Abstract

Sensory systems utilize temporal structure in the environment to build expectations about the timing of forthcoming events. We investigated the effects of rhythm-based temporal expectation on auditory responses measured with EEG recorded from the frontocentral sites implicated in auditory processing. By manipulating temporal expectation and the interonset interval (IOI) of tones, we examined how neural responses adapted to auditory rhythm and reacted to stimuli that violated the rhythm. Participants passively listened to the tones while watching a silent nature video. In Experiment 1 (n = 22), in the long-IOI block, tones were frequently presented (80%) with 1.7-sec IOI and infrequently presented (20%) with 1.2-sec IOI, generating unexpectedly early tones that violated temporal expectation. Conversely, in the short-IOI block, tones were frequently presented with 1.2-sec IOI and infrequently presented with 1.7-sec IOI, generating late tones. We analyzed the tone-evoked N1–P2 amplitude of ERPs and intertrial phase clustering in the theta–alpha band. The results provided evidence of strong delay-dependent adaptation effects (short-term, sensitive to IOI), weak cumulative adaptation effects (long-term, driven by tone repetition over time), and robust temporal-expectation violation effects over and above the adaptation effects. Experiment 2 (n = 22) repeated Experiment 1 with shorter IOIs of 1.2 and 0.7 sec. Overall, we found evidence of strong delay-dependent adaptation effects, weak cumulative adaptation effects (which may most efficiently accumulate at the tone presentation rate of ∼1 Hz), and robust temporal-expectation violation effects that substantially boost auditory responses to the extent of overriding the delay-dependent adaptation effects likely through mechanisms involved in exogenous attention.

INTRODUCTION

Temporal structure in the sensory environment—such as a predictable interval between two events or rhythmically presented events—can lead to the generation of temporal expectations and influence behavioral and neural responses to the events taking place within these temporal structures (for reviews, see Nobre & van Ede, 2018; Coull, 2009). For instance, behavioral responses are facilitated (becoming faster and/or more accurate) when targets fulfill temporal expectations that are generated by either predictable intervals (e.g., Griffin, Miniussi, & Nobre, 2002; Miniussi, Wilding, Coull, & Nobre, 1999; Coull & Nobre, 1998) or rhythmic events (e.g., Cravo, Rohenkohl, Wyart, & Nobre, 2013; Rohenkohl & Nobre, 2011; Lange, 2009, 2010) compared with when targets violate temporal expectations or when no temporal expectation is established. Studies using human EEG have shown that behavioral facilitation via fulfilled temporal expectations is accompanied by changes in the brain's responses to those events. Studies have reported EEG modulations at early sensory stages (e.g., Lampar & Lange, 2011; Rohenkohl & Nobre, 2011; Correa & Nobre, 2008; Sanders & Astheimer, 2008; Correa, Lupiáñez, Madrid, & Tudela, 2006; Doherty, Rao, Mesulam, & Nobre, 2005; Lange, Rösler, & Röder, 2003; Griffin et al., 2002) as well as at later decision- and/or response-related stages (e.g., Lange, 2009; Correa & Nobre, 2008; Sanders & Astheimer, 2008; Griffin et al., 2002; Miniussi et al., 1999). However, the direction of the modulations has been heterogeneous, especially regarding the early sensory auditory or visual ERPs, with evidence for both increased (e.g., Lampar & Lange, 2011; Rohenkohl & Nobre, 2011; Sanders & Astheimer, 2008; Correa et al., 2006; Griffin et al., 2002) and decreased (e.g., Lange, 2009, 2010; Correa & Nobre, 2008) amplitudes for temporally expected/attended events. This could be partially explained by task parameters and the relative involvement of temporal prediction and selective attention mechanisms (for a review, see Lange, 2013). Our goal was to carefully examine EEG correlates of temporal expectation at a basic level without task or attention demands. Specifically, we investigated the automatic (bottom–up) effects of rhythm-based temporal expectation on auditory-evoked EEG signals in relation to the effects of sensory adaptation (also referred to as repetition suppression, e.g., Todorovic & de Lange, 2012) in a simple paradigm in which participants passively heard periodic tones while their attention was directed elsewhere.

We recorded EEG while participants watched a muted nature video and passively heard rhythmic tones with occasional temporal perturbations. We manipulated temporal expectation across two blocks. In the long-interonset-interval (IOI) block, tones were frequently presented (80%) with the long IOI—the long-IOI-block standard tones—and infrequently presented (20%) with the short IOI—the long-IOI-block unexpectedly early tones, which would violate temporal expectation (Figure 1A). Conversely, in the short-IOI block, tones were frequently presented (80%) with the short IOI—the short-IOI-block standard tones—and infrequently presented (20%) with the long IOI—the short-IOI-block late tones, which would likely be anticipated after an unexpected absence of a short-IOI-block standard tone (Figure 1B). We compared the auditory-evoked EEG responses (computed as ERPs and intertrial phase clustering [ITPC]) across these four types of tones. We used intervals that are close to those observed in rhythms of human action (for reviews, see Repp, 2005; Moelants, 2002); the long and short IOIs were 1.7 and 1.2 sec in Experiment 1 and 1.2 and 0.7 sec in Experiment 2, respectively.

Figure 1. 

Experimental design. We presented rhythmic sequences of pure tones (100 msec, 440 Hz) with infrequent and unpredictable temporal perturbations. (A) The long-IOI block. The tones were presented frequently (80%) with the long IOI and infrequently (20%) with the short IOI. The expected long-IOI tones are termed the “long-IOI-block standard” tones (large gray circles), whereas the rare short-IOI tones, which could not be anticipated, are termed the “long-IOI-block unexpectedly early” tones (small light-blue circle). (B) The short-IOI block. The tones were presented frequently (80%) with the short IOI and infrequently (20%) with the long IOI. The expected short-IOI tones are termed the “short-IOI-block standard” tones (small gray circles), whereas the rare long-IOI tones, which could be anticipated after experiencing the absence of the short-standard tones, are termed “short-IOI-block late” tones (large blue circle). The short and long IOIs were 1.2 and 1.7 sec in Experiment 1 and 0.7 and 1.2 sec in Experiment 2, respectively.

Figure 1. 

Experimental design. We presented rhythmic sequences of pure tones (100 msec, 440 Hz) with infrequent and unpredictable temporal perturbations. (A) The long-IOI block. The tones were presented frequently (80%) with the long IOI and infrequently (20%) with the short IOI. The expected long-IOI tones are termed the “long-IOI-block standard” tones (large gray circles), whereas the rare short-IOI tones, which could not be anticipated, are termed the “long-IOI-block unexpectedly early” tones (small light-blue circle). (B) The short-IOI block. The tones were presented frequently (80%) with the short IOI and infrequently (20%) with the long IOI. The expected short-IOI tones are termed the “short-IOI-block standard” tones (small gray circles), whereas the rare long-IOI tones, which could be anticipated after experiencing the absence of the short-standard tones, are termed “short-IOI-block late” tones (large blue circle). The short and long IOIs were 1.2 and 1.7 sec in Experiment 1 and 0.7 and 1.2 sec in Experiment 2, respectively.

Given our experimental design, we could infer the presence of three distinct effects. We could observe auditory adaptation effects that include a short-term, delay-dependent component—a delay-dependent adaptation effect—and a long-term, cumulative component—a cumulative adaptation effect. A delay-dependent adaptation effect, dependent only on the latency from the immediately preceding tone, would reduce ERP and ITPC responses to the short-IOI tones relative to the long-IOI tones regardless of the block type (i.e., stronger delay-dependent adaptation for tones after shorter intervals). A cumulative adaptation effect would be observed as the responses to tones becoming gradually diminished with increasing tone repetitions as the experiment progresses. Furthermore, the strength of the cumulative adaptation effect might depend on the rate of tone presentation (i.e., a rate dependency of the cumulative adaptation effect); the effect could become stronger over time and/or persist longer in the short-IOI block relative to the long-IOI block (independently of delay-dependent adaptation). Such an effect would be observed as overall reduced neural responses to tones in the short-IOI block relative to the tones in the long-IOI block. Delay-dependent adaptation effects were expected from previous magnetoencephalography (MEG)/EEG studies reporting that the strength of neural responses to repeated tones decreased as IOIs were reduced (e.g., Pereira et al., 2014; Carver, Fuchs, Jantzen, & Kelso, 2002; Hari, Kaila, Katila, Tuomisto, & Varpula, 1982; Nelson & Lassman, 1973). Both delay-dependent and cumulative adaptation effects were expected based on evidence from animal single-unit studies (e.g., Ulanovsky, Las, Farkas, & Nelken, 2004; Ulanovsky, Las, & Nelken, 2003; Condon & Weinberger, 1991) reporting that neural responses adapt to repetitive tones in multiple time scales, ranging from a few milliseconds to hundreds of seconds.

The primary goal of the current study was to investigate how temporal expectation mechanisms operating under the condition of passive listening might respond to temporally deviant (i.e., potentially important) tones. To that end, in addition to adaptation effects, we sought to identify temporal-expectation violation effects reflected as differences between responses to standard tones and responses to their deviant counterparts. It is reasonable to expect that the auditory system automatically monitors rhythmic structures even under passive listening with attention primarily engaged in visual processing, so that when a tone is presented unexpectedly early, it may enhance responses to counter any sensory adaptation effects. The auditory system may also enhance responses to the “unexpectedly” late tones. However, when an expected short-IOI tone was absent in the short-IOI block, the auditory system would likely reorient its temporal expectation to the long IOI to anticipate the impending late tone (which occurred with 100% probability), so that any potential response enhancement to the late tones would likely be negligible. Indeed, previous studies (including ours) have reported that behavioral responses (RTs and accuracy) were similar whether a long-IOI stimulus was presented as a temporally expected stimulus in the long-IOI block or as a late stimulus in the short-IOI block (Menceloglu, Grabowecky, & Suzuki, 2017a, 2017b, 2019; for a review, see Nobre & van Ede, 2018). Thus, although reorientation might somewhat elevate responses to the short-IOI-block late tones, we assume that this effect would be negligible (as supported by our results; see below). This assumption was necessary for the planned analysis because our experimental design afforded sufficient degrees of freedom to dissociate only three of the four potential effects: a cumulative adaptation effect, a delay-dependent adaptation effect, a temporal-expectation violation effect, and a reorientation effect.

A cumulative adaptation effect can be inferred from a response decrement to tones as the experiment progresses (e.g., neural responses to tones in the first half of the experiment would be stronger than those in the second half of the experiment). Furthermore, a cumulative adaptation effect difference between blocks (i.e., a block effect; a faster presentation rate leading to a stronger cumulative adaptation effect) can be inferred from response decrements to the long-IOI tones in the short-IOI block relative to the long-IOI block (i.e., long-IOI-block standard tones vs. short-IOI-block late tones). This comparison controls for any delay-dependent adaptation effect and temporal-expectation violation effect as both tones are presented with the same IOI and are temporally expected; we assumed that any reorientation effect would be negligible (see above).

A delay-dependent adaptation effect (i.e., interval effect) can be inferred from a response decrement to the short-IOI tones relative to the long-IOI tones within the short-IOI block (i.e., short-IOI-block standard tones vs. short-IOI-block late tones). This within-block comparison controls for any cumulative adaptation effect difference between blocks and temporal-expectation violation effect as both tones are presented in the same block and are temporally expected; we assumed that any reorientation effect would be negligible (see above). Note that, if there is no cumulative adaptation effect difference between blocks (i.e., no block effect), the presence of a delay-dependent adaptation effect can also be inferred from a response decrement to the short-IOI tones in the short-IOI block relative to the long-IOI tones in the long-IOI block (i.e., short-IOI-block standard tones vs. long-IOI-block standard tones), without needing to control for block.

Finally, the degree to which the enhanced responses elicited by temporal-expectation violation may overcome delay-dependent adaptation can be inferred from the difference between the responses to the short-IOI (unexpectedly early) and long-IOI tones within the long-IOI block (i.e., long-IOI-block unexpectedly early tone vs. long-IOI-block standard tone). This within-block comparison controls for any cumulative adaptation effect difference between blocks. In particular, a nonsignificant difference would indicate that temporal-expectation violation enhances auditory responses to the extent of statistically overriding delay-dependent adaptation effects. Note that, if there is no cumulative adaptation effect difference between blocks (i.e., no block effect), a complementary analysis is possible where the presence of a temporal-expectation violation effect can be inferred from a response enhancement to short-IOI tones in the long-IOI block relative to the short-IOI block (i.e., long-IOI-block unexpectedly early tones vs. short-IOI-block standard tones), without needing to control for block.

We thus considered four sets of potential outcomes that would distinguish the cumulative adaptation, delay-dependent adaptation, and temporal-expectation violation effects, as illustrated in Figure 2. A cumulative adaptation (block) effect with no delay-dependent adaptation effect or temporal-expectation violation effect would be characterized by a block-based pattern of vertical separation (Figure 2A), with stronger responses to the long-IOI-block tones (the solid line in Figure 2A) than to the short-IOI-block tones (the dashed line in Figure 2A). A delay-dependent adaptation (IOI) effect with no cumulative adaptation effect or temporal-expectation violation effect would be characterized by a “crossover interaction” pattern (Figure 2B), with stronger responses to the long-IOI tones (the large circles in Figure 2B) than to the short-IOI tones (the small circles in Figure 2B) regardless of block. A linear (additive) combination of a cumulative adaptation effect and a delay-dependent adaptation effect with no temporal-expectation violation effect would be characterized by a combination of the vertical separation and opposing slopes (Figure 2C). Finally (see Figure 2D), a result dominated by a delay-dependent adaptation effect and a temporal-expectation violation effect with no cumulative adaptation effect would be characterized by a crossover interaction pattern (Figure 2B) plus elevated responses to the temporal-expectation violating (long-IOI-block unexpectedly early) tones (the small light-blue circle in Figure 2D). Note that any additional cumulative adaptation effect would lower the shapes connected with the gray dashed line (i.e., reduced responses in the short-IOI block).

Figure 2. 

An illustration of expected results assuming negligible reorientation effects (see text for details). (A) A “cumulative adaptation effect” would cause a vertical separation based on block type, reducing responses in the short-IOI block (dashed line) relative to the long-IOI block (solid line) irrespective of IOI (large vs. small circles) or temporal deviance (gray vs. blue circles). (B) A “delay-dependent adaptation effect” would cause a “crossover interaction” pattern, reducing responses to the short-IOI tones (small circles) relative to the long-IOI tones (large circles) irrespective of block or temporal deviance. (C) A linear combination of a cumulative adaptation effect and a delay-dependent adaptation effect would include a combination of the vertical separation from the cumulative adaptation effect (A) and the opposing slopes from the delay-dependent effect (B). The exact pattern would depend on the relative strengths of the two effects. (D) The actual results were dominated by a delay-dependent adaptation effect and a temporal-expectation violation effect, characterized by the crossover interaction pattern (B) plus the elevated (arrow) responses to the temporal-expectation violating tones (the long-IOI-block unexpectedly early tones, small light-blue circle).

Figure 2. 

An illustration of expected results assuming negligible reorientation effects (see text for details). (A) A “cumulative adaptation effect” would cause a vertical separation based on block type, reducing responses in the short-IOI block (dashed line) relative to the long-IOI block (solid line) irrespective of IOI (large vs. small circles) or temporal deviance (gray vs. blue circles). (B) A “delay-dependent adaptation effect” would cause a “crossover interaction” pattern, reducing responses to the short-IOI tones (small circles) relative to the long-IOI tones (large circles) irrespective of block or temporal deviance. (C) A linear combination of a cumulative adaptation effect and a delay-dependent adaptation effect would include a combination of the vertical separation from the cumulative adaptation effect (A) and the opposing slopes from the delay-dependent effect (B). The exact pattern would depend on the relative strengths of the two effects. (D) The actual results were dominated by a delay-dependent adaptation effect and a temporal-expectation violation effect, characterized by the crossover interaction pattern (B) plus the elevated (arrow) responses to the temporal-expectation violating tones (the long-IOI-block unexpectedly early tones, small light-blue circle).

In summary, we evaluated the degree to which the auditory system maintained rhythm-based temporal expectation in a baseline condition where people passively heard rhythmic tones while attending to a nature video. We determined the strength of cumulative and delay-dependent auditory adaptation effects as well as the degree to which the auditory system automatically enhanced responses to temporal-expectation violating tones. We obtained strong delay-dependent adaptation effects and some cumulative adaptation effects that may accumulate most efficiently at a tone presentation rate of ∼1 Hz. Importantly, our results provide strong evidence suggesting that the auditory system substantially enhances responses to temporal-expectation violating tones to the extent of overriding the robust delay-dependent adaptation effect even under passive listening; further analyses suggest that this response enhancement is not because of disruption of adaptive inhibition but of boosting of responses in downstream processing.

EXPERIMENT 1: 1.7-SEC (LONG) VERSUS 1.2-SEC (SHORT) IOI

Methods

Participants

Twenty-six volunteers from Northwestern University and the greater Chicago area were recruited to participate in the study. All were right-handed and had normal or corrected-to-normal vision and normal hearing. None had a history of neurological disorder or concussion. Participants gave informed consent and were treated according to the guidelines of the institutional review board at Northwestern University. Participants received monetary compensation ($10 per hour) for their participation in an approximately 2.5-hr session, which included several experiments. Data collection for this experiment lasted approximately 30 min. Data from four participants were excluded because of excessive EEG artifacts (see EEG Recording and Preprocessing section for details). The final sample included 22 participants (14 women, seven men, one nonbinary) between ages 18 and 22 years (M = 19.86, SD = 1.21).

Stimuli and Procedure

We recorded scalp EEG while participants watched a muted nature video and passively heard rhythmic tones with occasional temporal perturbations. The participants were informed of the characteristics of the tones so that they did not become curious about them. We instructed participants to pay attention to the video and ignore the tones. The nature video was played on a 13-in. 2017 MacBook Pro (Apple, Inc.), placed 100 cm away from the participant. The brief tones (60 dB, 400 Hz, and 100 msec long with 10-msec rise and fall times) were presented using two loudspeakers placed behind the computer screen. In separate experimental blocks, we manipulated the rhythmicity of the tones by varying their IOIs. In the long-IOI block, tones were frequently presented (80% of the tones) with a 1.7-sec IOI—long-IOI-block standard tones—and infrequently presented (20% of the tones) with a 1.2-sec IOI—long-IOI-block unexpectedly early tones (Figure 1A). In the short-IOI block, tones were frequently presented (80%) with a 1.2-sec IOI—short-IOI-block standard tones—and infrequently presented (20%) with a 1.7-sec IOI—short-IOI-block late tones (Figure 1B). The order of tone types was randomized. In each block, we presented 10 standard “practice” tones (which were excluded from the analysis except for the examination of cumulative adaptation effects; see below) followed by 400 standard or deviant experimental tones. We ran each IOI block twice in alternation, with the order counterbalanced across participants. We gave participants a short break after each block. In total, each participant experienced 640 long-IOI-block standard tones, 160 long-IOI unexpectedly early tones, 640 short-IOI-block standard tones, and 160 short-IOI-block late tones.

EEG Recording and Preprocessing

Continuous EEG was recorded with a sampling rate of 512 Hz from 64 scalp electrodes, using a BioSemi ActiveTwo system (see www.biosemi.com for details). EOG activity was recorded using four face electrodes, one placed lateral to each eye and one placed beneath each eye. Two additional electrodes were placed on the left and right mastoids. Data were preprocessed using the EEGLAB and ERPLAB toolboxes for MATLAB (Lopez-Calderon & Luck, 2014; Delorme & Makeig, 2004). The single-ended EEG signals (recorded with online references using the common mode sense electrode located between Pz and PO3 and the driven right leg electrode located between Pz and PO4) were converted to differential signals offline, referenced to the average of the left and right mastoids. Nevertheless, because we applied a surface Laplacian transform to our data (discussed below), the choice of reference electrodes made little difference (Tenke & Kayser, 2012). The EEG and EOG data were then filtered using a bandpass filter (noncausal Butterworth impulse response function, half-amplitude cutoffs at 0.01 and 80 Hz, 12 dB/octave roll-off) and an additional notch filter (Parks–McClellan notch filter) at 60 Hz to remove line noise. Blink artifacts were removed via independent component analysis (ICA) using EEGLAB's runica function (Makeig, Bell, Jung, & Sejnowski, 1996). Up to two components were removed per participant (M = 1.6, SD = 0.49).

The continuous data were epoched in 3-sec windows, time-locked to the onset of each tone with a prestimulus period of 0.5 sec. After baseline subtraction (with the −0.3- to −0.1-sec prestimulus interval as the baseline period), the epochs with artifacts were first removed using the ERPLAB's standard algorithm that rejects epochs containing signals outside the 200-μV threshold; the remaining epochs were visually examined for additional artifacts. For four (of 26) participants, the EEG data had excessive artifacts (more than 30% of the epochs containing artifacts), so that their data were excluded from the analysis. For the remaining participants (≤ 30% removal), this procedure resulted in the removal of 16% of the epochs on average (SD = 9%, range = 4–30%). To reduce the effects of volume conduction and reference electrode choices, as well as to facilitate data-driven EEG source discrimination, we applied a surface Laplacian transform to all retained EEG data (Tenke & Kayser, 2012; Kayser & Tenke, 2006; Hjorth, 1980), using Perrin and associates' method (e.g., Perrin, Pernier, Bertrand, & Echallier, 1989a, 1989b; Perrin, Pernier, Bertrand, Giard, & Echallier, 1987) with a typical set of parameter values (Cohen, 2014).

ERP Analysis

For the ERP analysis (and ITPC analysis; see below), we selected the epochs that were time-locked to the onsets of standard tones that immediately preceded deviant tones. In this way, the ERP analysis was based on an equal number of responses to the standard and deviant tones, and responses to the repeated deviant tones were excluded. The long-IOI blocks yielded epochs in which a long-IOI standard tone was followed by an unexpectedly early tone, whereas the short-IOI blocks yielded epochs in which a short-IOI standard tone was followed by a late tone. From these two types of epochs, we computed the ERPs to the long-IOI-block standard tones, long-IOI-block unexpectedly early tones, short-IOI-block standard tones, and short-IOI-block late tones. Each epoch type had the same epoch length and baseline window as discussed above. On average, the analysis included 113 repetitions per tone type per participant after artifact rejection.

We chose the scalp-site ROI, namely, F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2 (10–20 system), based on prior studies reporting maximal auditory-evoked ERP responses in the frontocentral electrodes (see Pratt, 2011, for a review). The group-averaged ERP responses to the short- and long-IOI standard tones (averaged across the two experiments; n = 44) yielded visually distinct N1 (at 0.09–0.14 sec after tone onset) and P2 (at 0.15–0.21 sec after tone onset) components (the ERP plot in Figure 3). The scalp topographies of the average amplitudes (in t values) of N1 and P2 components show that the ROI reasonably reflected the auditory-evoked ERPs (the topographic scalp maps in Figure 3).

Figure 3. 

The ERP responses from the frontocentral ROI (F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2; 10–20 system) to the standard tones (averaged across Experiments 1 and 2 with n = 44) containing the N1 (at 0.09–0.14 sec after tone onset) and P2 (at 0.15–0.21 sec after tone onset) components. The scalp maps show the topographies of the average N1 and P2 responses (in t values). The black rectangles indicate the ROI chosen based on the topographic scalp maps and the literature (see text). The gray-shaded area represents ±1 SEM, adjusted for the within-participants comparisons (Morey, 2008).

Figure 3. 

The ERP responses from the frontocentral ROI (F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2; 10–20 system) to the standard tones (averaged across Experiments 1 and 2 with n = 44) containing the N1 (at 0.09–0.14 sec after tone onset) and P2 (at 0.15–0.21 sec after tone onset) components. The scalp maps show the topographies of the average N1 and P2 responses (in t values). The black rectangles indicate the ROI chosen based on the topographic scalp maps and the literature (see text). The gray-shaded area represents ±1 SEM, adjusted for the within-participants comparisons (Morey, 2008).

As an index of the strength of the ERP responses to each tone type, we computed the N1–P2 amplitude. Specifically, we averaged the ERP waveform across the ROI scalp sites (per tone type per participant) and bandpass filtered it at 0.1–30 Hz to reliably extract the N1 and P2 peaks with minimal noise; we then measured the absolute difference between these peaks. We obtained the N1–P2 amplitudes (per participant) for the long-IOI-block standard, long-IOI-block unexpectedly early, short-IOI-block standard, and short-IOI-block late tones. The method of combining N1 and P2 into a single ERP complex reflecting auditory sensory processing has been used since the early ERP studies of auditory perception and attention (e.g., Ford & Hillyard, 1981; Picton, Hillyard, Galambos, & Schiff, 1971), especially when using continuously presented auditory stimuli with different IOIs, which makes the identification of single ERP peaks problematic because of potential differences in baseline window across different presentation rates. Furthermore, the N1 and P2 components behaved similarly in our study; that is, in all critical comparisons, when the N1 peak amplitude was less negative, the P2 peak amplitude was also less positive and vice versa (see the Results section). This provides further justification for the use of combined N1–P2 amplitudes as the ERP measure for the current study. Nevertheless, it is important to note that another EEG study has reported dissociable effects of stimulus repetition and temporal expectation on auditory N1 and P2 (Costa-Faidella, Baldeweg, Grimm, & Escera, 2011). However, repetition and expectation were manipulated differently in that study. We pseudorandomly embedded timing-violating tones within a long sequence of regular-interval tones to examine neural responses to sporadic temporal-expectation violating tones; in contrast, Costa-Faidella et al. (2011) used sequences with regular and irregular rhythms to examine the effects of temporal regularity on repetition effects, used fewer repetitions (a maximum of 12), and also included changes in tone pitch. Thus, the N1 and P2 components may respond similarly or differently to manipulations of repetition and expectation depending on specific stimulus and/or task contexts.

ITPC Analysis

ITPC can be used to index the cross-trial consistency of EEG spectral phase relative to a specific event. ITPC can take on values between 0 and 1, where 0 reflects an absence of phase clustering and 1 reflects perfect phase clustering at a single angle across trials. For the current study, ITPC was time-locked to each of the four tone types, namely, the long-IOI-block standard, long-IOI-block unexpectedly early, short-IOI-block standard, and short-IOI-block late. That is, the degree of spectral-phase clustering was computed for each tone type, with the individual tones corresponding to trials. Thus, a larger ITPC value for a given tone type would indicate that EEG responses were more strongly time-locked to the corresponding tone type. However, because ITPC values also depend on spectral power, especially when the signal-to-noise ratio is low, a larger ITPC value in practice would suggest a combination of increased phase clustering and/or increased phase-locked (or evoked) power. Phase-locked power (the dB ratio of standard spectral power to non-phase-locked spectral power [spectral power computed after subtracting ERPs]) is closely correlated with ITPC because both reflect the extent to which spectrally decomposed EEG responses are time- and phase-locked to a stimulus (Cohen, 2014). We chose to report ITPC in the main analyses as the interpretation of ITPC more closely aligns with the temporal precision of neural responses, which we expected to be sensitive to our manipulations. The results based on phase-locked power, which as expected were essentially the same as those based on ITPC, are provided in the Supplementary Materials.1 For our purposes, a larger ITPC for a given tone type would indicate more robust (stronger and/or more temporally precise) sensory responses to the corresponding tone type. We included both ERP and ITPC measures because demonstrating consistent effects of adaptation and temporal expectation on sensory-evoked neural responses in terms of both raw and spectrally decomposed EEG increases confidence in the results. The ITPC measure also provides an additional benefit of elucidating the frequency components relevant to the ERP measure.

To compute ITPC, we applied a standard time–frequency wavelet transform to the EEG data to extract spectral power (amplitude squared) and phase as a function of time. The ITPC analysis utilized the spectral phase information, whereas the phase-locked power analysis presented in the Supplementary Materials utilized the spectral power information. For each relevant epoch (per scalp site per participant), we extracted the complex spectral amplitude as a function of time by convolving the EEG data with 60 time–frequency Gabor wavelets with their center frequencies fc and the factor n (roughly the number of cycles per wavelet, which is related to the temporal standard deviation of the wavelet by, SD = n/2πfc) logarithmically spaced (because neural temporal-frequency tunings tend to be approximately logarithmic; e.g., Lui, Bourne, & Rosa, 2007; Hess & Snowden, 1992). fc spanned the range of 2–55 Hz and n spanned the range of 3–16, resulting in temporal resolutions of SD = 239 msec (at 2 Hz) to SD = 46 msec (at 55 Hz) and spectral resolutions of FWHM = 1.56 Hz (at 2 Hz) to FWHM = 8.09 Hz (at 55 Hz). These values struck a good balance for the temporal–spectral resolution trade-off and are typically used in the literature (e.g., Cohen, 2014).

The ITPC analysis paralleled the ERP (N1–P2 amplitude) analysis. The long-IOI blocks yielded epochs in which a long-IOI standard tone was followed by an unexpectedly early tone, whereas the short-IOI blocks yielded epochs in which a short-IOI standard tone was followed by a late tone. We computed ITPC per fc as a function of time for the long-IOI and short-IOI epochs by averaging the normalized complex spectral amplitudes across all instances of each epoch type per fc per time point and taking the magnitude (per participant). The ITPC baseline (per fc per participant) was computed by averaging the ITPC values across the −0.3- to −0.1-sec prestimulus interval for each epoch type and averaging across the two epoch types. This common baseline was subtractively applied per participant. This procedure yielded baselined ITPC values (per fc per participant) as a function of time for the long-IOI epochs, which included baselined ITPC responses to the long-IOI standard and unexpectedly early tones, and for the short-IOI epochs, which included baselined ITPC responses to the short-IOI standard and late tones.

To select the appropriate frequency band for the ITPC analysis, we first plotted the ITPC response to all standard tones averaged across the ROI sites as a function of time and frequency. We visually determined from the time–frequency plots (see Figure 4D for Experiment 1 and Figure 5D for Experiment 2) that above-baseline ITPC responses to the tones occurred within the time window of −0.04 to 0.4 sec. The ITPC values within this window were averaged for each fc, and the range of the most responsive (top quartile) fc was determined for each experiment, yielding 3.5–8.5 Hz for Experiment 1 and 4.5–10.5 Hz for Experiment 2. We note that the pattern of results would not change if we used a single broad fc range (3.5–10.5 Hz) for both experiments. ITPC responses to all tone types had a single-peaked temporal profile (see Figure 4E for Experiment 1 and Figure 5E for Experiment 2). We thus determined the peak ITPC value for each tone type (per participant) by fitting the corresponding temporal profile with a second-order polynomial. These peak ITPC values were used in all analyses except for the examination of cumulative adaptation effects.

Figure 4. 

ERP and ITPC results from Experiment 1 (1.2- vs. 1.7-sec IOI). (A) Scalp-site ROI (sites F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2 based on the 10–20 system) used for ERP and ITPC measurements is shown. (B) Gray and light-blue wave: mean ERPs recorded from the ROI, elicited by the long-IOI-block standard tones (gray N1–P2 component) and the immediately following long-IOI-block unexpectedly early tones (light-blue N1–P2 component), time-locked to the onset of the standard tone. Gray and blue wave: mean ERPs recorded from the ROI, elicited by the short-IOI-block standard tones (gray N1–P2 component) and the immediately following short-IOI-block late tones (blue N1–P2 component), time-locked to the onset of the standard tone. Vertical dashed lines (and black arrows) indicated tone onsets (0 sec for the standard tones, 1.2 sec for the long-IOI-block unexpectedly early tones, and 1.7 sec for the short-IOI-block late tones); black trace indicated the baseline period (−0.3 to −0.1 sec). (C) The mean N1–P2 peak-to-peak amplitude as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). (D) Time–frequency plot of ITPC at the ROI, elicited by the standard tones, time-locked to the onset of the tones and subtractively baselined relative to the −0.3- to −0.1-sec prestimulus period. The black rectangle indicated the frequency band yielding the strongest (top quartile) ITPC response (3.5–8.5 Hz). (E) Gray and light-blue wave: mean 3.5- to 8.5-Hz ITPC recorded from the ROI, elicited by the long-IOI-block standard tones (gray peak) and the immediately following long-IOI-block unexpectedly early tones (light-blue peak), time-locked to the onset of the standard tone. Gray and blue wave: mean 3.5- to 8.5-Hz ITPC recorded from the ROI, elicited by the short-IOI-block standard tones (gray peak) and the immediately following short-IOI-block late tones (blue peak), time-locked to the onset of the standard tone. (F) Mean ITPC as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008). In C and F, the black line and black bracket reflect the delay-dependent adaptation effect, the blue line reflects cumulative adaptation effect differences between blocks, the dashed light-blue line reflects the effect of delay-dependent adaptation that is not overridden by the temporal-expectation violation effect, and the solid light-blue line reflects the temporal-expectation violation effect with respect to the standard tone matched for IOI.

Figure 4. 

ERP and ITPC results from Experiment 1 (1.2- vs. 1.7-sec IOI). (A) Scalp-site ROI (sites F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2 based on the 10–20 system) used for ERP and ITPC measurements is shown. (B) Gray and light-blue wave: mean ERPs recorded from the ROI, elicited by the long-IOI-block standard tones (gray N1–P2 component) and the immediately following long-IOI-block unexpectedly early tones (light-blue N1–P2 component), time-locked to the onset of the standard tone. Gray and blue wave: mean ERPs recorded from the ROI, elicited by the short-IOI-block standard tones (gray N1–P2 component) and the immediately following short-IOI-block late tones (blue N1–P2 component), time-locked to the onset of the standard tone. Vertical dashed lines (and black arrows) indicated tone onsets (0 sec for the standard tones, 1.2 sec for the long-IOI-block unexpectedly early tones, and 1.7 sec for the short-IOI-block late tones); black trace indicated the baseline period (−0.3 to −0.1 sec). (C) The mean N1–P2 peak-to-peak amplitude as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). (D) Time–frequency plot of ITPC at the ROI, elicited by the standard tones, time-locked to the onset of the tones and subtractively baselined relative to the −0.3- to −0.1-sec prestimulus period. The black rectangle indicated the frequency band yielding the strongest (top quartile) ITPC response (3.5–8.5 Hz). (E) Gray and light-blue wave: mean 3.5- to 8.5-Hz ITPC recorded from the ROI, elicited by the long-IOI-block standard tones (gray peak) and the immediately following long-IOI-block unexpectedly early tones (light-blue peak), time-locked to the onset of the standard tone. Gray and blue wave: mean 3.5- to 8.5-Hz ITPC recorded from the ROI, elicited by the short-IOI-block standard tones (gray peak) and the immediately following short-IOI-block late tones (blue peak), time-locked to the onset of the standard tone. (F) Mean ITPC as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008). In C and F, the black line and black bracket reflect the delay-dependent adaptation effect, the blue line reflects cumulative adaptation effect differences between blocks, the dashed light-blue line reflects the effect of delay-dependent adaptation that is not overridden by the temporal-expectation violation effect, and the solid light-blue line reflects the temporal-expectation violation effect with respect to the standard tone matched for IOI.

Figure 5. 

ITPCrz3.5–8.5 Hz results from Experiment 1 (1.2- vs. 1.7-sec IOI). (A) Mean ITPCrz (Rayleigh's z-transformed ITPC) elicited by the standard tones in the long-IOI (large gray scales) and the short-IOI (small gray circles) blocks, plotted for every 10-trial bin over 41 bins. The lines show linear fits. (B) Mean linear slope for the long-IOI and short-IOI blocks. The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008).

Figure 5. 

ITPCrz3.5–8.5 Hz results from Experiment 1 (1.2- vs. 1.7-sec IOI). (A) Mean ITPCrz (Rayleigh's z-transformed ITPC) elicited by the standard tones in the long-IOI (large gray scales) and the short-IOI (small gray circles) blocks, plotted for every 10-trial bin over 41 bins. The lines show linear fits. (B) Mean linear slope for the long-IOI and short-IOI blocks. The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008).

For the cumulative adaptation analysis, we sequentially segmented the 410 EEG epochs (each time-locked to a tone; see above) in each block into 41 bins of 10 epochs (or tones). Because we ran both the long-IOI and short-IOI blocks twice, we had 20 epochs (tones) per bin per participant. Because we only analyzed EEG responses to standard tones (unaffected by temporal expectation effects) in the cumulative adaptation analysis and because some epochs were removed by artifact rejection, we actually had 13 epochs (or tones) per bin per participant on average. We only used the ITPC measure for this analysis because it was difficult to reliably determine the N1 and P2 peaks per participant with the small number of epochs (tones) per bin. We used Rayleigh's z-transformed ITPC, ITPCrz, because it is less sensitive to variations in the number of epochs (or tones) across bins especially as each bin contained a relatively small number of epochs (tones; Cohen, 2014). The small number of epochs (tones) per bin, generating noisy temporal profiles of ITPCrz, also made it difficult to determine the peak ITPCrz value per bin; instead, we averaged the ITPCrz values over the critical −0.04- to 0.4-sec interval (see above) per bin. We analyzed the resultant ITPCrz as a function of bin for the long-IOI and short-IOI blocks, using a statistically significant negative linear slope as evidence of a cumulative adaptation effect. Note that our bins are based on the serial positions of the tones rather than the elapsed time; thus, a larger negative slope would indicate stronger contributions of individual tones to cumulative adaptation.

Results and Discussion

The ERP and ITPC responses were analyzed according to the planned comparisons discussed in the Introduction section. The time courses of the ERP and ITPC responses are shown in Figure 4B and 4E, respectively.

Adaptation and Temporal-Expectation Violation Effects Based on ERP Results

A cumulative adaptation effect difference between blocks was not evident as the N1–P2 response was statistically equivalent for the slower rate (for the long-IOI-block standard tones; the large gray circle in Figure 4C) and faster rate (for the short-IOI-block late tones; the large blue circle in Figure 4C), t(21) = 0.29, p = .77, controlling for IOI and temporal expectation (the blue line in Figure 4C; see Cumulative Adaptation section for the results of bin analyses testing the presence/strength of cumulative adaptation effects).

A delay-dependent adaptation effect was observed as the N1–P2 responses were significantly larger for the short-IOI-block late tones (large blue circle) than for the short-IOI-block standard tones (small gray circle), t(21) = 3.78, p = .001, d = 0.81, indicating that the N1–P2 responses substantially recovered from adaptation to the preceding tone after a 1.7-sec delay relative to after a 1.2-sec delay, controlling for block and temporal expectation (the black line in Figure 4C). Given that we did not observe a cumulative adaptation effect difference between blocks (i.e., no block effect), without needing to control for block, the presence of a delay-dependent adaptation effect can also be measured as significantly greater N1–P2 responses to the long-IOI-block standard tones (the large gray circle in Figure 4C) relative to the short-IOI-block standard tones (the small gray circle in Figure 4C), t(21) = 2.12, p = .046, d = 0.45 (the black brace in Figure 4C).

Crucially, as for the temporal-expectation violation effect, the N1–P2 responses to the long-IOI-block standard tones (the large gray circle in Figure 4C) and the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 4C) were not significantly different, t(21) = 0.57, p = .57, suggesting that auditory responses to the temporal-expectation violating tones were boosted to the extent of virtually overriding delay-dependent adaptation, controlling for block (the dashed light-blue line in Figure 4C). Given that we did not observe a block difference in cumulative adaptation effect, without needing to control for block, the presence of a temporal-expectation violation effect could also be measured as significantly greater N1–P2 responses to the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 4C) relative to the short-IOI-block standard tones (the small gray circle in Figure 4C), t(21) = 2.29, p = .032, d = 0.49 (the solid light-blue line in Figure 4C).

Adaptation and Temporal-Expectation Violation Effects Based on ITPC Results

The ITPC results paralleled the ERP results. The absence of a cumulative adaptation effect difference between blocks was confirmed as the ITPC3.5–8.5 Hz responses (the baselined ITPC averaged across the fc range of 3.5–8.5 Hz; see the Methods section) were statistically equivalent for the slower rate (for the long-IOI-block standard tones; the large gray circle in Figure 4F) and the faster rate (for the short-IOI-block late tones; the large blue circle in Figure 4F), t(21) = 0.41, p = .69, controlling for IOI and temporal expectation (the blue line in Figure 4F).

The delay-dependent adaptation effect was confirmed as the ITPC3.5–8.5 Hz responses were significantly higher for the short-IOI-block late tones (the large blue circle in Figure 4F) than for the short-IOI-block standard tones (the small gray circle in Figure 4F), t(21) = 4.56, p < .001, d = 0.97, indicating that the ITPC3.5–8.5 Hz response substantially recovered from adaptation to the preceding tone after a 1.7-sec delay relative to after a 1.2-sec delay, controlling for block and temporal expectation (the black line in Figure 4F). Given that we did not observe a cumulative adaptation effect difference between blocks, without needing to control for block, the presence of a delay-dependent adaptation effect was also confirmed by significantly higher ITPC3.5–8.5 Hz responses to long-IOI-block standard tones (the large gray circle in Figure 4F) relative to short-IOI-block standard tones (the small gray circle in Figure 4F), t(21) = 3.33, p = .003, d = 0.71 (the black brace in Figure 4F).

The nonsignificant difference between the ITPC3.5–8.5 Hz responses to the long-IOI-block standard tones (the large gray circle in Figure 4F) and the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 4F), t(21) = 1.51, p = .15, confirmed that auditory responses to the temporal-expectation violating tones were boosted to the extent of virtually overriding delay-dependent adaptation, controlling for block (the dashed light-blue line in Figure 4F). The complementary comparison justified by the absence of a cumulative adaptation effect difference between blocks further revealed significantly higher ITPC3.5–8.5 Hz responses to the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 4F) relative to the short-IOI-block standard tones (the small gray circle in Figure 4F), t(21) = 2.71, p = .013, d = 0.58, confirming the presence of a temporal-expectation violation effect (the solid light-blue line in Figure 4F).

Cumulative Adaptation

We examined how the ITPC3.5–8.5 Hz values gradually diminished as a function of the number of presented tones by computing ITPC3.5–8.5 Hz values for sequential bins of 10 tones; note that the ITPC3.5–8.5 Hz value for each bin (per participant) was computed based on 20 tones because the long-IOI and short-IOI blocks were run twice per participant; the actual number of tones per bin were fewer because we only included standard tones in this analysis and numbers were also reduced because of artifact rejection; and Rayleigh's z-transformed ITPC3.5–8.5 Hz values, ITPCrz3.5–8.5 Hz, were used because of the relatively small and variable number of tones across bins (see Methods section). The strength of cumulative adaptation was assessed by the linear slope of ITPCrz3.5–8.5 Hz as a function of time bin. For the long (1.7 sec)-IOI block, the slope (M = −0.005, SE = 0.004) was not significantly different from zero, t(21) = −1.33, p = .20 (the large circles in Figure 5A). For the short (1.2 sec)-IOI block, the slope (M = −0.008, SE = 0.004) was marginally negative, t(21) = −2.04, p = .054, d = −0.44 (the small circles in Figure 5A). The results are numerically consistent with the expectation that any cumulative adaptation would accumulate faster as a function of the number of presented tones when the tone presentation rate is faster. However, the slopes for the faster-rate and slower-rate conditions were not significantly different, t(21) = 0.80, p = .43 (Figure 5B). Thus, any cumulative adaptation effects were weak based on this analysis.

From another perspective, this nearly null result helps to resolve the ambiguity inherent in our primary analysis because of the potentially nonnegligible effects of reorientation. Whereas a cumulative adaptation effect would reduce the ITPC3.5–8.5 Hz responses to the short-IOI-block late tones (the large blue circle in Figure 4F) relative to the long-IOI-block standard tones (the large gray circle in Figure 4F), a potential reorientation effect would increase the responses to the late tones. The nonsignificant difference in the cumulative adaptation slopes between the faster and slower rates suggests that any long-term adaptation-based response reduction to the late tones was minimal. Because the ITPC3.5–8.5 Hz responses were statistically equivalent for the short-IOI-block late tones (the large blue circle in Figure 4F) and the long-IOI-block standard tones (the large gray circle in Figure 4F), a logical conclusion would be that any reorientation effect was minimal.

In summary, the ERP and ITPC results demonstrated a strong delay-dependent adaptation effect from which the auditory system substantially recovered after a 1.7-sec delay (relative to a 1.2-sec delay) and a weak-to-negligible cumulative adaptation effect (although the linear adaptation slope of cumulative adaptation was numerically stronger for the faster rate). Importantly, the unexpectedly early tones elicited strong responses virtually overriding the strong delay-dependent adaptation effect. In contrast, the late tones did not noticeably influence the responses consistent with the interpretation that the auditory system automatically reoriented to the longer delay upon experiencing the absence of an expected tone.

EXPERIMENT 2: 1.2-SEC (LONG) VERSUS 0.7-SEC (SHORT) IOI

This experiment was the same as Experiment 1 except that the IOIs were reduced so that the long IOI was 1.2 sec (equivalent to the short IOI in Experiment 1) and the short IOI was 0.7 sec.

Methods

Participants

Twenty-three different volunteers from Northwestern University and the greater Chicago area were recruited to participate in the study. All were right-handed and had normal or corrected-to-normal vision and normal hearing. None had a history of neurological disorder or concussion. Participants gave informed consent and were treated according to the guidelines of the institutional review board at Northwestern University. Participants received monetary compensation ($10 per hour) for their participation in an approximately 2.5-hr session, which included several experiments. Data collection for this experiment lasted approximately 30 min. Data from one participant were excluded because of excessive EEG artifacts using the same criterion as in Experiment 1 (more than 30% of epochs containing artifacts; see EEG Recording and Preprocessing section for Experiment 1). The final sample included 22 participants (15 women, seven men) between ages 18 and 29 years (M = 22.18, SD = 3.89). The effective sample size in this experiment was the same as that in Experiment 1.

Stimuli and Procedure

These were the same as in Experiment 1 except that the long and short IOIs were reduced to 1.2 sec (the short IOI in Experiment 1) and 0.7 sec, respectively.

EEG Recording and Preprocessing: EEG Data Analysis

These were the same as in Experiment 1 except that the baselined ITPC was averaged across the fc range of 4.5–10.5 Hz (see the ITPC Analysis section of the Methods section for Experiment 1). In this experiment, the artifact detection procedure resulted in the removal of 7% of the epochs on average (SD = 7%, range = 1–24%). The blink artifact removal procedure using ICA resulted in the removal of up to two components per participant (M = 1.2, SD = 0.39).

Results and Discussion

As in Experiment 1, the ERP and ITPC responses were analyzed according to the planned comparisons discussed in the Introduction section. The time courses of the ERP and ITPC responses are shown in Figure 6B and 6E, respectively.

Figure 6. 

ERP and ITPC results from Experiment 2 (0.7- vs. 1.2-sec IOI). (A) Scalp-site ROI (sites F1, Fz, FC1, FCz, FC2, C1, Cz, and C2 based on the 10–20 system) used from ERP and ITPC measurements is shown. (B) Gray and light-blue wave: mean ERPs recorded from the ROI, elicited by the long-IOI-block standard tones (gray N1–P2 component) and the immediately following long-IOI-block unexpectedly early tones (light-blue N1–P2 component), time-locked to the onset of the standard tone. Gray and blue wave: mean ERPs recorded from the ROI, elicited by the short-IOI-block standard tones (gray N1–P2 component) and the immediately following short-IOI-block late tones (blue N1–P2 component), time-locked to the onset of the standard tone. Vertical dashed lines (and black arrows) indicate the tone onsets (0 sec for the standard tones, 0.7 sec for the long-IOI-block unexpectedly early-tones, and 1.2 for the short-IOI-block late tones); black trace indicated the baseline period (−0.3 to −0.1 sec). (C) The mean N1–P2 peak-to-peak amplitude as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). (D) Time–frequency plot of ITPC at the ROI, elicited by the standard tones, time-locked to the onset of the tones and subtractively baselined relative to the −0.3- to −0.1-sec prestimulus period. The black rectangle indicates the frequency band yielding the strongest (top quartile) ITPC response (4.5–10.5 Hz). (E) Gray and light-blue wave: mean 4.5- to 10.5-Hz ITPC recorded from the ROI, elicited by the long-IOI-block standard tones (gray peak) and the immediately following long-IOI-block unexpectedly early tones (light-blue peak), time-locked to the onset of the standard tone. Gray and blue wave: mean 4.5- to 10.5-Hz ITPC recorded from the ROI, elicited by the short-IOI-block standard tones (gray peak) and the immediately following short-IOI-block late tones (blue peak), time-locked to the onset of the standard tone. (F) Mean ITPC as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008). In C and F, the black line and black bracket reflect the delay-dependent adaptation effect, the blue line reflects the cumulative adaptation effect differences between blocks, the dashed light-blue line reflects the effect of delay-dependent adaptation that is not overridden by the temporal-expectation violation effect, and the solid light-blue line reflects the temporal-expectation violation effect with respect to the standard tone matched for IOI.

Figure 6. 

ERP and ITPC results from Experiment 2 (0.7- vs. 1.2-sec IOI). (A) Scalp-site ROI (sites F1, Fz, FC1, FCz, FC2, C1, Cz, and C2 based on the 10–20 system) used from ERP and ITPC measurements is shown. (B) Gray and light-blue wave: mean ERPs recorded from the ROI, elicited by the long-IOI-block standard tones (gray N1–P2 component) and the immediately following long-IOI-block unexpectedly early tones (light-blue N1–P2 component), time-locked to the onset of the standard tone. Gray and blue wave: mean ERPs recorded from the ROI, elicited by the short-IOI-block standard tones (gray N1–P2 component) and the immediately following short-IOI-block late tones (blue N1–P2 component), time-locked to the onset of the standard tone. Vertical dashed lines (and black arrows) indicate the tone onsets (0 sec for the standard tones, 0.7 sec for the long-IOI-block unexpectedly early-tones, and 1.2 for the short-IOI-block late tones); black trace indicated the baseline period (−0.3 to −0.1 sec). (C) The mean N1–P2 peak-to-peak amplitude as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). (D) Time–frequency plot of ITPC at the ROI, elicited by the standard tones, time-locked to the onset of the tones and subtractively baselined relative to the −0.3- to −0.1-sec prestimulus period. The black rectangle indicates the frequency band yielding the strongest (top quartile) ITPC response (4.5–10.5 Hz). (E) Gray and light-blue wave: mean 4.5- to 10.5-Hz ITPC recorded from the ROI, elicited by the long-IOI-block standard tones (gray peak) and the immediately following long-IOI-block unexpectedly early tones (light-blue peak), time-locked to the onset of the standard tone. Gray and blue wave: mean 4.5- to 10.5-Hz ITPC recorded from the ROI, elicited by the short-IOI-block standard tones (gray peak) and the immediately following short-IOI-block late tones (blue peak), time-locked to the onset of the standard tone. (F) Mean ITPC as a function of temporal deviance (standard vs. deviant) and IOI (short vs. long). The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008). In C and F, the black line and black bracket reflect the delay-dependent adaptation effect, the blue line reflects the cumulative adaptation effect differences between blocks, the dashed light-blue line reflects the effect of delay-dependent adaptation that is not overridden by the temporal-expectation violation effect, and the solid light-blue line reflects the temporal-expectation violation effect with respect to the standard tone matched for IOI.

Adaptation and Temporal-Expectation Violation Effects Based on ERP Results

We replicated the ERP results of Experiment 1 for the shorter pair of IOIs (1.2 and 0.7 sec). A cumulative adaptation effect difference between blocks was not evident as the N1–P2 responses were statistically equivalent for the slower rate (for the long-IOI-block standard tones; the large gray circle in Figure 6C) and faster rate (for the short-IOI-block late tones; the large blue circle in Figure 6C), t(21) = 0.17, p = .87, controlling for IOI and temporal expectation (the blue line in Figure 6C).

A delay-dependent adaptation effect was observed as the N1–P2 responses were significantly larger for the short-IOI-block late tones (the large blue circle in Figure 6C) than for the short-IOI-block standard tones (the small gray circle in Figure 6C), t(21) = 4.58, p < .001, d = 0.98, indicating that the N1–P2 responses substantially recovered from adaptation to the preceding tone after a 1.2-sec delay relative to after a 0.7-sec delay, controlling for block and temporal expectation (the black line in Figure 6C). Given that we did not observe a cumulative adaptation effect difference between blocks, without needing to control for block, the presence of delay-dependent adaptation effect can also be measured as significantly greater N1–P2 responses to the long-IOI-block standard tones (large gray circle) relative to the short-IOI-block standard tones (the small gray circle in Figure 6C), t(21) = 5.97, p < .001, d = 1.27 (the black brace in Figure 6C).

Crucially, as for the temporal-expectation violation effect, the N1–P2 responses to the long-IOI-block standard tones (the large gray circle in Figure 6C) and the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 6C) were not significantly different, t(21) = 0.92, p = .37, suggesting that auditory responses to the temporal-expectation violating tones were boosted to the extent of virtually overriding delay-dependent adaptation, controlling for block (the light-blue line in Figure 6C). Given that we did not observe a block difference in cumulative adaptation effect, without needing to control for block, the presence of a temporal-expectation violation effect can also be measured as significantly greater N1–P2 responses to the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 6C) relative to the short-IOI-block standard tones (the small gray circle in Figure 6C), t(21) = 3.79, p = .001, d = 0.81 (the solid light-blue line in Figure 6C).

Adaptation and Temporal-Expectation Violation Effects Based on ITPC Results

We also replicated the ITPC results of Experiment 1 for the shorter pair of IOIs (1.2 and 0.7 sec). A marginal cumulative adaptation effect difference between blocks was observed as the ITPC4.5–10.5 Hz responses were somewhat reduced for the faster rate (for the short-IOI-block late tones; the large blue circle in Figure 6F) relative to the slower rate (for the long-IOI-block standard tones; the large gray circle in Figure 6F), t(21) = 1.76, p = .093, controlling for IOI and temporal expectation (the blue line in Figure 6F).

The delay-dependent adaptation effect was confirmed as the ITPC4.5–10.5 Hz responses were significantly higher for the short-IOI-block late tones (the large blue circle in Figure 6F) than for the short-IOI-block standard tones (the small gray circle in Figure 6F), t(21) = 5.71, p < .0001, d = 1.22, indicating that the ITPC4.5–10.5 Hz responses substantially recovered from adaptation to the preceding tone after a 1.2-sec delay relative to after a 0.7-sec delay, controlling for block and temporal expectation (the black line in Figure 6F). Given that we did not observe a strong cumulative adaptation effect difference between blocks, without needing to control for block, the presence of a delay-dependent adaptation effect was also confirmed by significantly higher ITPC4.5–10.5 Hz responses to the long-IOI-block standard tones (the large gray circle in Figure 6F) relative to the short-IOI-block standard tones (the small gray circle in Figure 6F), t(21) = 6.80, p < .0001, d = 1.45 (the black brace in Figure 6F).

The ITPC4.5–10.5 Hz responses were marginally reduced for the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 6F) relative to the long-IOI-block standard tones (the large gray circle in Figure 6F), t(21) = 2.00, p = .058, suggesting that auditory responses to the temporal-expectation violating tones were boosted to the extent of nearly (but not fully) overriding delay-dependent adaptation, controlling for block (the dashed light-blue line in Figure 6F). The complementary comparison justified by the absence of a strong cumulative adaptation effect difference between blocks further revealed significantly higher ITPC4.5–10.5 Hz responses to the long-IOI-block unexpectedly early tones (the small light-blue circle in Figure 6F) relative to the short-IOI-block standard tones (the small gray circle in Figure 6F), t(21) = 3.40, p = .003, d = 0.72, confirming the presence of a temporal-expectation violation effect (the solid light-blue line in Figure 6F).

Cumulative Adaptation

Paralleling Experiment 1, we examined how the ITPC4.5–10.5 Hz values gradually diminished as a function of the overall number of presented tones by computing the ITPCrz4.5–10.5 Hz values for sequential bins of 10 tones, with the linear slope of ITPCrz4.5–10.5 Hz (per time bin) indicating the strength of cumulative adaptation. For the long (1.2 sec)-IOI block (presenting tones at the slower rate), the slope (M = −0.011, SE = 0.004) was significantly negative, t(21) = −3.14, p = .005, d = 0.67 (the large circles in Figure 7A). For the short (0.7 sec)-IOI block (presenting tones at the faster rate), the slope (M = −0.008, SE = 0.004) was not significantly different from zero, t(21) = −1.65, p = .12 (the small circles in Figure 7A). Furthermore, the slope for the slower-rate condition was significantly more negative than the slope for the faster-rate condition, t(21) = 2.56, p = .018, d = 0.55 (Figure 7B).

Figure 7. 

ITPCrz4.5–10.5 Hz results from Experiment 2 (1.2 vs. 1.7 sec). (A) Mean ITPCrz (Rayleigh's z-transformed ITPC) elicited by the standard tones in the long-IOI (large gray circles) and the short-IOI (small gray circles) blocks, plotted for every 10-trial bin over 41 bins. The lines show linear fits. (B) Mean linear slope for the long-IOI and short-IOI blocks. The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008).

Figure 7. 

ITPCrz4.5–10.5 Hz results from Experiment 2 (1.2 vs. 1.7 sec). (A) Mean ITPCrz (Rayleigh's z-transformed ITPC) elicited by the standard tones in the long-IOI (large gray circles) and the short-IOI (small gray circles) blocks, plotted for every 10-trial bin over 41 bins. The lines show linear fits. (B) Mean linear slope for the long-IOI and short-IOI blocks. The error bars represent ±1 SEM, adjusted for within-participants comparisons (Morey, 2008).

Given that the cumulative adaptation slope was marginally negative for the 1.2-sec-IOI block in Experiment 1, the significantly negative slope obtained for the same tone presentation rate in this experiment may suggest that a ∼1-Hz rhythm optimally induces auditory cumulative adaptation. However, additional research is necessary to investigate this speculation because the adaptation slope for the 1.7-sec-IOI block was only numerically less negative in Experiment 1 and because the significantly less negative adaptation slope for the 0.7-sec-IOI block obtained in this experiment (Figure 7) could reflect a floor effect because of the strong delay-dependent adaptation effect.

There appears to be a contradiction between this cumulative adaptation slope analysis and the preceding main analysis. Although the adaptation slope was more negative for the slower tone rate than for the faster tone rate (Figure 7), the main analysis yielded a seemingly opposite marginal effect where the ITPC responses were lower for the faster rate than for the slower rate (the blue line in Figure 6F), controlling for the tone delay. Note that we used bins of 10 tones to conduct the cumulative adaptation slope analysis. Thus, the slope analysis did not include any adaptation or entrainment that rapidly accumulated over the first 10 tones. Because a 0.7-sec IOI is within the range of typical rates of spontaneous rhythmic tapping (Repp, 2005), it is possible that the 0.7-sec-IOI condition rapidly generated rhythmic habituation that reduced auditory responses over the first 10 tones. This is feasible as it has been shown that 10 rhythmic repetitions are sufficient to induce automatic (bottom–up) rhythm-based temporal expectations (Lange, 2010). The cumulative adaptation effect revealed in the main analysis (the blue line in Figure 6F) reflects the sum of any potential rhythmic habituation developed over the first 10 trials and the cumulative adaptation effects indicated by the adaptation slopes. Thus, if our line of reasoning is correct, the overall results suggest that the total cumulative adaptation effect as the sum of the rapid rhythmic habituation and cumulative adaptation was marginally stronger for the 0.7-sec-IOI condition than for the 1.2-sec-IOI condition.

Potential Mechanisms Underlying the Enhanced Response to Temporal-Expectation Violating Tones

Our design allowed us to consider two potential mechanisms underlying the ERP and ITPC response enhancements to the unexpectedly early tones. One is that, when the mechanisms that monitor rhythm-based temporal expectations detect a deviant (i.e., potentially behaviorally significant) stimulus, the sensory response to it is boosted by temporarily disengaging the inhibitory mechanisms of sensory adaptation. If so, delay-dependent adaptation after an unexpectedly early tone would be weak. The other possibility is that, akin to selective attention, the sensory response to a deviant stimulus is actively boosted via top–down feedback. If so, delay-dependent adaptation after an unexpectedly early tone would be just as strong as that after a standard (expectation fulfilling) tone or be potentially stronger given that, in the visual modality, selective attention has been shown to increase sensory adaptation (e.g., Suzuki, 2001, 2003; Suzuki & Grabowecky, 2003; Spivey & Spirn, 2000).

To evaluate these possibilities, we compared the N1–P2 and ITPC responses between the two consecutive long-IOI-block standard tones that immediately followed an unexpectedly early tone. If rhythm-based temporal-expectation monitoring mechanisms temporarily disengaged delay-dependent adaptation upon detecting a deviant tone, responses to the immediately following standard tone should be stronger than those to the next standard tone (which would be subjected to the regular delay-dependent adaptation effects from the immediately preceding standard tone). In contrast, if expectation-monitoring mechanisms boosted sensory responses to a deviant tone via top–down feedback (akin to attention mechanisms), responses to both standard tones would be equivalent, or the responses to the first standard tones may be reduced because of stronger adaptation to the “attention-capturing” deviant tone. In Experiment 1, the N1–P2 responses were significantly weaker for the first than second standard tones (M = 14.13 and SE = 1.56 for the first standard tones; M = 15.26 and SE = 1.75 for the second standard tones), t(21) = 2.49, p = .02, d = 0.53, whereas the ITPC3.5–8.5 Hz responses were not significantly different between them (M = 0.14 and SE = 0.02 for the first standard tones; M = 0.15 and SE = 0.02 for the second standard tones), t(21) = 1.66, p = .11. In Experiment 2, neither the N1–P2 responses (M = 9.90 and SE = 0.87 for the first standard tones; M = 9.00 and SE = 0.79 for the second standard tones), t(21) = −1.32, p = .2, nor the ITPC4.5–10.5 Hz responses (M = 0.09 and SE = 0.01 for the first standard tones; M = 0.10 and SE = 0.01 for the second standard tones), t(21) = −1.37, p = .18, were significantly different between the first and second standard tones. These results are consistent with the interpretation that rhythm-based temporal-expectation monitoring mechanisms actively boost responses to temporal-expectation violating stimuli via top–down feedback akin to auditory attention capture (e.g., Schmitt, Postma, & De Haan, 2000; Spence & Driver, 1994) rather than indirectly boosting responses by disengaging the delay-dependent adaptation mechanisms.

GENERAL DISCUSSION

In two EEG experiments, we investigated how auditory sensory-evoked neural responses (measured as ERPs and ITPCs) adapted to auditory rhythm and reacted to stimuli that violated the rhythm when participants passively listened to the tones. The design illustrated in Figure 2 allowed us to assess delay-dependent adaptation effects, cumulative adaptation effects, and potential auditory response boosts to temporal-expectation violating tones.

We observed reliable delay-dependent adaptation effects across the two experiments as the reduced ERP/ITPC responses at a 1.2-sec delay relative to a 1.7-sec delay (Experiment 1) and the reduced responses at a 0.7-sec delay relative to a 1.2-sec delay (Experiment 2), controlling for any effects of cumulative adaptation (the black line in Figures 4C, 4F, 6C, and 6F). We obtained little evidence of cumulative adaptation effects for the slowest (1.7-sec IOI) and fastest (0.7-sec IOI) tone presentation rates. The cumulative adaptation slope was marginally negative (Experiment 1) and significantly negative (Experiment 2) for the intermediate (1.2-sec IOI) tone presentation rate. This result may suggest that rhythmic auditory presentation at ∼1 Hz optimizes the growth of cumulative adaptation. Nevertheless, the lack of cumulative adaptation at the faster rate (0.7-sec IOI) may reflect a floor effect potentially from rapid habituation to rhythm that occurred over the initial ∼10 tones especially with 0.7-sec IOI representing a typical rate of spontaneous tapping (Repp, 2005). The lack of strong cumulative adaptation effects (especially in Experiment 1) and the consistent results of nonsignificant differences in the ERP/ITPC responses between the long-IOI-block standard tones and the short-IOI-block late tones (the blue line in Figures 4C, 4F, 6C, and 6F) provide little evidence of any effects of a tone being “unexpectedly” late, suggesting that the auditory system successfully reorients to the longer delay in the rare cases where a tone did not occur at the expected delay, responding as if the late tones were expected.

Crucially, we have demonstrated that the ERP/ITPC responses were significantly elevated for the temporally deviant tones relative to the standard (expected) tones over and above any adaptation effects in both experiments. The ERP/ITPC responses in both experiments were nonsignificantly (or, at most, marginally) different between the long-IOI-block standard tones and the long-IOI-block unexpectedly early tones (the dashed light-blue line in Figures 4C, 4F, 6C, and 6F), suggesting that the unexpectedly early tones elicited strong auditory responses, fully or nearly overriding the strong delay-dependent adaptation effects. Note that this particular interpretation is independent of whether or not the temporal reorientation effects are actually minimal.

Our delay-dependent adaptation findings are consistent with and extend the findings of several MEG/EEG studies of auditory perception that show interval-dependent adaptation effects for repeating tones under a variety of task demands (e.g., Pereira et al., 2014; Carver et al., 2002; Hari et al., 1982; Nelson & Lassman, 1973). For instance, Nelson and Lassman (1973) found that the peak-to-peak amplitude of the auditory N1–P2 complex elicited by tones (18 msec, 1000 Hz) increased as a logarithmic function of the IOIs using intervals of 0.5, 1, 2, 4, and 8 sec, while participants read a fiction text. Hari et al. (1982) further showed that the amplitude of the auditory N1 as well as its MEG counterpart N1m in response to the tones (20 msec, 1000 Hz) decreased with faster presentation rates using intervals of 1, 2, 4, 8, and 16 sec, while participants mentally counted the tones. Carver et al. (2002) demonstrated that N1m to tones (60 msec, 1000 Hz) linearly decreased as IOIs decreased, using faster presentation rates ranging between 0.6 and 8.1 Hz while participants passively listened to the tones. More recently, Pereira et al. (2014) showed similar interval-dependent effects on the auditory N1 and P2 components elicited by tones (60 msec, 1000 Hz) using intervals of 0.6, 1, 3, and 6 sec. This effect was present regardless of whether participants were passively listening to the tones or monitoring for rare frequency-deviant tones. In our study, we replicated these findings using the tone (100 msec, 440 Hz) intervals of 0.7, 1.2, and 1.7 sec, while clearly dissociating delay-dependent adaptation effects from cumulative adaptation effects.

Our cumulative adaptation findings provided tentative evidence suggesting that a ∼1-Hz presentation rate optimizes cumulative auditory adaptation and that a ∼2-Hz presentation rate establishes habituation to rhythm within ∼10 repetitions. These interpretations are consistent with animal studies that investigated auditory adaptation in longer time scales (e.g., Ulanovsky et al., 2003, 2004). For instance, Ulanovsky et al. (2004) investigated the time scales of frequency-specific neural adaptation to repetitive tones (230-msec tones presented with 0.736-sec IOI) in cat primary auditory cortex (A1). They observed a fast adaptation taking place within hundreds of milliseconds, potentially corresponding to our delay-dependent adaptation effects, a slow cumulative adaptation that lasted up to tens of seconds, potentially corresponding to our habituation-to-rhythm effect, and a very slow and small adaptation up to a few hundred seconds, potentially corresponding to our cumulative adaptation effects.

Early ERP studies investigated the auditory responses to temporally deviant tones in paradigms similar to the current study (e.g., Nordby, Roth, & Pfefferbaum, 1988; Ford & Hillyard, 1981). For instance, Ford and Hillyard (1981) tested the sensitivity of the auditory N1–P2 responses to temporal deviants in rhythmic tone sequences. In their first experiment, for one group of participants (n = 6), they presented standard tones (50 msec) with a regular 0.6-sec IOI and rare, early tones with a 0.3-sec IOI. For another group of participants (n = 6), they presented the standard tones with a regular 0.3-sec IOI and rare, late tones with a 0.6-sec IOI. Participants passively listened to the tones while reading a book. The early tones elicited larger N1–P2 responses than the standard tones at Cz and Fz, whereas the late and standard tones elicited smaller and similar N1–P2 responses. However, the authors noted that their standard tones did not elicit responses reliably above the noise level for many of their participants. In their second experiment, participants (n = 12) listened to standard tones with a regular 0.6-sec IOI and rare, early tones with a 0.3-sec IOI, while either attending to the tones (enforced by a pitch-deviant tone detection task) or reading a book. The early tones elicited larger N1–P2 responses than the standard tones in both conditions, but the effect was larger in the attend-to-tones condition. In their third experiment (n = 7), they further replicated this effect by having rare, early tones with varying IOIs of 0.3, 0.45, or 0.6 sec embedded among the standard tones with a regular 1.2-sec IOI. The temporal-expectation violating early tones with the shorter IOIs elicited larger N1–P2 responses than the standard tones with the longest IOI, whereas their prior experiments showed that the N1–P2 responses to the standard (expected) tones were attenuated for shorter IOIs, suggesting that temporal-expectation violation enhances auditory responses. Similarly, Nordby et al. (1988) reported a larger N1 component elicited by rare, early tones (50 msec) with a 400-msec IOI relative to standard tones with a 800-msec IOI (observed at Cz and Fz electrode sites) and whether participants attended to the tones to detect pitch- or time-deviant tones or passively listened to the tones while reading.

The current results extend these earlier findings. First, given that we had a fully orthogonal within-participant design crossing temporal deviance (standard vs. deviant) with tone presentation IOI (long vs. short), we were able to evaluate the effects of delay-dependent adaptation, cumulative adaptation, and temporal-expectation violation relatively independently. This allowed us to show that response enhancements to temporal-expectation violating tones fully (or nearly fully) override the large delay-dependent adaptation effect. Furthermore, our experimental design allowed us to support the hypothesis that the mechanisms that monitor rhythm-based temporal expectations enhance responses to temporal-expectation violating tones by actively boosting responses through top–down feedback rather than by temporarily disengaging the inhibitory effects of delay-dependent adaptation. Second, we included a large number of trials so that we were able to compare the standard-tone and deviant-tone conditions with matched numbers of trials (∼113 trials per participant per condition; the trial numbers were also matched in Experiment 1 in Ford & Hillyard, 1981, with 32 trials per condition in a between-participants design); note that unmatched trial numbers across conditions could inflate ERP peak amplitude measurements for conditions with fewer trials (e.g., Luck, 2005). Our large number of trials also allowed us to meaningfully analyze the effects of cumulative adaptation. Third, although the longest IOI previously used was close to 1 sec, we included IOIs that were substantially shorter than 1 sec (0.7 sec), about 1 sec (1.2 sec), and substantially longer than 1 sec (1.7 sec), allowing us to observe that a tone presentation rate of ∼1 Hz may be optimal for the growth of cumulative adaptation. Fourth, by including parallel analyses using ITPC, we were able to show that the phase locking of theta/low-alpha bands may potentially play a role in tracking auditory rhythm and detecting temporal-expectation violations.

Although the primary focus of the current study was to investigate the effects of temporal expectation and sensory adaptation on auditory evoked responses, our experimental design allowed for potentially observing an “omission” response elicited by the absence of an expected tone. Indeed, early ERP studies using repetitive sounds have reported ERPs to omitted sounds, indicative of the brain's sensory predictions (e.g., Picton, Hillyard, & Galambos, 1976; Klinke, Fruhstorfer, & Finkenzeller, 1968). More recently, such neural responses have been localized to the auditory cortex (e.g., SanMiguel, Widmann, Bendixen, Trujillo-Barreto, & Schröger, 2013) and have been shown to exhibit specificity for tone frequency (Berlot, Formisano, & De Martino, 2018). Although omission responses have been observed when sounds are task relevant or closely attended, they have been typically absent in studies when participants passively listened to sounds (e.g., Ford & Hillyard, 1981), similar to the experiments reported here. We did not observe any omission-related responses (potentially occurring before the short-IOI-block late tones) in our data based on analyses of ERP, ITPC, or non-stimulus-locked (induced) spectral power (not shown). Additional research is needed to understand the conditions under which robust omission responses occur and to develop EEG analysis techniques that detect such responses with high sensitivity.

Finally, our results are consistent with previous studies that examined the effects of expectation and sensory adaptation to repetition. Todorovic, van Ede, Maris, and de Lange (2011) examined top–down effects of the knowledge of repetition probability on auditory repetition suppression. On each trial, they presented a brief tone (5 m, 1000 Hz), which was either repeated after 500 msec (repetition trial) or was presented in isolation (nonrepetition trial). In two blocks, they manipulated the expectation of repetition; in the repetition-expected block, the repetition trials occurred frequently (75%) and nonrepetition trials occurred infrequently (25%), whereas in the repetition-unexpected block, the probabilities were reversed. They recorded MEG while participants detected rare, pitch-deviant tones (1200 Hz), which could occur at either the first or second temporal position. The MEG responses to the repeated tones (measured as auditory evoked fields as well as power in the theta and gamma frequency bands) were more suppressed (relative to the first tone) in the repetition-expected block relative to the repetition-unexpected block. Todorovic et al. (2011) inferred that repetition suppression—which, they note, could be driven by stimulus-induced adaptation and/or a reduction in perceptual prediction error—was reduced by a violation of top–down expectation. Furthermore, Todorovic, Schoffelen, van Ede, Maris, and de Lange (2015) showed that, in the absence of task-specific attention, the expectation effect was also sensitive to timing (as opposed to the presence/absence of repetition per se). They presented tone pairs where the timing of the second (and task-irrelevant) tone was manipulated to generate temporal expectation and found that the MEG responses (measured as auditory beta-band power) to the second tone were more suppressed when it was temporally expected relative to when it was temporally unexpected using 500-msec IOI. Whereas tone repetitions were continuously expected in our experiments based on the rhythmic structure of the mostly periodic tone presentations, the unexpectedly early tones violated the timing of presentation, which countered the strong delay-dependent adaptation. Overall, these set of results suggest that the top–down knowledge-based expectation effect (Todorovic et al., 2011), the interval-based temporal expectation effect (Todorovic et al., 2015), and our rhythm-based temporal expectation effect may be mediated by similar mechanisms.

In summary, by manipulating rhythm-based temporal expectation and tone presentation rate, we were able to distinguish the effects of delay-dependent adaptation, cumulative adaptation, and temporal-expectation violation on auditory-evoked EEG responses (analyzed as N1–P2 amplitude and ITPC). The results have demonstrated a strong delay-dependent adaptation effect, a weak cumulative adaptation effect (which may most efficiently accumulate at the tone presentation rate of ∼1 Hz), and a robust expectation-violation effect that substantially boosts responses to override delay-dependent adaptation effects likely through mechanisms involved in exogenous attention capture.

Acknowledgments

This study was supported by a National Institutes of Health grant (T32 NS047987).

Reprint requests should be sent to Melisa Menceloglu, Department of Psychology, Northwestern University, 2029 Sheridan Rd., Evanston, IL 60208, or via e-mail: mencel@u.northwestern.edu.

Note

1. 

Supplementary Materials for this paper can be retrieved from https://osf.io/jvzc2/.

REFERENCES

Berlot
,
E.
,
Formisano
,
E.
, &
De Martino
,
F.
(
2018
).
Mapping frequency-specific tone predictions in the human auditory cortex at high spatial resolution
.
Journal of Neuroscience
,
38
,
4934
4942
.
Carver
,
F. W.
,
Fuchs
,
A.
,
Jantzen
,
K. J.
, &
Kelso
,
J. S.
(
2002
).
Spatiotemporal analysis of the neuromagnetic response to rhythmic auditory stimulation: Rate dependence and transient to steady-state transition
.
Clinical Neurophysiology
,
113
,
1921
1931
.
Cohen
,
M. X.
(
2014
).
Analyzing neural time series data: Theory and practice
.
Cambridge, MA
:
MIT Press
.
Condon
,
C. D.
, &
Weinberger
,
N. M.
(
1991
).
Habituation produces frequency-specific plasticity of receptive fields in the auditory cortex
.
Behavioral Neuroscience
,
105
,
416
430
.
Correa
,
Á.
,
Lupiáñez
,
J.
,
Madrid
,
E.
, &
Tudela
,
P.
(
2006
).
Temporal attention enhances early visual processing: A review and new evidence from event-related potentials
.
Brain Research
,
1076
,
116
128
.
Correa
,
Á.
, &
Nobre
,
A. C.
(
2008
).
Neural modulation by regularity and passage of time
.
Journal of Neurophysiology
,
100
,
1649
1655
.
Costa-Faidella
,
J.
,
Baldeweg
,
T.
,
Grimm
,
S.
, &
Escera
,
C.
(
2011
).
Interactions between “what” and “when” in the auditory system: Temporal predictability enhances repetition suppression
.
Journal of Neuroscience
,
31
,
18590
18597
.
Coull
,
J. T.
(
2009
).
Neural substrates of mounting temporal expectation
.
PLoS Biology
,
7
,
e1000166
.
Coull
,
J. T.
, &
Nobre
,
A. C.
(
1998
).
Where and when to pay attention: The neural systems for directing attention to spatial locations and to time intervals as revealed by both PET and fMRI
.
Journal of Neuroscience
,
18
,
7426
7435
.
Cravo
,
A. M.
,
Rohenkohl
,
G.
,
Wyart
,
V.
, &
Nobre
,
A. C.
(
2013
).
Temporal expectation enhances contrast sensitivity by phase entrainment of low-frequency oscillations in visual cortex
.
Journal of Neuroscience
,
33
,
4002
4010
.
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
.
Doherty
,
J. R.
,
Rao
,
A.
,
Mesulam
,
M. M.
, &
Nobre
,
A. C.
(
2005
).
Synergistic effect of combined temporal and spatial expectations on visual attention
.
Journal of Neuroscience
,
25
,
8259
8266
.
Ford
,
J. M.
, &
Hillyard
,
S. A.
(
1981
).
Event-related potentials (ERPs) to interruptions of a steady rhythm
.
Psychophysiology
,
18
,
322
330
.
Griffin
,
I. C.
,
Miniussi
,
C.
, &
Nobre
,
A. C.
(
2002
).
Multiple mechanisms of selective attention: Differential modulation of stimulus processing by attention to space or time
.
Neuropsychologia
,
40
,
2325
2340
.
Hari
,
R.
,
Kaila
,
K.
,
Katila
,
T.
,
Tuomisto
,
T.
, &
Varpula
,
T.
(
1982
).
Interstimulus interval dependence of the auditory vertex response and its magnetic counterpart: implications for their neural generation
.
Electroencephalography and Clinical Neurophysiology
,
54
,
561
569
.
Hess
,
R. F.
, &
Snowden
,
R. J.
(
1992
).
Temporal properties of human visual filters: Number, shapes and spatial covariance
.
Vision Research
,
32
,
47
60
.
Hjorth
,
B.
(
1980
).
Source derivation simplifies topographical EEG interpretation
.
American Journal of EEG Technology
,
20
,
121
132
.
Kayser
,
J.
, &
Tenke
,
C. E.
(
2006
).
Principal components analysis of Laplacian waveforms as a generic method for identifying ERP generator patterns: I. Evaluation with auditory oddball tasks
.
Clinical Neurophysiology
,
117
,
348
368
.
Klinke
,
R.
,
Fruhstorfer
,
H.
, &
Finkenzeller
,
P.
(
1968
).
Evoked responses as a function of external and stored information
.
Electroencephalography and Clinical Neurophysiology
,
25
,
119
122
.
Lampar
,
A.
, &
Lange
,
K.
(
2011
).
Effects of temporal trial-by-trial cuing on early and late stages of auditory processing: evidence from event-related potentials
.
Attention, Perception, & Psychophysics
,
73
,
1916
1933
.
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.
(
2010
).
Can a regular context induce temporal orienting to a target sound?
International Journal of Psychophysiology
,
78
,
231
238
.
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
.
Lange
,
K.
,
Rösler
,
F.
, &
Röder
,
B.
(
2003
).
Early processing stages are modulated when auditory stimuli are presented at an attended moment in time: An event-related potential study
.
Psychophysiology
,
40
,
806
817
.
Lopez-Calderon
,
J.
, &
Luck
,
S. J.
(
2014
).
ERPLAB: An open-source toolbox for the analysis of event-related potentials
.
Frontiers in Human Neuroscience
,
8
,
213
.
Luck
,
S. J.
(
2005
).
Ten simple rules for designing ERP experiments
. In
T. C.
Handy
(Ed.),
Eventrelated potentials: A methods handbook
.
Cambridge, MA
:
MIT Press
.
Lui
,
L. L.
,
Bourne
,
J. A.
, &
Rosa
,
M. G. P.
(
2007
).
Spatial and temporal frequency selectivity of neurons in the middle temporal visual area of new world monkeys (Callithrix jacchus)
.
European Journal of Neuroscience
,
25
,
1780
1792
.
Makeig
,
S.
,
Bell
,
A. J.
,
Jung
,
T.-P.
, &
Sejnowski
,
T. J.
(
1996
).
Independent component analysis of electroencephalographic data
. In
D.
Touretzky
,
M.
Mozer
, &
M.
Hasselmo
(Eds.),
Advances in neural information processing systems
(
Vol. 8
, pp.
145
151
).
Cambridge, MA
:
MIT Press
.
Menceloglu
,
M.
,
Grabowecky
,
M.
, &
Suzuki
,
S.
(
2017a
).
Comparing the effects of implicit and explicit temporal expectation on choice response time and response conflict
.
Attention, Perception, & Psychophysics
,
79
,
169
179
.
Menceloglu
,
M.
,
Grabowecky
,
M.
, &
Suzuki
,
S.
(
2017b
).
Temporal expectation weights visual signals over auditory signals
.
Psychonomic Bulletin & Review
,
24
,
416
422
.
Menceloglu
,
M.
,
Grabowecky
,
M.
, &
Suzuki
,
S.
(
2019
).
Probability-driven and stimulus-driven orienting of attention to time and sensory modality
.
Attention, Perception, & Psychophysics
,
81
,
2732
2744
.
Miniussi
,
C.
,
Wilding
,
E. L.
,
Coull
,
J. T.
, &
Nobre
,
A. C.
(
1999
).
Orienting attention in time: Modulation of brain potentials
.
Brain
,
122
,
1507
1518
.
Moelants
,
D.
(
2002
).
Preferred tempo reconsidered
. In
C.
Stevens
,
D.
Burnham
,
G.
McPherson
,
E.
Schubert
, &
J.
Renwick
(Eds.),
Proceedings of the 7th International Conference on Music Perception and Cognition
(pp.
580
583
).
Sydney, Australia
:
Causal Productions
.
Morey
,
R. D.
(
2008
).
Confidence intervals from normalized data: A correction to Cousineau (2005)
.
Reason
,
4
,
61
64
.
Nelson
,
D. A.
, &
Lassman
,
F. M.
(
1973
).
Combined effects of recovery period and stimulus intensity on the human auditory evoked vertex response
.
Journal of Speech and Hearing Research
,
16
,
297
308
.
Nobre
,
A. C.
, &
van Ede
,
F.
(
2018
).
Anticipated moments: Temporal structure in attention
.
Nature Reviews Neuroscience
,
19
,
34
48
.
Nordby
,
H.
,
Roth
,
W. T.
, &
Pfefferbaum
,
A.
(
1988
).
Event-related potentials to time-deviant and pitch-deviant tones
.
Psychophysiology
,
25
,
249
261
.
Pereira
,
D. R.
,
Cardoso
,
S.
,
Ferreira-Santos
,
F.
,
Fernandes
,
C.
,
Cunha-Reis
,
C.
,
Paiva
,
T. O.
, et al
(
2014
).
Effects of inter-stimulus interval (ISI) duration on the N1 and P2 components of the auditory event-related potential
.
International Journal of Psychophysiology
,
94
,
311
318
.
Perrin
,
F.
,
Pernier
,
J.
,
Bertrand
,
O.
, &
Echallier
,
J. F.
(
1989a
).
Corrigenda EEG 02274
.
Electroencephalography and Clinical Neurophysiology
,
76
,
565
.
Perrin
,
F.
,
Pernier
,
J.
,
Bertrand
,
O.
, &
Echallier
,
J. F.
(
1989b
).
Spherical splines for scalp potential and current density mapping
.
Electroencephalography and Clinical Neurophysiology
,
72
,
184
187
.
Perrin
,
F.
,
Pernier
,
J.
,
Bertrand
,
O.
,
Giard
,
M. H.
, &
Echallier
,
J. F.
(
1987
).
Mapping of scalp potentials by surface spline interpolation
.
Electroencephalography and Clinical Neurophysiology
,
66
,
75
81
.
Picton
,
T. W.
,
Hillyard
,
S. A.
, &
Galambos
,
R.
(
1976
).
Habituation and attention in the auditory system
. In
W. D.
Keidel
, &
W. D.
Neff
(Eds.),
Handbook of Sensory Physiology
(pp.
343
389
).
Berlin, Heidelberg
:
Springer
.
Picton
,
T. W.
,
Hillyard
,
S. A.
,
Galambos
,
R.
, &
Schiff
,
M.
(
1971
).
Human auditory attention: A central or peripheral process?
Science
,
173
,
351
353
.
Pratt
,
H.
(
2011
).
Sensory ERP components
. In
E. S.
Kappenman
&
S. J.
Luck
(Eds.),
The Oxford handbook of event-related potential components
(pp.
89
114
).
Oxford, UK
:
Oxford University Press
.
Repp
,
B. H.
(
2005
).
Sensorimotor synchronization: A review of the tapping literature
.
Psychonomic Bulletin & Review
,
12
,
969
992
.
Rohenkohl
,
G.
, &
Nobre
,
A. C.
(
2011
).
Alpha oscillations related to anticipatory attention follow temporal expectations
.
Journal of Neuroscience
,
31
,
14076
14084
.
Sanders
,
L. D.
, &
Astheimer
,
L. B.
(
2008
).
Temporally selective attention modulates early perceptual processing: Event-related potential evidence
.
Perception & Psychophysics
,
70
,
732
742
.
SanMiguel
,
I.
,
Widmann
,
A.
,
Bendixen
,
A.
,
Trujillo-Barreto
,
N.
, &
Schröger
,
E.
(
2013
).
Hearing silences: Human auditory processing relies on preactivation of sound-specific brain activity patterns
.
Journal of Neuroscience
,
33
,
8633
8639
.
Schmitt
,
M.
,
Postma
,
A.
, &
De Haan
,
E.
(
2000
).
Interactions between exogenous auditory and visual spatial attention
.
Quarterly Journal of Experimental Psychology Section A
,
53
,
105
130
.
Spence
,
C. J.
, &
Driver
,
J.
(
1994
).
Covert spatial orienting in audition: Exogenous and endogenous mechanisms
.
Journal of Experimental Psychology: Human Perception and Performance
,
20
,
555
574
.
Spivey
,
M. J.
, &
Spirn
,
M. J.
(
2000
).
Selective visual attention modulates the direct tilt aftereffect
.
Perception & Psychophysics
,
62
,
1525
1533
.
Suzuki
,
S.
(
2001
).
Attention-dependent brief adaptation to contour orientation: A high-level aftereffect for convexity?
Vision Research
,
41
,
3883
3902
.
Suzuki
,
S.
(
2003
).
Attentional selection of overlapped shapes: A study using brief shape aftereffects
.
Vision Research
,
43
,
549
561
.
Suzuki
,
S.
, &
Grabowecky
,
M.
(
2003
).
Attention during adaptation weakens negative afterimages
.
Journal of Experimental Psychology: Human Perception and Performance
,
29
,
793
807
.
Tenke
,
C. E.
, &
Kayser
,
J.
(
2012
).
Generator localization by current source density (CSD): Implications of volume conduction and field closure at intracranial and scalp resolutions
.
Clinical Neurophysiology
,
123
,
2328
2345
.
Todorovic
,
A.
, &
de Lange
,
F. P.
(
2012
).
Repetition suppression and expectation suppression are dissociable in time in early auditory evoked fields
.
Journal of Neuroscience
,
32
,
13389
13395
.
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
.
Todorovic
,
A.
,
van Ede
,
F.
,
Maris
,
E.
, &
de Lange
,
F. P.
(
2011
).
Prior expectation mediates neural adaptation to repeated sounds in the auditory cortex: An MEG study
.
Journal of Neuroscience
,
31
,
9118
9123
.
Ulanovsky
,
N.
,
Las
,
L.
,
Farkas
,
D.
, &
Nelken
,
I.
(
2004
).
Multiple time scales of adaptation in auditory cortex neurons
.
Journal of Neuroscience
,
24
,
10440
10453
.
Ulanovsky
,
N.
,
Las
,
L.
, &
Nelken
,
I.
(
2003
).
Processing of low-probability sounds by cortical neurons
.
Nature Neuroscience
,
6
,
391
398
.