Abstract

Predicting the timing of incoming information allows the brain to optimize information processing in dynamic environments. Behaviorally, temporal expectations have been shown to facilitate processing of events at expected time points, such as sounds that coincide with the beat in musical rhythm. Yet, temporal expectations can develop based on different forms of structure in the environment, not just the regularity afforded by a musical beat. Little is still known about how different types of temporal expectations are neurally implemented and affect performance. Here, we orthogonally manipulated the periodicity and predictability of rhythmic sequences to examine the mechanisms underlying beat-based and memory-based temporal expectations, respectively. Behaviorally and using EEG, we looked at the effects of beat-based and memory-based expectations on auditory processing when rhythms were task-relevant or task-irrelevant. At expected time points, both beat-based and memory-based expectations facilitated target detection and led to attenuation of P1 and N1 responses, even when expectations were task-irrelevant (unattended). For beat-based expectations, we additionally found reduced target detection and enhanced N1 responses for events at unexpected time points (e.g., off-beat), regardless of the presence of memory-based expectations or task relevance. This latter finding supports the notion that periodicity selectively induces rhythmic fluctuations in neural excitability and furthermore indicates that, although beat-based and memory-based expectations may similarly affect auditory processing of expected events, their underlying neural mechanisms may be different.

INTRODUCTION

To optimize sensory processing and perception in a changing environment, the human brain continuously tries to predict incoming information (Clark, 2013; Friston, 2005). Being able to not only predict the content of sensory input (“what”) but also its timing (“when”) allows the system to prepare for and focus on time points when useful information is likely to occur (Nobre & van Ede, 2018; Large & Jones, 1999). Indeed, temporal expectations have been shown to improve processing of events at expected time points (Haegens & Zion-Golumbic, 2018; Nobre & van Ede, 2018; ten Oever, Schroeder, Poeppel, van Atteveldt, & Zion-Golumbic, 2014; Henry & Obleser, 2012; Rohenkohl, Cravo, Wyart, & Nobre, 2012). Additionally, temporal expectations allow us to align our actions to sensory input, enabling complex behaviors such as dancing and synchronizing to musical rhythm (Honing & Bouwer, 2019; McGarry, Sternin, & Grahn, 2019; Merchant, Grahn, Trainor, Rohrmeier, & Fitch, 2015).

Temporal expectations are often studied in the context of some form of periodic input, such as a regular beat in music (“beat-based expectations”). However, temporal expectations can be formed based on different types of structure in the environment, which need not necessarily be periodic. For example, temporal expectations can result from learning the relationship between a cue and a particular temporal interval or from learning (nonperiodic) sequences of temporal intervals (Nobre & van Ede, 2018). In the latter two cases, expectations rely on memory of absolute durations. We will refer to these as “memory-based expectations.” Note that, elsewhere, the terms duration-based timing, absolute timing, and interval-based timing have also been used (Breska & Ivry, 2018; Merchant & Honing, 2014; Teki, Grube, Kumar, & Griffiths, 2011).

Temporal expectations have been explained by entrainment models, such as dynamic attending theory (DAT). Such models propose that temporal expectations result from synchronization between internal oscillations and external rhythmic stimulation (Haegens & Zion-Golumbic, 2018; Henry & Herrmann, 2014; Large & Jones, 1999; Jones & Boltz, 1989). On a neural level, the internal oscillations can be thought of as fluctuations in low-frequency oscillatory activity or cortical excitability, such that the high-excitability phase of low-frequency neural oscillations coincides with the timing of expected events, facilitating their processing by increasing sensory gain (Schroeder & Lakatos, 2009; Lakatos, Karmos, Mehta, Ulbert, & Schroeder, 2008).

It is unclear whether entrainment, thought to underlie beat-based expectations, can account for memory-based expectations, which do not rely on periodic input (Rimmele, Morillon, Poeppel, & Arnal, 2018; Breska & Deouell, 2017b; Morillon, Schroeder, Wyart, & Arnal, 2016). Arguably, memory-based expectations can better be explained within a predictive processing framework. Notably, beat-based expectations, although often thought to be dependent on entrainment, have also been modeled using hierarchical predictive coding models (van der Weij, Pearce, & Honing, 2017; Forth, Agres, Purver, & Wiggins, 2016). In such models, temporal expectations reflect the learned probability of an event at a given time, similar to learning the probability of content (“what”; Koelsch, Vuust, & Friston, 2019). Thus, whether beat-based and memory-based temporal expectations are based on shared or separate underlying mechanisms—be it entrainment, leading to changes in sensory gain, or predictive processing based on learned probabilistic information—is presently a matter of active debate. In the current EEG study, we addressed this outstanding question by directly comparing the effects of beat-based and memory-based expectations on auditory processing and behavior.

Previously, beat-based and memory-based timing have been differentiated in terms of their occurrence in different species (Honing, Bouwer, Prado, & Merchant, 2018; Honing & Merchant, 2014), the neural networks involved (Teki et al., 2011), and in how they are affected by neuropsychological disorders (Breska & Ivry, 2018), suggesting that beat-based and memory-based expectations are subserved by separate mechanisms. However, some have argued for one integrated system for timing (Schwartze & Kotz, 2013; Teki, Grube, & Griffiths, 2012), based on neuropsychological evidence (Cope, Grube, Singh, Burn, & Griffiths, 2014), and for reasons of parsimony (Rimmele et al., 2018). Complicating this discussion, to date, many studies examining temporal expectations used isochronous sequences of events to elicit temporal expectations in general and beat-based expectations in particular (Breska & Ivry, 2018; Breska & Deouell, 2016; Arnal, Doelling, & Poeppel, 2015; Lawrance, Harper, Cooke, & Schnupp, 2014; Lakatos et al., 2013; Schwartze, Farrugia, & Kotz, 2013; Henry & Obleser, 2012; Rohenkohl et al., 2012; Teki et al., 2011). Isochronous sequences are fully predictable, both in terms of their absolute intervals and in terms of their ongoing periodicity, and therefore do not allow for differentiation between beat-based and memory-based expectations.

A handful of studies attempted to directly compare beat-based and memory-based expectations by using both isochronous stimuli and stimuli that were predictable but not periodic. In a behavioral experiment, responses to sequences that were isochronous (affording both beat-based and memory-based expectations), predictably speeding up or slowing down (affording only memory-based expectations), or with random timing (no expectations) were compared (Morillon et al., 2016). Beat-based expectations improved both perceptual sensitivity and response speed, whereas memory-based expectations only affected perceptual sensitivity, which was suggested to result from a special relationship between beat-based expectations and the motor system (Morillon et al., 2016). However, in another behavioral study, both beat-based and memory-based expectations improved response speed (Breska & Ivry, 2018) and phase coherence of delta oscillations, which is often used as a proxy for neural entrainment, was shown to be similarly enhanced by memory-based and beat-based expectations (Breska & Deouell, 2017b), suggesting either that phase coherence is not a good measure of entrainment, as suggested by the authors, or that entrainment is a general, rather than a context-specific, mechanism of temporal expectations (Rimmele et al., 2018). Note that in all three studies described above (Breska & Ivry, 2018; Breska & Deouell, 2017b; Morillon et al., 2016), whereas care was taken to design stimuli that elicited only memory-based but not beat-based expectations, responses to isochronous stimuli were used as a proxy for beat-based expectations. In addition to affording beat-based expectations, isochronous stimuli are, by definition, more predictable in terms of learning their intervals than any other type of memory-based sequence (e.g., only one interval needs to be learned in an isochronous sequence). Thus, in these studies, the effects of beat-based expectations were always confounded with increases in temporal predictability based on memory. In addition, in the latter two studies (Breska & Ivry, 2018; Breska & Deouell, 2017b), static visual stimuli were used. The auditory system has repeatedly been found to be superior over the visual system in eliciting temporal expectations in humans (Grahn, 2012; Grahn, Henry, & McAuley, 2011; Zarco, Merchant, Prado, & Mendez, 2009). Although it is possible to achieve equal synchronization performance (a measure of temporal expectations) with visual and auditory stimuli alike, this requires the use of moving, rather than static visual input (Iversen, Patel, Nicodemus, & Emmorey, 2015; Grahn, 2012). Thus, the stimuli used by previous studies to study beat-based expectations not only allowed for memory-based strategies to form expectations, they were arguably not optimal for creating beat-based expectations at all.

Here, to gain a better understanding of how beat-based and memory-based expectations may (differentially) influence early sensory processing, we orthogonally manipulated beat-based and memory-based expectations using auditory stimuli and examined their effects on both behavioral and auditory ERP responses. Effects on ERPs have not been studied before in this context and may provide a potentially fruitful way of differentiating between beat-based and memory-based expectations. Temporal expectations have been reported to both enhance and attenuate the auditory P1 and N1 responses. Enhancement of sensory responses at expected time points (Bouwer & Honing, 2015; Escoffier, Herrmann, & Schirmer, 2015; Fitzroy & Sanders, 2015; Hsu, Hämäläinen, & Waszak, 2013; Tierney & Kraus, 2013; Rimmele, Jolsvai, & Sussman, 2011) is in line with entrainment models of temporal expectations, which assume increased sensory gain at expected time points (Large & Jones, 1999). By contrast, attenuation of sensory responses at expected time points (Sherwell, Garrido, & Cunnington, 2017; Paris, Kim, & David, 2016; van Atteveldt et al., 2015; Sanabria & Correa, 2013; Schwartze et al., 2013; Lange, 2009) is in line with predictive models of brain function that assert more efficient processing of incoming information when predicted information is suppressed (Marzecová, Widmann, SanMiguel, Kotz, & Schröger, 2017; Schröger, Kotz, & SanMiguel, 2015; Schröger, Marzecová, & SanMiguel, 2015; Friston, 2005). Whether temporal expectations lead to enhancement or attenuation of sensory responses may depend on the type of temporal structure that affords an expectation. Periodic input, affording beat-based expectations, may lead to entrainment and thus increased neural excitability at expected time points, resulting in enhancement of early sensory responses (Haegens & Zion-Golumbic, 2018). By contrast, learned probabilistic information about timing, affording memory-based expectations, may lead to the suppression of predicted information, resulting in attenuation of early sensory responses. Thus, if based on separate mechanisms, one subserved by entrainment and one by predictive processing, beat-based and memory-based expectations may have opposing effects on sensory responses, enhancement, and attenuation of responses, respectively.

In addition to the effects of beat-based and memory-based expectations on behavioral and auditory responses, we examined the effects of task relevance on both types of expectations. Entrainment and beat-based processing have been shown to be somewhat independent of task relevance (Bouwer, Werner, Knetemann, & Honing, 2016; Bouwer, Van Zuijen, & Honing, 2014; Breska & Deouell, 2014; Rohenkohl, Coull, & Nobre, 2011). Predictive processing, however, has been shown to depend on and interact with task relevance in both the visual and the auditory domain (Hsu, Hämäläinen, & Waszak, 2018; Paris et al., 2016; Todorovic, Schoffelen, van Ede, Maris, & de Lange, 2015; Kok, Rahnev, Jehee, Lau, & de Lange, 2012). Thus, if beat-based expectations rely on entrainment while memory-based expectations rely on learning probabilistic information, they may be affected by task relevance differently.

In the current study, in two experiments, we compared responses to auditory sequences that were either periodic, affording beat-based expectations, or aperiodic, thus not affording beat-based expectations. Also, sequences could either consist of fully predictable temporal intervals, affording memory-based expectations, or unpredictable, randomly concatenated intervals. The responses to events in the rhythmic sequences thus depended on their expectedness, with expectations coming from two distinct sources: the periodicity of the sequence (beat-based) and the repetition of the pattern (memory-based). We not only examined responses for events on the beat (e.g., in phase with the periodicity, at expected times) but also off the beat (e.g., out of phase with the periodicity, at less expected times), as entrainment theories predict not only increased sensory gain at expected moments but also reduced sensory gain in between (Breska & Deouell, 2014). Thus, on the beat, when comparing periodic with aperiodic sequences, we could assess the facilitating effects of the presence of beat-based expectations, whereas off the beat, we could assess the effects of events occurring at times that were “mispredicted” in terms of the expected beat (Hsu, Le Bars, Hämäläinen, & Waszak, 2015).

In Experiment 1, we examined the effects of beat-based and memory-based expectations at two positions (on and off the beat) on behavioral performance, by measuring the speed and accuracy of the detection of targets in the form of rare softer tones in the rhythmic sequences. In Experiment 2, we additionally recorded ERP responses to all nontarget sounds to examine the effects of beat-based and memory-based expectations at two positions on P1 and N1 responses. Additionally, in Experiment 2, ERP responses were examined both when sequences were task relevant and task irrelevant.

In terms of behavioral outcomes, we expected faster and more accurate detection of targets with expected than unexpected timing, both for beat-based and memory-based expectations. Additionally, beat-based and memory-based expectations may interact in two distinct ways. First, the presence of beat-based expectations may lead to an increase in sensory gain for sounds on the beat, as predicted by entrainment models. Increased gain should increase the precision of memory-based predictions (Feldman & Friston, 2010), leading to enhanced effects of memory-based expectations in the presence of beat-based expectations. Thus, interacting effects of beat-based and memory-based expectations, which are larger than could be expected based on additivity, could indicate separate mechanisms, with beat-based expectations influencing the gain or precision of sensory processing through entrainment and memory-based expectations affecting the probabilistic predictions themselves. Alternatively, if beat-based and memory-based expectations rely on shared mechanisms, the simultaneous presence of both types of expectations may lead to interference, with smaller effects of either type when both need to be engaged. In terms of ERP responses, qualitative differences between the effects of beat-based and memory-based expectations—the former leading to enhancement and the latter to attenuation of sensory responses—could indicate different underlying mechanisms. However, if both types of expectations rely on similar mechanisms, be it based on entrainment or predictive processing, we expected them to affect sensory responses similarly.

METHODS

Experiment 1

Participants

Thirty-four participants (26 women), aged between 19 and 45 years (M = 24.6, SD = 5.7) with no history of neurological or hearing disorders, took part in the experiment. Data from two participants were removed because of technical problems, leaving 32 participants for the analysis. All participants provided written consent before the study, and participants were reimbursed with either a monetary fee or course credit. The study was approved by the ethics review board of the Faculty of Social and Behavioral Sciences of the University of Amsterdam. The statistical analysis of Experiment 1 was preregistered (https://aspredicted.org/az8kr.pdf).

Stimuli

We created sound patterns of five or six consecutive temporal intervals (Figure 1A), marked by identical woodblock sounds of 60 msec length, generated in GarageBand (Apple, Inc.). Patterns of five or six intervals are short enough to allow for learning of the temporal intervals (Schultz, Stevens, Keller, & Tillmann, 2013) and to not make too large demands on working memory (Grahn & Schuit, 2012). At the same time, with a total length of 1800 msec, patterns were long enough to avoid the perception of a regular, periodic beat when patterns were concatenated into sequences, as people do not readily perceive a beat with a period of 1800 msec (London, 2012). Patterns were concatenated into sequences of 128 patterns, with a final tone added to each sequence. Each sequence thus lasted for 3 min 51 sec.

Figure 1. 

Schematic overview of the stimuli. (A) Patterns of five (Patterns 1 and 2) and six (Patterns 3 and 4) temporal intervals were created. All patterns were 12 units length, equaling 1800 msec (1 unit = 150 msec). Periodic patterns (left, orange shades) consisted of intervals related by integer ratios 1:2:3:4. In periodic patterns, groups of intervals always added up to unit length 4, creating a beat with an interbeat onset of 600 msec (red dotted lines). Aperiodic patterns (right, purple shades) consisted of intervals related by noninteger ratios, not allowing for the perception of a regular beat. (B) Patterns were concatenated into sequences of either 128 identical patterns (predictable sequences) or 128 randomly chosen patterns (unpredictable sequences). For the unpredictable sequences, not only patterns starting at Beat 1 (Patterns 1.1, 2.1, 3.1, and 4.1) were used, but also cyclic permutations of these patterns, starting at Beats 2 or 3 (Patterns 1.2, 1.3, 2.2, etc.). For all analyses, only events after intervals with unit lengths 1 and 3 were used, to control for acoustic context.

Figure 1. 

Schematic overview of the stimuli. (A) Patterns of five (Patterns 1 and 2) and six (Patterns 3 and 4) temporal intervals were created. All patterns were 12 units length, equaling 1800 msec (1 unit = 150 msec). Periodic patterns (left, orange shades) consisted of intervals related by integer ratios 1:2:3:4. In periodic patterns, groups of intervals always added up to unit length 4, creating a beat with an interbeat onset of 600 msec (red dotted lines). Aperiodic patterns (right, purple shades) consisted of intervals related by noninteger ratios, not allowing for the perception of a regular beat. (B) Patterns were concatenated into sequences of either 128 identical patterns (predictable sequences) or 128 randomly chosen patterns (unpredictable sequences). For the unpredictable sequences, not only patterns starting at Beat 1 (Patterns 1.1, 2.1, 3.1, and 4.1) were used, but also cyclic permutations of these patterns, starting at Beats 2 or 3 (Patterns 1.2, 1.3, 2.2, etc.). For all analyses, only events after intervals with unit lengths 1 and 3 were used, to control for acoustic context.

Beat-based expectations.

For the periodic, beat-based patterns, temporal intervals were related by the integer ratios of 1:2:2:3:4 (five intervals) and 1:1:1:2:3:4 (six intervals). The shortest interval was set at 150 msec, leading to interonset intervals for the other intervals of 300, 450, and 600 msec. In the periodic patterns, temporal intervals were organized to form groups of four units in length (600 msec) and grouped such that a perceptually accented tone was present at the start of each group (Grahn & Brett, 2007; Povel & Essens, 1985). In these patterns, a beat could be perceived with an interbeat interval of 600 msec (100 BPM or 1.7 Hz), the optimal rate for human beat perception (London, 2012). These patterns could be regarded as strictly metric, with the periodicity of the pattern always being marked by a sound (Grahn & Brett, 2007). Each pattern consisted of 12 units in length or three beats of four units in length.

To create aperiodic equivalents of the beat-based patterns, we changed the ratios by which the temporal intervals were related. For the aperiodic patterns, intervals were related by noninteger ratios of 1:1.4:1.4:3:5.2 (five intervals) and 1:1:1:1.4:3:4.6 (six intervals). The aperiodic patterns were equal to their periodic counterparts in terms of length, grouping, and number of tones. However, the aperiodic patterns did not contain a periodic beat at unit length four. A pilot confirmed that aperiodic patterns were rated as less beat inducing than periodic patterns.

Note that it is impossible to create sequences of sounds that are not to some extent (quasi)periodic (Breska & Deouell, 2017a; Obleser, Henry, & Lakatos, 2017). However, the sequences clearly differed in the presence of periodicity at a rate that afforded beat-based expectations. First, in the aperiodic patterns, contrary to the periodic patterns, events did not align with the intended beat at a rate close to the ideal tempo for beat perception (1.6 Hz). This was also apparent from a spectral analysis of the waveforms (results not shown). Peaks at 1.6 Hz (the beat frequency) were larger for the beat-based sequences than the memory-based sequences. Second, although periodicity was present in the aperiodic sequences at the level of concatenated patterns (with a period of 1800 msec or 33 BPM or 0.6 Hz), this was too slow for humans to readily perceive a beat in (London, 2012). Finally, we performed an informal pilot experiment in which we asked 17 participants to rate on a scale from 1 to 10 how strongly they heard a beat in the aperiodic and periodic patterns. On average, each periodic pattern was rated as containing more beat (on average 9.3) than each aperiodic pattern (on average 6.7), confirming that our manipulation of periodicity indeed affected perception at the level of hearing a beat. Thus, although we are aware that the aperiodic patterns could be classified as quasi or weakly periodic, for clarity, we will refer to them as aperiodic.

Memory-based expectations.

Fully predictable sequences were created by concatenating 128 identical patterns into a sequence (Figure 1B). The surface structure of temporal intervals in these sequences could easily be predicted based on probabilistic information alone. Unpredictable sequences were created by concatenating 128 semirandomly chosen patterns. Patterns were chosen both from the original patterns, which were also used for the predictable sequences (patterns starting at Beat 1; see Figure 1), and from cyclic permutations of these (patterns starting at Beats 2 or 3; see Figure 1). The cyclic permutations were identical to the original patterns when looped (as in the predictable sequences), but not when concatenated in random order (as in the unpredictable sequences). Within an unpredictable sequence, only patterns with either five or six intervals were concatenated. The longest interval had a different length in the aperiodic patterns with five and six intervals (unit lengths 5.2 and 4.6, respectively). This was necessary to keep the overall length of each pattern identical. However, combining the two sets would lead to a larger number of possible temporal intervals in the aperiodic than periodic sequences, possibly confounding the effects of beat-based expectations with differences in entropy. Thus, to keep both the number of events per pattern (event density) and the number of possible intervals (entropy) identical between conditions, we did not combine the sets of rhythms with five and six intervals. Within unpredictable sequences, each pattern could occur maximally twice consecutively.

Position.

With manipulations of periodicity and predictability, we were able to compare responses to events that were expected based on a beat or on learned interval structure with responses that could not be predicted based on their timing. However, we also wanted to examine how beat-based expectations affected responses to events with unexpected timing (e.g., not unpredicted, but rather mispredicted; see also Hsu et al., 2018). Therefore, we not only probed events that were in phase with the periodicity (e.g., on the beat) but also events that were out of phase with the periodicity (e.g., off the beat; see Figure 1). The off-beat events fell at quarter phase (150 or 450 msec) relative to the beat. We did not include events at antiphase (300 msec), as possibly, participants could perceive an additional beat at the faster subdivision rate, making these events ambiguous in terms of their metrical salience.

We assumed that people would not perceive a beat in the aperiodic patterns and, therefore, that the distinction between on the beat and off the beat would be meaningless for these patterns in terms of temporal expectations. However, grouping effects could lead to differences in the perception of events on and off the beat. For example, the auditory system tends to group events in groups of two (Potter, Fenwick, Abecasis, & Brochard, 2009; Abecasis, Brochard, Granot, & Drake, 2005; Brochard, Abecasis, Potter, Ragot, & Drake, 2003), and perceptual accents are known to arise based on the surface structure of a rhythm (Povel & Essens, 1985). To be able to assess the effects of temporal expectations at different positions without confounding these with differences in grouping, we classified events in the aperiodic patterns as on-beat or off-beat, depending on their grouping in the periodic counterpart, even when we assumed no beat would be perceived in response to the aperiodic patterns (i.e., when we refer to an event as “on-beat” in an aperiodic pattern, it is an event that falls on the beat in the periodic equivalent and thus is matched to its periodic equivalent in terms of grouping).

Targets.

In the behavioral task, temporal expectations were probed implicitly by introducing infrequent intensity decrements as targets. Based on previous experiments, we expected that temporal expectations would improve the detection of these targets (Bouwer et al., 2014, 2016; Bouwer & Honing, 2015; Potter et al., 2009). Intensity decrements of 6 dB were used (Bouwer & Honing, 2015). In each sequence of 128 patterns, 32 patterns (25%) contained a target. Half of the targets appeared on the beat, and half of the targets appeared off the beat. In each sequence, 26 targets were in positions after temporal intervals with unit lengths 1 and 3, present in both periodic and aperiodic patterns. Only these targets were used for the analysis to equate their acoustic context. At least two standard patterns separated a pattern containing a target.

Procedure

A total of 16 sequences were presented to each participant, four of each type. Sequences of different types were semirandomized, with each type appearing once every four sequences and, therefore, a maximum of two sequences of the same type in a row. Upon arrival, participants completed a consent form and were allowed to practice the task. They were instructed to avoid movement, listen to the rhythm carefully, and press a button as fast as possible when they heard a target. Participants were allowed breaks between sequences. An entire experimental session lasted for about 2 hr. Participants were tested individually in a dedicated lab at the University of Amsterdam. Sounds were presented at 70 dB SPL with one Logitech speaker positioned in front of the participants, using Presentation software (Version 19.0, www.neurobs.com).

Data Analysis

In total, each participant was presented with 52 targets for each condition that was included in the analysis. All responses made within 2000 msec of a target were recorded. Responses faster than 150 msec were discarded, as were responses that were more than 2.5 SDs from the mean of a participant's RT within each condition. Removal of outliers led to the exclusion of 2.9% of the responses in Experiment 1 and 3.1% of the responses in Experiment 2. Participants who did not achieve a hit rate of higher than 50% in any of the conditions were excluded from the analysis. In Experiment 1, on this ground, two participants were excluded, leaving 30 participants for the analysis of hit rates. One additional participant was excluded for the analysis of RTs, as this participant had less than five valid RTs in one condition (i.e., less than 5 out of a possible 52 targets were hits). In Experiment 2, no participants were excluded from the analysis. Hit rates for each condition and participant and average RTs for each condition and participant were entered into three-way repeated-measures ANOVAs, with Periodicity (periodic, aperiodic), Predictability (predictable, unpredictable), and Position (on the beat, off the beat) as within-subject factors.

The meaning of the factor position can be regarded as somewhat ambiguous. In the beat-based sequences, differences between on-beat and off-beat positions could be due to both differences in grouping and differences in metrical position (assuming that participants would hear a beat in these sequences). In the memory-based sequences, grouping differences could still affect the results for the factor position, whereas there would presumably not be an effect of metrical position. Thus, the meaning of the factor position depends on the periodicity of a sequence, potentially limiting the conclusions that can be drawn from an ANOVA that assumes this factor to be orthogonal to the other factors of interest. Therefore, in addition to the (preregistered) full ANOVA, we also ran separate ANOVAs for on-beat and off-beat positions, to check whether our results would hold up when taking into account the possible confounded position factor. For all ANOVAs, for significant interactions (p < .05), post hoc tests of simple effects were performed. For all simple effects, we report uncorrected p values. Effect sizes are reported as partial eta squared. All statistical analyses were performed in SPSS 24.

Experiment 2

Participants

Thirty-two participants (22 women), aged between 19 and 44 years (M = 23.4, SD = 4.9) with no history of neurological or hearing disorders, took part in Experiment 2. Data of one participant were removed because of excess noise in the EEG signal, leaving 31 participants for the analysis. All participants provided written consent before the study, and participants were reimbursed with either a monetary fee or course credit. The study was approved by the ethics review board of the Faculty of Social and Behavioral Sciences of the University of Amsterdam.

Stimuli and Procedure

The materials and procedure for Experiment 2 were identical to those for Experiment 1. In Experiment 2, participants additionally completed an unattended version of the experiment. In the unattended condition, participants were asked to ignore the rhythmic sequences and focus on a self-selected muted movie, rendering the stimuli task irrelevant. All participants first completed the unattended EEG experiment and subsequently the attended EEG experiment. The EEG recording lasted for around 2 hr. Participants were encouraged to take breaks if needed. During the recording, data quality was assessed online by the experimenter, and if needed, channels were refitted and extra electrode gel was applied. For Experiment 2, one experimental session was about 4 hr, including breaks, practice, and setting up equipment.

EEG Recording

EEG was recorded using a 64-channel Biosemi Active-Two acquisition system, with a standard 10/20 configuration and additional electrodes for EOG channels, on the nose, on both mastoids, and on both earlobes. The EEG signal was recorded at 1 kHz.

EEG Analysis

Preprocessing was performed in MATLAB (The MathWorks, Inc) and EEGLAB (Delorme & Makeig, 2004). Data were offline re-referenced to linked mastoids, bad channels were removed, and independent component analysis was used to remove eye blinks. Subsequently, bad channels were replaced by values interpolated from the surrounding channels. Visual inspection of the ERPs revealed a postauricular muscle (PAM) response in several participants. The auditory evoked potential can be easily contaminated by the PAM response (Bell, Smith, Allen, & Lutman, 2004; Picton, Hillyard, Krausz, & Galambos, 1974). To avoid contamination, we re-referenced the data to earlobes for all further analyses. For completeness, we also report results from the mastoid-referenced data.

Data were offline down-sampled to 512 Hz and filtered using 0.1-Hz high-pass and 40-Hz low-pass finite impulse response filters. Epochs for each condition separately were extracted for nontarget sounds, from 200 msec preceding the onset of each event till 500 msec after the onset of each event. Only epochs for events following an interval of unit length 1 or unit length 3 (150 and 450 msec, respectively) were included to equate the acoustic context of events used in the analysis. Also, the first 12 sounds of each sequence (two whole patterns) were excluded from the analysis to allow for the build-up of expectations. Epochs with a voltage change of more than 150 μV in a 200-msec sliding window were rejected from further analysis. For each condition and participant, epochs were averaged to obtain ERPs and baseline-corrected using the average voltage of the 50 msec window preceding each sound. Finally, ERPs were averaged over participants to obtain grand-averaged waveforms.

Peak latencies for the P1 and N1 responses were determined independent from the statistical analysis, from the average waveform collapsed over all conditions. P1 peaked at 58 msec after tone onset. We defined P1 amplitude as the average amplitude in a 20-msec window around the peak (48–68 msec). N1 peaked at 124 msec and was more distributed in time. Thus, we defined N1 amplitude as the average amplitude from a 40-msec window around the peak (104–144 msec). Auditory evoked potentials are known to be maximal over frontocentral electrodes (Ruhnau, Herrmann, Maess, & Schröger, 2011; Picton et al., 1974), which was also observed in the current data set. Therefore, ERP amplitudes were computed from the average of a cluster of 15 frontocentral electrodes: F3, F1, Fz, F2, F4, FC3, FC1, FCz, FC2, FC4, C3, C1, Cz, C2, and C4. All statistics and figures reported here are based on the average amplitude from this ROI.

Statistical Analysis

Amplitudes for P1 and N1 were entered into repeated-measures ANOVAs, with four within-subject factors: Periodicity (periodic, aperiodic), Predictability (predictable, unpredictable), Position (on the beat, off the beat), and Attention (attended, unattended). As for the behavioral results, we also included separate ANOVAs for on-beat and off-beat positions to account for the possible collinearity of position with periodicity. For significant interactions (p < .05), subsequent tests of simple effects were performed. For all simple effects, we report uncorrected p values. Effect sizes are reported as partial eta squared. Analyses were performed in SPSS 24.

Comparing Beat-based and Memory-based Expectations Directly

Although the factorial design included both periodicity and predictability as factors, the ANOVAs do not allow for a direct comparison of their effects. To directly compare the effects of beat-based and memory-based expectations, we performed two additional analyses. In both analyses, the effect of beat-based expectations was quantified as the difference between responses on the beat in periodic and aperiodic sequences. The effect of memory-based expectations was quantified as the difference between responses in predictable and unpredictable sequences. For the latter, we only included responses on the beat to make sure that possible differences between beat-based and memory-based expectations could not be attributed to differences in grouping.

First, we compared the effects of beat-based and memory-based expectations in the P1 and N1 windows directly using paired-samples t tests. To quantify the possibility that no differences were present, we performed both traditional and Bayesian t tests using JASP (JASP Team, 2019; Wagenmakers et al., 2018). We estimated Bayes factors using a Cauchy prior distribution (r = .71), with as a null hypothesis no differences between the effects of beat-based and memory-based expectations. In addition, we performed a robustness check as implemented in JASP to assess whether the results would change with a different prior (r = 1) as proposed originally for Bayesian t tests (Wagenmakers et al., 2018; Jeffreys, 1961).

Second, we directly compared the effects of memory-based and beat-based expectations using cluster-based permutation tests. With this approach we could examine potential differences at all time points and at all electrodes while taking into account the multiple comparisons along both the spatial and time axes. As ERP components often overlap in time, their real peaks may be obscured in grand-averaged waveforms (Luck, 2005). The use of cluster-based permutation testing allowed us to make sure we did not miss potential differences between beat-based and memory-based expectations by selectively examining peak time windows and selected clusters of electrodes. Cluster-based permutation tests were performed using the Fieldtrip toolbox (Oostenveld, Fries, Maris, & Schoffelen, 2011). All time points in the 150 msec time window following the onset of sound events were included. We limited the analysis to the first 150 msec to avoid contamination of subsequent sounds, which in some but not all conditions could occur 150 msec following an event. Clusters were formed based on adjacent time–electrode samples that survived a statistical threshold of p < .01 when comparing the conditions of interest with dependent samples t tests. Clusters were subsequently evaluated with permutation tests using 2000 iterations.

RESULTS

Behavioral Results

Hit rates and RTs for all conditions from both Experiments 1 and 2 are shown in Tables 1 and 2 and Figure 2. In Experiment 2, all results from Experiment 1 were replicated with comparable effect sizes.

Table 1. 
Hit Rates for All Conditions in Experiments 1 and 2
 Experiment 1 (n = 30)Experiment 2 (n = 31)
PeriodicAperiodicEffect of Beat-based ExpectationsPeriodicAperiodicEffect of Beat-based Expectations
Beat 
Predictable 82 (14) 81 (15) +1 79 (17) 79 (17) 
Unpredictable 73 (19) 69 (23) +4 72 (17) 67 (20) +5** 
Effect of memory-based expectations +9** +12**   +7** +12**   
  
Offbeat 
Predictable 63 (20) 68 (21) −5* 64 (21) 72 (19) −8** 
Unpredictable 52 (22) 61 (26) −9** 51 (20) 60 (21) −9** 
Effect of memory-based expectations +11** +7*   +13** +12**   
 Experiment 1 (n = 30)Experiment 2 (n = 31)
PeriodicAperiodicEffect of Beat-based ExpectationsPeriodicAperiodicEffect of Beat-based Expectations
Beat 
Predictable 82 (14) 81 (15) +1 79 (17) 79 (17) 
Unpredictable 73 (19) 69 (23) +4 72 (17) 67 (20) +5** 
Effect of memory-based expectations +9** +12**   +7** +12**   
  
Offbeat 
Predictable 63 (20) 68 (21) −5* 64 (21) 72 (19) −8** 
Unpredictable 52 (22) 61 (26) −9** 51 (20) 60 (21) −9** 
Effect of memory-based expectations +11** +7*   +13** +12**   

Hit rates are reported as percentage hits. Standard deviations are in parentheses.

*

Simple effect significant at p < .05 (uncorrected).

**

Simple effect significant at p < .01 (uncorrected).

Table 2. 
RTs (in msec) for All Conditions in Experiments 1 and 2
 Experiment 1 (n = 29)Experiment 2 (n = 31)
PeriodicAperiodicEffect of Beat-based ExpectationsPeriodicAperiodicEffect of Beat-based Expectations
Beat 
Predictable 538 (78) 541 (78) −3 517 (59) 509 (60) +8 
Unpredictable 557 (74) 571 (80) −14 542 (66) 554 (75) −12 
Effect of memory-based expectations −19* −30**   −25** −45**   
  
Offbeat 
Predictable 573 (84) 571 (81) +2 550 (93) 531 (76) +19 
Unpredictable 601 (92) 591 (96) +10 569 (81) 571 (96) −2 
Effect of memory-based expectations −28* −20*   −19 −40**   
 Experiment 1 (n = 29)Experiment 2 (n = 31)
PeriodicAperiodicEffect of Beat-based ExpectationsPeriodicAperiodicEffect of Beat-based Expectations
Beat 
Predictable 538 (78) 541 (78) −3 517 (59) 509 (60) +8 
Unpredictable 557 (74) 571 (80) −14 542 (66) 554 (75) −12 
Effect of memory-based expectations −19* −30**   −25** −45**   
  
Offbeat 
Predictable 573 (84) 571 (81) +2 550 (93) 531 (76) +19 
Unpredictable 601 (92) 591 (96) +10 569 (81) 571 (96) −2 
Effect of memory-based expectations −28* −20*   −19 −40**   

Standard deviations are in parentheses.

*

Simple effect significant at p < .05 (uncorrected).

**

Simple effect significant at p < .01 (uncorrected).

Figure 2. 

Hit rates (top) and RTs (bottom) show improved target detection for events at expected time points. Memory-based expectations lead to better and faster target detection in Experiment 1 (left) and Experiment 2 (right). Beat-based expectations only lead to improvements in the absence of memory-based expectations in Experiment 2, but consistently lead to deteriorated detection of targets off the beat in both experiments.

Figure 2. 

Hit rates (top) and RTs (bottom) show improved target detection for events at expected time points. Memory-based expectations lead to better and faster target detection in Experiment 1 (left) and Experiment 2 (right). Beat-based expectations only lead to improvements in the absence of memory-based expectations in Experiment 2, but consistently lead to deteriorated detection of targets off the beat in both experiments.

Memory-based Expectations

In general, targets were detected more often and faster in predictable than in unpredictable sequences, as reflected in a significant main effect of Predictability for hit rates (Experiment 1: F(1, 29) = 39.4, p < .001, ηp2 = .58; Experiment 2: F(1, 30) = 46.4, p < .001, ηp2 = .61) and RTs (Experiment 1: F(1, 28) = 19.6, p < .001, ηp2 = .41; Experiment 2: F(1, 30) = 25.0, p < .001, ηp2 = .46). The simple effect of Predictability was significant for all comparisons for hit rates (all uncorrected ps < .016; see Table 1) and RTs (all uncorrected ps < .023, except for targets off the beat in periodic sequences in Experiment 2; p = .055). Thus, as expected, memory-based expectations led to improved detection of targets at expected time points. Results from the separate ANOVAs for on-beat and off-beat positions confirmed the main effect of Predictability for hit rates (Experiment 1, on-beat: F(1, 29) = 31.0, p < .001, ηp2 = .52; off-beat: F(1, 29) = 27.29, p < .001, ηp2 = .49; Experiment 2, on-beat: F(1, 30) = 35.68, p < .001, ηp2 = .54; off-beat: F(1, 30) = 34.86, p < .001, ηp2 = .54) and RTs (Experiment 1, on-beat: F(1, 28) = 16.6, p < .001, ηp2 = .37; off-beat: F(1, 28) = 12.4, p = .001, ηp2 = .31; Experiment 2, on-beat: F(1, 30) = 39.44, p < .001, ηp2 = .57; off-beat: F(1, 30) = 10.27, p = .003, ηp2 = .26). Figure 3 (top) shows the main effect of Predictability on target detection performance collapsed over the levels of periodicity and position.

Figure 3. 

Temporal expectations improve target detection. Memory-based expectations (A) led to both higher hit rates and faster RTs (main effect of Predictability). Beat-based expectations mainly affected hit rates by deteriorating detection for mispredicted events (off-beat; bottom of B). Thick lines represent the mean, and thin lines are individual participants.

Figure 3. 

Temporal expectations improve target detection. Memory-based expectations (A) led to both higher hit rates and faster RTs (main effect of Predictability). Beat-based expectations mainly affected hit rates by deteriorating detection for mispredicted events (off-beat; bottom of B). Thick lines represent the mean, and thin lines are individual participants.

Beat-based Expectations

In Experiment 2, we found a main effect of Periodicity on hit rates (F(1, 30) = 6.59, p = .016, ηp2 = .18). However, as expected, the effect of periodicity on hit rates in both experiments depended on the position of the target. On the beat, targets were detected more often in periodic than aperiodic sequences, whereas off the beat, targets were detected more often in aperiodic than periodic sequences, as reflected in a significant interaction in the overall ANOVA between Periodicity and Position (Experiment 1: F(1, 29) = 35.0, p < .001, ηp2 = .54; Experiment 2: F(1, 30) = 39.3, p < .001, ηp2 = .57). In line with the results for hit rates, numerically, detection of targets off the beat, but not on the beat, was slower in periodic than aperiodic sequences. However, the interaction between Periodicity and Position did not reach significance for the RTs (Experiment 1: F(1, 28) = 3.74, p = .063, ηp2 = .12; Experiment 2: F(1, 30) = 1.91, p = .18, ηp2 = .06). The separate ANOVAs for on-beat and off-beat positions showed that, although the facilitating effect of Periodicity on the beat did not reach significance for hit rates nor for RTs in either experiment (all ps > .11), the decrease in performance for targets presented out of phase with the periodicity (e.g., mispredicted) was significant for hit rates, apparent from a main effect of Periodicity for off-beat targets (Experiment 1: F(1, 29) = 19.71, p < .001, ηp2 = .41; Experiment 2: F(1, 30) = 30.28, p < .001, ηp2 = .50). Figure 3 (bottom) shows the main effect of Periodicity on target detection performance separate for on-beat and off-beat targets and collapsed over the levels of predictability.

Interaction between Beat-based and Memory-based Expectations

The predictability and periodicity of the sequences also affected hit rates in interaction, though depending on the position of the target, as reflected in a significant three-way interaction in the overall ANOVA between Periodicity, Position, and Predictability (Experiment 1: F(1, 29) = 6.69, p = .015, ηp2 = .19; Experiment 2: F(1, 30) = 5.59, p = .025, ηp2 = .16) and a significant two-way interaction between Periodicity and Predictability in the separate ANOVA for on-beat targets (Experiment 2: F(1, 30) = 6.67, p = .015, ηp2 = .18). Off the beat, the simple effect of Periodicity was significant for all comparisons (all uncorrected ps < .029). Thus, beat-based expectations led to decreased detection of targets when they were presented out of phase with the periodicity and therefore occurred at time points that were mispredicted with regard to the beat. On the beat, the simple effect of Periodicity only reached significance for unpredictable sequences and only in Experiment 2 (uncorrected p = .003), showing improved detection through beat-based expectations only in the absence of memory-based expectations. In Experiment 2, we also found an interaction between Periodicity and Predictability for RTs, with a stronger effect of Predictability in aperiodic than periodic sequences (F(1, 30) = 6.67, p = .015, ηp2 = .18). The interaction between Periodicity and Predictability did not reach significance in the separate ANOVAs split over position for RTs (all ps > .06).

In addition to effects of memory-based and beat-based expectations, in the overall ANOVA, we found a main effect of Position for hit rates (Experiment 1: F(1, 29) = 99.9, p < .001, ηp2 = .78; Experiment 2: F(1, 30) = 100.3, p < .001, ηp2 = .77) and RTs (Experiment 1: F(1, 28) = 37.1, p < .001, ηp2 = .57; Experiment 2: F(1, 30) = 14.4, p = .001, ηp2 = .32). Targets on the beat were detected more often and faster than targets off the beat, even in aperiodic sequences, most likely reflecting differences in surface grouping.

Thus, behaviorally, memory-based expectations improved target detection both in terms of response accuracy and response speed (see Figure 3). Beat-based expectations similarly led to improved target detection for targets at expected time points (on-beat), but only when no memory-based expectations were present and only in terms of response accuracy, not response speed. The improvements in target detection afforded by beat-based expectations were thus small and dependent on memory-based expectations. In contrast, at unexpected time points (off-beat, “mispredicted”), beat-based expectations led to decrements in performance, both in the absence and presence of memory-based expectations. Interestingly, although we found an interaction between beat-based and memory-based expectations, it was in the opposite direction of what we expected: Rather than enhancing each other, when both types of expectations were present, their effects were diminished.

EEG Results

Figure 4 shows the auditory evoked potentials for all conditions. Average amplitudes as extracted from the P1 and N1 time windows and frontocentral ROI are depicted in Figure 5.

Figure 4. 

Temporal expectations reduce auditory responses. Temporally expected events evoked smaller auditory P1 and N1 responses, regardless of whether they were attended (task-relevant) or not. Grand-averaged waveforms are shown for each condition separately. The time windows used for the ANOVA analyses of P1 and N1 are indicated with rectangles. Both memory-based and beat-based expectations attenuated P1 and N1 responses to expected events. N1 responses were additionally enhanced for events off the beat in periodic sequences, that is, when they were unexpected with regard to the periodicity.

Figure 4. 

Temporal expectations reduce auditory responses. Temporally expected events evoked smaller auditory P1 and N1 responses, regardless of whether they were attended (task-relevant) or not. Grand-averaged waveforms are shown for each condition separately. The time windows used for the ANOVA analyses of P1 and N1 are indicated with rectangles. Both memory-based and beat-based expectations attenuated P1 and N1 responses to expected events. N1 responses were additionally enhanced for events off the beat in periodic sequences, that is, when they were unexpected with regard to the periodicity.

Figure 5. 

Temporal expectations reduce auditory responses. Amplitudes of P1 and N1 averaged over the time windows used for analysis (see Figure 4) and frontocentral ROI are shown.

Figure 5. 

Temporal expectations reduce auditory responses. Amplitudes of P1 and N1 averaged over the time windows used for analysis (see Figure 4) and frontocentral ROI are shown.

Memory-based Expectations

As expected, both P1 and N1 responses were smaller for predictable than unpredictable events, reflected in a main effect of Predictability in the overall ANOVA (P1: F(1, 30) = 10.3, p = .003, ηp2 = .26; N1: F(1, 30) = 4.52, p = .042, ηp2 = .13), showing attenuation through memory-based expectations. Although significant in the overall ANOVA, the attenuating effect of Predictability on the P1 and N1 responses did not reach significance for the separate on-beat ANOVAs (both ps > .12) and only for the P1 for the off-beat positions (F(1, 30) = 12.3, p = .001, ηp2 = .29), possibly due to a lack of power when splitting up the data. The overall ANOVA suggested that the effect of Predictability did not depend on task relevance (e.g., none of the interactions including Predictability and Attention reached significance: all ps > .09). However, off-beat, the interaction between Attention and Predictability did reach significance (F(1, 30) = 4.86, p = .035, ηp2 = .14). Figure 6 shows a summary of the main effects of memory-based expectations on auditory-evoked potentials, collapsed over the different levels of position and periodicity. The attenuation of the auditory P1 and N1 responses associated with memory-based expectations suggests processes related to suppression of predicted information, rather than memory-based expectations leading to changes in sensory gain. Numerically, for the P1 response, the effects of memory-based expectations were larger in attended than unattended conditions. Although statistically the effects of memory-based expectations were independent for task relevance when including all data in one ANOVA, the split ANOVA for off-beat positions suggests that, possibly, task relevance does affect the effects of memory-based expectations, at least for the P1.

Figure 6. 

Memory-based expectations attenuated auditory responses. Auditory evoked potentials showing the main effect of Predictability on P1 and N1 responses are shown on the top. The main effect of Predictability is depicted as the difference between responses to events in predictable and unpredictable sequences, collapsed over the levels of periodicity and position (i.e., independent of beat-based expectations). In both attended and unattended conditions, P1 and N1 responses were larger for unpredictable than for predictable events. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top.

Figure 6. 

Memory-based expectations attenuated auditory responses. Auditory evoked potentials showing the main effect of Predictability on P1 and N1 responses are shown on the top. The main effect of Predictability is depicted as the difference between responses to events in predictable and unpredictable sequences, collapsed over the levels of periodicity and position (i.e., independent of beat-based expectations). In both attended and unattended conditions, P1 and N1 responses were larger for unpredictable than for predictable events. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top.

Beat-based Expectations

Figures 7 and 8 show a summary of the effects of beat-based expectations on auditory-evoked potentials, collapsed over the different levels of predictability and separate for responses to events on the beat (Figure 7) and off the beat (Figure 8). Unexpectedly, for P1, the effect of Periodicity did not depend on the position of an event (no interaction between Periodicity and Position: F(1, 30) = 0.036, p = .85, ηp2 = .001). Instead, we found a main effect of Periodicity (F(1, 30) = 11.3, p = .002, ηp2 = .27), with smaller P1 responses to events in periodic than aperiodic sequences, significant both on-beat (F(1, 30) = 5.06, p = .032, ηp2 = .14) and off-beat (F(1, 30) = 6.72, p = .015, ηp2 = .18). For the N1 response, we did find a significant interaction between Periodicity and Position (F(1, 30) = 12.7, p = .001, ηp2 = .30). On the beat, N1 responses were smaller to events in periodic than aperiodic sequences (though only marginally so for the N1: F(1, 30) = 3.99, p = .055, ηp2 = .12). Off the beat, N1 responses were larger to events in periodic than aperiodic sequences (F(1, 30) = 17.32, p < .001, ηp2 = .37).

Figure 7. 

Beat-based expectations attenuated auditory responses (on-beat). Auditory evoked potentials showing the effect of Periodicity on the beat on P1 and N1 responses are shown on the top. The effect of Periodicity is depicted as the difference between responses to on-beat events in periodic and aperiodic sequences, collapsed over the levels of predictability (i.e., independent of memory-based expectations). In both attended and unattended conditions, P1 and N1 responses were smaller in amplitude for periodic than aperiodic events on the beat. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top. Note that here, the confidence intervals around ERP waveforms represent the variability of the difference waves, in which most of the between-subject variance has been eliminated through subtraction.

Figure 7. 

Beat-based expectations attenuated auditory responses (on-beat). Auditory evoked potentials showing the effect of Periodicity on the beat on P1 and N1 responses are shown on the top. The effect of Periodicity is depicted as the difference between responses to on-beat events in periodic and aperiodic sequences, collapsed over the levels of predictability (i.e., independent of memory-based expectations). In both attended and unattended conditions, P1 and N1 responses were smaller in amplitude for periodic than aperiodic events on the beat. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top. Note that here, the confidence intervals around ERP waveforms represent the variability of the difference waves, in which most of the between-subject variance has been eliminated through subtraction.

Figure 8. 

Beat-based expectations enhanced N1 responses for unexpected sounds (off-beat). Auditory evoked Potentials showing the effect of Periodicity off the beat on P1 and N1 responses are shown on the top. The effect of Periodicity is depicted as the difference between responses to off-beat events in periodic and aperiodic sequences, collapsed over the levels of predictability (i.e., independent of memory-based expectations). In both attended and unattended conditions, N1 responses were larger in amplitude for periodic than aperiodic events off the beat, whereas P1 responses were larger for aperiodic events. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top. Note that here, the confidence intervals around ERP waveforms represent the variability of the difference waves, in which most of the between-subject variance has been eliminated through subtraction.

Figure 8. 

Beat-based expectations enhanced N1 responses for unexpected sounds (off-beat). Auditory evoked Potentials showing the effect of Periodicity off the beat on P1 and N1 responses are shown on the top. The effect of Periodicity is depicted as the difference between responses to off-beat events in periodic and aperiodic sequences, collapsed over the levels of predictability (i.e., independent of memory-based expectations). In both attended and unattended conditions, N1 responses were larger in amplitude for periodic than aperiodic events off the beat, whereas P1 responses were larger for aperiodic events. On the bottom, P1 and N1 amplitudes are shown for all participants separately, with the mean plotted on top. Note that here, the confidence intervals around ERP waveforms represent the variability of the difference waves, in which most of the between-subject variance has been eliminated through subtraction.

Like for memory-based expectations, the effects of beat-based expectations were statistically independent of task relevance in the overall ANOVA (e.g., none of the interactions including Periodicity and Attention reached significance: all ps > .06). Interestingly, similar to the effects of memory-based expectations, though statistically independent of task relevance, the effects of beat-based expectations on the P1 response were numerically driven by the attended condition. Indeed, like for memory-based expectations, when splitting up the ANOVA over the two positions, a significant interaction between Attention and Periodicity was present on the beat for the P1 response (F(1, 30) = 5.86, p = .022, ηp2 = .16), indicating a small, but significant influence of task relevance on the attenuation of the P1 response caused by beat-based expectations.

Smaller responses to events in periodic (i.e., beat-based expected) than aperiodic (no beat-based expectations present) sequences, as we found for both the P1 and N1 responses (though the latter did not reach significance), could indicate attenuation of expected events through beat-based expectations, in line with predictive processing, and similar to the effects of memory-based expectations on the P1 and N1 response. In line with this, N1 responses were smallest for events that were expected (i.e., on-beat, in phase with the periodicity), largest for events that were unexpected (i.e., off-beat, “mispredicted,” out of phase with the periodicity), with the amplitude of the N1 to events that were neither expected nor unexpected (i.e., in aperiodic sequences where no beat-based expectations were present) in between. Attenuation of responses to expected events and enhancement of responses to unexpected events suggest that first-order predictions (cf. Koelsch et al., 2019), rather than changes in sensory gain, underlie the effects of beat-based expectations on perception, similar to the effects of memory-based expectations. However, off the beat, we also found smaller P1 responses to events in periodic than aperiodic sequences, in line with resource withdrawal and smaller responses off the beat in the presence of an underlying periodic beat, as proposed by DAT. This divergent result for P1 responses off the beat is hard to interpret and warrants further research.

In addition to the effects of memory-based and beat-based expectations, P1 responses were larger in amplitude for attended than unattended events (main effect of attention: F(1, 30) = 74.3, p < .001, ηp2 = .71, present both on-beat: F(1, 30) = 39.03, p < .001, ηp2 = .57, and off-beat: F(1, 30) = 25.85, p < .001, ηp2 = .46). Finally, we found a main effect of position on N1 responses (F(1, 30) = 10.3, p = .003, ηp2 = .26). However, the effect of position depended on attention (interaction between position and attention: F(1, 30) = 15.9, p < .001, ηp2 = .35). In the attended condition, N1 responses to events off the beat were larger than to events on the beat (p < .001), whereas in the unattended condition, there was no difference between responses on and off the beat (p = .74). Likely, these effects of position, similar to the effects in the behavioral experiments, were due to grouping differences. In line with the interaction between Attention and Position in the overall ANOVA, in the separate ANOVAs for the two levels of position, the effect of Attention was significant for off-beat positions (F(1, 30) = 10.0, p = .004, ηp2 = .25), with larger responses in the attended than the unattended condition and, curiously, for on-beat positions (F(1, 30) = 6.09, p = .02, ηp2 = .17), for which the responses in the unattended condition were slightly larger than in the attended condition.

To check whether observed results were not confounded by our choice of earlobe reference, we repeated the analysis with mastoid reference, as is more customary in auditory ERP analyses. Note that these results may have been confounded with the PAM response, which was present to some degree in several participants. With mastoid reference, results generally were the same as with earlobe reference. However, with mastoid reference, we found a four-way interaction between Attention, Predictability, Periodicity, and Position for the P1 (F(1, 30) = 5.18, p = .030, ηp2 = .15). This four-way interaction did not reach significance for the data referenced to earlobes (F(1, 30) = 2.64, p = .11, ηp2 = .081). To pursue the interaction in the mastoid-referenced data, we split the data between attended and unattended conditions. In the unattended condition, no significant effects were present. In the attended condition, we found a three-way interaction between Periodicity, Predictability, and Position (F(1, 30) = 4.50, p = .04, ηp2 = .13). The effect of Periodicity on the P1 response was significant only for events off the beat in predictable sequences (uncorrected p = .005) and for events on the beat in unpredictable sequences (uncorrected p = .005). Although there were thus small differences in significance between the data sets, the main findings—the effect of Predictability for both P1 and N1 and interaction between Periodicity and Position for N1—were robust to changes in reference.

Comparing Beat-based and Memory-based Expectations Directly

Finally and crucial to our main question, we compared the effects of memory-based and beat-based temporal expectations on auditory-evoked responses directly. We subtracted responses to events on the beat in unpredictable from predictable sequences to index the effects of memory-based expectations, and we subtracted responses to events on the beat in aperiodic from periodic sequences to index the effects of beat-based expectations. Subsequently, we compared the results of this subtraction procedure, both in the restricted time windows around the P1 and N1 peaks, using Bayesian and traditional t tests, and in a window from 0 to 150 msec after the onset of sound events using cluster-based permutation testing. As the ANOVA yielded somewhat inconclusive results regarding the effects of Task Relevance, here we report analyses split over attended and unattended conditions, as well as analyses in which we collapsed over attention conditions.

As is visible in Figure 9, the difference waves indexing the two different types of temporal expectations examined have a very similar morphology. Indeed, traditional t tests showed no significant differences between the effects of Predictability and Periodicity in the P1 window (attended: t(30) = 1.24, p = .23; unattended: t(30) = −0.88, p = .39; combined: t(30) = 0.33, p = .74), nor in the N1 window (attended: t(30) = 0.57, p = .57; unattended: t(30) = −0.97, p = .34; combined: t(30) = −0.35, p = .73). The Bayes factors indicated evidence in favor of the null hypothesis (no difference between conditions) for the P1 window (attended: BF01 = 2.6; unattended: BF01 = 3.7; combined: BF01 = 5.0) and the N1 window (attended: BF01 = 4.5; unattended: BF01 = 3.4; combined: BF01 = 4.9). In other words, the data were 2.6–5.0 times more likely to occur under the null hypothesis than in the presence of differences between the effects of periodicity and predictability. Bayes factors of >3 are regarded as moderate evidence in favor of the null hypothesis (Wagenmakers et al., 2018). The robustness check indicated that the results did not change as a function of the prior used. With the more traditional prior (r = 1), Bayes factors in favor of the null hypothesis ranged from 3.5 to 6.8.

Figure 9. 

Beat-based and memory-based expectations similarly affected auditory stimulus processing. Effects of beat-based expectations are defined as the difference between responses to events on the beat in periodic and aperiodic sequences. Effects of memory-based expectations are defined as the difference between responses to events on the beat predictable and unpredictable sequences. When comparing the effects of beat-based and memory-based expectations directly using t tests and cluster-based permutation tests, no significant differences were found.

Figure 9. 

Beat-based and memory-based expectations similarly affected auditory stimulus processing. Effects of beat-based expectations are defined as the difference between responses to events on the beat in periodic and aperiodic sequences. Effects of memory-based expectations are defined as the difference between responses to events on the beat predictable and unpredictable sequences. When comparing the effects of beat-based and memory-based expectations directly using t tests and cluster-based permutation tests, no significant differences were found.

Cluster-based testing on the ERP waveforms showed no significant differences between the effects of periodicity and predictability on the auditory evoked potential, neither in attended nor in unattended conditions, nor when we collapsed the data over attended and unattended conditions. Thus, using traditional t tests, Bayesian t tests, and cluster-based testing, we could not find evidence suggesting differences between the effects of memory-based and beat-based expectations on early auditory responses.

DISCUSSION

Temporal expectations facilitate sensory processing and perception in dynamic environments and play an important role in synchronizing our actions to regularities in the outside world, for example, when dancing to a beat in music (Honing & Bouwer, 2019; McGarry et al., 2019). Currently, little is still known about whether shared or separate mechanisms contribute to temporal expectations based on different types of structure in the environment. Specifically, it is unclear whether beat-based expectations (based on periodicity of the input) and memory-based expectations (based on learned predictions of absolute intervals) affect sensory processing and performance in similar ways, independently of each other or in interaction (Breska & Deouell, 2017b; Nobre & Rohenkohl, 2014). Moreover, to what extent these effects rely on task relevance is also still unclear.

In the current study, we show that beat-based and memory-based expectations cannot be differentiated in terms of their effects on auditory processing and/or performance. Both types of expectations lead to enhanced detection of expected events and to attenuation of auditory responses to those events. Also, beat-based and memory-based expectations interacted, with smaller behavioral effects when both were present. These findings are in line with the notion that beat-based and memory-based expectations are subserved by a shared mechanism for temporal predictive processing (Rimmele et al., 2018; Breska & Deouell, 2017b; Teki et al., 2012). Yet, memory-based expectations enhanced target detection even for events that were fully predictable based on the beat. Also, beat-based expectations lead to reduced target detection and enhanced auditory responses to events out of phase with the periodicity, even when these events were fully predictable based on memory. These findings suggest that, although the effects of beat-based and memory-based expectations on sensory processing and performance are the same, the underlying computation may in fact be separate, with beat-based expectations relying (in part) on a rhythmic processing mode characterized by withdrawal of resources off the beat (Breska & Deouell, 2017b). Below, we discuss these findings and their theoretical implications in detail.

Behaviorally, both beat-based and memory-based expectations facilitated target detection in a rhythmic sound sequence, as reflected by enhanced hit rates for targets with expected timing. Although only memory-based expectations improved response speed, the results for beat-based expectations were numerically in the same direction. Thus, unlike previous research (Morillon et al., 2016), we did not find a qualitative difference between beat-based and memory-based expectations on performance. However, in previous work, beat-based expectations were elicited by isochronous sequences, which also elicit memory-based expectations, rendering interpretation of their findings difficult.

That the effects of beat-based and memory-based expectations on auditory processing may rely (in part) on shared mechanisms may further be supported by the observed interaction between the two types of expectations. Beat-based facilitation of detection rates only reached significance in the absence of memory-based expectations. Given that detection rates on average did not exceed 82%, even in the fully predictable sequences, it is unlikely that the absence of a beat-based effect here was due to a ceiling effect. Moreover, we expected beat-based expectations to facilitate memory-based expectations, similar to the facilitation of content predictions (“what”) afforded by temporal expectations (Auksztulewicz et al., 2018; Selchenkova, Jones, & Tillmann, 2014; Hoch, Tyler, & Tillmann, 2013; Schwartze, Rothermich, Schmidt-Kassow, & Kotz, 2011). However, if anything, the effect of predictability on target detection was smaller in the periodic than the aperiodic sequences. Thus, when both types of expectations were present, their effects on auditory processing and performance were diminished. This may suggest that beat-based and memory-based expectations also to some extent compete for limited capacity temporal processing to form expectations, leading to interference when both need to be engaged.

In correspondence to our behavioral findings, our ERP results also do not suggest qualitative differences between the effects of beat-based and memory-based expectations. Both P1 and N1 responses to expected events were attenuated, in line with theories about predictive processing, and explained by assuming that the brain only processes input that is not predicted (Schröger, Marzecová, et al., 2015; Friston, 2005). Effects of beat-based expectations are often explained by entrainment models and DAT, which assume increases in sensory gain at expected time points (Henry & Herrmann, 2014; Large & Jones, 1999), leading to enhanced, rather than attenuated sensory responses (Haegens & Zion-Golumbic, 2018). Our results, however, do not show enhanced auditory responses due to beat-based expectations.

Although the changes in sensory gain suggested by entrainment have often been equated with fluctuations in attention (Large, Herrera, & Velasco, 2015; Henry & Herrmann, 2014; Lakatos et al., 2008), other studies have also shown that entrainment leads to attenuation rather than enhancement of sensory responses (O'Connell et al., 2015; van Atteveldt et al., 2015), like the beat-based expectations in the current study. Moreover, the effects of periodicity, which guides expectations bottom–up, can be dissociated from the effects of task relevance and general top–down attention mechanisms (Kunert & Jongman, 2017). Thus, although entrainment may lead to fluctuations in neural excitability, these may be related to predictions, rather than attention, as proposed by DAT.

Several studies have found enhancement of sensory responses when manipulating temporal expectations, which may seem contradictory to the current findings. However, first, the observed enhancement may depend on the use of different manipulations of temporal expectations. For example, when temporal expectations are cue-based (Hsu et al., 2013), they may indeed be manipulations of the relevance of a time point, rather than prediction (Lange, 2013). Second, when comparing responses to sounds in phase and out of phase with some periodicity, differences in evoked responses may be due to grouping differences between sounds on and off the beat (Rajendran, Harper, Garcia-Lazaro, Lesica, & Schnupp, 2017). Finally, the perception of hierarchical structure in music (meter) may lead to perceived illusory metrical accents (Bouwer, Burgoyne, Odijk, Honing, & Grahn, 2018; Repp, 2010), causing enhanced responses on the beat when compared with off the beat. Indeed, two studies that specifically manipulated beat-based expectations by asking participants to imagine accents on the beat found enhancement of sensory responses (Schaefer, Vlek, & Desain, 2011; Iversen, Repp, & Patel, 2009). To sum up, it may be that entrainment and beat-based expectations are not necessarily related to fluctuations in attention and that previously reported enhancement of auditory responses is due to grouping and metrical accenting, which we controlled for in the current study by including the aperiodic sequences and by comparing only responses within one level of grouping. Which specific task- and stimulus-related factors cause auditory responses to be enhanced and whether attention plays a role in this at all remains an interesting question for further research.

In addition to considering the effects of beat-based expectations at expected time points, we also looked at how beat-based expectations affected processing at time points that were unexpected or rather mispredicted based on the beat (i.e., off-beat). On these positions, N1 responses were larger for events in periodic sequences than for events with similar grouping structure in aperiodic sequences. Enhancement of auditory processing of events off the beat when periodicity is present is in line with these events being more unexpected than their aperiodic counterparts. Behaviorally, we found deteriorated target detection for events off the beat in periodic sequences when compared with aperiodic sequences, in line with reduced processing of events off the beat. Importantly, detection was hampered even if event timing was fully predictable based on learning the sequence of intervals.

Our findings illustrate the importance of also examining effects of beat-based expectations off the beat in distinguishing between beat-based and memory-based expectations. They support the view that beat-based expectations may allow the brain to go into the more efficient “rhythmic mode” of processing, instead of a continuous “vigilance mode” associated with nonperiodic input. Rhythmic mode may be accompanied by automatic suppression of out of phase input when entrainment occurs (Zoefel & VanRullen, 2017; Schroeder & Lakatos, 2009). Beat-based, but not memory-based, expectations have indeed been associated with withdrawal of resources from unexpected moments in time, as apparent from immediate CNV resolution after expected time points for beat-based, but not memory-based, expectations (Breska & Deouell, 2017b). A rhythmic processing mode characterized by withdrawal of resources off the beat, rather than focusing of resources on the beat, could also explain why beat-based expectations only weakly affected responses on the beat, especially when memory-based expectations were also present. With the current design, we could not probe events that were mispredicted in terms of memory-based expectations. Thus, we cannot be sure that memory-based expectations do not show similar withdrawal of resources from time points where no event is expected. However, focusing on the effects of expectations at unexpected instead of expected time points may be an interesting way to compare beat-based and memory-based expectations in the future.

Off the beat, the enhancement of the N1 through beat-based expectations was independent of task relevance. This may be related to the fact that, off the beat, events were mispredicted, rather than unpredicted (e.g., the aperiodic sequences did not allow for the formation of beat-based expectations, making all events unpredicted based on a beat, whereas in the periodic sequences, events off the beat were not in line with the beat-based expectations that could be formed). The effects of mispredicted stimuli have been shown to be independent of task relevance (Hsu et al., 2018), reminiscent of the independence of the MMN, an ERP response to mispredicted sounds, from task relevance (Näätänen, Paavilainen, Rinne, & Alho, 2007).

For the facilitating effects of both beat-based (on-beat) and memory-based expectations, no interaction with task relevance was present in the overall ANOVA. However, based on the current data, numerically, effects were larger in the attended than unattended condition. Indeed, the interaction with attention did reach significance for the attenuation of the P1 response when only considering beat-based expectations on the beat and memory-based expectations off the beat and, additionally, for the full ANOVA with mastoid reference. We can therefore not draw strong conclusions about the interaction between task relevance and temporal expectations, other than noting that the absence of an interaction with attention in the full ANOVA suggests that task relevance does not seem to be a prerequisite for expectations to develop. Whether task relevance interacts with expectations to shape perception is currently a matter of active debate. Recently, several studies have found interactions between prediction and attention, when looking at feature predictions (Smout, Tang, Garrido, & Mattingley, 2019; Auksztulewicz & Friston, 2015) and spatial predictions (Fardo et al., 2017; Marzecová et al., 2017). However, none of these studies looked at temporal predictions, and none found interacting effects as early as the P1 and N1, as studied here. Indeed, a recent study specifically aimed at uncovering earlier effects of attention and prediction found interacting effects of attention and prediction on the P3 in the absence of effects earlier in the processing stream (Alilović, Timmermans, Reteig, van Gaal, & Slagter, 2019). Thus, the presence of an effect of task relevance on expectations may be due to the type of predictions and the latency of the effect studied, with effects as early as the P1, as found in the current study, possibly only present for temporal expectations. It has been suggested that temporal expectations are fundamentally different from feature and spatial expectations (Auksztulewicz et al., 2018; Sherwell et al., 2017), which would hint that the presence of a possible interaction between temporal expectations and task relevance may indeed be informative of the underlying process. Of note to the current study, although the results with regard to the effects of task relevance were somewhat mixed, they did not in any way differentiate between beat-based and memory-based expectations. Both were to some extent affected by task relevance, and for both, this effect was mainly visible in the P1 response.

In the current study, we strictly controlled the acoustic context preceding the sounds of interest (i.e., the last time point before the events used in the analysis that differed in terms of the occurrence of preceding sounds was at −300 msec relative to onset). Yet, we cannot rule out that the longer latency portion of responses to sounds preceding the sounds of interest bled into the ERP waveforms to some extent. In particular, this could have affected the results by affecting the baseline. In addition, if beat-based expectations elicit low-frequency entrainment, differences in the phase of the underlying oscillation could lead to a baseline offset. Issues with differences in the baseline can be avoided by using stringent high-pass filtering, eliminating both the beat frequency and the longer latency portion of the auditory ERP (cf. Chang, Gavin, & Davies, 2012), but such filters have been shown to lead to significant distortions of the P1 waveform and can even shift low-frequency information from the N1 range into the P1 response (Liljander, Holm, Keski-Säntti, & Partanen, 2016; Tanner, Morgan-Short, & Luck, 2015). While filtering as such is not optimal to eliminate potential confounds in the baseline, here, for the attenuating effects of expectations on both P1 and N1, such a confound is unlikely. Differences in baseline activity would result in a general shift of the ERP waveform, but the attenuation for expected events was present both for P1 and N1—components of opposite polarity. Thus, our conclusions with regard to the qualitative effects of temporal expectations, attenuation instead of enhancement, are independent of baseline shifts.

For the N1 enhancement we observed for off-beat events in periodic sequences, the effect of baseline differences is harder to rule out, as here, the direction of the effect was opposite for the P1 (attenuation; see Figure 8). This touches upon a fundamental issue when examining temporal expectations in general and beat-based expectations in particular. Humans are most apt at perceiving a beat in auditory stimuli (Grahn, 2012), at a tempo of around 100–120 beats per min (London, 2012). When moving away from isochronous stimuli, this will automatically result in rapid successions of sounds, making it difficult to completely eliminate effects of previous sounds in ERPs, even in a paradigm as highly controlled as ours. Also, this prevents the analysis of ERPs after subsequent sounds are presented, in our case 150 msec after the onset of the sounds of interest, even if some effects of expectations may only be visible at longer latencies (Alilović et al., 2019). Balancing the need for ecologically valid stimuli, optimal for eliciting (beat-based) temporal expectations, with the need for strict experimental control allowing for easy assessment of separate ERP components, remains a big challenge for future research (Bouwer et al., 2014, 2016; Honing, Bouwer, & Háden, 2014).

Conclusion

To sum up, in the current study, we could not differentiate between beat-based and memory-based temporal expectations in terms of their effects on the detection of, and auditory responses to targets with predictable timing, nor in terms of how they were affected by task relevance. Also, beat-based and memory-based expectations to some extent interfered with each other. Similar effects on ERPs and behavior and the presence of interference may point at a shared underlying mechanism, which may be surprising given the evidence from clinical studies (Breska & Ivry, 2018) and research in nonhuman animals (Honing et al., 2018) that suggests that beat-based expectations are distinct from other types of temporal expectations. Indeed, we also found evidence for distinct processing of beat-based and memory-based expectations: When the timing of events was fully predictable based on memory, the presence of beat-based expectations still deteriorated target detection and enhanced sensory responses for events off the beat (at unexpected moments), and when events were fully predictable based on the beat, memory-based expectations still improved target detection.

To reconcile these findings, we propose that beat-based and memory-based expectations overlap in their effect on auditory responses and behavior, in line with both types of expectations serving the same function but, nonetheless, have partly separate underlying mechanisms to form expectations. Future research will have to focus on distinguishing between possible separate underlying mechanisms by examining the neural dynamics and neural networks involved in different types of temporal expectations. This may also elucidate at which point in the processing stream different types of temporal expectations interact, be it at early stages, during the formation of expectations, or at later stages, where expectations exert their effect on perception and behavior. The special status of beat-based expectations, which has been linked with evolutionary advantages of music (Honing, ten Cate, Peretz, & Trehub, 2015), thus remains an open question. We contribute to this question by being the first to look at the orthogonal effects of beat-based and memory-based expectations on responses to auditory rhythm, and by showing how focusing on the effects of expectations on processing at unexpected, rather than expected time points, may provide a fruitful way to differentiate beat-based from other types of expectations in future research.

Acknowledgments

F. L. B. was supported by an ABC Talent Grant awarded by Amsterdam Brain and Cognition. H. A. S. was supported by a European Research Council starting grant (679399). We would like to thank Peter Saalbrink for his assistance with the data collection.

Reprint requests should be sent to Fleur L. Bouwer, Department of Psychology, Amsterdam Brain and Cognition, University of Amsterdam, Nieuwe Achtergracht 129-B, Amsterdam, The Netherlands, 1018 WS, or via e-mail: fleurbouwer@hotmail.com or Heleen A. Slagter, Department of Experimental and Applied Psychology, Vrije Universiteit Amsterdam, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands, or via e-mail: h.a.slagter@vu.nl.

REFERENCES

Abecasis
,
D.
,
Brochard
,
R.
,
Granot
,
R.
, &
Drake
,
C.
(
2005
).
Differential brain response to metrical accents in isochronous auditory sequences
.
Music Perception
,
22
,
549
562
.
Alilović
,
J.
,
Timmermans
,
B.
,
Reteig
,
L. C.
,
van Gaal
,
S.
, &
Slagter
,
H. A.
(
2019
).
No evidence that predictions and attention modulate the first feedforward sweep of cortical information processing
.
Cerebral Cortex
,
29
,
2261
2278
.
Arnal
,
L. H.
,
Doelling
,
K. B.
, &
Poeppel
,
D.
(
2015
).
Delta–beta coupled oscillations underlie temporal prediction accuracy
.
Cerebral Cortex
,
25
,
3077
3085
.
Auksztulewicz
,
R.
, &
Friston
,
K. J.
(
2015
).
Attentional enhancement of auditory mismatch responses: A DCM/MEG study
.
Cerebral Cortex
,
25
,
4273
4283
.
Auksztulewicz
,
R.
,
Schwiedrzik
,
C. M.
,
Thesen
,
T.
,
Doyle
,
W.
,
Devinsky
,
O.
,
Nobre
,
A. C.
, et al
(
2018
).
Not all predictions are equal: “What” and “when” predictions modulate activity in auditory cortex through different mechanisms
.
Journal of Neuroscience
,
38
,
8680
8693
.
Bell
,
S. L.
,
Smith
,
D. C.
,
Allen
,
R.
, &
Lutman
,
M. E.
(
2004
).
Recording the middle latency response of the auditory evoked potential as a measure of depth of anaesthesia. A technical note
.
British Journal of Anaesthesia
,
92
,
442
445
.
Bouwer
,
F. L.
,
Burgoyne
,
J. A.
,
Odijk
,
D.
,
Honing
,
H.
, &
Grahn
,
J. A.
(
2018
).
What makes a rhythm complex? The influence of musical training and accent type on beat perception
.
PLoS One
,
13
,
e0190322
.
Bouwer
,
F. L.
, &
Honing
,
H.
(
2015
).
Temporal attending and prediction influence the perception of metrical rhythm: Evidence from reaction times and ERPs
.
Frontiers in Psychology
,
6
,
1094
.
Bouwer
,
F. L.
,
Van Zuijen
,
T. L.
, &
Honing
,
H.
(
2014
).
Beat processing is pre-attentive for metrically simple rhythms with clear accents: An ERP study
.
PLoS One
,
9
,
e97467
.
Bouwer
,
F. L.
,
Werner
,
C. M.
,
Knetemann
,
M.
, &
Honing
,
H.
(
2016
).
Disentangling beat perception from sequential learning and examining the influence of attention and musical abilities on ERP responses to rhythm
.
Neuropsychologia
,
85
,
80
90
.
Breska
,
A.
, &
Deouell
,
L. Y.
(
2014
).
Automatic bias of temporal expectations following temporally regular input independently of high-level temporal expectation
.
Journal of Cognitive Neuroscience
,
26
,
1555
1571
.
Breska
,
A.
, &
Deouell
,
L. Y.
(
2016
).
When synchronizing to rhythms is not a good thing: Modulations of preparatory and post-target neural activity when shifting attention away from on-beat times of a distracting rhythm
.
Journal of Neuroscience
,
36
,
7154
7166
.
Breska
,
A.
, &
Deouell
,
L. Y.
(
2017a
).
Dance to the rhythm, cautiously: Isolating unique indicators of oscillatory entrainment
.
PLoS Biology
,
15
,
e2003534
.
Breska
,
A.
, &
Deouell
,
L. Y.
(
2017b
).
Neural mechanisms of rhythm-based temporal prediction: Delta phase-locking reflects temporal predictability but not rhythmic entrainment
.
PLoS Biology
,
15
,
e2001665
.
Breska
,
A.
, &
Ivry
,
R. B.
(
2018
).
Double dissociation of single-interval and rhythmic temporal prediction in cerebellar degeneration and Parkinson's disease
.
Proceedings of the National Academy of Sciences, U.S.A.
,
115
,
12283
12288
.
Brochard
,
R.
,
Abecasis
,
D.
,
Potter
,
D. D.
,
Ragot
,
R.
, &
Drake
,
C.
(
2003
).
The “ticktock” of our internal clock: Direct brain evidence of subjective accents in isochronous sequences
.
Psychological Science
,
14
,
362
366
.
Chang
,
W.-P.
,
Gavin
,
W. J.
, &
Davies
,
P. L.
(
2012
).
Bandpass filter settings differentially affect measurement of P50 sensory gating in children and adults
.
Clinical Neurophysiology
,
123
,
2264
2272
.
Clark
,
A.
(
2013
).
Whatever next? Predictive brains, situated agents, and the future of cognitive science
.
Behavioral and Brain Sciences
,
36
,
181
204
.
Cope
,
T. E.
,
Grube
,
M.
,
Singh
,
B.
,
Burn
,
D. J.
, &
Griffiths
,
T. D.
(
2014
).
The basal ganglia in perceptual timing: Timing performance in Multiple System Atrophy and Huntington's disease
.
Neuropsychologia
,
52
,
73
81
.
Delorme
,
A.
, &
Makeig
,
S.
(
2004
).
EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis
.
Journal of Neuroscience Methods
,
134
,
9
21
.
Escoffier
,
N.
,
Herrmann
,
C. S.
, &
Schirmer
,
A.
(
2015
).
Auditory rhythms entrain visual processes in the human brain: Evidence from evoked oscillations and event-related potentials
.
Neuroimage
,
111
,
267
276
.
Fardo
,
F.
,
Auksztulewicz
,
R.
,
Allen
,
M.
,
Dietz
,
M. J.
,
Roepstorff
,
A.
, &
Friston
,
K. J.
(
2017
).
Expectation violation and attention to pain jointly modulate neural gain in somatosensory cortex
.
Neuroimage
,
153
,
109
121
.
Feldman
,
H.
, &
Friston
,
K. J.
(
2010
).
Attention, uncertainty, and free-energy
.
Frontiers in Human Neuroscience
,
4
,
215
.
Fitzroy
,
A. B.
, &
Sanders
,
L. D.
(
2015
).
Musical meter modulates the allocation of attention across time
.
Journal of Cognitive Neuroscience
,
27
,
2339
2351
.
Forth
,
J.
,
Agres
,
K.
,
Purver
,
M.
, &
Wiggins
,
G. A.
(
2016
).
Entraining IDyOT: Timing in the Information Dynamics of Thinking
.
Frontiers in Psychology
,
7
,
1575
.
Friston
,
K. J.
(
2005
).
A theory of cortical responses
.
Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences
,
360
,
815
836
.
Grahn
,
J. A.
(
2012
).
See what I hear? Beat perception in auditory and visual rhythms
.
Experimental Brain Research
,
220
,
51
61
.
Grahn
,
J. A.
, &
Brett
,
M.
(
2007
).
Rhythm and beat perception in motor areas of the brain
.
Journal of Cognitive Neuroscience
,
19
,
893
906
.
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
.
Grahn
,
J. A.
, &
Schuit
,
D.
(
2012
).
Individual differences in rhythmic ability: Behavioral and neuroimaging investigations
.
Psychomusicology: Music, Mind, and Brain
,
22
,
105
121
.
Haegens
,
S.
, &
Zion-Golumbic
,
E.
(
2018
).
Rhythmic facilitation of sensory processing: A critical review
.
Neuroscience & Biobehavioral Reviews
,
86
,
150
165
.
Henry
,
M. J.
, &
Herrmann
,
B.
(
2014
).
Low-frequency neural oscillations support dynamic attending in temporal context
.
Timing & Time Perception
,
2
,
62
86
.
Henry
,
M. J.
, &
Obleser
,
J.
(
2012
).
Frequency modulation entrains slow neural oscillations and optimizes human listening behavior
.
Proceedings of the National Academy of Sciences, U.S.A.
,
109
,
20095
20100
.
Hoch
,
L.
,
Tyler
,
M. D.
, &
Tillmann
,
B.
(
2013
).
Regularity of unit length boosts statistical learning in verbal and nonverbal artificial languages
.
Psychonomic Bulletin & Review
,
20
,
142
147
.
Honing
,
H.
, &
Bouwer
,
F. L.
(
2019
).
Rhythm
. In
P. J.
Rentfrow
&
D. J.
Levitin
(Eds.),
Foundations in music psychology: Theory and research
(pp.
33
70
).
Cambridge, MA
:
MIT Press
.
Honing
,
H.
,
Bouwer
,
F. L.
, &
Háden
,
G. P.
(
2014
).
Perceiving temporal regularity in music: The role of auditory event-related potentials (ERPs) in probing beat perception
. In
H.
Merchant
&
V.
de Lafuente
(Eds.),
Neurobiology of interval timing
(
Vol. 829
, pp.
305
323
).
New York
:
Springer
.
Honing
,
H.
,
Bouwer
,
F. L.
,
Prado
,
L.
, &
Merchant
,
H.
(
2018
).
Rhesus monkeys (Macaca mulatta) sense isochrony in rhythm, but not the beat: Additional support for the gradual audiomotor evolution hypothesis
.
Frontiers in Neuroscience
,
12
,
475
.
Honing
,
H.
, &
Merchant
,
H.
(
2014
).
Differences in auditory timing between human and nonhuman primates
.
Behavioral and Brain Sciences
,
37
,
557
558
.
Honing
,
H.
,
ten Cate
,
C.
,
Peretz
,
I.
, &
Trehub
,
S. E.
(
2015
).
Without it no music: Cognition, biology and evolution of musicality
.
Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences
,
370
,
20140088
.
Hsu
,
Y.-F.
,
Hämäläinen
,
J. A.
, &
Waszak
,
F.
(
2013
).
Temporal expectation and spectral expectation operate in distinct fashion on neuronal populations
.
Neuropsychologia
,
51
,
2548
2555
.
Hsu
,
Y.-F.
,
Hämäläinen
,
J. A.
, &
Waszak
,
F.
(
2018
).
The processing of mispredicted and unpredicted sensory inputs interact differently with attention
.
Neuropsychologia
,
111
,
85
91
.
Hsu
,
Y.-F.
,
Le Bars
,
S.
,
Hämäläinen
,
J. A.
, &
Waszak
,
F.
(
2015
).
Distinctive representation of mispredicted and unpredicted prediction errors in human electroencephalography
.
Journal of Neuroscience
,
35
,
14653
14660
.
Iversen
,
J. R.
,
Patel
,
A. D.
,
Nicodemus
,
B.
, &
Emmorey
,
K.
(
2015
).
Synchronization to auditory and visual rhythms in hearing and deaf individuals
.
Cognition
,
134
,
232
244
.
Iversen
,
J. R.
,
Repp
,
B. H.
, &
Patel
,
A. D.
(
2009
).
Top–down control of rhythm perception modulates early auditory responses
.
Annals of the New York Academy of Sciences
,
1169
,
58
73
.
JASP Team
. (
2019
).
JASP (version 0.9.2) [computer software]
.
Retrieved from https://jasp-stats.org
.
Jeffreys
,
H.
(
1961
).
Theory of probability
(3rd ed.).
Oxford
:
Oxford University Press
.
Jones
,
M. R.
, &
Boltz
,
M. G.
(
1989
).
Dynamic attending and responses to time
.
Psychological Review
,
96
,
459
491
.
Koelsch
,
S.
,
Vuust
,
P.
, &
Friston
,
K. J.
(
2019
).
Predictive processes and the peculiar case of music
.
Trends in Cognitive Sciences
,
23
,
63
77
.
Kok
,
P.
,
Rahnev
,
D.
,
Jehee
,
J. F. M.
,
Lau
,
H. C.
, &
de Lange
,
F. P.
(
2012
).
Attention reverses the effect of prediction in silencing sensory signals
.
Cerebral Cortex
,
22
,
2197
2206
.
Kunert
,
R.
, &
Jongman
,
S. R.
(
2017
).
Entrainment to an auditory signal: Is attention involved?
Journal of Experimental Psychology: General
,
146
,
77
88
.
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
.
Lakatos
,
P.
,
Musacchia
,
G.
,
O'Connell
,
M. N.
,
Falchier
,
A. Y.
,
Javitt
,
D. C.
, &
Schroeder
,
C. E.
(
2013
).
The spectrotemporal filter mechanism of auditory selective attention
.
Neuron
,
77
,
750
761
.
Lange
,
K.
(
2009
).
Brain correlates of early auditory processing are attenuated by expectations for time and pitch
.
Brain and Cognition
,
69
,
127
137
.
Lange
,
K.
(
2013
).
The ups and downs of temporal orienting: A review of auditory temporal orienting studies and a model associating the heterogeneous findings on the auditory N1 with opposite effects of attention and prediction
.
Frontiers in Human Neuroscience
,
7
,
263
.
Large
,
E. W.
,
Herrera
,
J. A.
, &
Velasco
,
M. J.
(
2015
).
Neural networks for beat perception in musical rhythm
.
Frontiers in Systems Neuroscience
,
9
,
159
.
Large
,
E. W.
, &
Jones
,
M. R.
(
1999
).
The dynamics of attending: How people track time-varying events
.
Psychological Review
,
106
,
119
159
.
Lawrance
,
E. L. A.
,
Harper
,
N. S.
,
Cooke
,
J. E.
, &
Schnupp
,
J. W. H.
(
2014
).
Temporal predictability enhances auditory detection
.
Journal of the Acoustical Society of America
,
135
,
EL357
EL363
.
Liljander
,
S.
,
Holm
,
A.
,
Keski-Säntti
,
P.
, &
Partanen
,
J. V.
(
2016
).
Optimal digital filters for analyzing the mid-latency auditory P50 event-related potential in patients with Alzheimer's disease
.
Journal of Neuroscience Methods
,
266
,
50
67
.
London
,
J.
(
2012
).
Hearing in time: Psychological aspects of musical meter
(2nd ed.).
Oxford
:
Oxford University Press
.
Luck
,
S. J.
(
2005
).
An introduction to the event-related potential technique
.
Cambridge, MA
:
MIT Press
.
Marzecová
,
A.
,
Widmann
,
A.
,
SanMiguel
,
I.
,
Kotz
,
S. A.
, &
Schröger
,
E.
(
2017
).
Interrelation of attention and prediction in visual processing: Effects of task-relevance and stimulus probability
.
Biological Psychology
,
125
,
76
90
.
McGarry
,
L. M.
,
Sternin
,
A.
, &
Grahn
,
J. A.
(
2019
).
Music and movement
. In
P. J.
Rentfrow
&
D. J.
Levitin
(Eds.),
Foundations in music psychology: Theory and research
(pp.
609
640
).
Cambridge, MA
:
MIT Press
.
Merchant
,
H.
,
Grahn
,
J. A.
,
Trainor
,
L. J.
,
Rohrmeier
,
M. A.
, &
Fitch
,
W. T.
(
2015
).
Finding the beat: A neural perspective across humans and non-human primates
.
Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences
,
370
,
20140093
.
Merchant
,
H.
, &
Honing
,
H.
(
2014
).
Are non-human primates capable of rhythmic entrainment? Evidence for the gradual audiomotor evolution hypothesis
.
Frontiers in Neuroscience
,
7
,
274
.
Morillon
,
B.
,
Schroeder
,
C. E.
,
Wyart
,
V.
, &
Arnal
,
L. H.
(
2016
).
Temporal prediction in lieu of periodic stimulation
.
Journal of Neuroscience
,
36
,
2342
2347
.
Näätänen
,
R.
,
Paavilainen
,
P.
,
Rinne
,
T.
, &
Alho
,
K.
(
2007
).
The mismatch negativity (MMN) in basic research of central auditory processing: A review
.
Clinical Neurophysiology
,
118
,
2544
2590
.
Nobre
,
A. C.
, &
Rohenkohl
,
G.
(
2014
).
Time for the fourth dimension in attention
. In
A. C.
Nobre
&
S.
Kastner
(Eds.),
The Oxford handbook of attention
(pp.
676
721
).
New York
:
Oxford University Press
.
Nobre
,
A. C.
, &
van Ede
,
F.
(
2018
).
Anticipated moments: Temporal structure in attention
.
Nature Reviews Neuroscience
,
19
,
34
48
.
Obleser
,
J.
,
Henry
,
M. J.
, &
Lakatos
,
P.
(
2017
).
What do we talk about when we talk about rhythm?
PLoS Biology
,
15
,
e2002794
.
O'Connell
,
M. N.
,
Barczak
,
A.
,
Ross
,
D.
,
McGinnis
,
T.
,
Schroeder
,
C. E.
, &
Lakatos
,
P.
(
2015
).
Multi-scale entrainment of coupled neuronal oscillations in primary auditory cortex
.
Frontiers in Human Neuroscience
,
9
,
655
.
Oostenveld
,
R.
,
Fries
,
P.
,
Maris
,
E.
, &
Schoffelen
,
J.-M.
(
2011
).
FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data
.
Computational Intelligence and Neuroscience
,
2011
,
156869
.
Paris
,
T.
,
Kim
,
J.
, &
David
,
C.
(
2016
).
The processing of attended and predicted sounds in time
.
Journal of Cognitive Neuroscience
,
28
,
158
165
.
Picton
,
T. W.
,
Hillyard
,
S. A.
,
Krausz
,
H. I.
, &
Galambos
,
R.
(
1974
).
Human auditory evoked potentials. I: Evaluation of components
.
Electroencephalography and Clinical Neurophysiology
,
36
,
179
190
.
Potter
,
D. D.
,
Fenwick
,
M.
,
Abecasis
,
D.
, &
Brochard
,
R.
(
2009
).
Perceiving rhythm where none exists: Event-related potential (ERP) correlates of subjective accenting
.
Cortex
,
45
,
103
109
.
Povel
,
D.-J.
, &
Essens
,
P.
(
1985
).
Perception of temporal patterns
.
Music Perception
,
2
,
411
440
.
Rajendran
,
V. G.
,
Harper
,
N. S.
,
Garcia-Lazaro
,
J. A.
,
Lesica
,
N. A.
, &
Schnupp
,
J. W. H.
(
2017
).
Midbrain adaptation may set the stage for the perception of musical beat
.
Proceedings of the Royal Society of London, Series B: Biological Sciences
,
284
,
20171455
.
Repp
,
B. H.
(
2010
).
Do metrical accents create illusory phenomenal accents?
Attention, Perception, & Psychophysics
,
72
,
1390
1403
.
Rimmele
,
J. M.
,
Jolsvai
,
H.
, &
Sussman
,
E.
(
2011
).
Auditory target detection is affected by implicit temporal and spatial expectations
.
Journal of Cognitive Neuroscience
,
23
,
1136
1147
.
Rimmele
,
J. M.
,
Morillon
,
B.
,
Poeppel
,
D.
, &
Arnal
,
L. H.
(
2018
).
Proactive sensing of periodic and aperiodic auditory patterns
.
Trends in Cognitive Sciences
,
22
,
870
882
.
Rohenkohl
,
G.
,
Coull
,
J. T.
, &
Nobre
,
A. C.
(
2011
).
Behavioural dissociation between exogenous and endogenous temporal orienting of attention
.
PLoS One
,
6
,
e14620
.
Rohenkohl
,
G.
,
Cravo
,
A. M.
,
Wyart
,
V.
, &
Nobre
,
A. C.
(
2012
).
Temporal expectation improves the quality of sensory information
.
Journal of Neuroscience
,
32
,
8424
8428
.
Ruhnau
,
P.
,
Herrmann
,
B.
,
Maess
,
B.
, &
Schröger
,
E.
(
2011
).
Maturation of obligatory auditory responses and their neural sources: Evidence from EEG and MEG
.
Neuroimage
,
58
,
630
639
.
Sanabria
,
D.
, &
Correa
,
Á.
(
2013
).
Electrophysiological evidence of temporal preparation driven by rhythms in audition
.
Biological Psychology
,
92
,
98
105
.
Schaefer
,
R. S.
,
Vlek
,
R. J.
, &
Desain
,
P.
(
2011
).
Decomposing rhythm processing: Electroencephalography of perceived and self-imposed rhythmic patterns
.
Psychological Research
,
75
,
95
106
.
Schroeder
,
C. E.
, &
Lakatos
,
P.
(
2009
).
Low-frequency neuronal oscillations as instruments of sensory selection
.
Trends in Neurosciences
,
32
,
9
18
.
Schröger
,
E.
,
Kotz
,
S. A.
, &
SanMiguel
,
I.
(
2015
).
Bridging prediction and attention in current research on perception and action
.
Brain Research
,
1626
,
1
13
.
Schröger
,
E.
,
Marzecová
,
A.
, &
SanMiguel
,
I.
(
2015
).
Attention and prediction in human audition: A lesson from cognitive psychophysiology
.
European Journal of Neuroscience
,
41
,
641
664
.
Schultz
,
B. G.
,
Stevens
,
C. J.
,
Keller
,
P. E.
, &
Tillmann
,
B.
(
2013
).
The implicit learning of metrical and nonmetrical temporal patterns
.
Quarterly Journal of Experimental Psychology
,
66
,
360
380
.
Schwartze
,
M.
,
Farrugia
,
N.
, &
Kotz
,
S. A.
(
2013
).
Dissociation of formal and temporal predictability in early auditory evoked potentials
.
Neuropsychologia
,
51
,
320
325
.
Schwartze
,
M.
, &
Kotz
,
S. A.
(
2013
).
A dual-pathway neural architecture for specific temporal prediction
.
Neuroscience & Biobehavioral Reviews
,
37
,
2587
2596
.
Schwartze
,
M.
,
Rothermich
,
K.
,
Schmidt-Kassow
,
M.
, &
Kotz
,
S. A.
(
2011
).
Temporal regularity effects on pre-attentive and attentive processing of deviance
.
Biological Psychology
,
87
,
146
151
.
Selchenkova
,
T.
,
Jones
,
M. R.
, &
Tillmann
,
B.
(
2014
).
The influence of temporal regularities on the implicit learning of pitch structures
.
Quarterly Journal of Experimental Psychology
,
67
,
2360
2380
.
Sherwell
,
C.
,
Garrido
,
M. I.
, &
Cunnington
,
R.
(
2017
).
Timing in predictive coding: The roles of task relevance and global probability
.
Journal of Cognitive Neuroscience
,
29
,
780
792
.
Smout
,
C. A.
,
Tang
,
M. F.
,
Garrido
,
M. I.
, &
Mattingley
,
J. B.
(
2019
).
Attention promotes the neural encoding of prediction errors
.
PLoS Biology
,
17
,
e2006812
.
Tanner
,
D.
,
Morgan-Short
,
K.
, &
Luck
,
S. J.
(
2015
).
How inappropriate high-pass filters can produce artifactual effects and incorrect conclusions in ERP studies of language and cognition
.
Psychophysiology
,
52
,
997
1009
.
Teki
,
S.
,
Grube
,
M.
, &
Griffiths
,
T. D.
(
2012
).
A unified model of time perception accounts for duration-based and beat-based timing mechanisms
.
Frontiers in Integrative Neuroscience
,
5
,
90
.
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
.
ten Oever
,
S.
,
Schroeder
,
C. E.
,
Poeppel
,
D.
,
van Atteveldt
,
N.
, &
Zion-Golumbic
,
E.
(
2014
).
Rhythmicity and cross-modal temporal cues facilitate detection
.
Neuropsychologia
,
63
,
43
50
.
Tierney
,
A.
, &
Kraus
,
N.
(
2013
).
Neural responses to sounds presented on and off the beat of ecologically valid music
.
Frontiers in Systems Neuroscience
,
7
,
14
.
Todorovic
,
A.
,
Schoffelen
,
J.-M.
,
van Ede
,
F.
,
Maris
,
E.
, &
de Lange
,
F. P.
(
2015
).
Temporal expectation and attention jointly modulate auditory oscillatory activity in the beta band
.
PLoS One
,
10
,
e0120288
.
van Atteveldt
,
N.
,
Musacchia
,
G.
,
Zion-Golumbic
,
E.
,
Sehatpour
,
P.
,
Javitt
,
D. C.
, &
Schroeder
,
C. E.
(
2015
).
Complementary fMRI and EEG evidence for more efficient neural processing of rhythmic vs. unpredictably timed sounds
.
Frontiers in Psychology
,
6
,
1663
.
van der Weij
,
B.
,
Pearce
,
M. T.
, &
Honing
,
H.
(
2017
).
A probabilistic model of meter perception: Simulating enculturation
.
Frontiers in Psychology
,
8
,
824
.
Wagenmakers
,
E.-J.
,
Love
,
J.
,
Marsman
,
M.
,
Jamil
,
T.
,
Ly
,
A.
,
Verhagen
,
J.
, et al
(
2018
).
Bayesian inference for psychology. Part II: Example applications with JASP
.
Psychonomic Bulletin & Review
,
25
,
58
76
.
Zarco
,
W.
,
Merchant
,
H.
,
Prado
,
L.
, &
Mendez
,
J. C.
(
2009
).
Subsecond timing in primates: Comparison of interval production between human subjects and rhesus monkeys
.
Journal of Neurophysiology
,
102
,
3191
3202
.
Zoefel
,
B.
, &
VanRullen
,
R.
(
2017
).
Oscillatory mechanisms of stimulus processing and selection in the visual and auditory systems: State-of-the-art, speculations and suggestions
.
Frontiers in Neuroscience
,
11
,
296
.

Author notes

This article is part of a Special Focus deriving from a symposium at the 2018 annual meeting of Cognitive Neuroscience Society, entitled, “What makes musical rhythm special: Cross-species, developmental and social perspectives.”