Neuroscience 290 (2015) 389–397
NEURAL OSCILLATORY CORRELATES OF DURATION MAINTENANCE IN WORKING MEMORY Y. G. CHEN, X. CHEN, C. W. KUANG AND X. T. HUANG *
can be temporally held in the WM after it is retrieved from long-term memory (Gibbon, 1977; Gibbon et al., 1984; Allan, 1998; Coull et al., 2008). Previous studies have confirmed that visual durations are retained in the visuospatial sketchpad, and auditory durations are retained in the phonological loop (Franssen et al., 2006; Rattat, 2010; Rattat and Picard, 2012). The neural networks underlying duration maintenance in WM are mainly cortico-striatal circuits involving frontal, parietal, and striatal regions (Sakurai et al., 2004; Coull et al., 2008; Genovesio et al., 2009; Harrington et al., 2010). However, more research is needed to understand the internal representation of duration retained in WM. Some researchers have argued that separate mechanisms process durations below and above about 3 s. Perception of duration below about 3 s is based on the subjective present (Fraisse, 1984). A low-frequency binding mechanism integrates sensory inputs into a coherent experience or temporal gestalt (Po¨ppel, 1997). Durations above about 3 s can no longer be perceived as a unit, and estimation of longer duration is based on memory and cognitive reconstruction (Fraisse, 1984). There is some evidence to support this theory (Kagerer et al., 2002; Szelag et al., 2002; Ulbrich et al., 2007). One event-related potential (ERP) study found that temporal reproductions were accurate for durations up to 3 s, and accompanied by a slow negative wave named contingent negative variation (CNV), whereas the CNV was reduced or absent when durations longer than 3 s were processed (Elbert et al., 1991). Furthermore, a functional magnetic resonance imaging (fMRI) study revealed that the motor system and the default mode network process durations below and above 2 s, respectively (Morillon et al., 2009). Based on these theoretical and experimental findings, we hypothesized that different internal representations were retained in WM for durations below and above about 3 s. The synchronous activity of neural oscillation is a critical link between single-neuron activity and behavior (Engel et al., 2001; Buzsa´ki and Draguhn, 2004). Theta and alpha band frequencies serve distinct functional roles during WM maintenance. It is widely reported that the oscillatory power in the theta band (4–8 Hz) over the prefrontal region increases with the number of items maintained in WM, such as locations (Gevins et al., 1997), letters (Gevins et al., 1997; Onton et al., 2005), and digits (Jensen and Tesche, 2002; Meltzer et al., 2007). Research shows that theta oscillation reflects the organization of sequentially ordered WM items (Hsieh et al.,
Key Laboratory of Cognition and Personality (Ministry of Education), School of Psychology, Southwest University, Chongqing 400715, China
Abstract—Working memory (WM) is a core element of temporal information processing, but little is known about the internal representation and neuronal underpinnings of the duration maintenance in WM. The neural oscillations during maintenance of duration in WM were examined using electroencephalogram (EEG) recordings. The EEG results showed that theta amplitude was not modulated by the length of duration retained in WM, while alpha amplitude decreased in a 4-s duration condition compared with 1-s, 2-s, and 3-s duration conditions. The amplitude of alpha power positively correlated with accuracy for the 3-s duration condition. The results suggest that alpha activity is involved in duration maintenance in WM. Our study provides electrophysiological evidence that different internal representations are retained in WM for durations below and above about 3 s. Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved.
Key words: temporal information, duration, working memory, neural oscillation.
INTRODUCTION Although there is no specific biological system that senses time, as there is for sight, hearing, and taste, we can perceive and process temporal information, and effectively use timing in our daily activities(Buhusi and Meck, 2005; Meck et al., 2013; Merchant et al., 2013). Working memory (WM), the short-term storage and online manipulation of information (Baddeley, 1992), is possibly a core element of temporal information processing. The internal representation of the external event duration must be stored in WM before it is transferred into long-term reference memory; the representation of reference duration *Corresponding author. Address: School of Psychology, Southwest University, Beibei, Chongqing 400715, China. Tel: +86-23-825-3630 (O); fax: +86-23-6825-3084. E-mail addresses: [email protected] (Y. G. Chen), [email protected] (X. T. Huang). Abbreviations: ACC, accuracy; ANOVA, analysis of variance; CNV, contingent negative variation; EEG, electroencephalogram; EOG, electrooculogram ; ERPs, event-related potentials; ERSP, eventrelated spectral perturbation; ICs, independent components; WM, working memory. http://dx.doi.org/10.1016/j.neuroscience.2015.01.036 0306-4522/Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved. 389
390
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
2011; Roberts et al., 2013; Roux and Uhlhaas, 2014). Alpha band activity (8–12 Hz) over the parieto-occipital region increases with the number of items maintained in WM, such as locations (Gevins et al., 1997; Johnson et al., 2011), letters (Jensen et al., 2002), digits (Meltzer et al., 2007), shapes (Johnson et al., 2011), and faces (Jokisch and Jensen, 2007; Tuladhar et al., 2007). However, the precise role of the alpha band in WM is still under debate. Some scholars have proposed that oscillatory alpha activity reflects the inhibition of cortical areas representing task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014), while others have suggested that alpha oscillation is related to successful maintenance of item information (Jensen et al., 2002; Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011). Duration maintenance in WM has received much less attention compared with non-temporal stimulus properties, and the relationship between neural oscillation and duration maintenance in WM has yet to be identified. In the present study, a matching-to-sample task was adopted to study the neural oscillatory correlates of duration maintenance in WM. This experimental paradigm is widely used to investigate the brain substrates of retention of duration in WM (Sakurai et al., 2004; Coull et al., 2008; Genovesio et al., 2009; Harrington et al., 2010). A sample stimulus and a probe stimulus are presented in succession, separated by a delay. Subjects are asked to judge whether the durations of the probe stimulus and the sample stimulus are identical or different. An advantage of this paradigm is that the encoding (timing), maintenance, and decision for temporal information can be separated (Coull et al., 2008; Harrington et al., 2010). To our knowledge, no previous study has reported the oscillatory correlates of duration maintenance in WM. We focused on the relationship between duration maintenance in WM and neural oscillations in theta and alpha bands, because previous studies on WM have delineated these bands well. In the present study, participants were required to retain one duration (1 s, 2 s, 3 s, or 4 s) in WM. As the theta band reflects the organization of sequentially ordered WM items (Hsieh et al., 2011; Roberts et al., 2013; Roux and Uhlhaas, 2014), the temporal order was equal for four duration conditions; thus, we predicted that the amplitude of the theta band would not be modulated by the length of duration retained in WM. As previously stated, we hypothesized that different internal representations would be retained in WM for durations below and above about 3 s. We inferred that if alpha oscillation is related to successful maintenance of item information (Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011), then the alpha band activity would reflect different internal representations of durations below and above about 3 s retained in WM. If alpha oscillation reflects inhibition of task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014), and there is no relationship between alpha band activity and internal representation of duration retained in WM, we would observe no significant alpha band difference between durations above and below about 3 s.
EXPERIMENTAL PROCEDURES Participants Eighteen right-handed undergraduates (six male), 20.67 ± 1.33 years of age, were paid for participation in the experiment. Each participant had normal or corrected-to-normal visual acuity and reported having normal hearing. Participants were not taking medications and did not suffer from any central nervous system abnormality or injury. The study was approved by the local institutional review board (RIB), and written informed consent was obtained from each participant. The experimental procedure was in accordance with the Declaration of Helsinki (World Medical Association, 2013). Experimental material and apparatus The visual stimuli were displayed on a black background in the center of a computer screen. There were two types of visual stimuli: a red circle and a question mark. The red circle was 3 cm in size (2.29°), its RGB values were [255, 0, 0], and its luminance was 48.5 cd/m2. The question mark was white and 2 cm in size (1.53°). The refresh rate of the computer monitor was 85 Hz, and the computer screen was placed about 75 cm in front of participants. Procedure There were seven blocks of a matching-to-sample task in the experiment. Each block contained 48 trials. There were 84 trials for one duration condition in the experiment. As shown in Fig. 1, in each trial, two red circles (the sample stimulus and the probe stimulus) were presented in tandem, separated by an interstimulus interval of 3 s (the delay/maintenance phase). Each circle was presented randomly for one of four durations (1 s, 2 s, 3 s, or 4 s). After an interval of 1 s, a question mark (response signal) was presented in the center of the screen, and terminated by a key press, or after 2 s had elapsed. Participants had to estimate the duration of the second red circle (probe) as shorter, equal to, or longer than the first circle (sample). During the response period, participants pressed 1, 2, or 3 on the keyboard using the index, middle, or ring finger, respectively (1 denoted shorter, 2 denoted equal to, 3 denoted longer). Half of the participants responded with the left hand, and the other half responded with the right hand. An intertrial interval of 2 s was used. Electrophysiological recording Continuous electroencephalogram (EEG) was acquired from Ag/AgCl electrodes mounted in an elastic cap (Brain Products GmbH, Gilching, Germany). Sixty-four electrodes were positioned according to the extended 10–20 system. Additional electrodes were placed on the mastoids. The horizontal electrooculogram (EOG) was acquired using a bipolar pair of electrodes positioned at the external ocular canthi, and vertical EOG was recorded from electrodes placed below the left eye. The EEG and EOG were digitized at 500 Hz with an
391
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
Fig. 1. Individual trial timeline with durations of each screen presentation.
amplifier bandpass of 0.01–100 Hz, including a 50-Hz notch filter, and were stored for offline analysis. All electrode impedances were maintained below 5 kX. EEG analysis EEGLAB (Delorme and Makeig, 2004) and MATLAB (The MathWorks, Inc., Natick, Massachusetts, USA) were used for offline EEG data processing. The continuous EEG data were re-referenced to the average of the right and left mastoids, and high-pass filtered at 0.5 Hz (Hsieh et al., 2011). EEG epochs were extracted for the period 1 s prior to the onset of the sample stimulus to 8 s following the onset of the sample stimulus (the first red circle). Each EEG epoch was then baseline-corrected by subtracting the mean voltage before the onset of the sample stimulus. Segmentations with EOG artifacts (mean EOG voltage exceeding ±80 lV) and those contaminated with artifacts due to amplifier clipping or peakto-peak deflection exceeding ±80 lV were excluded. The remaining EOG artifacts were visually identified and removed using independent component (IC) analysis according to their scalp maps and activity profile (Jung et al., 2000a,b). The ICs related to eye movements had a large EOG channel contribution and a frontal scalp distribution. A total of 90.71% of trials remained, and a repeated measures analysis of variance (ANOVA) revealed no significant main effect of Duration (four duration conditions) on the remaining number of trials [F(3,51) = 1.033, p > 0.05, gp2 = 0.57]. The segmented and artifact-free data were used for the time–frequency analysis. Trial-by-trial event-related spectral power modulation of the EEG rhythms was studied using EEGLAB. The time–frequency analysis used Hanning-windowed sinusoidal wavelets of three cycles at 3 Hz, rising linearly to about 15 cycles at 30 Hz (Makeig et al., 2004). Changes in event-related spectral power response (in dB) were computed using the event-related spectral perturbation (ERSP) index (Makeig, 1993). As shown in Fig. 2, the ERSP during sample duration encoding, delay, and probe phases was displayed, and corrected by subtracting the mean oscillatory power of the pre-stimulus baseline interval ( 0.4 s to 0.1 s) before the onset of the sample stimulus. To identify the temporal dynamics of the theta band (4–8 Hz) and the alpha band (8–12 Hz) powers during WM maintenance, the theta and alpha band powers were averaged across frequency, and were extracted during the delay phase (Figs. 3 and 4). In the new coordinate system, zero was the onset of delay, and the
baseline was 0.4 s to 0.1 s interval before the onset of the sample stimulus (the first red circle). Following a previous study (Hsieh et al., 2011), electrodes were grouped into nine different clusters: left-frontal cluster (AF7, F7, F5), middle-frontal cluster (F1, Fz, F2), rightfrontal cluster (AF8, F8, F6), left-central cluster (C3, C5, T7), middle-central cluster (C1, Cz, C2), right-central cluster (C4, C6, T8), left-posterior cluster (P5, P7, PO7), middle-posterior cluster (O1, O2, Oz), and right-posterior cluster (P6, P8, PO8). As shown in Fig. 4, steady alpha band activity was elicited from 1 s to 3 s after the onset of delay; the statistical time window (1–3 s time interval after the onset of the delay) was chosen and averaged. To assess whether the statistical analysis was contaminated by different time intervals between baseline and statistical time window, an additional statistical time window from 0.5 s to 0.1 s before the onset of delay was chosen. Two-way repeated measures ANOVAs were conducted on the mean power of theta and alpha bands in the 0.5 s to 0.1 s, and 1 s to 3 s time intervals, respectively. The ANOVA factors were Duration (1 s, 2 s, 3 s, and 4 s) and Region (nine clusters). The Greenhouse–Geisser correction was used to correct for any violations of sphericity (Greenhouse and Geisser, 1959), and the partial eta squared (gp2) was used to estimate the ANOVA effect size (Levine and Hullett, 2002).
RESULTS Behavioral data A repeated measures ANOVA of Accuracy (ACC) with Duration (1 s, 2 s, 3 s, and 4 s) as a within-participant factor revealed a significant main effect of Duration [F(3,51) = 17.851, p < 0.001, gp2 = 0.512]. Post hoc tests revealed that the ACC of the 1-s duration condition (Mean ± SE: 0.836 ± 0.036) was significantly higher than that of the 2-s (0.776 ± 0.026), 3s (0.711 ± 0.039), and 4-s (0.679 ± 0.043) duration conditions (t(17) = 3.172–5.485, p values 0.05, gp2 = 0.021].
392
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
Fig. 2. Average EEG spectrum power as a function of time for the whole epoch at the middle-frontal cluster in the 1-s (A), 2-s (B), 3-s (C), and 4-s (D) duration conditions.
Fig. 3. Grand average theta band power during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
EEG data Fig. 2 shows the dynamic activity of EEG spectrum power during the encoding (sample), delay, and probe phases. The offset of stimuli evoked distinct theta band oscillation. The amplitude of the alpha band decreased
compared with baseline during the encoding and probe phases, and it increased compared with baseline during the delay phase. Figs. 3 and 4 show the time course of the delay phase indexed by dynamic activity of theta and alpha bands (zero represents the onset of the delay phase). The first
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
393
Fig. 4. Grand average alpha band power during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
stage was sensory processing indexed by the theta band (Fig. 3). The offset of sample stimulus elicited theta activity, which was peaking at about 0.35 s, and the theta activity reduced to the minimum value at about 0.6 s. The second stage was duration maintenance in WM indexed by the alpha band (Fig. 4). Alpha band activity increased abruptly from about 0.4–1 s, which reflected an increase in the neural activity of duration maintenance in WM. The result was consistent with the time course of an event-related potential component, P300, that is an index of the WM process (Figs. 3 and 4 do not show the event-related potentials, but they show oscillatory powers averaged across frequency for theta and alpha bands respectively) (Polich and Criado, 2006; Polich, 2007). Then, the alpha band maintained a steady activation from 1 s to 3 s. The third stage was sensory processing indexed by the theta band (Fig. 3). The theta band was elicited by onset of the probe stimulus from 3 s to 3.5 s; at the same time, alpha band activity decreased abruptly (Fig. 4). For theta band power from the 0.5-s to 0.1-s time interval before the onset of the delay phase, a repeated measures ANOVA did not show a significant effect of
Duration, or a Duration Region interaction (p values >0.05). For the theta band power from the 1-s to 3-s interval after the onset of delay, an ANOVA revealed a significant main effect of Region [F(4.197,71.350) = 5.606, p < 0.001, gp2 = 0.248]. The theta power was greatest over the middle-frontal (0.032 ± 0.155 dB) and middle-central regions (-0.011 ± 0.107 dB). The main effect of Duration [F(2.058,34.986) = 0.929, p > 0.05, gp2 = 0.052] and the Duration Region interaction [F(6.360,108.128) = 1.024, p > 0.05, gp2 = 0.057] were not significant. For alpha band power from the 0.5-s to 0.1-s interval before the onset of delay, a repeated measures ANOVA did not show a significant effect of Duration or a Duration Region interaction (p values > 0.05). For alpha band power from the 1-s to 3-s interval after the onset of delay, an ANOVA revealed a significant main effect of Duration [F(2.697,45.843) = 3.813, p < 0.05, gp2 = 0.183]. Post hoc tests revealed that the differences of alpha power between 1-s (0.272 ± 0.135 dB), 2-s (0.315 ± 0.143 dB), and 3-s (0.235 ± 0.140 dB) duration conditions were not significant (t(17) = 0.568–0.820, p values >0.05), and
394
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
that the alpha power of 4-s (0.018 ± 0.129 dB) was lower than that of the 1-s, 2-s, and 3-s duration conditions (t(17) = 2.307–2.790, p values 0.05, gp2 = 0.088] and the Duration Region interaction [F(5.767,98.042) = 0.777, p > 0.05, gp2 = 0.044] were not significant. As shown in Fig. 5, theta and alpha band powers were both greatest over the frontal region during the 1–3-s time interval after the onset of the delay phase. To further assess the role of alpha band power in duration maintenance in WM, the relationship between accuracy and amplitude of alpha power at the middle-frontal cluster during the delay phase (1–3 s) was analyzed for 1 s, 2 s, 3 s, and 4 s, respectively. As shown in Fig. 6, the amplitude of alpha power positively correlated with accuracy for the 3-s duration condition (Pearson’s r = 0.546, p < 0.05). The correlations were not significant for 1-s (r = 0.354, p > 0.05), 2-s (r = 0.124, p > 0.05), and 4-s (r = 0.395, p > 0.05) conditions.
DISCUSSION The present study used a matching-to-sample task to study the neural basis of duration maintenance in WM. EEG data showed that the theta band amplitude in the delay phase was not significantly modulated by the length of duration. The result confirmed our hypothesis that theta band frequencies are not engaged in maintenance of duration in WM. We also found that the alpha band amplitude of the 4-s duration condition in the delay phase was lowest among four duration conditions, and the alpha band amplitude positively correlated with accuracy in the 3-s duration condition. The findings emphasized the role of the alpha band in duration maintenance in WM. As shown in Fig. 2, the time intervals between baseline and delay phase were different in the 1-s, 2-s, 3-s, and 4-s duration conditions, and this factor might have affected the alpha amplitude during the delay phase. If this factor had been an influence, it would have affected the alpha band amplitude between 0.5 and 0.1 s before the onset of the delay phase,
Fig. 6. The correlation between accuracy and amplitude of alpha band power during duration maintenance in the 3-s duration condition.
because the time intervals between baseline and this time window were different for the four duration conditions. An ANOVA revealed no significant main effect of Duration and no Duration Region interaction for amplitude of theta and alpha band power from the 0.5 to 0.1 s before the onset of the delay phase. It indicated that the difference in time intervals from baseline to the delay phase for four duration conditions did not significantly affect the theta and alpha band amplitude during the delay phase. It is still debated whether alpha oscillation is related to successful maintenance of item information (Jensen et al., 2002; Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011), or whether it reflects inhibition of task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014). Our study provided three pieces of evidence to support the maintenance rather than inhibition. First, alpha band amplitude in the delay phase was significantly lower in the 4-s duration condition than in the 1-s, 2-s, and 3-s duration conditions. Previous
Fig. 5. The topographies of oscillatory activities of theta and alpha bands during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
studies suggest that short time intervals (below about 3 s) can be perceived as a unit, whereas long time intervals (above about 3 s) have to be divided into subunits and depend on a memory-based cognitive reconstruction (Fraisse, 1984; Po¨ppel, 1997; Ulbrich et al., 2007). Our findings support previous work, as they indicate that durations below about 3 s are encoded as a unit, and can therefore be retained in WM; thus, the reproduction and discrimination of durations below about 3 s are accurate, and the alpha band amplitude was high during the delay phase. In contrast, the encoding of durations above about 3 s is based on memory and cognitive reconstruction, and their internal representations cannot be retained in WM. Thus, the reproduction and discrimination of durations above 3 s is less accurate, and the alpha band amplitude was low during the delay phase. Therefore, the EEG results support our hypothesis that alpha band activity reflected duration maintenance in WM, and different internal representations were retained in WM for durations below and above about 3 s. Second, alpha band activity was greatest over the frontal region in the present study (Fig. 5). Previous studies indicate that alpha band activity is typically found over the parieto-occipital region, so some scholars suggest it reflects suppression of processing in visual areas (Jokisch and Jensen, 2007; Tuladhar et al., 2007; Freunberger et al., 2011). However, the finding of frontal distributed alpha band is consistent with previous findings that the frontal region plays an important role in duration maintenance in WM (Olton, 1989; Sakurai et al., 2004). Furthermore, one EEG study reported that alpha band activity appeared over the frontal regions when visuospatial information was retained and demanded manipulation in WM; these authors suggested that an increase of prefrontal alpha amplitude may not be interpreted in terms of inhibition, but may enable a tight functional coupling between prefrontal cortical areas (Sauseng et al., 2005). Therefore, the present study indicates that the alpha band activity distributed over the frontal region does not reflect the inhibition of the frontal region during maintenance of duration in WM, but suggests that the frontal distributed alpha band reflects the duration maintenance in WM. Third, alpha band amplitude positively correlated with accuracy in the 3-s duration condition (Fig. 6). We hypothesized that 3 s is near the boundary between short and long durations based on previous studies. It is natural to expect individual differences in the boundary. Subjects with a longer boundary (for whom 3 s is a short duration) are likely to show high accuracy and high alpha band amplitude, and subjects with a shorter boundary (for whom 3 s is a long duration) are likely to show low accuracy and low alpha band amplitude. This hypothesis was confirmed by the significant correlation between alpha band amplitude and accuracy in the 3-s duration condition. In addition, two issues need to be explained in the present study. Firstly, the behavioral and EEG results revealed different boundaries between short and long durations. Our results were in line with previous
395
studies. Behavioral studies support the concept of a boundary between 2 s and 3 s (Szelag et al., 2002; Ulbrich et al., 2007), but an EEG study indicates a boundary between 3 s and 4 s (Elbert et al., 1991). Research suggests that variance in timing behavior is mainly due to encoding (clock), memory, and decision processes (Gibbon and Church, 1984; Allan, 1998). In the present study, the behavioral results were not only influenced by WM, but also by encoding and decision processes. However, the encoding (timing), maintenance, and decision processes for temporal information can be distinguished by EEG. Thus, EEG data are more suitable than behavioral data to investigate the boundary between short and long durations. This explains the slight difference in boundary between the behavioral and EEG results. Lastly, the correlations between alpha band amplitude and accuracy were not significant in 1-s, 2-s and 4-s duration conditions. The results do not simply imply that alpha band power has nothing to do with accuracy in 1-s, 2-s and 4-s duration conditions. A correlation with larger sample size is easier to be statistical significant (Lowry, 2001), and a larger sample size is adopted by typical correlation studies (e.g. Adelabu, 2007; Hayashi et al., 2014). Given the ideal statistical power (1 b) for any study is considered to be 0.8 (Suresh and Chandrashekara, 2012), and the significance level (a) is 0.05, we can compute the ideal sample size as 60, 508, and 48 for 1-s, 2-s and 4-s duration conditions respectively (Cohen, 1988). The present experiment was designed to test the hypothesis of different internal representations retained in WM for durations below and above about 3 s, but it was not suitable to evaluate the correlations between alpha band amplitude and accuracy in 1-s, 2-s, and 4-s duration conditions because of the small sample size.
CONCLUSIONS The present study is the first to explore the neural oscillatory correlates of duration maintenance in WM. We found that theta activity was not modulated by increasing length of duration retained in WM, whereas alpha power was lower in the 4-s duration condition than in the 1-s, 2-s, and 3-s duration conditions. The alpha power amplitude was positively correlated with accuracy for the 3-s duration condition, emphasizing the important role of frontal distributed alpha band oscillation in maintenance of duration in WM. We provided evidence of different internal representations retained in WM for durations below and above about 3 s. The precise nature of the internal representation of duration is still unclear, and further work is necessary to elucidate this. Acknowledgments—This study was supported by a grant from the National Natural Science Foundation of China (31200855, 31300845), the Key Research Institute of Humanities and Social Science in Chongqing (10SKB23), the Doctoral Foundation of Southwest University (SWU110037).
396
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
REFERENCES Adelabu DH (2007) Time perspective and school membership as correlates to academic achievement among African American adolescents. Adolescence 42:525–538. Allan LG (1998) The influence of the scalar timing model on human timing research. Behav Process 44:101–117. Baddeley A (1992) Working memory. Science 255:556–559. Buhusi CV, Meck WH (2005) What makes us tick? Functional and neural mechanisms of interval timing. Nat Rev Neurosci 6: 755–765. Buzsa´ki G, Draguhn A (2004) Neuronal oscillations in cortical networks. Science 304:1926–1929. Cohen J (1988) Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Coull JT, Nazarian B, Vidal F (2008) Timing, storage, and comparison of stimulus duration engage discrete anatomical components of a perceptual timing network. J Cogn Neurosci 20:2185–2197. Delorme A, Makeig S (2004) EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 134:9–21. Elbert T, Ulrich R, Rockstroh B, Lutzenberger W (1991) The processing of temporal intervals reflected by CNV-like brain potentials. Psychophysiology 28:648–655. Engel AK, Fries P, Singer W (2001) Dynamic predictions: oscillations and synchrony in top-down processing. Nat Rev Neurosci 2:704–716. Fraisse P (1984) Perception and estimation of time. Annu Rev Psychol 35:1–36. Franssen V, Vandierendonck A, Van Hiel A (2006) Duration estimation and the phonological loop: articulatory suppression and irrelevant sounds. Psychol Res 70:304–316. Freunberger R, Werkle-Bergner M, Griesmayr B, Lindenberger U, Klimesch W (2011) Brain oscillatory correlates of working memory constraints. Brain Res 1375:93–102. Genovesio A, Tsujimoto S, Wise SP (2009) Feature- and order-based timing representations in the frontal cortex. Neuron 63:254–266. Gevins A, Smith ME, McEvoy L, Yu D (1997) High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing, and practice. Cereb Cortex 7: 374–385. Gibbon J (1977) Scalar expectancy theory and Weber’s law in animal timing. Psychol Rev 84:279–325. Gibbon J, Church RM (1984) Sources of variance in an information processing theory of timing. In: Animal cognition (Roitblat HL, et al., eds), pp 465–488 Hillsdale, NJ: Erlbaum. Gibbon J, Church RM, Meck W (1984) Scalar timing in memory. In: Annals of the New York Academy of Sciences: timing and time perception, vol. 423 (Gibbon J, Allan L, eds), pp 52–77 New York: New York Academy of Sciences. Greenhouse SW, Geisser S (1959) On methods in the analysis of profile data. Psychometrika 24:95–112. Harrington DL, Zimbelman JL, Hinton SC, Rao SM (2010) Neural modulation of temporal encoding, maintenance, and decision processes. Cereb Cortex 20:1274–1285. Hayashi MJ, Kantele M, Walsh V, Carlson S, Kanai R (2014) Dissociable neuroanatomical correlates of subsecond and suprasecond time perception. J Cogn Neurosci 26:1685–1693. Hsieh LT, Ekstrom AD, Ranganath C (2011) Neural oscillations associated with item and temporal order maintenance in working memory. J Neurosci 31:10803–10810. Jensen O, Tesche CD (2002) Frontal theta activity in humans increases with memory load in a working memory task. Eur J Neurosci 15:1395–1399. Jensen O, Gelfand J, Kounios J, Lisman JE (2002) Oscillations in the alpha band (9–12 Hz) increase with memory load during retention in a short-term memory task. Cereb Cortex 12:877–882. Johnson JS, Sutterer DW, Acheson DJ, Lewis-Peacock JA, Postle BR (2011) Increased alpha-band power during the retention of shapes and shape-location associations in visual short-term memory. Front Psychol 2:128.
Jokisch D, Jensen O (2007) Modulation of gamma and alpha activity during a working memory task engaging the dorsal or ventral stream. J Neurosci 27:3244–3251. Jung TP, Makeig S, Humphries C, Lee TW, McKeown MJ, Iragui V, Sejnowski TJ (2000a) Removing electroencephalographic artifacts by blind source separation. Psychophysiology 37: 163–178. Jung TP, Makeig S, Westerfield M, Townsend J, Courchesne E, Sejnowski TJ (2000b) Removal of eye activity artifacts from visual event-related potentials in normal and clinical subjects. Clin Neurophysiol 111:1745–1758. Kagerer FA, Wittmann M, Szelag E, Von Steinbuchel N (2002) Cortical involvement in temporal reproduction: evidence for differential roles of the hemispheres. Neuropsychologia 40: 357–366. Klimesch W, Sauseng P, Hanslmayr S (2007) EEG alpha oscillations: the inhibition-timing hypothesis. Brain Res Rev 53:63–88. Levine TR, Hullett CR (2002) Eta squared, partial eta squared, and misreporting of effect size in communication research. Hum Commun Res 28:612–625. Lowry R (2001) Significance of a correlation coefficient. Available from: http://vassarstats.net/rsig.html. Makeig S (1993) Auditory event-related dynamics of the EEG spectrum and effects of exposure to tones. Electroencephalogr Clin Neurophysiol 86:283–293. Makeig S, Delorme A, Westerfield M, Jung TP, Townsend J, Courchesne E, Sejnowski TJ (2004) Electroencephalographic brain dynamics following manually responded visual targets. PLoS Biol 2:e176. Manza P, Hau CL, Leung HC (2014) Alpha power gates relevant information during working memory updating. J Neurosci 34:5998–6002. Meck WH, Vatakis A, van Rijn H (2013) Timing & time perception enters a new dimension. Timing Time Percept 1:1–2. Meltzer JA, Negishi M, Mayes LC, Constable RT (2007) Individual differences in EEG theta and alpha dynamics during working memory correlate with fMRI responses across subjects. Clin Neurophysiol 118:2419–2436. Merchant H, Harrington DL, Meck WH (2013) Neural basis of the perception and estimation of time. Annu Rev Neurosci 36: 313–336. Morillon B, Kell CA, Giraud AL (2009) Three stages and four neural systems in time estimation. J Neurosci 29:14803–14811. Olton DS (1989) Frontal cortex, timing and memory. Neuropsychologia 27:121–130. Onton J, Delorme A, Makeig S (2005) Frontal midline EEG dynamics during working memory. NeuroImage 27:341–356. Palva S, Palva JM (2007) New vistas for alpha-frequency band oscillations. Trends Neurosci 30:150–158. Polich J (2007) Updating p300: an integrative theory of P3a and P3b. Clin Neurophysiol 118:2128–2148. Polich J, Criado JR (2006) Neuropsychology and neuropharmacology of P3a and P3b. Int J Psychophysiol 60:172–185. Po¨ppel E (1997) A hierarchical model of temporal perception. Trends Cogn Sci 1:56–61. Rattat AC (2010) Bidirectional interference between timing and concurrent memory processing in children. J Exp Child Psychol 106:145–162. Rattat AC, Picard D (2012) Short-term memory for auditory and visual durations: evidence for selective interference effects. Psychol Res 76:32–40. Roberts BM, Hsieh LT, Ranganath C (2013) Oscillatory activity during maintenance of spatial and temporal information in working memory. Neuropsychologia 51:349–357. Roux F, Uhlhaas PJ (2014) Working memory and neural oscillations: alpha-gamma versus theta-gamma codes for distinct WM information? Trends Cogn Sci 18:16–25. Sakurai Y, Takahashi S, Inoue M (2004) Stimulus duration in working memory is represented by neuronal activity in the monkey prefrontal cortex. Eur J Neurosci 20:1069–1080.
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397 Sauseng P, Klimesch W, Doppelmayr M, Pecherstorfer T, Freunberger R, Hanslmayr S (2005) EEG alpha synchronization and functional coupling during top-down processing in a working memory task. Hum Brain Mapp 26:148–155. Suresh K, Chandrashekara S (2012) Sample size estimation and power analysis for clinical research studies. J Hum Reprod Sci 5:7–13. Szelag E, Kowalska J, Rymarczyk K, Poppel E (2002) Duration processing in children as determined by time reproduction: implications for a few seconds temporal window. Acta Psychol 110:1–19.
397
Tuladhar AM, ter Huurne N, Schoffelen JM, Maris E, Oostenveld R, Jensen O (2007) Parieto-occipital sources account for the increase in alpha activity with working memory load. Hum Brain Mapp 28:785–792. Ulbrich P, Churan J, Fink M, Wittmann M (2007) Temporal reproduction: further evidence for two processes. Acta Psychol 125:51–65. World Medical Association (2013). WMA declaration of Helsinki – ethical principles for medical research involving human subjects. Available from: http://www.wma.net/en/30publications/10policies/ b3/index.html.
(Accepted 8 January 2015) (Available online 28 January 2015)
NEURAL OSCILLATORY CORRELATES OF DURATION MAINTENANCE IN WORKING MEMORY Y. G. CHEN, X. CHEN, C. W. KUANG AND X. T. HUANG *
can be temporally held in the WM after it is retrieved from long-term memory (Gibbon, 1977; Gibbon et al., 1984; Allan, 1998; Coull et al., 2008). Previous studies have confirmed that visual durations are retained in the visuospatial sketchpad, and auditory durations are retained in the phonological loop (Franssen et al., 2006; Rattat, 2010; Rattat and Picard, 2012). The neural networks underlying duration maintenance in WM are mainly cortico-striatal circuits involving frontal, parietal, and striatal regions (Sakurai et al., 2004; Coull et al., 2008; Genovesio et al., 2009; Harrington et al., 2010). However, more research is needed to understand the internal representation of duration retained in WM. Some researchers have argued that separate mechanisms process durations below and above about 3 s. Perception of duration below about 3 s is based on the subjective present (Fraisse, 1984). A low-frequency binding mechanism integrates sensory inputs into a coherent experience or temporal gestalt (Po¨ppel, 1997). Durations above about 3 s can no longer be perceived as a unit, and estimation of longer duration is based on memory and cognitive reconstruction (Fraisse, 1984). There is some evidence to support this theory (Kagerer et al., 2002; Szelag et al., 2002; Ulbrich et al., 2007). One event-related potential (ERP) study found that temporal reproductions were accurate for durations up to 3 s, and accompanied by a slow negative wave named contingent negative variation (CNV), whereas the CNV was reduced or absent when durations longer than 3 s were processed (Elbert et al., 1991). Furthermore, a functional magnetic resonance imaging (fMRI) study revealed that the motor system and the default mode network process durations below and above 2 s, respectively (Morillon et al., 2009). Based on these theoretical and experimental findings, we hypothesized that different internal representations were retained in WM for durations below and above about 3 s. The synchronous activity of neural oscillation is a critical link between single-neuron activity and behavior (Engel et al., 2001; Buzsa´ki and Draguhn, 2004). Theta and alpha band frequencies serve distinct functional roles during WM maintenance. It is widely reported that the oscillatory power in the theta band (4–8 Hz) over the prefrontal region increases with the number of items maintained in WM, such as locations (Gevins et al., 1997), letters (Gevins et al., 1997; Onton et al., 2005), and digits (Jensen and Tesche, 2002; Meltzer et al., 2007). Research shows that theta oscillation reflects the organization of sequentially ordered WM items (Hsieh et al.,
Key Laboratory of Cognition and Personality (Ministry of Education), School of Psychology, Southwest University, Chongqing 400715, China
Abstract—Working memory (WM) is a core element of temporal information processing, but little is known about the internal representation and neuronal underpinnings of the duration maintenance in WM. The neural oscillations during maintenance of duration in WM were examined using electroencephalogram (EEG) recordings. The EEG results showed that theta amplitude was not modulated by the length of duration retained in WM, while alpha amplitude decreased in a 4-s duration condition compared with 1-s, 2-s, and 3-s duration conditions. The amplitude of alpha power positively correlated with accuracy for the 3-s duration condition. The results suggest that alpha activity is involved in duration maintenance in WM. Our study provides electrophysiological evidence that different internal representations are retained in WM for durations below and above about 3 s. Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved.
Key words: temporal information, duration, working memory, neural oscillation.
INTRODUCTION Although there is no specific biological system that senses time, as there is for sight, hearing, and taste, we can perceive and process temporal information, and effectively use timing in our daily activities(Buhusi and Meck, 2005; Meck et al., 2013; Merchant et al., 2013). Working memory (WM), the short-term storage and online manipulation of information (Baddeley, 1992), is possibly a core element of temporal information processing. The internal representation of the external event duration must be stored in WM before it is transferred into long-term reference memory; the representation of reference duration *Corresponding author. Address: School of Psychology, Southwest University, Beibei, Chongqing 400715, China. Tel: +86-23-825-3630 (O); fax: +86-23-6825-3084. E-mail addresses: [email protected] (Y. G. Chen), [email protected] (X. T. Huang). Abbreviations: ACC, accuracy; ANOVA, analysis of variance; CNV, contingent negative variation; EEG, electroencephalogram; EOG, electrooculogram ; ERPs, event-related potentials; ERSP, eventrelated spectral perturbation; ICs, independent components; WM, working memory. http://dx.doi.org/10.1016/j.neuroscience.2015.01.036 0306-4522/Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved. 389
390
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
2011; Roberts et al., 2013; Roux and Uhlhaas, 2014). Alpha band activity (8–12 Hz) over the parieto-occipital region increases with the number of items maintained in WM, such as locations (Gevins et al., 1997; Johnson et al., 2011), letters (Jensen et al., 2002), digits (Meltzer et al., 2007), shapes (Johnson et al., 2011), and faces (Jokisch and Jensen, 2007; Tuladhar et al., 2007). However, the precise role of the alpha band in WM is still under debate. Some scholars have proposed that oscillatory alpha activity reflects the inhibition of cortical areas representing task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014), while others have suggested that alpha oscillation is related to successful maintenance of item information (Jensen et al., 2002; Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011). Duration maintenance in WM has received much less attention compared with non-temporal stimulus properties, and the relationship between neural oscillation and duration maintenance in WM has yet to be identified. In the present study, a matching-to-sample task was adopted to study the neural oscillatory correlates of duration maintenance in WM. This experimental paradigm is widely used to investigate the brain substrates of retention of duration in WM (Sakurai et al., 2004; Coull et al., 2008; Genovesio et al., 2009; Harrington et al., 2010). A sample stimulus and a probe stimulus are presented in succession, separated by a delay. Subjects are asked to judge whether the durations of the probe stimulus and the sample stimulus are identical or different. An advantage of this paradigm is that the encoding (timing), maintenance, and decision for temporal information can be separated (Coull et al., 2008; Harrington et al., 2010). To our knowledge, no previous study has reported the oscillatory correlates of duration maintenance in WM. We focused on the relationship between duration maintenance in WM and neural oscillations in theta and alpha bands, because previous studies on WM have delineated these bands well. In the present study, participants were required to retain one duration (1 s, 2 s, 3 s, or 4 s) in WM. As the theta band reflects the organization of sequentially ordered WM items (Hsieh et al., 2011; Roberts et al., 2013; Roux and Uhlhaas, 2014), the temporal order was equal for four duration conditions; thus, we predicted that the amplitude of the theta band would not be modulated by the length of duration retained in WM. As previously stated, we hypothesized that different internal representations would be retained in WM for durations below and above about 3 s. We inferred that if alpha oscillation is related to successful maintenance of item information (Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011), then the alpha band activity would reflect different internal representations of durations below and above about 3 s retained in WM. If alpha oscillation reflects inhibition of task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014), and there is no relationship between alpha band activity and internal representation of duration retained in WM, we would observe no significant alpha band difference between durations above and below about 3 s.
EXPERIMENTAL PROCEDURES Participants Eighteen right-handed undergraduates (six male), 20.67 ± 1.33 years of age, were paid for participation in the experiment. Each participant had normal or corrected-to-normal visual acuity and reported having normal hearing. Participants were not taking medications and did not suffer from any central nervous system abnormality or injury. The study was approved by the local institutional review board (RIB), and written informed consent was obtained from each participant. The experimental procedure was in accordance with the Declaration of Helsinki (World Medical Association, 2013). Experimental material and apparatus The visual stimuli were displayed on a black background in the center of a computer screen. There were two types of visual stimuli: a red circle and a question mark. The red circle was 3 cm in size (2.29°), its RGB values were [255, 0, 0], and its luminance was 48.5 cd/m2. The question mark was white and 2 cm in size (1.53°). The refresh rate of the computer monitor was 85 Hz, and the computer screen was placed about 75 cm in front of participants. Procedure There were seven blocks of a matching-to-sample task in the experiment. Each block contained 48 trials. There were 84 trials for one duration condition in the experiment. As shown in Fig. 1, in each trial, two red circles (the sample stimulus and the probe stimulus) were presented in tandem, separated by an interstimulus interval of 3 s (the delay/maintenance phase). Each circle was presented randomly for one of four durations (1 s, 2 s, 3 s, or 4 s). After an interval of 1 s, a question mark (response signal) was presented in the center of the screen, and terminated by a key press, or after 2 s had elapsed. Participants had to estimate the duration of the second red circle (probe) as shorter, equal to, or longer than the first circle (sample). During the response period, participants pressed 1, 2, or 3 on the keyboard using the index, middle, or ring finger, respectively (1 denoted shorter, 2 denoted equal to, 3 denoted longer). Half of the participants responded with the left hand, and the other half responded with the right hand. An intertrial interval of 2 s was used. Electrophysiological recording Continuous electroencephalogram (EEG) was acquired from Ag/AgCl electrodes mounted in an elastic cap (Brain Products GmbH, Gilching, Germany). Sixty-four electrodes were positioned according to the extended 10–20 system. Additional electrodes were placed on the mastoids. The horizontal electrooculogram (EOG) was acquired using a bipolar pair of electrodes positioned at the external ocular canthi, and vertical EOG was recorded from electrodes placed below the left eye. The EEG and EOG were digitized at 500 Hz with an
391
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
Fig. 1. Individual trial timeline with durations of each screen presentation.
amplifier bandpass of 0.01–100 Hz, including a 50-Hz notch filter, and were stored for offline analysis. All electrode impedances were maintained below 5 kX. EEG analysis EEGLAB (Delorme and Makeig, 2004) and MATLAB (The MathWorks, Inc., Natick, Massachusetts, USA) were used for offline EEG data processing. The continuous EEG data were re-referenced to the average of the right and left mastoids, and high-pass filtered at 0.5 Hz (Hsieh et al., 2011). EEG epochs were extracted for the period 1 s prior to the onset of the sample stimulus to 8 s following the onset of the sample stimulus (the first red circle). Each EEG epoch was then baseline-corrected by subtracting the mean voltage before the onset of the sample stimulus. Segmentations with EOG artifacts (mean EOG voltage exceeding ±80 lV) and those contaminated with artifacts due to amplifier clipping or peakto-peak deflection exceeding ±80 lV were excluded. The remaining EOG artifacts were visually identified and removed using independent component (IC) analysis according to their scalp maps and activity profile (Jung et al., 2000a,b). The ICs related to eye movements had a large EOG channel contribution and a frontal scalp distribution. A total of 90.71% of trials remained, and a repeated measures analysis of variance (ANOVA) revealed no significant main effect of Duration (four duration conditions) on the remaining number of trials [F(3,51) = 1.033, p > 0.05, gp2 = 0.57]. The segmented and artifact-free data were used for the time–frequency analysis. Trial-by-trial event-related spectral power modulation of the EEG rhythms was studied using EEGLAB. The time–frequency analysis used Hanning-windowed sinusoidal wavelets of three cycles at 3 Hz, rising linearly to about 15 cycles at 30 Hz (Makeig et al., 2004). Changes in event-related spectral power response (in dB) were computed using the event-related spectral perturbation (ERSP) index (Makeig, 1993). As shown in Fig. 2, the ERSP during sample duration encoding, delay, and probe phases was displayed, and corrected by subtracting the mean oscillatory power of the pre-stimulus baseline interval ( 0.4 s to 0.1 s) before the onset of the sample stimulus. To identify the temporal dynamics of the theta band (4–8 Hz) and the alpha band (8–12 Hz) powers during WM maintenance, the theta and alpha band powers were averaged across frequency, and were extracted during the delay phase (Figs. 3 and 4). In the new coordinate system, zero was the onset of delay, and the
baseline was 0.4 s to 0.1 s interval before the onset of the sample stimulus (the first red circle). Following a previous study (Hsieh et al., 2011), electrodes were grouped into nine different clusters: left-frontal cluster (AF7, F7, F5), middle-frontal cluster (F1, Fz, F2), rightfrontal cluster (AF8, F8, F6), left-central cluster (C3, C5, T7), middle-central cluster (C1, Cz, C2), right-central cluster (C4, C6, T8), left-posterior cluster (P5, P7, PO7), middle-posterior cluster (O1, O2, Oz), and right-posterior cluster (P6, P8, PO8). As shown in Fig. 4, steady alpha band activity was elicited from 1 s to 3 s after the onset of delay; the statistical time window (1–3 s time interval after the onset of the delay) was chosen and averaged. To assess whether the statistical analysis was contaminated by different time intervals between baseline and statistical time window, an additional statistical time window from 0.5 s to 0.1 s before the onset of delay was chosen. Two-way repeated measures ANOVAs were conducted on the mean power of theta and alpha bands in the 0.5 s to 0.1 s, and 1 s to 3 s time intervals, respectively. The ANOVA factors were Duration (1 s, 2 s, 3 s, and 4 s) and Region (nine clusters). The Greenhouse–Geisser correction was used to correct for any violations of sphericity (Greenhouse and Geisser, 1959), and the partial eta squared (gp2) was used to estimate the ANOVA effect size (Levine and Hullett, 2002).
RESULTS Behavioral data A repeated measures ANOVA of Accuracy (ACC) with Duration (1 s, 2 s, 3 s, and 4 s) as a within-participant factor revealed a significant main effect of Duration [F(3,51) = 17.851, p < 0.001, gp2 = 0.512]. Post hoc tests revealed that the ACC of the 1-s duration condition (Mean ± SE: 0.836 ± 0.036) was significantly higher than that of the 2-s (0.776 ± 0.026), 3s (0.711 ± 0.039), and 4-s (0.679 ± 0.043) duration conditions (t(17) = 3.172–5.485, p values 0.05, gp2 = 0.021].
392
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
Fig. 2. Average EEG spectrum power as a function of time for the whole epoch at the middle-frontal cluster in the 1-s (A), 2-s (B), 3-s (C), and 4-s (D) duration conditions.
Fig. 3. Grand average theta band power during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
EEG data Fig. 2 shows the dynamic activity of EEG spectrum power during the encoding (sample), delay, and probe phases. The offset of stimuli evoked distinct theta band oscillation. The amplitude of the alpha band decreased
compared with baseline during the encoding and probe phases, and it increased compared with baseline during the delay phase. Figs. 3 and 4 show the time course of the delay phase indexed by dynamic activity of theta and alpha bands (zero represents the onset of the delay phase). The first
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
393
Fig. 4. Grand average alpha band power during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
stage was sensory processing indexed by the theta band (Fig. 3). The offset of sample stimulus elicited theta activity, which was peaking at about 0.35 s, and the theta activity reduced to the minimum value at about 0.6 s. The second stage was duration maintenance in WM indexed by the alpha band (Fig. 4). Alpha band activity increased abruptly from about 0.4–1 s, which reflected an increase in the neural activity of duration maintenance in WM. The result was consistent with the time course of an event-related potential component, P300, that is an index of the WM process (Figs. 3 and 4 do not show the event-related potentials, but they show oscillatory powers averaged across frequency for theta and alpha bands respectively) (Polich and Criado, 2006; Polich, 2007). Then, the alpha band maintained a steady activation from 1 s to 3 s. The third stage was sensory processing indexed by the theta band (Fig. 3). The theta band was elicited by onset of the probe stimulus from 3 s to 3.5 s; at the same time, alpha band activity decreased abruptly (Fig. 4). For theta band power from the 0.5-s to 0.1-s time interval before the onset of the delay phase, a repeated measures ANOVA did not show a significant effect of
Duration, or a Duration Region interaction (p values >0.05). For the theta band power from the 1-s to 3-s interval after the onset of delay, an ANOVA revealed a significant main effect of Region [F(4.197,71.350) = 5.606, p < 0.001, gp2 = 0.248]. The theta power was greatest over the middle-frontal (0.032 ± 0.155 dB) and middle-central regions (-0.011 ± 0.107 dB). The main effect of Duration [F(2.058,34.986) = 0.929, p > 0.05, gp2 = 0.052] and the Duration Region interaction [F(6.360,108.128) = 1.024, p > 0.05, gp2 = 0.057] were not significant. For alpha band power from the 0.5-s to 0.1-s interval before the onset of delay, a repeated measures ANOVA did not show a significant effect of Duration or a Duration Region interaction (p values > 0.05). For alpha band power from the 1-s to 3-s interval after the onset of delay, an ANOVA revealed a significant main effect of Duration [F(2.697,45.843) = 3.813, p < 0.05, gp2 = 0.183]. Post hoc tests revealed that the differences of alpha power between 1-s (0.272 ± 0.135 dB), 2-s (0.315 ± 0.143 dB), and 3-s (0.235 ± 0.140 dB) duration conditions were not significant (t(17) = 0.568–0.820, p values >0.05), and
394
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
that the alpha power of 4-s (0.018 ± 0.129 dB) was lower than that of the 1-s, 2-s, and 3-s duration conditions (t(17) = 2.307–2.790, p values 0.05, gp2 = 0.088] and the Duration Region interaction [F(5.767,98.042) = 0.777, p > 0.05, gp2 = 0.044] were not significant. As shown in Fig. 5, theta and alpha band powers were both greatest over the frontal region during the 1–3-s time interval after the onset of the delay phase. To further assess the role of alpha band power in duration maintenance in WM, the relationship between accuracy and amplitude of alpha power at the middle-frontal cluster during the delay phase (1–3 s) was analyzed for 1 s, 2 s, 3 s, and 4 s, respectively. As shown in Fig. 6, the amplitude of alpha power positively correlated with accuracy for the 3-s duration condition (Pearson’s r = 0.546, p < 0.05). The correlations were not significant for 1-s (r = 0.354, p > 0.05), 2-s (r = 0.124, p > 0.05), and 4-s (r = 0.395, p > 0.05) conditions.
DISCUSSION The present study used a matching-to-sample task to study the neural basis of duration maintenance in WM. EEG data showed that the theta band amplitude in the delay phase was not significantly modulated by the length of duration. The result confirmed our hypothesis that theta band frequencies are not engaged in maintenance of duration in WM. We also found that the alpha band amplitude of the 4-s duration condition in the delay phase was lowest among four duration conditions, and the alpha band amplitude positively correlated with accuracy in the 3-s duration condition. The findings emphasized the role of the alpha band in duration maintenance in WM. As shown in Fig. 2, the time intervals between baseline and delay phase were different in the 1-s, 2-s, 3-s, and 4-s duration conditions, and this factor might have affected the alpha amplitude during the delay phase. If this factor had been an influence, it would have affected the alpha band amplitude between 0.5 and 0.1 s before the onset of the delay phase,
Fig. 6. The correlation between accuracy and amplitude of alpha band power during duration maintenance in the 3-s duration condition.
because the time intervals between baseline and this time window were different for the four duration conditions. An ANOVA revealed no significant main effect of Duration and no Duration Region interaction for amplitude of theta and alpha band power from the 0.5 to 0.1 s before the onset of the delay phase. It indicated that the difference in time intervals from baseline to the delay phase for four duration conditions did not significantly affect the theta and alpha band amplitude during the delay phase. It is still debated whether alpha oscillation is related to successful maintenance of item information (Jensen et al., 2002; Palva and Palva, 2007; Hsieh et al., 2011; Johnson et al., 2011), or whether it reflects inhibition of task-irrelevant information (Jokisch and Jensen, 2007; Klimesch et al., 2007; Tuladhar et al., 2007; Manza et al., 2014). Our study provided three pieces of evidence to support the maintenance rather than inhibition. First, alpha band amplitude in the delay phase was significantly lower in the 4-s duration condition than in the 1-s, 2-s, and 3-s duration conditions. Previous
Fig. 5. The topographies of oscillatory activities of theta and alpha bands during the delay phase in the 1-s, 2-s, 3-s, and 4-s duration conditions.
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
studies suggest that short time intervals (below about 3 s) can be perceived as a unit, whereas long time intervals (above about 3 s) have to be divided into subunits and depend on a memory-based cognitive reconstruction (Fraisse, 1984; Po¨ppel, 1997; Ulbrich et al., 2007). Our findings support previous work, as they indicate that durations below about 3 s are encoded as a unit, and can therefore be retained in WM; thus, the reproduction and discrimination of durations below about 3 s are accurate, and the alpha band amplitude was high during the delay phase. In contrast, the encoding of durations above about 3 s is based on memory and cognitive reconstruction, and their internal representations cannot be retained in WM. Thus, the reproduction and discrimination of durations above 3 s is less accurate, and the alpha band amplitude was low during the delay phase. Therefore, the EEG results support our hypothesis that alpha band activity reflected duration maintenance in WM, and different internal representations were retained in WM for durations below and above about 3 s. Second, alpha band activity was greatest over the frontal region in the present study (Fig. 5). Previous studies indicate that alpha band activity is typically found over the parieto-occipital region, so some scholars suggest it reflects suppression of processing in visual areas (Jokisch and Jensen, 2007; Tuladhar et al., 2007; Freunberger et al., 2011). However, the finding of frontal distributed alpha band is consistent with previous findings that the frontal region plays an important role in duration maintenance in WM (Olton, 1989; Sakurai et al., 2004). Furthermore, one EEG study reported that alpha band activity appeared over the frontal regions when visuospatial information was retained and demanded manipulation in WM; these authors suggested that an increase of prefrontal alpha amplitude may not be interpreted in terms of inhibition, but may enable a tight functional coupling between prefrontal cortical areas (Sauseng et al., 2005). Therefore, the present study indicates that the alpha band activity distributed over the frontal region does not reflect the inhibition of the frontal region during maintenance of duration in WM, but suggests that the frontal distributed alpha band reflects the duration maintenance in WM. Third, alpha band amplitude positively correlated with accuracy in the 3-s duration condition (Fig. 6). We hypothesized that 3 s is near the boundary between short and long durations based on previous studies. It is natural to expect individual differences in the boundary. Subjects with a longer boundary (for whom 3 s is a short duration) are likely to show high accuracy and high alpha band amplitude, and subjects with a shorter boundary (for whom 3 s is a long duration) are likely to show low accuracy and low alpha band amplitude. This hypothesis was confirmed by the significant correlation between alpha band amplitude and accuracy in the 3-s duration condition. In addition, two issues need to be explained in the present study. Firstly, the behavioral and EEG results revealed different boundaries between short and long durations. Our results were in line with previous
395
studies. Behavioral studies support the concept of a boundary between 2 s and 3 s (Szelag et al., 2002; Ulbrich et al., 2007), but an EEG study indicates a boundary between 3 s and 4 s (Elbert et al., 1991). Research suggests that variance in timing behavior is mainly due to encoding (clock), memory, and decision processes (Gibbon and Church, 1984; Allan, 1998). In the present study, the behavioral results were not only influenced by WM, but also by encoding and decision processes. However, the encoding (timing), maintenance, and decision processes for temporal information can be distinguished by EEG. Thus, EEG data are more suitable than behavioral data to investigate the boundary between short and long durations. This explains the slight difference in boundary between the behavioral and EEG results. Lastly, the correlations between alpha band amplitude and accuracy were not significant in 1-s, 2-s and 4-s duration conditions. The results do not simply imply that alpha band power has nothing to do with accuracy in 1-s, 2-s and 4-s duration conditions. A correlation with larger sample size is easier to be statistical significant (Lowry, 2001), and a larger sample size is adopted by typical correlation studies (e.g. Adelabu, 2007; Hayashi et al., 2014). Given the ideal statistical power (1 b) for any study is considered to be 0.8 (Suresh and Chandrashekara, 2012), and the significance level (a) is 0.05, we can compute the ideal sample size as 60, 508, and 48 for 1-s, 2-s and 4-s duration conditions respectively (Cohen, 1988). The present experiment was designed to test the hypothesis of different internal representations retained in WM for durations below and above about 3 s, but it was not suitable to evaluate the correlations between alpha band amplitude and accuracy in 1-s, 2-s, and 4-s duration conditions because of the small sample size.
CONCLUSIONS The present study is the first to explore the neural oscillatory correlates of duration maintenance in WM. We found that theta activity was not modulated by increasing length of duration retained in WM, whereas alpha power was lower in the 4-s duration condition than in the 1-s, 2-s, and 3-s duration conditions. The alpha power amplitude was positively correlated with accuracy for the 3-s duration condition, emphasizing the important role of frontal distributed alpha band oscillation in maintenance of duration in WM. We provided evidence of different internal representations retained in WM for durations below and above about 3 s. The precise nature of the internal representation of duration is still unclear, and further work is necessary to elucidate this. Acknowledgments—This study was supported by a grant from the National Natural Science Foundation of China (31200855, 31300845), the Key Research Institute of Humanities and Social Science in Chongqing (10SKB23), the Doctoral Foundation of Southwest University (SWU110037).
396
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397
REFERENCES Adelabu DH (2007) Time perspective and school membership as correlates to academic achievement among African American adolescents. Adolescence 42:525–538. Allan LG (1998) The influence of the scalar timing model on human timing research. Behav Process 44:101–117. Baddeley A (1992) Working memory. Science 255:556–559. Buhusi CV, Meck WH (2005) What makes us tick? Functional and neural mechanisms of interval timing. Nat Rev Neurosci 6: 755–765. Buzsa´ki G, Draguhn A (2004) Neuronal oscillations in cortical networks. Science 304:1926–1929. Cohen J (1988) Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Coull JT, Nazarian B, Vidal F (2008) Timing, storage, and comparison of stimulus duration engage discrete anatomical components of a perceptual timing network. J Cogn Neurosci 20:2185–2197. Delorme A, Makeig S (2004) EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 134:9–21. Elbert T, Ulrich R, Rockstroh B, Lutzenberger W (1991) The processing of temporal intervals reflected by CNV-like brain potentials. Psychophysiology 28:648–655. Engel AK, Fries P, Singer W (2001) Dynamic predictions: oscillations and synchrony in top-down processing. Nat Rev Neurosci 2:704–716. Fraisse P (1984) Perception and estimation of time. Annu Rev Psychol 35:1–36. Franssen V, Vandierendonck A, Van Hiel A (2006) Duration estimation and the phonological loop: articulatory suppression and irrelevant sounds. Psychol Res 70:304–316. Freunberger R, Werkle-Bergner M, Griesmayr B, Lindenberger U, Klimesch W (2011) Brain oscillatory correlates of working memory constraints. Brain Res 1375:93–102. Genovesio A, Tsujimoto S, Wise SP (2009) Feature- and order-based timing representations in the frontal cortex. Neuron 63:254–266. Gevins A, Smith ME, McEvoy L, Yu D (1997) High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing, and practice. Cereb Cortex 7: 374–385. Gibbon J (1977) Scalar expectancy theory and Weber’s law in animal timing. Psychol Rev 84:279–325. Gibbon J, Church RM (1984) Sources of variance in an information processing theory of timing. In: Animal cognition (Roitblat HL, et al., eds), pp 465–488 Hillsdale, NJ: Erlbaum. Gibbon J, Church RM, Meck W (1984) Scalar timing in memory. In: Annals of the New York Academy of Sciences: timing and time perception, vol. 423 (Gibbon J, Allan L, eds), pp 52–77 New York: New York Academy of Sciences. Greenhouse SW, Geisser S (1959) On methods in the analysis of profile data. Psychometrika 24:95–112. Harrington DL, Zimbelman JL, Hinton SC, Rao SM (2010) Neural modulation of temporal encoding, maintenance, and decision processes. Cereb Cortex 20:1274–1285. Hayashi MJ, Kantele M, Walsh V, Carlson S, Kanai R (2014) Dissociable neuroanatomical correlates of subsecond and suprasecond time perception. J Cogn Neurosci 26:1685–1693. Hsieh LT, Ekstrom AD, Ranganath C (2011) Neural oscillations associated with item and temporal order maintenance in working memory. J Neurosci 31:10803–10810. Jensen O, Tesche CD (2002) Frontal theta activity in humans increases with memory load in a working memory task. Eur J Neurosci 15:1395–1399. Jensen O, Gelfand J, Kounios J, Lisman JE (2002) Oscillations in the alpha band (9–12 Hz) increase with memory load during retention in a short-term memory task. Cereb Cortex 12:877–882. Johnson JS, Sutterer DW, Acheson DJ, Lewis-Peacock JA, Postle BR (2011) Increased alpha-band power during the retention of shapes and shape-location associations in visual short-term memory. Front Psychol 2:128.
Jokisch D, Jensen O (2007) Modulation of gamma and alpha activity during a working memory task engaging the dorsal or ventral stream. J Neurosci 27:3244–3251. Jung TP, Makeig S, Humphries C, Lee TW, McKeown MJ, Iragui V, Sejnowski TJ (2000a) Removing electroencephalographic artifacts by blind source separation. Psychophysiology 37: 163–178. Jung TP, Makeig S, Westerfield M, Townsend J, Courchesne E, Sejnowski TJ (2000b) Removal of eye activity artifacts from visual event-related potentials in normal and clinical subjects. Clin Neurophysiol 111:1745–1758. Kagerer FA, Wittmann M, Szelag E, Von Steinbuchel N (2002) Cortical involvement in temporal reproduction: evidence for differential roles of the hemispheres. Neuropsychologia 40: 357–366. Klimesch W, Sauseng P, Hanslmayr S (2007) EEG alpha oscillations: the inhibition-timing hypothesis. Brain Res Rev 53:63–88. Levine TR, Hullett CR (2002) Eta squared, partial eta squared, and misreporting of effect size in communication research. Hum Commun Res 28:612–625. Lowry R (2001) Significance of a correlation coefficient. Available from: http://vassarstats.net/rsig.html. Makeig S (1993) Auditory event-related dynamics of the EEG spectrum and effects of exposure to tones. Electroencephalogr Clin Neurophysiol 86:283–293. Makeig S, Delorme A, Westerfield M, Jung TP, Townsend J, Courchesne E, Sejnowski TJ (2004) Electroencephalographic brain dynamics following manually responded visual targets. PLoS Biol 2:e176. Manza P, Hau CL, Leung HC (2014) Alpha power gates relevant information during working memory updating. J Neurosci 34:5998–6002. Meck WH, Vatakis A, van Rijn H (2013) Timing & time perception enters a new dimension. Timing Time Percept 1:1–2. Meltzer JA, Negishi M, Mayes LC, Constable RT (2007) Individual differences in EEG theta and alpha dynamics during working memory correlate with fMRI responses across subjects. Clin Neurophysiol 118:2419–2436. Merchant H, Harrington DL, Meck WH (2013) Neural basis of the perception and estimation of time. Annu Rev Neurosci 36: 313–336. Morillon B, Kell CA, Giraud AL (2009) Three stages and four neural systems in time estimation. J Neurosci 29:14803–14811. Olton DS (1989) Frontal cortex, timing and memory. Neuropsychologia 27:121–130. Onton J, Delorme A, Makeig S (2005) Frontal midline EEG dynamics during working memory. NeuroImage 27:341–356. Palva S, Palva JM (2007) New vistas for alpha-frequency band oscillations. Trends Neurosci 30:150–158. Polich J (2007) Updating p300: an integrative theory of P3a and P3b. Clin Neurophysiol 118:2128–2148. Polich J, Criado JR (2006) Neuropsychology and neuropharmacology of P3a and P3b. Int J Psychophysiol 60:172–185. Po¨ppel E (1997) A hierarchical model of temporal perception. Trends Cogn Sci 1:56–61. Rattat AC (2010) Bidirectional interference between timing and concurrent memory processing in children. J Exp Child Psychol 106:145–162. Rattat AC, Picard D (2012) Short-term memory for auditory and visual durations: evidence for selective interference effects. Psychol Res 76:32–40. Roberts BM, Hsieh LT, Ranganath C (2013) Oscillatory activity during maintenance of spatial and temporal information in working memory. Neuropsychologia 51:349–357. Roux F, Uhlhaas PJ (2014) Working memory and neural oscillations: alpha-gamma versus theta-gamma codes for distinct WM information? Trends Cogn Sci 18:16–25. Sakurai Y, Takahashi S, Inoue M (2004) Stimulus duration in working memory is represented by neuronal activity in the monkey prefrontal cortex. Eur J Neurosci 20:1069–1080.
Y. G. Chen et al. / Neuroscience 290 (2015) 389–397 Sauseng P, Klimesch W, Doppelmayr M, Pecherstorfer T, Freunberger R, Hanslmayr S (2005) EEG alpha synchronization and functional coupling during top-down processing in a working memory task. Hum Brain Mapp 26:148–155. Suresh K, Chandrashekara S (2012) Sample size estimation and power analysis for clinical research studies. J Hum Reprod Sci 5:7–13. Szelag E, Kowalska J, Rymarczyk K, Poppel E (2002) Duration processing in children as determined by time reproduction: implications for a few seconds temporal window. Acta Psychol 110:1–19.
397
Tuladhar AM, ter Huurne N, Schoffelen JM, Maris E, Oostenveld R, Jensen O (2007) Parieto-occipital sources account for the increase in alpha activity with working memory load. Hum Brain Mapp 28:785–792. Ulbrich P, Churan J, Fink M, Wittmann M (2007) Temporal reproduction: further evidence for two processes. Acta Psychol 125:51–65. World Medical Association (2013). WMA declaration of Helsinki – ethical principles for medical research involving human subjects. Available from: http://www.wma.net/en/30publications/10policies/ b3/index.html.
(Accepted 8 January 2015) (Available online 28 January 2015)
