity in sensory neurons, making it unnecessary to represent the interval explicitly (Buhusi & Meck, 2005). Instead, it has been suggested that temporal expectations are generated because of entrainment of internal oscillatory processes to the rhythmic stimulation. As a result, when upcoming stimuli appear in phase with the entrained oscillation, their processing is facilitated (Dynamic Attending Theory; Large & Jones, 1999; Jones & Boltz, 1989). Recent electrophysiological studies support entrainment models by showing the existence of periodicity in sensitivity of perceptual processes (Busch & VanRullen, 2010; Busch, Dubois, & VanRullen, 2009; Mathewson, Gratton, Fabiani, Beck, & Ro, 2009) as well as for entrainment of neuronal activity to external rhythmic stimulation (Besle et al., 2011; Lakatos, Karmos, Mehta, Ulbert, & Schroeder, 2008; Martin, Houck, Kicic, & Tesche, 2008).

However, whether temporal expectations are indeed created unintentionally and passively, whenever one is exposed to rhythmic stimulation, has not been clearly established. That is, the benefit gained from periodicity could depend on additional involvement of high-level, intentional mechanisms. For example, it is possible that, although entrainment of low-level processes occurs nonintentionally, entrainment of higher systems (e.g., attention; Schroeder & Lakatos, 2009) is what constitutes expectations and occurs only intentionally. Moreover, because even without rhythmic stimulation it possible to form expectations based on memorized intervals between a preparatory warning signal (WS) and an imperative signal (Bevan, Hardesty, & Avant, 1965; see Niemi & Näätänen, 1981) and their application is intentional (endogenous temporal orienting; Coull & Nobre, 2008; Correa, Lupiañez, & Tudela, 2006), it is possible that in rhythmic context expectations are also formed by intentionally extracting and rehearsing the fixed interval (Drake & Botte, 1993; Keele, Nicoletti, Ivry, & Pokorny, 1989). Indeed, similar EEG modulation of CNV and target-evoked potentials were found in temporal cuing paradigms with no rhythmic context (Griffin, Miniussi, & Nobre, 2002; Miniussi, Wilding, Coull, & Nobre, 1999).

The question of whether the effect of rhythmic stimulation is indeed automatic lingers because most previous studies that examined the effect of rhythms were not designed to address the question of automaticity and thus used rhythms that predicted the target time. In some studies, there was high probability that the target would appear in-phase with the rhythm (e.g., Praamstra et al., 2006), whereas in others whenever the sequence was rhythmic (comparing to nonrhythmic) the target appeared in-phase with it (Miller, Carlson, & McAuley, 2012 [Experiment 2]; Rohenkohl et al., 2012; Doherty et al., 2005). In both cases, it is advantageous for the participant to make use of the rhythm, even if the explicit instructions did not require it. Thus, it is hard to attribute the effects observed for in-phase stimuli to purely automatic formation of expectation. To establish the automaticity of rhythmic expectations, it must be ascertained that high-level processes or intention are not involved in their formation.

In an attempt to reduce the motivation of the participants to use the rhythm, in some studies the rhythm was made nonpredictive by presenting the target at a random IOI after the rhythm (e.g., Mathewson et al., 2010; Jones et al., 2002, 2006). Although this design is more suitable to test automaticity, it still has two shortcomings. First, the distribution of target times in such paradigms is usually symmetric around the in-phase time, which might cause subjective time estimations to regress to the mean (Jazayeri & Shadlen, 2010), making the in-phase target seem more probable again. This caveat was recently addressed by Sanabria and colleagues (2011; Experiment 3), who found that even when the IOIs were random and their distribution not symmetric around the in-phase time, there was relative facilitation for targets that appeared in-phase with the preceding rhythm. Second, even in the latter study (as in the other nonpredictive paradigms mentioned above), it cannot be ruled out that the participants were intentionally creating expectation based on the rhythm, as it provided the only temporal reference frame for the onset of the next stimulus, even if it was nonpredictive. Furthermore, participants may choose to adhere to the rhythmic sequence because it may easier to cling to some expectation than having none at all. Stronger support for nonintentional effects of the rhythmic stimulation can be gained by controlling participants' incentive to use the rhythm.

In the current study, two experiments assessed the ability of rhythmic visual input to create temporal expectations automatically and dissociated them from high-level, intentional expectations. In the main condition, the color of a nonpredictive rhythm was used to form a primary high-level temporal expectation, which was either congruent or incongruent with the interval of the rhythm. If rhythmic context creates temporal expectations unintentionally, in a stimulus-driven manner, there should be facilitations for in-phase targets regardless of the color-based intentional expectations. In another condition, neither the color nor the rhythm were predictive (cf. Sanabria et al., 2011, in the auditory modality), allowing us to compare the effect of nonpredictive rhythm in the main condition to that obtained without concurrent color cuing. In a third condition, the rhythm was predictive and was intentionally used to create expectations, allowing us to compare such expectations to those observed when the rhythm was nonpredictive. Our paradigm is similar to that used by Rohenkohl, Coull, and Nobre (2011), but note that in the present experiment, both predictive and nonpredictive sequences were rhythmic, giving no incentive for the participant to use the rhythms. Finally, we also measured EEG and examined whether the pattern of modulation of brain activity would differ when expectations are created intentionally or unintentionally.

METHODS

Participants

Twenty-nine students (Experiment 1: 11 women, 4 men, mean age = 24.1 years, 11 right-handed; Experiment 2: 8 women, 6 men, mean age = 24.7 years, 11 right-handed) from the Hebrew University of Jerusalem participated in a 2-hr experiment, in return for course credit or monetary compensation. All participants reported normal or corrected-to-normal vision and normal color vision. Participants did not have professional music training and did not play an instrument during the 3 years before the experiment. The study was approved by the institutional ethics committee, and all participants provided written consent.

Stimuli and Task

Experimental stimuli were filled color circles (diameter = 1.2°), with a duration of 100 msec. The stimuli were presented at the center of a 17-in. CRT screen (ViewSonic G75f, with 100 Hz refresh rate), on gray background. Each trial began with a fixation point (black “+” symbol, 0.6 visual angles) for 500 msec, followed by an isochronous sequence of four, five, or six stimuli (uniform probability) with a within-trial IOI of 700 or 1300 msec (fast/slow rhythm, respectively; Figure 1). Within a trial, the stimuli in the rhythmic sequence were all red (50% of trials) or all green (50%), except for the final stimulus, which was white and served as a WS. We used a different number of rhythmic stimuli to prevent the WS from being fully predictable, thus reducing anticipatory brain activity in the pre-WS period, which was used as baseline (see below). In most trials, the WS was followed by a target, a black circle. The target appeared with an IOI of either 700 msec (“short IOI”) or 1300 msec (“long IOI”) from the WS or jittered around these IOIs (between 500 and 1500 msec; see design below). The task was to respond as quickly as possible to the onset of the black circle, which remained on screen until a response was made. In “catch” trials, there was no target. Instead, a blank screen was displayed after the WS for 1500 msec, followed by the next trial. We used catch trials to prevent expectations to be created based on conditional probability that the stimulus will occur given that it has not occurred yet (Correa et al., 2006).

Figure 1. 

Trial sequence and design of the predictive blocks. Trials consisted of a sequence of flickering colored stimuli (filled circles), followed by a white WS and a black target, to which participants had to respond. Each row represents a possible trial sequence, and each colored line represents a stimulus within a trial. The predictivity of each cue was manipulated through the probabilities of the eight trial sequences. In Color-predictive blocks, Sequences 1, 2, 7, and 8 had a 75% probability, whereas the other sequences had a 12.5% probability, and the color predictive of target timing (green = short interval to target, red = long interval to target). In Rhythm-Predictive blocks, Sequences 1, 3, 6, and 8 had a 75% probability, and the other sequences had a 12.5% probability, making the rhythm predictive of target timing (high probability that the target appears in-phase with the rhythm). The table details the validity of each trial sequence when using color or rhythm to predict target timing.

Figure 1. 

Trial sequence and design of the predictive blocks. Trials consisted of a sequence of flickering colored stimuli (filled circles), followed by a white WS and a black target, to which participants had to respond. Each row represents a possible trial sequence, and each colored line represents a stimulus within a trial. The predictivity of each cue was manipulated through the probabilities of the eight trial sequences. In Color-predictive blocks, Sequences 1, 2, 7, and 8 had a 75% probability, whereas the other sequences had a 12.5% probability, and the color predictive of target timing (green = short interval to target, red = long interval to target). In Rhythm-Predictive blocks, Sequences 1, 3, 6, and 8 had a 75% probability, and the other sequences had a 12.5% probability, making the rhythm predictive of target timing (high probability that the target appears in-phase with the rhythm). The table details the validity of each trial sequence when using color or rhythm to predict target timing.

Design

The experiment consisted of three conditions, presented in separate blocks (Figure 1). In the “Color-Predictive” condition, the color of sequence stimuli predicted the target IOI (green = short IOI, red = long IOI), whereas in the “Rhythm-Predictive” condition, the IOI of the sequence predicted the target IOI (i.e., target IOI was identical to rhythm IOI). In both conditions, the target appeared at the predicted time in 75% of the trials (valid trials), at the other IOI in 12.5% of the trials (invalid trials), and not at all in 12.5% of the trials (catch trials). The rhythm in Color-Predictive condition and the color in Rhythm-Predictive condition were nonpredictive: They were valid in 50% of the trials and invalid in 50%, orthogonally to the validity of the predictive cue and to the target IOI. In the “Noninformative” condition, half of the targets were assigned to short IOI and half to long, independent of the rhythm IOI and of the color, making both cues nonpredictive. However, in this situation, the target would appear in-phase with the rhythm in 50% of the trials, which could cause participants to use the rhythm intentionally to create expectations, having no other source of prediction. To prevent this, the actual target IOI was jittered around the short/long target IOI (±200 msec in 50 msec steps, uniform distribution), resulting in target IOIs of 500–1500 msec. In this condition, 25% of the trials were catch trials.

Procedure

The experiment was conducted in a sound attenuated chamber (Eckel C-26, UK). Participants were seated in a comfortable chair at a viewing distance of 90 cm from the screen. Stimulus presentation and response acquisition were handled using Psychophysics toolbox (Brainard, 1997; Pelli, 1997) for MATLAB (version 7.5.0, Mathworks, Natick, MA). Participants conducted 14 blocks of 32 trials each. The two Noninformative blocks were conducted first to prevent carryover effects from the predictive blocks, followed by six Color-Predictive and six Rhythm-Predictive blocks, alternating between conditions (the starting condition was counterbalanced across participants). At the beginning of each block, participants were instructed about the task, WS, and catch trials and also about the predictive information (if available). Participants were encouraged to use the information to be prepared for target timing, but to respond only when the target actually appeared (and refrain from anticipatory responses). At the beginning of each block, participants performed a practice session of at least eight trials on the first time a condition was performed and at least four trials otherwise. Practice continued until participants could correctly identify the target IOI to be short or long and explain the contingency of the current block. In addition, at the beginning of each trial in the two predictive conditions, the respective instructions “use color” or “use rhythm” were presented for 1 sec. No instructions were presented in the Noninformative condition. Participants provided manual responses using a custom-made response box.1 Short breaks were given between blocks. At the end of the experiment, participants were debriefed, verifying that they had used the predictive cues to create expectations.

Behavioral Analysis

Within each participant, trials were discarded if the RT was larger than three standard deviations from the mean RT, separately for each condition and validity of the predictive cue (in the Noninformative condition separately for rhythm valid and invalid) or if the RT was smaller than 50 msec, presumably reflecting anticipatory response to the expected timing of the target. Within each predictive condition, the effects of target IOI (short/long), rhythm validity (valid/invalid), and color validity (valid/invalid) on RT were analyzed using a repeated-measures ANOVA. Because of the small number of trials at the exact in-phase IOI in the Noninformative condition, trials with IOI of up to 100 msec around the short- and long-target IOIs were used as representing these IOIs. The choice of this interval did not affect the final results. Results in this condition were collapsed across the color factor because no instructions about color were yet given (this condition was always run first), and the color was not predictive. For the comparison of the effect of rhythmic expectations in the three conditions, we calculated the rhythm validity effect (invalid − valid) in each condition across the target IOI and color factors and analyzed the effect of condition on these scores using a one-way repeated-measures ANOVA. Violations of the sphericity assumption were corrected using the Greenhouse–Geisser methods. The corrected p value and the uncorrected degrees of freedom are reported.

To examine the extent to which the facilitation effects of rhythm in the three conditions reflect a common process of engagement with rhythmic input, the rhythm validity effect scores were modeled using factor analysis (using the principle component method). The validity effect of the color in the Color-Predictive condition was also included in the model as an estimate of nonrhythmic high-level temporal expectations. Because of the small number of participants, a common factor was considered reliable only if its eigenvalue was larger than 95% of the eigenvalues that would be obtained by chance. The distribution of eigenvalues under chance was estimated using a permutation analysis, with 10,000 iterations. In each iteration, the validity effect scores of all participants were permuted, separately within each condition. On this permuted data, the same factor analysis model was computed, and the maximal eigenvalue was taken (Good, 2005). To further improve the stability of the model, results of participants with extreme scores (outliers) were eliminated from the analysis. Outliers were identified if they had extreme leverage in one of the validity effects. The cutoff point was set at 3(k − 1)/n, where k represents the number of predictors and n represents the number of data points (as suggested by Cohen, Cohen, West, & Aiken, 2003, pp. 394–398). We used k = 3, assuming a multiple regression structure in which three validity effects predict the fourth.

EEG Recording and Preprocessing

EEG was recorded continuously from 66 preamplified Ag/AgCl electrodes using an Active 2 system (BioSemi, Amsterdam, The Netherlands). Sixty-four electrodes were mounted on a plastic cap positioned according to the international 10–20 system, and two were positioned over the left and right mastoids. Four additional electrodes recorded EOG activity from the outer canthi of the right and left eyes (horizontal EOG) and from above and below the center of the right eye (vertical EOG). The EEG and EOG signal was amplified and digitized at a sampling rate of 1024 Hz with an on-line antialiasing 208 Hz low-pass filter.

Analysis of EEG was conducted off-line using BrainVision Analyzer 2.0 (Brain Products, Munich, Germany) and MATLAB. The unsegmented data were referenced to the nose and digitally filtered between 0.1 and 30 Hz using a zero-shift 24 dB/octave Butterworth filter. Vertical and horizontal bipolar channels were calculated as the difference between vertical EOG channels and the difference between horizontal EOG channels, respectively. Blinks and eye movement artifacts were removed using independent component analysis method (Jung et al., 2000) based on typical scalp topography and time course that resembled either the vertical or the horizontal bipolar channels. Other artifacts were identified when absolute activity was larger than 100 μV or when a change of more than 100 μV was observed in a 200-msec interval and rejected from analysis.

ERP Analysis

Analysis of ERPs was conducted separately for pretarget expectation-related activity and posttarget responses. Analysis of pretarget expectation-related activity focused on the CNV. Artifact-clean data were segmented into epochs extending from 200 msec before to 800 msec after the WS. These epochs were referenced to the mean of a period of 100 msec before the WS. The CNV was analyzed in a cluster of central electrodes around Cz (FC1, FCz, FC2, C1, Cz, C2, CP1, CPz, CP2), referenced to the average of the left and right mastoids and at a time window of 350–650 msec after the WS. This window was chosen as it starts later than the visual-evoked potential elicited by the WS and ends before the evoked potential elicited by the target at the shorter IOI of 700 msec. On the basis of previous findings (e.g., Praamstra et al., 2006), the CNV in this window should be more negative when expecting a target at the short versus long IOI. The epochs were averaged separately for each type of predictive cue (color/rhythm), the target IOI signaled by the rhythm (short/long) and the target IOI signaled by the color (short/long). The effects of the latter two factors on CNV amplitude were analyzed separately within each condition using a repeated-measures ANOVA with factors color interval and rhythm interval.

For analysis of posttarget responses, data were segmented into epochs extending from 200 msec before to 700 msec after the target onset. These epochs were referenced to a period of 100 msec pretarget. The P1 and N1 components were analyzed at a cluster of occipital electrodes (O1, PO3, PO7, O2, PO4, PO8), referenced to the nose. The P3 was analyzed at a cluster of midline-parietal electrodes (Pz, P1, P2), referenced to mastoids. The time window for each of these potentials was defined based on unbiased visual inspection of average activity across all participants and conditions (P1: 110–160 msec; N1: 160–210 msec; P3: 200–550 msec). To prevent bias in estimation of the P1 and N1 amplitude because of extended effect of pretarget processes and to the overlap of posttarget components like the P3, the P1 amplitude was defined peak-to-peak between the peak voltage in the predefined P1 window and the preceding minimum and the N1 amplitude was defined as the peak-to-peak difference between the minimum in the N1 window and the preceding maximum. The peak amplitude of the P3 was defined as the average activity in a symmetrical window of 30 msec around the maximal positive amplitude within the predefined interval. As the P3 has a broad peak, the P3 latency was estimated by averaging the latencies in which the amplitude reached 30%, 50%, and 70% of the maximum activity, allowing a stable estimation of the half-maximum point (Kiesel, Miller, Jolicoeur, & Brisson, 2008). For all analyzed components, reliable estimation of peaks and latencies were achieved using the Jackknifing method (Miller, Patterson, & Ulrich, 1998), with appropriate correction for statistical tests (Ulrich & Miller, 2001).

We examined whether the visual P1, N1, and P3 were affected by whether the target appeared at the expected time or not (validity effect). Epochs were averaged separately for each type of predictive cue (color/rhythm), target IOI2 (short/long), and three levels of predictability: both predictive and nonpredictive cues valid, predictive cue valid but nonpredictive invalid, predictive cue invalid. Because of the small number of trials where the predictive cues were invalid, these trials were not further divided according to the validity of the nonpredictive cues. The amplitudes and latencies of posttarget ERPs were analyzed using two planned subanalyses, which separately examined (1) the validity effect of predictive cues, collapsing across the validity of the nonpredictive cue, and (2) the validity effect of nonpredictive cues, but only for trials where predictive cues were valid. Note that these two analyses constitute two independent contrasts on the validity type factor of the full model; thus, the effect of predictive and nonpredictive cues is not directly comparable.

Time–Frequency Analysis

Time–frequency decompositions were calculated using a complex Morlet wavelet transform, with central frequencies of 1–30 Hz in 1-Hz steps (Tallon-Baudry & Bertrand, 1999). We used a constant ratio of 8 between the central frequency and the standard deviation of the Gaussian-shaped wavelet in the frequency domain. This procedure was applied to single-trial segments starting with the rhythmic stimulus that preceded the WS (i.e., 700 msec before WS for short rhythm and 1300 msec for long) and ending 1600 msec after the WS (300 msec after the long IOI target). This long window was used to compare the activity that occurred before and after the WS. Additional margins of 400 msec in the beginning and end of each segment were added to absorb edge artifacts but were then discarded from analysis. The resulting time–frequency representations were averaged within each condition and baseline-corrected using a range of 0–100 msec after the WS. We did not use the prewarning period as baseline as the interval preceding the WS contained activity elicited by the previous stimulus, which was time-locked to the WS, and, moreover, this residual activity differed between short and long IOI. Instead, we reasoned that the occurrence of the WS is the only time point that is identical both in anticipation and in perceptual input between the short and long IOI conditions.

On the basis of previous studies (Rohenkohl & Nobre, 2011; Praamstra et al., 2006), we focused on alpha activity (8–12 Hz) expressed in occipital sites (PO3/PO4, PO7/PO8, O1/O2), in two time bins: one just preceding and one at the time of the anticipated short IOI target (600–700 and 700–800 msec after WS, respectively). We limited the analysis to trials in which the targets actually appeared in the long IOI. This way, we could measure the power modulation around the time a stimulus was expected or not expected, avoiding the overlap with smeared response to the target itself (note that time-resolved spectral analysis acts like a filter in which the onset of an apparent response antedates the actual onset). Directional a priori comparisons testing for reduced power when expecting a target were conducted using one-tailed statistical tests, addressing the effect of predictive and nonpredictive rhythms.

Experiment 2

In Experiment 1, because blocks of predictive and nonpredictive rhythm alternated, an effect of the nonpredictive rhythms could reflect carryover from the blocks in which the rhythm was predictive because of task set maintenance or confusion. To address this, a different set of participants performed Experiment 2, which included only the conditions in which the rhythm was nonpredictive. The stimuli, task, and procedure were otherwise identical to Experiment 1. Participants conducted two blocks of the Noninformative condition, followed by six blocks of the Color-Predictive condition.

RESULTS

Behavioral Results

Mean RTs for the three experimental conditions are presented in Figure 2A. As the target IOI factor (short/long) had no main effect or interactions with the other factors in any of the conditions (all ps > .2), results are presented across this factor. In the Noninformative condition, in which both rhythm and color were not predictive, there was a main effect for Rhythm validity, with faster responses when the target appeared in-phase with the rhythm compared with when it was out-of-phase, F(1, 14) = 14.36, p < .005, MSE = 1164. In the Color-Predictive condition, participants created explicit expectations intentionally based on a predictive cue (color), although the rhythm was nonpredictive. As expected, there was a main effect for Color validity, with faster responses when the predictive color was valid than when it was invalid, F(1, 14) = 12.13, p < .01, MSE = 1689. Nevertheless, despite the fact that the rhythm was not predictive in this condition, there was also a main effect for Rhythm validity, with faster responses when the target appeared in-phase with the rhythm, F(1, 14) = 25.93, p < .01, MSE = 225. Importantly, there was no interaction between the Rhythm validity and Color validity (F < 1); planned comparisons revealed a validity effect for rhythm in this condition both when the color was valid, t(14) = −5.04, p < .01, and when it was invalid, t(14) = −2.32, p < .03. Finally, in the Rhythm-Predictive condition, there was a main effect for Rhythm validity, with faster responses when the predictive rhythm was valid than when it was invalid, F(1, 14) = 52.38, p < .0001, MSE = 2311, but there was no Color validity effect (in fact, when the color was invalid, responses were nominally faster than when it was valid) and no interaction, F(1, 14) = 1.03, p = .33. Thus, although both color and rhythm created a validity effect when they were predictive, only rhythm created a validity effect even when it was not predictive and, moreover, when participants intentionally predicted target timing based on an orthogonal cue.

Figure 2. 

Behavioral results of Experiment 1. (A) Mean RTs in each combination of color and rhythm validity in the three experimental conditions. To make error bars informative in the within-subject design, error bars represent standard errors for rhythm validity effect (rhythm valid − rhythm invalid, within each level of color validity). (B) Correlations between validity effects obtained for the rhythm in the three experimental conditions and for the color in the Color-Predictive condition. (C) Percentage of intersubject variance that was explained by the common factor extracted by the factor analysis in each of the validity effects obtained for the rhythm in the three experimental conditions and for the color in the Color-Predictive condition.

Figure 2. 

Behavioral results of Experiment 1. (A) Mean RTs in each combination of color and rhythm validity in the three experimental conditions. To make error bars informative in the within-subject design, error bars represent standard errors for rhythm validity effect (rhythm valid − rhythm invalid, within each level of color validity). (B) Correlations between validity effects obtained for the rhythm in the three experimental conditions and for the color in the Color-Predictive condition. (C) Percentage of intersubject variance that was explained by the common factor extracted by the factor analysis in each of the validity effects obtained for the rhythm in the three experimental conditions and for the color in the Color-Predictive condition.

The above results revealed that rhythm affected temporal expectations in all three conditions. However, the rhythm validity effects differed significantly between the three conditions, F(2, 14) = 19.24, Greenhouse–Geisser corrected p < .001, MSE = 657. Planned contrasts revealed that the Rhythm validity effect was larger in the Rhythm-Predictive condition than in the two other conditions in which the rhythm was nonpredictive, t(14) = 5.48, p < .001; there was no significant difference between the Rhythm validity effects in these two conditions, t(14) = 1.11, p = .29, 95% confidence interval: −3.62–11.35). The validity effect of rhythm was not modulated by the number of stimuli in the rhythmic sequence in any condition (F < 1).

We used exploratory factor analysis to identify a common factor in the intersubject variance of the Rhythm validity effects in the different conditions and of the Color validity effect in the Color-Predictive condition as a nonrhythmic, intentional control. One participant was eliminated from the analysis based on the extreme leverage criterion (see Methods). Significant correlation across participants was found only between the validity effects of rhythm in Noninformative and Color-Predictive conditions (r = 0.86, t(12) = 5.84, p < .001; Figure 2B). Weaker positive correlations were found between these two validity effects and the validity effect of rhythm in Rhythm-Predictive condition. The factor analysis extracted one reliable factor (eigenvalue = 2.05, 95% cutoff value of eigenvalue chance distribution = 1.82). This factor explained most of the variance in the effects of rhythm in the Noninformative and Color-Predictive conditions, but almost none of the variance in the effect of color in the Color-Predictive condition (Figure 2C), suggesting that it reflected the magnitude of engagement with nonpredictive rhythm alone. Interestingly, the proportion of variance in the effect of rhythm that was explained by the factor in the Rhythm-Predictive condition was smaller than in the nonpredictive rhythms but larger than the proportion of variance explained in the effect of color in the color-predictive condition.

ERP Results: WS

We tested the effect of temporal expectations on pretarget anticipatory brain activity by analyzing the CNV following the WSs. In both conditions, brain activity before target onset demonstrated a typical frontocentral CNV distribution (Figure 3A). In the Rhythm-Predictive condition, there was a main effect of the Rhythm interval factor, such that the CNV amplitude was more negative when the rhythm signaled a short compared with a long interval, F(1, 14) = 32.76, p < .001, MSE = 7.85 (Figure 3B). This suggests that preparatory activity evolved faster when the target was expected at an earlier time based on the rhythm. Similar to the behavioral results, there was no main effect for the Color interval factor in this condition, F(1, 14) < 1(Figure 3B, middle and right column), nor was there an interaction between the factors, F(1, 14) < 1.

Figure 3. 

Brain activity evoked by the WS in the two predictive conditions. (A) Scalp topographies measured 200 msec before target onset, 100 msec before target onset, and at target onset in the Rhythm-Predictive condition (across colors and expected intervals) and the Color-Predictive condition (across rhythms and expected intervals). (B) Grand-averaged waveforms in the Rhythm-Predictive condition. The CNV trajectory is affected by the interval signaled by the predictive rhythm (left column) but not by that signaled by the nonpredictive color (middle and right columns). (C) Grand-averaged waveforms in the Color-Predictive condition. The CNV trajectory is hardly affected by the interval signaled by the predictive color (left column) but is affected by the interval signaled by the nonpredictive rhythm (middle and right columns). The predefined time window that was used for statistical analysis of the CNV is illustrated in the bottom left panel.

Figure 3. 

Brain activity evoked by the WS in the two predictive conditions. (A) Scalp topographies measured 200 msec before target onset, 100 msec before target onset, and at target onset in the Rhythm-Predictive condition (across colors and expected intervals) and the Color-Predictive condition (across rhythms and expected intervals). (B) Grand-averaged waveforms in the Rhythm-Predictive condition. The CNV trajectory is affected by the interval signaled by the predictive rhythm (left column) but not by that signaled by the nonpredictive color (middle and right columns). (C) Grand-averaged waveforms in the Color-Predictive condition. The CNV trajectory is hardly affected by the interval signaled by the predictive color (left column) but is affected by the interval signaled by the nonpredictive rhythm (middle and right columns). The predefined time window that was used for statistical analysis of the CNV is illustrated in the bottom left panel.

In contrast to the effect of the rhythm in the Rhythm-predictive condition, there was only a trend for an effect of the Color interval factor on CNV amplitude in the Color-Predictive condition, F(1, 14) = 2.61, one-tailed p = .07, MSE = 2.75 (Figure 3C). This surprising result becomes clear when noting that there was a main effect for the (nonpredictive) Rhythm interval factor, F(1, 14) = 26.33, p < .01, MSE = 3.77, in this condition, with no interaction between the factors, F(1, 14) = 0.5, p > .4, MSE = 3.43. Detailed analyses (see Figure 3C, middle and right columns) revealed that, both when the color predicted a target at the short interval and when it predicted a long interval, the CNV was less negative when the rhythm signaled a long interval compared with a short interval (for color short condition: t(14) = −3.87, p < .01; for color long condition: t(14) = −3.55, p < .01). In summary, the CNV was driven by the rhythm, regardless of whether the rhythm was predictive or not.

ERP Results: Targets

The ERPs locked to the target were dominated by a parietal-maximum P3 (Figure 4A). In the analysis of predictive cues (Figure 4B, left column), the P3 latency was shorter when the target appeared at the cued interval (i.e., was validly cued), rather than in the other interval (main effect of Validity; corrected F(1, 14) = 10.72, p < .05). Although the interaction between cue type and validity showed only a trend (corrected F(1, 14) = 3.46, p = .084), we note that the validity effect was driven mainly by the Rhythm Predictive condition, F(1, 14) = 89.75, p < .05, and not so much by the Color-Predictive condition, F(1, 14) = 1.12, p = .31. There was no main effect for the Cue type factor or other interactions (all corrected Fs < 1). In contrast to the modulation by predictive cues, the validity of the nonpredictive cues (Figure 4B, right column) had no effect on the P3 latency nor did this factor interact with the Cue type or IOI (all corrected Fs < 1.55, all ps > .1). To address the concern that the results were affected by differences in the baseline period because of the CNV, we repeated this analysis using the 100 msec before the WS as baseline. The pattern of results was similar, with validity effect for predictive cues: F(1, 14) = 6.83, p < .05 and no validity effect for nonpredictive cues.

Figure 4. 

Brain activity evoked by the target. (A) Scalp topographies measured 350 msec after target onset in the two conditions. (B) Grand-averaged waveforms evoked by the target at a cluster of parietal electrodes (P1, Pz, P2) in the two conditions, demonstrating the modulation of P3 latency by cue validity when each cue type is either predictive or nonpredictive. The predefined time window for the P3 is illustrated in pink in the bottom left panel.

Figure 4. 

Brain activity evoked by the target. (A) Scalp topographies measured 350 msec after target onset in the two conditions. (B) Grand-averaged waveforms evoked by the target at a cluster of parietal electrodes (P1, Pz, P2) in the two conditions, demonstrating the modulation of P3 latency by cue validity when each cue type is either predictive or nonpredictive. The predefined time window for the P3 is illustrated in pink in the bottom left panel.

P3 amplitudes were analyzed separately for each IOI, as the baseline could be different in short and long IOI conditions because of the pretarget CNV (see Woldorff, 1993). No effects of validity or cue type were found for either interval (all ps > .1), neither when the cues were predictive nor nonpredictive.

The effects of temporal expectations on posttarget activity were analyzed also in the P1 and N1 components, presumably reflecting early sensory processing (Figure 5). Note that these results should be treated with caution, because of clear overlap with the preceding CNV and the ensuing P3. The P1 was larger for longer IOI (predictive: corrected F(1, 14) = 5.59, p < .05; nonpredictive: corrected F(1, 14) = 13.09, p < .05) but was not affected by any other factor (all ps > .1). Similar to previous findings (e.g., Correa et al., 2008), the N1 amplitude was significantly reduced for validly predicted targets in the predictive (corrected F(1, 14) = 8.53, p < .05), but not in the nonpredictive condition (F < 1). No other main effects or interactions were found.

Figure 5. 

Grand-averaged waveforms evoked by the target at a cluster of parietal electrodes (O1, PO3, PO7, O2, PO4, PO8), demonstrating modulations of P1 and N1 peak-to-peak amplitude by cue validity and target IOI when each cue type is either predictive or nonpredictive. The P1 and N1 components are marked in the bottom left panel.

Figure 5. 

Grand-averaged waveforms evoked by the target at a cluster of parietal electrodes (O1, PO3, PO7, O2, PO4, PO8), demonstrating modulations of P1 and N1 peak-to-peak amplitude by cue validity and target IOI when each cue type is either predictive or nonpredictive. The P1 and N1 components are marked in the bottom left panel.

Time–Frequency Analysis

Our main interest in this analysis was whether alpha-band (8–12 Hz) oscillatory activity will be reduced before or at expected stimulus time (as found by Rohenkohl & Nobre, 2011; Praamstra et al., 2006). We examined the power of alpha activity in the 100 msec just preceding the time of the short IOI target (early bin) and in the 100 msec during the short IOI target time (late bin) in conditions in which the target actually appeared in the long IOI. To assess the effect of expectations, we compared conditions in which the rhythm signaled a short or long IOI target, separately for when the rhythm was predictive or not.

In all conditions, stimuli presentations resulted with a reduction of alpha-band power, followed by a gradual power increase toward the time of the next stimulus. The WS was followed by a stronger reduction in power. Following this, alpha power remained low around the short IOI time when a stimulus was expected at the short IOI more than when the stimulus was expected at the long IOI (Figure 6). When rhythm-based expectations were directed to the short IOI intentionally (Figure 6A), this reduction of alpha power was significant both in the pretarget bin, t(14) = −2.05, p < .05, one-tailed, and in the target bin, t(14) = −2.16, p < .05, one-tailed. Modulation of alpha-band power by nonpredictive rhythmic stimulation was examined in the condition where intentional expectations were directed to the long IOI based on the predictive color (Figure 6B). Although alpha power was reduced around the short-interval target time when the nonpredictive rhythm had a short IOI compared with when the rhythm IOI was long, this difference was not significant in the pretarget bin, t(14) = −1.23, p = .12, one-tailed, and only marginally significant in the target bin, t(14) = −1.63, p = .06, one-tailed.

Figure 6. 

Time course of alpha-band oscillatory activity. Gray squares mark stimuli that were presented when the rhythm had long IOI, black squares mark stimuli that were presented when the rhythm had short IOI (corresponding to the line colors), and an empty square marks the presumed time of the short IOI target (which never appeared at that time in the displayed conditions). The pretarget and target time bins that were used for analysis are marked with red and blue shading, respectively. (A) Rhythm-Predictive condition in Experiment 1, separately for “short rhythm–long target” condition (black) and “long rhythm–long target” condition (gray). (B) Color-Predictive condition in Experiment 1, separately for “short rhythm–long target” condition (black) and “long rhythm–long target” condition (gray). In both conditions, the color predicts a long IOI target. (C) Same as B, but in Experiment 2.

Figure 6. 

Time course of alpha-band oscillatory activity. Gray squares mark stimuli that were presented when the rhythm had long IOI, black squares mark stimuli that were presented when the rhythm had short IOI (corresponding to the line colors), and an empty square marks the presumed time of the short IOI target (which never appeared at that time in the displayed conditions). The pretarget and target time bins that were used for analysis are marked with red and blue shading, respectively. (A) Rhythm-Predictive condition in Experiment 1, separately for “short rhythm–long target” condition (black) and “long rhythm–long target” condition (gray). (B) Color-Predictive condition in Experiment 1, separately for “short rhythm–long target” condition (black) and “long rhythm–long target” condition (gray). In both conditions, the color predicts a long IOI target. (C) Same as B, but in Experiment 2.

Experiment 2

A key result in Experiment 1 was the behavioral (RT) and electrophysiological (CNV) effect of nonpredictive rhythm cues in the Noninformative and Color-Predictive conditions, suggesting that expectations are created automatically in rhythmic context. The goal of Experiment 2 was to overrule the possibility that the effect of nonpredictive rhythm in the color-predictive blocks of Experiment 1 was because of the fact that the rhythm was predictive in other blocks of that experiment. Thus, in this experiment, we tested this effect in a new group of participants who were never exposed to any conditions in which the rhythm is predictive, that is, there were no Rhythm-predictive blocks.

Behavioral Results

As in Experiment 1, the results were collapsed across the target IOI factor, which had no main effect or interactions (all ps > .1, Figure 7A). There was a main effect of rhythm validity in the Noninformative condition, with faster responses when it was valid compared with invalid, F(1, 13) = 6.19, p < .05, MSE = 1660. In the Color-Predictive condition, as expected, there was a main effect of color validity, with faster responses when the predictive color was valid than when it was invalid, F(1, 13) = 6.18, p < .05, MSE = 1122, but critically, as in Experiment 1, there was also a main effect for the validity of the nonpredictive rhythm, with responses being faster when it was valid compared with invalid, F(1, 13) = 7.49, p < .05, MSE = 454. The interaction between color validity and rhythm validity was not significant (F < 1). The rhythm validity effect in the Noninformative condition was numerically, but not significantly, larger than in the Color-Predictive condition, but not significantly so, F(1, 13) = 1.53, p = .24. Also, there was a marginally significant correlation between the two validity effects (r = 0.52, p = .054). Comparison of the validity effects between the two experiments found no difference in the effect of nonpredictive rhythm in the Noninformative blocks, F(1, 27) = 1.76, p = .2, or in the Color-Predictive blocks, F(1, 27) = 0.2, p > .5. Although the validity effect of color in Experiment 2 was numerically smaller compared with Experiment 1, this difference was not significant, F(1, 27) = 3.15, p = .09. Thus, Experiment 2 replicated all the major behavioral findings of Experiment 1.

Figure 7. 

Experiment 2 results. (A) Behavioral results. Mean RTs in each combination of color and rhythm validity in the two experimental conditions. Error bars represent standard errors for rhythm validity effect within each level of color validity. (B) EEG results. Grand-averaged waveforms evoked by the WS in the two predictive conditions. The predefined time window for the CNV is illustrated in the bottom left panel. The CNV trajectory is hardly affected by the interval signaled by the predictive color (left column) but is affected by the interval signaled by the nonpredictive rhythm (middle and right columns).

Figure 7. 

Experiment 2 results. (A) Behavioral results. Mean RTs in each combination of color and rhythm validity in the two experimental conditions. Error bars represent standard errors for rhythm validity effect within each level of color validity. (B) EEG results. Grand-averaged waveforms evoked by the WS in the two predictive conditions. The predefined time window for the CNV is illustrated in the bottom left panel. The CNV trajectory is hardly affected by the interval signaled by the predictive color (left column) but is affected by the interval signaled by the nonpredictive rhythm (middle and right columns).

ERP Results and Time–Frequency Analysis

The CNV in the Color-Predictive condition was modulated by the interval of the nonpredictive rhythm, F(1, 13) = 10.36, p < .01, MSE = 5.4 (Figure 7B). Replicating the results of Experiment 1, regardless of whether the color predicted a target at the short or long interval, the CNV was less negative at the window of interest when the rhythm signaled a long compared with a short interval (for color short condition: t(13) = −2.52, p < .05, for color long condition: t(13) = −2.7, p < .05). There was no main effect for the interval signaled by the color or for the interaction between rhythm and color (both Fs < 1). Thus, although the rhythm was never predictive in the current experiment, it had a significant effect on both performance and the CNV.

The posttarget effects found in Experiment 1 were also replicated. Specifically, there was larger P1 for longer IOI (predictive color: corrected F(1, 13) = 14.7, p < .05; nonpredictive rhythm: corrected F(1, 13) = 41.7, p < .05) and attenuated N1 for valid compared with invalid targets for predictive color (corrected F(1, 13) = 6, p < .05), but not for nonpredictive rhythm, F(1, 13) = 2.56, p > .13. As in Experiment 1, no validity effect was found in P3 latency for both predictive color and nonpredictive rhythm (all corrected Fs < 1).

As in Experiment 1, we used trials in the Color-Predictive condition in which intentional expectations were directed to the long IOI based on color to test unintentional modulation of alpha-band power by rhythmic stimulation (Figure 6C). Although when the nonpredictive rhythm IOI was short, there was reduced alpha power compared with when it was long, this difference was not significant in both the pretarget, t(13) = −0.7, and the target bin, t(13) = −0.92.

GENERAL DISCUSSION

By applying stringent control over high-level factors, the current study provides behavioral and electrophysiological evidence for unintentional creation of temporal expectations with exposure to rhythmic simulation. We found that rhythmic sequences affected behavior and the anticipatory CNV potential even when it was strategically disadvantageous to use the information embedded in the rhythm. We further differentiated unintentional expectations from those that are created by intentional usage of the rhythm: When the rhythm was intentionally used to predict the timing of the targets, the effect of prediction was significantly increased relative to when predictions were formed unintentionally. Moreover, only when expectations were formed intentionally, the latency of the P3 was reduced when targets appeared at the expected time.

Rhythmic Context Induces Temporal Expectations Unintentionally

The results of our Noninformative condition replicate previous studies, which found faster responses for events that occurred in phase with rhythms (Miller et al., 2012; Mathewson et al., 2010; Jones et al., 2002, 2006; Barnes & Jones, 2000), even when the rhythm did not predict the timing of the event (e.g., Sanabria et al., 2011). However, without controlling intentional temporal expectations, it is difficult to attribute this effect to unintentional processing, as participants could be using the rhythm intentionally to form expectations as long as there was no reason not to. Thus, in our Color-Predictive condition, the participants were instructed to use the color cue to predict the target timing, and the validity effect observed for the color cue in the this condition confirms that intentional allocation of temporal expectations was indeed driven by the color cue. The fact that there was still a validity effect for the rhythm although the interval embedded in the nonpredictive rhythm was orthogonal to that of the color cue and could even detract from participants' performance allows us to conclude that temporal expectations were biased automatically by the rhythmic stimulation.

When a cue–target interval is known, it is reflected in the time course of the CNV (Pfeuty et al., 2005; Macar & Vidal, 2004), a phenomenon that was also found in the context of predictive rhythms (Praamstra et al., 2006). By showing that the CNV was directionally modulated by the rhythm in the Color-Predictive condition where intentional expectations were controlled, we extend these findings to nonpredictive rhythms. Furthermore, although previous studies found modulations of the CNV by symbolic cuing when presented in the absence of any temporal context (e.g., Miniussi et al., 1999), no such modulation was observed in the context of rhythmic stimulation in our study, although the rhythm was nonpredictive. Thus, in the context of rhythmic stimulation, anticipatory brain activity is driven unintentionally by the rhythm and not by cue-driven intentional expectations.

Our results are consistent with previous findings of sequential effects not involving rhythmic stimulation (Capizzi, Correa, & Sanabria, 2013; Capizzi, Sanabria, & Correa, 2012; Los & Heslenfeld, 2005; Los & Van Den Heuevl, 2001). For example, when the IOI in a specific trial of a temporal-orienting S1–S2 paradigm with variable intervals was shorter than the IOI of the preceding trial, responses were slowed (Los & Van Den Heuevl, 2001), and the CNV time course was modulated according to the preceding interval (Los & Heslenfeld, 2005). Bolstering the automatic nature of the effect, it was found to be resistant to dual-task interference (Capizzi et al., 2012). The current findings suggest that automatic effects of temporal information are formed even when participants are not required to respond to preceding stimuli, as was the case in S1–S2 paradigms, if these stimuli appear rhythmically. Moreover, comparison of the pattern of results obtained here with those previously reported suggests that the rhythm effect does not merely reflect repeating sequential effect as found in the S1–S2 paradigms. Specifically, in the current study, nonpredictive rhythms and symbolic cuing affected behavior additively but affected the CNV underadditively (in fact, the CNV was completely driven by the nonpredictive rhythmic stimulation). In contrast, in the studies of Los et al. (Los & Heslenfeld, 2005; Los & Van Den Heuevl, 2001) targeting sequential effects and symbolic cuing, the effects combined underadditively in behavior (sequential effects were seen only when the symbolic cue was invalid), and additively in the CNV. These different patterns imply that sequential effects and the effects of rhythms rely on mechanisms that are not entirely overlapping.

One possible criticism of the results of Experiment 1 is that participants could have intentionally used the (nonpredictive) rhythm in the Color-Predictive condition, because of confusion, task set inertia, or strategy, because rhythm did carry valid temporal information in some of the blocks. This possibility seems unlikely for the following reasons. First, the lack of color validity effect in the Rhythm-Predictive condition precludes the possibility that there was general confusion between tasks. Second, intentional usage of the rhythm in the Color-Predictive condition should have resulted in a pattern of results that is similar to the effect of rhythm in the Rhythm-Predictive condition, but it was significantly smaller, even when only the trials in which the rhythm happened to be congruent with the color were considered, t(1, 14) = 2.68, p < .05. Finally, and most conclusively, the fact that the two main findings of Experiment 1 (behavioral facilitation by nonpredictive rhythm when using color and directional modulation of CNV) were replicated in Experiment 2, in which rhythms were never predictive, supports the attribution of the nonpredictive rhythm effect to automatic, unintentional, processing.

It is commonly assumed that the auditory system is more suited for temporal processing than the visual system (see Näätänen & Winkler, 1999). Several lines of research demonstrate superiority of the auditory system in temporal processing (Repp & Penel, 2002; Fendrich & Corballis, 2001), especially of rhythmic stimuli (Grahn, Henry, & McAuley, 2011; Rencanzone, 2003). Indeed, the effect of truly nonpredictive rhythms was previously examined only for auditory rhythms (Sanabria et al., 2011; for discussion of related visual studies, see Introduction). Our results demonstrate that, although the visual system may have a lower temporal resolution, temporal predictions take advantage of temporal regularities in visual input, at least with the rather crude temporal resolution, on the order of hundreds of milliseconds, used here. This does not preclude the possibility, raised by some, that the effect of visual rhythms is mediated by creation of auditory rhythm representations (Grahn et al., 2011; Guttman, Gilroy, & Blake, 2005). Future research is required to attribute the automatic effects to purely visual processes, for example, by introducing a secondary auditory task hindering implicit reliance on auditory representations.

Forming Expectations Intentionally Affects Behavior and the P3 Latency

When the rhythm was predictive and expectations were intentionally based on it, the behavioral validity effect was considerably larger than that of nonpredictive rhythms (cf. Sanabria et al., 2011, although they did not directly compare the two conditions).3 This increase can be explained in several ways. One possibility is that intentional usage of the rhythm results in amplification of the same representation as the one involved in the unintentional effect of rhythm, mediated by the same mechanism. An opposite extreme alternative is that intentional usage of the rhythm engages a different mechanism that is more effective but is unrelated to the mechanism, which mediates the unintentional effect of rhythm. These alternatives are not consistent with the results of the factor analysis, namely that the effect of intentionally using the rhythm was explained by the common factor reflecting unintentional engagement with the rhythm but only mildly so. Instead, the factor analysis results suggest that the expectation created unintentionally exerts an effect, but that additional mechanisms are involved when the rhythm is used intentionally. However, this finding by itself does not allow us to determine whether the intentional and unintentional effects act at the same locus or not and whether they interact or are additive.

A hint may be derived from the dissociation between the two effects on the CNV and P3 components. The CNV was modulated by the expected interval based on the rhythm, both when the rhythm was used intentionally (in the Rhythm-Predictive condition; Praamstra et al., 2006) and when its effect was unintentional (in the Color-Predictive condition). In contrast, modulation of the N1 and P3 components, previously reported for temporal expectations (Correa et al., 2008; Doherty et al., 2005; Griffin et al., 2002; Miniussi et al., 1999), was observed only for intentional usage of the rhythm (i.e., when it was predictive). Under the premise that the P3 reflects processes related to stimulus evaluation and response selection vis a vis the task (Verleger, Jaskowski, & Wascher, 2005; Verleger, 1997; Donchin & Coles, 1988), the effect of temporal expectation on the P3 latency implies that intentional temporal preparation affects these late processes. Whether this is related to the attenuation of the late visual response reflected in the N1 (Doherty et al., 2005), remains to be explored.

Together, although the current design did not allow for a direct comparison between intentional and automatic effects, the overall pattern of results implies a dissociation. Exposure to rhythmic input is sufficient to automatically bias anticipatory activity to the in-phase time, possibly by preparing the motor system (Nagai et al., 2004; Macar & Vidal, 2002), but modulation of target evaluation (as reflected by the P3) depends on intentional use of the temporal predictions.

Previous studies found that when stimuli appear rhythmically, alpha-band activity becomes desynchronized just preceding (Rohenkohl & Nobre, 2011) and at the time (Praamstra et al., 2006) of a temporally expected visual event. However, as we reasoned above, the effects found in those studies cannot be attributed to unintentional formation of temporal expectations by the rhythm, as the rhythm was irrelevant but predictive. In our data, significant reduction in alpha power was observed before and during expected target time when the rhythm was predictive, but only a nonsignificant difference in this direction was found when the rhythm was nonpredictive. This pattern may suggest that the alpha desynchronization that was found in previous studies of rhythmic temporal expectations reflects the intentional component of using the rhythm. This interpretation is also consistent with findings that relate alpha desynchronization to intentional factors such as cue-directed attending to location (Thut, Nietzel, Brandt, & Pascual-Leone, 2006) or modality (Foxe, Simpson, & Ahlfors, 1998) and cue certainty (Gould, Rushworth, & Nobre, 2011). However, unlike the P3 latency validity effect, which was completely absent in the nonpredictive condition, in Experiment 1 there was still a trend for some modulation of alpha band power even when the rhythm was nonpredictive. Future research should examine whether under some conditions nonpredictive rhythmic stimulation can bias alpha-band activity.

Disentangling Sources of Temporal Expectation and Levels of Voluntary Control

It is common to distinguish between exogenous temporal expectations, in which the temporal structure is embedded in the stimulation pattern, and endogenous expectations based on prelearned arbitrary association between a cue and an ensuing event. It is appealing (and, we argue, is frequently implied) to see a parallel between this dissociation and the dissociation between automatic and intentional processes (exogenous ≈ automatic, endogenous ≈ intentional). Nevertheless, the pattern of our results suggests that the source of temporal expectations (exo/endo) and level of voluntary control (automatic/controlled; Shiffrin & Schneider, 1977) should be seen as interacting. Our findings show that expectations based on a clearly exogenous cue (rhythm) differ substantially, depending on whether they are applied intentionally or induced passively merely by exposure to the rhythm, both in the magnitude of the effect on behavior and in the stage in which they are manifested in brain activity. These findings highlight a common principle regarding intentional and automatic effects, which was seen in other domains. Namely, the system is hard-wired to be passively driven by input properties, such that predictions reflect physical regularities in the world. Yet, goal-oriented systems that rely on high-level information can modulate these effects, possibly at different stages of processing (see Bacon & Egeth, 1994, for a discussion on intentional magnification of stimulus-driven representations in spatial orienting, and Haroush, Hochstein, & Deouell, 2010, regarding interaction between attention and the automaticity of MMN, another measure of regularity extraction).

The distinction between the exogenous–endogenous and automatic–intentional dimensions can be seen also in reference to endogenous cues based on prelearned memorized interval, triggered by a warning cue (Correa et al., 2006). As the memorized interval is an endogenous source for temporal expectations, utilizing it seems to be an intentional process. However, expectations based on exposure to nonrhythmic intervals may also be formed automatically, presumably based on a process akin to trace conditioning (Los & Heslenfeld, 2005; Los & Van Den Heuevl, 2001). Thus, it is appropriate to see the two dimensions of endogenous–exogenous and automatic–intentional as separable.

Multiple Mechanisms of Temporal Predictions

Several studies found dissociations between the neural mechanisms involved in explicit time estimation based on absolute duration memorization, which was related to activity in the cerebellum, and estimation based on rhythmic sequences, which was related to activity in the striatum, supplementary, and premotor cortex (Teki, Grube, Kumar, & Griffiths, 2011; Grube, Cooper, Chinnery, & Griffiths, 2010). In our results, the behavioral effects of nonpredictive rhythms and predictive symbolic cuing were additive, implying that this segregation exists not only in explicit duration estimation, but also in forming temporal expectations. However, when both rhythmic and memory-based predictions are present, anticipatory activity was dominated by the temporal regularity, to the extent that previously shown effects of symbolic cuing on the CNV (e.g., Miniussi et al., 1999) were not observed. This further emphasizes that these two mechanisms operate at different stages of processing.

Our results are agnostic regarding the mechanism underlying the unintentional effect of rhythm. It has been shown that exposure to rhythmic stimulation results in entrainment (Besle et al., 2011; Lakatos et al., 2008) and that some of the effects of rhythmic context are better explained by entrainment models than by interval-based models (McAuley & Jones, 2003; McAuley & Kidd, 1998). Under the premise that any effect of rhythmic stimulation is actually the outcome of entrainment, the current results could be taken to imply that entrainment occurs automatically with exposure to rhythmic stimulation. However, the fact that neural activity along the sensory processing stream becomes entrained to rhythmic stimulation does not mean that all the phenomena that occur in rhythmic context are necessarily attributed to sensory entrainment. Instead, they could reflect involvement of rhythm processing systems such as striatal or premotor regions (see also Martin et al., 2008) or modulation of some components in the mechanism that is used by the symbolic cuing task (e.g., rhythmic stimulation could speed or slow the pacemaker in a pacemaker-accumulator model; Block & Zakay, 1996; Gibbon, Church, & Meck, 1984).

In conclusion, as the dynamics of our sensory environment is often non-random, this information can be used to create expectations and facilitate behavior. The current study highlights two components of temporal expectations in rhythmic context. First, the presence of temporally regular context is enough to passively and unintentionally create an expectation for the predictable time, reflecting an implicit assumption implemented in the system that stimulus dynamics will maintain their regularity. Second, when it is known that an event of interest will occur in sync with stimulus dynamics, this expectation is enhanced such that it also affects posttarget activity. Together, these principles constitute a balance between effortless compatibility with regularities determined by physical constraints and flexible, though effortful, modulation of behavior according to current goals.

Acknowledgments

This research was funded by a grant from the Israel Science Foundation to Leon Y. Deouell.

Reprint requests should be sent to Assaf Breska, The Department of Psychology, The Hebrew University of Jerusalem, Mount Scopus, Jerusalem, 91905 Israel, or via e-mail: assaf.breska@mail.huji.ac.il.

Notes

1. 

Participants were instructed to respond with their dominant hand, but some of them switched hands during the experiment if it was more comfortable for them, and these changes were not tracked. Although we believe that this does not affect the validity of the findings we report, we refrained from analyzing lateralized brain responses.

2. 

Although the specific target IOI is not expected to influence the posttarget responses, because of their different “histories” the baselines of short and long IOI target trials is different and requires that these trials will be treated separately (see Woldorff, 1993).

3. 

It could be argued that in our Color-Predictive condition the validity effect of rhythm was reduced due to dual-task interference from the color task (see Capizzi and colleagues, 2012, 2013, who found that working memory load disrupted intentional temporal orienting based on a symbolic cue). However, the fact that this effect was similar to that observed in the Noninformative condition, when the color conveyed no information, is not consistent with this possibility.

REFERENCES

REFERENCES
Bacon
,
W. F.
, &
Egeth
,
H. E.
(
1994
).
Overriding stimulus-driven attentional capture.
Perception & Psychophysics
,
55
,
485
496
.
Barnes
,
R.
, &
Jones
,
M. R.
(
2000
).
Expectancy, attention, and time.
Cognitive Psychology
,
41
,
254
311
.
Besle
,
J.
,
Schevon
,
C. A.
,
Mehta
,
A. D.
,
Lakatos
,
P.
,
Goodman
,
R. R.
,
McKhann
,
G. M.
,
et al
(
2011
).
Tuning of the human neocortex to the temporal dynamics of attended events.
Journal of Neuroscience
,
31
,
3176
3185
.
Bevan
,
W.
,
Hardesty
,
D. L.
, &
Avant
,
L. L.
(
1965
).
Response latency with constant and variable interval schedules.
Perceptual and Motor Skills
,
20
,
969
972
.
Block
,
R. A.
, &
Zakay
,
D.
(
1996
).
Models of psychological time revisited.
In H. Helfrich (Ed.)
,
Time and mind
(pp.
171
195
).
Bern
:
Hogrefe and Huber
.
Brainard
,
D. H.
(
1997
).
The psychophysics toolbox.
Spatial Vision
,
10
,
433
436
.
Buhusi
,
C. V.
, &
Meck
,
W. H.
(
2005
).
What makes us tick? Functional and neural mechanisms of interval timing.
Nature Reviews Neuroscience
,
6
,
755
765
.
Busch
,
N.
,
Dubois
,
J.
, &
VanRullen
,
R.
(
2009
).
The phase of ongoing oscillations predicts visual perception.
Journal of Neuroscience
,
29
,
7869
7876
.
Busch
,
N. A.
, &
VanRullen
,
R.
(
2010
).
Spontaneous EEG oscillations reveal periodic sampling of visual attention.
Proceedings of the National Academy of Sciences, U.S.A.
,
107
,
16048
16053
.
Capizzi
,
M.
,
Correa
,
A.
, &
Sanabria
,
D.
(
2013
).
Temporal orienting of attention is interfered by concurrent working memory updating.
Neuropsychologia
,
51
,
326
339
.
Capizzi
,
M.
,
Sanabria
,
D.
, &
Correa
,
A.
(
2012
).
Dissociating controlled from automatic processing in temporal preparation.
Cognition
,
123
,
293
302
.
Cohen
,
J.
,
Cohen
,
P.
,
West
,
S. G.
, &
Aiken
,
L. S.
(
2003
).
Applied multiple regression/correlation analysis for the behavioral sciences.
Mahwah, NJ
:
Lawrence Erlbaum
.
Correa
,
A.
,
Lupiañez
,
J.
, &
Tudela
,
P.
(
2006
).
The attentional mechanism of temporal orienting: Determinants and attributes.
Experimental Brain Research
,
169
,
58
68
.
Correa
,
A.
, &
Nobre
,
A. C.
(
2008
).
Neural modulation by regularity and the passage of time.
Journal of Neurophysiology
,
100
,
1649
1655
.
Correa
,
A.
,
Rao
,
A.
, &
Nobre
,
A. C.
(
2008
).
Anticipating conflict facilitates controlled stimulus-specific selection.
Journal of Cognitive Neuroscience
,
21
,
1461
1472
.
Coull
,
J. T.
, &
Nobre
,
A. C.
(
2008
).
Dissociating explicit timing from temporal expectations with fMRI.
Current Opinion in Neurobiology
,
18
,
137
144
.
Doherty
,
J. R.
,
Rao
,
A.
,
Mesulam
,
M. M.
, &
Nobre
,
A. C.
(
2005
).
Synergistic effect of combined temporal and spatial expectations on visual attention.
Journal of Neuroscience
,
25
,
8259
8266
.
Donchin
,
E.
, &
Coles
,
M. G. H.
(
1988
).
Is the P300 component a manifest of context updating.
Behavioral and Brain Sciences
,
11
,
357
374
.
Drake
,
C.
, &
Botte
,
M. C.
(
1993
).
Tempo sensitivity in auditory sequences: Evidence for a multiple-look model.
Perception & Psychophysics
,
54
,
277
286
.
Fendrich
,
R.
, &
Corballis
,
P. M.
(
2001
).
The temporal cross-capture of audition and vision.
Perception & Psychophysics
,
63
,
719
725
.
Foxe
,
J. J.
,
Simpson
,
G. V.
, &
Ahlfors
,
S. P.
(
1998
).
Parieto-occipital approximately 10 Hz activity reflects anticipatory state of visual attention mechanisms.
NeuroReport
,
9
,
3929
3933
.
Gibbon
,
J.
,
Church
,
R. M.
, &
Meck
,
W. H.
(
1984
).
Scalar timing in memory.
Annals of the New York Academy of Sciences
,
423
,
52
77
.
Good
,
P. I.
(
2005
).
Resampling methods: A practical guide for data analysis.
Boston
:
Birkhauser
.
Gould
,
I. C.
,
Rushworth
,
M. F.
, &
Nobre
,
A. C.
(
2011
).
Indexing the graded allocation of visuospatial attention using anticipatory alpha oscillations.
Journal of Neurophysiology
,
105
,
1318
1326
.
Grahn
,
J. A.
,
Henry
,
M. J.
, &
McAuley
,
J. D.
(
2011
).
fMRI investigation of cross-modal interactions in beat perception: Audition primes vision, but not vice versa.
Neuroimage
,
54
,
1231
1243
.
Griffin
,
I. C.
,
Miniussi
,
C.
, &
Nobre
,
A. C.
(
2002
).
Multiple mechanisms of selective attention: Differential modulation of stimulus processing by attention to space or time.
Neuropsychologia
,
40
,
2325
2340
.
Grube
,
M.
,
Cooper
,
F. E.
,
Chinnery
,
P. F.
, &
Griffiths
,
T. D.
(
2010
).
Dissociation of duration-based and beat-based auditory timing in cerebellar degeneration.
Proceedings of the National Academy of Sciences, U.S.A.
,
107
,
11597
11601
.
Guttman
,
S. E.
,
Gilroy
,
L. A.
, &
Blake
,
R.
(
2005
).
Hearing what the eyes see: Auditory encoding of visual temporal sequences.
Psychological Science
,
16
,
228
235
.
Haroush
,
K.
,
Hochstein
,
S.
, &
Deouell
,
L. Y.
(
2010
).
Momentary fluctuations in allocation of attention: Cross-modal effects of visual task load on auditory discrimination.
Journal of Cognitive Neuroscience
,
22
,
1440
1451
.
Jazayeri
,
M.
, &
Shadlen
,
M. N.
(
2010
).
Temporal context calibrates interval timing.
Nature Neuroscience
,
13
,
1020
1028
.
Jones
,
M. R.
(
2010
).
Attending to sound patterns and the role of entrainment.
In A. C. Nobre & J. T. Coull (Eds.)
,
Attention and time
(pp.
317
330
).
New York
:
Oxford University Press
.
Jones
,
M. R.
, &
Boltz
,
M.
(
1989
).
Dynamic attending and responses to time.
Psychological Review
,
96
,
459
491
.
Jones
,
M. R.
,
Moynihan
,
H.
,
MacKenzie
,
N.
, &
Puente
,
J.
(
2002
).
Temporal aspects of stimulus-driven attending in dynamic arrays.
Psychological Science
,
13
,
313
319
.
Jones
,
M. R.
,
Moynihan-Johnston
,
H.
, &
Puente
,
J.
(
2006
).
Effects of auditory pattern structure on anticipatory and reactive attending.
Cognitive Psychology
,
53
,
59
96
.
Jung
,
T.
,
Makeig
,
S.
,
Humphries
,
C.
,
Lee
,
T.
,
McKeown
,
M. J.
,
Iragui
,
V.
,
et al
(
2000
).
Removing electroencephalographic artifacts by blind source separation.
Psychophysiology
,
37
,
163
178
.
Keele
,
S. W.
,
Nicoletti
,
R.
,
Ivry
,
R. I.
, &
Pokorny
,
R. A.
(
1989
).
Mechanisms of perceptual timing: Beat-based or interval-based judgments?
Psychological Research
,
50
,
251
256
.
Kiesel
,
A.
,
Miller
,
J.
,
Jolicoeur
,
P.
, &
Brisson
,
B.
(
2008
).
Measurement of ERP latency differences: A comparison of single-participant and jackknife-based scoring methods.
Psychophysiology
,
45
,
250
274
.
Lakatos
,
P.
,
Karmos
,
G.
,
Mehta
,
A. D.
,
Ulbert
,
I.
, &
Schroeder
,
C. E.
(
2008
).
Entrainment of neuronal oscillations as a mechanism of attentional selection.
Science
,
320
,
110
113
.
Large
,
E. W.
, &
Jones
,
M. R.
(
1999
).
The dynamics of entrainment: How people track time-varying events.
Psychological Review
,
106
,
119
159
.
Los
,
S. A.
, &
Heslenfeld
,
D. J.
(
2005
).
Intentional and unintentional contributions to nonspecific preparation: Electrophysiological evidence.
Journal of Experimental Psychology: General
,
134
,
52
72
.
Los
,
S. A.
, &
Van Den Heuevl
,
C. E.
(
2001
).
Intentional and unintentional contributions to nonspecific preparation during reaction time foreperiods.
Journal of Experimental Psychology: Human Perception and Performance
,
27
,
370
386
.
Macar
,
F.
, &
Vidal
,
F.
(
2002
).
Time processing reflected by EEG surface Laplacians.
Experimental Brain Research
,
145
,
403
406
.
Macar
,
F.
, &
Vidal
,
F.
(
2004
).
Event-related potentials as indices of time processing: A review.
Journal of Psychophysiology
,
18
,
89
104
.
Martin
,
T.
,
Egly
,
R.
,
Houck
,
J. M.
,
Bish
,
J. P.
,
Barrera
,
B. D.
,
Lee
,
D. C.
,
et al
(
2005
).
Chronometric evidence for entrained attention.
Perception & Psychophysics
,
67
,
168
184
.
Martin
,
T.
,
Houck
,
J. M.
,
Kicic
,
D.
, &
Tesche
,
C. D.
(
2008
).
Interval timers and coupled oscillators both mediate the effect of temporally structured cuing.
Neuroimage
,
40
,
1798
1806
.
Mathewson
,
K. E.
,
Fabiani
,
M.
,
Gratton
,
G.
,
Beck
,
D. M.
, &
Lleras
,
A.
(
2010
).
Rescuing stimuli from invisibility: Inducing a momentary release from visual masking with pre-target entrainment.
Cognition
,
115
,
186
191
.
Mathewson
,
K. E.
,
Gratton
,
G.
,
Fabiani
,
M.
,
Beck
,
D. M.
, &
Ro
,
T.
(
2009
).
To see or not to see: Prestimulus alpha phase predicts visual awareness.
Journal of Neuroscience
,
29
,
2725
2732
.
McAuley
,
J. D.
, &
Jones
,
M. R.
(
2003
).
Modeling effects of rhythmic context on perceived duration: A comparison of interval and entrainment approaches to short-interval timing.
Journal of Experimental Psychology: Human Perception & Performance
,
29
,
1102
1125
.
McAuley
,
J. D.
, &
Kidd
,
G. R.
(
1998
).
Effects of deviations from temporal expectations on tempo discriminations of isochronous tone sequences.
Journal of Experimental Psychology: Human Perception & Performance
,
24
,
1786
1800
.
Miller
,
J. E.
,
Carlson
,
L. A.
, &
McAuley
,
J. D.
(
2012
).
When what you hear influences when you see: Listening to an auditory rhythm influences the temporal allocation of visual attention.
Psychological Science
,
24
,
11
18
.
Miller
,
J.
,
Patterson
,
T.
, &
Ulrich
,
R.
(
1998
).
Jackknife-based method for measuring LRP onset latency differences.
Psychophysiology
,
35
,
99
115
.
Miniussi
,
C.
,
Wilding
,
E. L.
,
Coull
,
J. T.
, &
Nobre
,
A. C.
(
1999
).
Orienting attention in time: Modulation of brain potentials.
Brain
,
122
,
1507
1518
.
Näätänen
,
R.
, &
Winkler
,
I.
(
1999
).
The concept of auditory stimulus representation in cognitive neuroscience.
Psychological Bulletin
,
125
,
826
859
.
Nagai
,
Y.
,
Critchley
,
H. D.
,
Featherstone
,
E.
,
Fenwick
,
P. B. C.
,
Trimble
,
M. R.
, &
Dolan
,
R. J.
(
2004
).
Brain activity related with the contingent negative variation: An fMRI investigation.
Neuroimage
,
21
,
1232
1241
.
Niemi
,
P.
, &
Näätänen
,
R.
(
1981
).
Foreperiod and simple reaction time.
Psychological Bulletin
,
89
,
133
162
.
Nobre
,
A. C.
,
Correa
,
A.
, &
Coull
,
J. T.
(
2007
).
The hazards of time.
Current Opinion in Neurobiology
,
17
,
465
470
.
Pelli
,
D. G.
(
1997
).
The VideoToolbox software for visual psychophysics: Transforming numbers into movies.
Spatial Vision
,
10
,
437
442
.
Pfeuty
,
M.
,
Ragot
,
R.
, &
Pouthas
,
V.
(
2005
).
Relationship between CNV and timing of an upcoming event.
Neuroscience Letters
,
382
,
106
111
.
Posner
,
M. I.
,
Snyder
,
C. R. R.
, &
Davidson
,
B. J.
(
1980
).
Attention and the detection of signals.
Journal of Experimental Psychology: General
,
109
,
160
174
.
Praamstra
,
P.
,
Kourtis
,
D.
,
Kwok
,
H. F.
, &
Oostenveld
,
R.
(
2006
).
Neurophysiology of implicit timing in serial choice reaction-time performance.
Journal of Neuroscience
,
26
,
5448
5455
.
Rencanzone
,
G. H.
(
2003
).
Auditory influences on visual temporal rate perception.
Journal of Neurophysiology
,
89
,
1078
1093
.
Repp
,
B.
, &
Penel
,
A.
(
2002
).
Auditory dominance in temporal processing: New evidence from synchronization with simultaneous visual and auditory sequences.
Journal of Experimental Psychology: Human Perception and Perception
,
28
,
1085
1099
.
Rohenkohl
,
G.
,
Carvo
,
A. M.
,
Wyart
,
V.
, &
Nobre
,
A. C.
(
2012
).
Temporal expectation improves the quality of sensory information.
Journal of Neuroscience
,
32
,
8424
8428
.
Rohenkohl
,
G.
,
Coull
,
J. T.
, &
Nobre
,
A. C.
(
2011
).
Behavioral dissociation between exogenous and endogenous temporal orienting of attention.
PLoS One
,
6
,
e14620
.
Rohenkohl
,
G.
, &
Nobre
,
A. C.
(
2011
).
Alpha oscillations related to anticipatory attention follow temporal expectations.
Journal of Neuroscience
,
31
,
14076
14084
.
Sanabria
,
D.
,
Capizzi
,
M.
, &
Correa
,
A.
(
2011
).
Rhythms that speed you up.
Journal of Experimental Psychology: Human Perception & Performance
,
37
,
236
244
.
Schroeder
,
C. E.
, &
Lakatos
,
P.
(
2009
).
Low-frequency neuronal oscillations as instruments of sensory selection.
Trends in Neurosciences
,
32
,
9
18
.
Shiffrin
,
R. M.
, &
Schneider
,
W.
(
1977
).
Controlled and automatic human information processing. 2. Perceptual learning, automatic attending, and a general theory.
Psychological Review
,
84
,
127
190
.
Sterzer
,
P.
,
Frith
,
C.
, &
Petrovic
,
P.
(
2008
).
Believing is seeing: Expectations alter visual awareness.
Current Biology
,
18
,
R697
R698
.
Tallon-Baudry
,
C.
, &
Bertrand
,
O.
(
1999
).
Oscillatory gamma activity in humans and its role in object representation.
Trends in Cognitive Sciences
,
3
,
151
162
.
Teki
,
S.
,
Grube
,
M.
,
Kumar
,
S.
, &
Griffiths
,
T. D.
(
2011
).
Distinct neural substrates of duration-based and beat-based auditory timing.
Journal of Neuroscience
,
31
,
3805
3812
.
Thut
,
G.
,
Nietzel
,
A.
,
Brandt
,
S. A.
, &
Pascual-Leone
,
A.
(
2006
).
Alpha-band electro-encephalographic activity over occipital cortex indexes visuospatial attention bias and predicts visual target detection.
Journal of Neuroscience
,
26
,
9494
9502
.
Ulrich
,
R.
, &
Miller
,
J.
(
2001
).
Using the jackknife-based scoring method for measuring LRP onset effects in factorial designs.
Psychophysiology
,
38
,
816
827
.
Verleger
,
R.
(
1997
).
On the utility of P3 latency as an index of mental chronometry.
Psychophysiology
,
34
,
134
156
.
Verleger
,
R.
,
Jaskowski
,
P.
, &
Wascher
,
E.
(
2005
).
Evidence for an integrative role of P3b in linking reaction to perception.
Journal of Psychophysiology
,
19
,
165
181
.
Walter
,
W. G.
,
Cooper
,
R.
,
Aldridge
,
V. J.
,
McCallum
,
W. C.
, &
Winter
,
A. L.
(
1964
).
Contingent negative variation: And electric signal of sensorimotor association and expectancy in the human brain.
Nature
,
203
,
380
384
.
Woldorff
,
M. G.
(
1993
).
Distortion of ERP averages due to overlap from temporally adjacent ERPs: Analysis and correction.
Psychophysiology
,
30
,
98
119
.
Yantis
,
S.
, &
Jonides
,
J.
(
1990
).
Abrupt visual onsets and selective attention: Evidence from visual search.
Journal of Experimental Psychology: Human Perception & Performance
,
16
,
121
134
.