Cerebral Cortex June 2011;21:1223--1230 doi:10.1093/cercor/bhq233 Advance Access publication November 10, 2010 FEATURE ARTICLE Endogenous Auditory Spatial Attention Modulates Obligatory Sensory Activity in Auditory Cortex Alan J. Power1, Edmund C. Lalor1,2 and Richard B. Reilly1,2 Trinity Centre for Bioengineering and 2Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin 2, Ireland 1 Address correspondence to Edmund C. Lalor. Email: edlalor@tcd.ie. Keywords: AESPA, auditory evoked potential, continuous and competing stimuli, dichotic listening Introduction In natural everyday environments, we are typically bombarded with a convoluted mixture of competing sound sources. The auditory system has evolved, however, to be able to complete the highly complex task of allowing us to focus, with relative ease, on a constituent source among the combination of sources reaching the ears. There are 2 broad classes of attentional mechanisms: exogenous attention and endogenous attention (e.g., Hopfinger and West 2006). Exogenous attention is the attraction of attentional focus by an environmentally salient stimulus, whereas endogenous attention is the selfdirected focus of attention to a particular region or feature of the environment. Extensive research investigating the intriguing phenomenon of endogenous auditory attention has been carried out using the electroencephalogram (EEG) and the magnetoencephalogram (Hillyard et al. 1973; Näätänen et al. 1992; Woldorff et al. 1993). These studies throw up an interesting debate as to the generators affected by endogenous attention. Hillyard et al. (1973), in their seminal study, found the N1 wave of the auditory evoked potential (AEP) to be enhanced by auditory selective attention. They suggested that this was due to increased sensory processing of the attended stimulus. This view is supported by others (Woldorff et al. Ó The Author 2010. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com 1993). However, the idea that obligatory sensory components of the AEP are affected by endogenous attention has been challenged by Näätänen and Picton (1987) and Näätänen et al. (1992). They suggest that the N1 enhancement is likely due to the engagement or enhancement of separate components of the N1 wave that are not related to obligatory processes but to a matching process between the sensory input and an ‘‘attentional trace.’’ Although they suggest that under some conditions, obligatory components may possibly be affected by endogenous attention (Näätänen and Picton 1987), they remain somewhat skeptical about obligatory sensory involvement in endogenous auditory attention (Näätänen et al. 1992; Alho et al. 1994). Research into this question is ongoing (Bidet-Caulet et al. 2007; Ross et al. 2010). The majority of EEG studies investigating endogenous auditory attention, such as those mentioned above, have employed the standard event-related potential (ERP) technique. These studies, although informative, are somewhat hampered by the fact that they use discrete stimuli that may not be best suited to the investigation of endogenous attention. Specifically, the assertion that attentional effects to discrete onset stimuli are entirely due to endogenous top-down attention is complicated by the exogenous attention grabbing effects of onset stimuli. It has been suggested, in the visual domain, that endogenous and exogenous attention may be separate systems (Hopfinger and West 2006). Whether these proposed separate systems affect sensory processing in the same manner, however, is still unknown, as is how they interact when both are engaged. That said, despite the fact that the attention grabbing nature of discrete stimuli is well established in both visual (Jonides 1981) and auditory (Escera et al. 2000) modalities there have been very few studies that attempt to eliminate the possible confounding exogenous attentional influence on the investigation of endogenous attention. The continuous nature of steady-state response (SSR) stimuli likely minimizes the influence of exogenous attention on the SSRs. Studies employing SSR to investigate sustained auditory attention have had mixed results, however: One study using simultaneously presented click trains presented at a rate of either 37 or 40 Hz found a transient N1 effect to stimulus onset but no steady-state effect (Linden et al. 1987), which suggests that sustained responses are not affected by endogenous attention. A more recent study using amplitude modulated tone bursts of 800 ms duration, however, found a steady-state effect but no transient N1 effect (Ross et al. 2004), which suggests that sustained responses are affected by endogenous attention. The latter study did not employ a stimulus competition design however and used a visual task in the unattended condition. Thus, we are unable to discern, from this Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 Endogenous attention is the self-directed focus of attention to a region or feature of the environment. In this study, we assess the effects of endogenous attention on temporally detailed responses to continuous and competing auditory stimuli obtained using the novel auditory evoked spread spectrum analysis (AESPA) method. There is some debate as to whether an enhancement of sensory processing is involved in endogenous attention. It has been suggested that attentional effects are not due to increased sensory activity but are due to engagement of separate temporally overlapping nonsensory attention-related activity. There are also issues with the fact that the influence of exogenous attention grabbing mechanisms may hamper studies of endogenous attention. Due to the nature of the AESPA method, the obtained responses represent activity directly related to the stimulus envelope and thus predominantly correspond to obligatory sensory processing. In addition, the continuous nature of the stimuli minimizes exogenous attentional influence. We found attentional modulations at ~136 ms (during the Nc component of the AESPA response) and localized this to auditory cortex. Although the involvement of separate nonsensory attentional centers cannot be ruled out, these findings clearly demonstrate that endogenous attention does modulate obligatory sensory activity in auditory cortex. Figure 1. An AESPA response obtained, at Fz, using a modulated broadband noise carrier with equal power in the range 0--22.05 kHz. 1224 Endogenous Auditory Spatial Attention d Power et al. continuous auditory streams concurrently, one to the left and one to the right ear, and subjects are instructed to attend either to the left or right. Materials and Methods Subjects and Data Acquisition Seventeen subjects aged 21--33 (15 males) participated in the study. The experiment was undertaken in accordance with the Declaration of Helsinki. The Ethics Committee of the School of Psychology at Trinity College Dublin approved the experimental procedures, and each subject provided written informed consent. Subjects reported no history of hearing impairment or neurological disorder. EEG data were recorded from 130 electrode positions, filtered over the range 0--134 Hz, and digitized at the rate of 512 Hz using a BioSemi Active Two system. EEG data were then digitally filtered off-line with a high-pass filter, where the passband was above 2 Hz and with a –60 dB response at 1 Hz and a low-pass filter with passband below 35 Hz and a –50 dB response at 45 Hz. The data at each channel were re-referenced to the average of the responses at the left and right mastoids. Stimuli The AESPA stimulus consists of a carrier stimulus amplitude modulated by a spread spectrum signal. In this case, 2 root mean square (RMS) normalized band-pass noise carriers of bandwidth 1 kHz centered at 1 kHz (LOW stream) and 5 kHz (HIGH stream), respectively, were employed. These center frequencies were chosen on the basis that 1 kHz and 5 kHz tones are perceived with approximately the same loudness (ISO:226) and also because they are far enough apart in frequency that they are perceived separately. These carriers were then modulated by independent spread spectrum signals, and the low- and high-frequency streams were concurrently presented to the left and right ears, respectively (Fig. 2a,b,c). The reason that the stimuli were separated in location as well as carrier stimulus center frequency is because, 1) it is spatial attention that is under investigation and 2) if both the left and right streams had identical carriers, the stimuli may have been perceived as 1 auditory object varying in interaural intensity difference as opposed to 2 spatially separate objects. Figure 2. The stimulation setup. (a) LOW and HIGH streams were presented dichotically. A segment of the (b) LOW and (c) HIGH frequency carrier stimuli, respectively. Examples of (d) target and (e) distracter events. Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 experimental protocol, whether one competing stimulus would result in increased activity relative to another. The somewhat contradictory results suggest that the effects of sustained endogenous auditory attention on competitive continuous stimuli have yet to be adequately elucidated. A further issue relates to the simultaneity of stimuli in most ERP-based studies. In such studies, the discrete stimuli are not presented truly simultaneously in the attended and unattended channels (Hillyard et al. 1973; Näätänen et al. 1992). Thus, the argument that the effects found are due to competing stimuli is somewhat weakened. In an attempt to overcome this drawback, paradigms where simultaneous stimulation is employed have been carried out. A recent study assessed the transient onset responses of subjects who were asked to detect an occasional change in modulation frequency of amplitude modulated sounds presented to one ear while ignoring concurrent sounds in the other ear (Ross et al. 2010). Interpretation of these results, however, is complicated by the fact that any increase in the transient response to the attended stimulus would be superimposed on the unaffected (or even possibly inhibited) response to the unattended stimulus perhaps diluting any attention effect. Indeed, researchers investigating auditory scene analysis have been forced to go to great lengths to create the perception of auditory streaming using discrete stimuli (Sussman et al. 1999; Ritter et al. 2006; De Sanctis et al. 2008). Recently, however, there has been a move toward employing more natural stimulus paradigms to assess auditory function (Lalor et al. 2009; Kerlin et al. 2010; Lalor and Foxe 2010). Lalor and Foxe (2010) obtained temporally detailed responses to natural speech stimuli and Kerlin et al. (2010) found attentional modulation of activity in auditory cortex (AC) when using competing speech stimuli and a template-matching analysis method. In an attempt to further address some of the concerns outlined above, we employ a novel method for obtaining temporally detailed responses to continuous stimuli: The auditory evoked spread spectrum analysis (AESPA) method (Lalor et al. 2009). Figure 1 shows a typical AESPA response obtained using a continuous amplitude modulated broadband noise stimulus. The AESPA response represents obligatory sensory activity directly related to the input stimulus (Lalor et al. 2009). The AESPA method also allows for the extraction of separate responses to simultaneously presented continuous stimuli. Furthermore, the continuous nature of the stimuli minimizes exogenous attentional influence. Exploiting this method, we employed a cocktail party-like experimental approach to investigate the effects of endogenous attention on sensory processing in the auditory system. We present 2 The spread spectrum signals consisted of Gaussian noise with energy uniformly distributed between 0 and 30 Hz. Taking into account, the logarithmic nature of auditory stimulus intensity perception, the values of these modulating signals, x, were then mapped to the amplitude of the audio stimulus, x#, using the following exponential relationship: x# = 102x ; and normalized to between 0 and 1. It was expected that this would result in a more linear perception of audio intensity modulation. Transitions between levels were smoothed by using a 5-ms ramp consisting of half a period of a 100-Hz sine wave. Using the Nyquist sampling theorem and given that EEG power above 30 Hz is low, the modulation rate of each signal was set to be 60 Hz. Signal Processing We obtain the AESPA by performing a linear least squares fit of the response model Quantification of Results When calculating task performance any response occurring within a 1 s period after an event was considered to be a response to that event. We calculated the percentage of correct responses, percentage of responses to distracters in the attended stream, and also percentage of responses to events in the to-be ignored stream, which are shown in Table 1. To test the behavioral results statistically, they were submitted to a 2 3 2 repeated measures analysis of variance (ANOVA) using factors of stimulus (LOW-LEFT vs. HIGH-RIGHT) and event in the attended stream (target vs. distracter). The Global Field Power (GFP, Lehmann and Skrandies 1980) was obtained for the grand average responses to each stimulus condition. The GFP is a reference free measure of the response power over the whole scalp. The GFP was used for preliminary visualization of the data and to indicate possible periods of attentional modulation of the responses. The periods of interest were obtained by submitting the GFPs to running t-tests. A component was considered to be of interest if responses were significantly different (P < 0.01) for a period of at least 11 consecutive samples (~20 ms). In order to investigate how different areas are affected by attention, we also partitioned the scalp into 4 regions (Frontal, Central, Left Temporal, and Right Temporal) and averaged over responses at electrodes within each region. Later, AESPA components (i.e., Nc and Pd) have been shown to have very low signalto-noise ratio in parietal--occipital regions (Lalor et al. 2009), and thus this region was not included in analysis. Figure 3 shows the responses in the included regions as well as the GFPs for the HIGH-RIGHT and LOW-LEFT responses when attended and unattended. In the GFP plots, the areas identified as being of significant interest by the running t-tests are shaded gray. Topographic maps of the Nc component when attended and unattended are also shown, as are difference topographies. Statistical differences in components identified as being possibly effected by attention were tested using a repeated measures ANOVA. The RMS values in a ~10 ms window around the component peak were used in the statistical analysis. Topographic maps of affected components were also obtained using the average amplitude in the relevant windows. Component topographies were compared using the topographic ANOVA (TANOVA) method (Murray et al. 2008). This method assesses whether 2 topographies are statistically different using a nonparametric randomization procedure. Given that topographic maps reflect source configuration, this was done in order to assess whether the attentional modulation of any component was due to an engagement of additional generators or whether it was merely due to enhanced activity of generators already engaged in the unattended condition (i.e., if topographies are not statistically different, then we can assert that they are likely due to the same generators). A further more detailed investigation of the component generators was also carried out using the BESA (5.2) source analysis package. Table 1 Behavioral results yðt Þ = wðsÞ
xðt Þ + noise; where y(t) is the measured EEG response, x(t) is the amplitude envelope of the stimulus, the symbol * indicates convolution, w(s) is the impulse-response function to the amplitude of the stimulus, and the noise is assumed to be Gaussian (Lalor et al. 2009). In other words, the AESPA response, w (s), is analogous to a filter that describes how the brain transforms the auditory input into the EEG output. Keeping p(T) ± standard deviation% p(D) ± standard deviation% p(TBI) ± standard deviation% LOW-LEFT HIGH-RIGHT 82.89 ± 11.59 57.83 ± 26.81 1.79 ± 1.29 86.75 ± 9.35 65.41 ± 26.84 1.41 ± 1.07 Note: Percentage of correct responses (p(T)), percentage of responses to distracters in the attended stream (p(D)), and percentage of events responded to in the to-be-ignored stream (p(TBI)) are shown for each condition. Cerebral Cortex June 2011, V 21 N 6 1225 Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 Experimental Procedure and Tasks While abstaining completely from eyeblinks is not possible for long periods, subjects were instructed to keep the number of eyeblinks to a minimum during each session. Subjects were also instructed to keep all other types of motor activity to a minimum during testing. Testing was carried out in a dark room, and in order to minimize eye movements, subjects were asked to keep their eyes open and to fixate on a small cross presented in the center of their visual field. Each subject undertook 10 trials where they were asked to attend to the HIGH stream in the right ear (attend-HIGH-RIGHT condition) and 10 trials where they were asked to attend to the LOW stream in the left ear (attend-LOW-LEFT condition). Each trial was 120 s in duration. The sequence of conditions was randomized for each subject. Stimuli were presented at an intensity level deemed comfortable by the subject before beginning the experiment. In order to monitor each subject’s progress, targets and distracters were inserted randomly in each stream. These events consisted of a specific pattern of amplitude modulation imposed on the random process. Targets consisted of a modulation level of –2.5 dBfs for 25.5 ms followed by –12 dBfs for 16 ms followed by –2.5 dBfs for 25.5 ms, giving a total length of 67 ms, whereas distracters consisted of a flat modulation of –6 dBfs for 67 ms. dBfs refers to decibels full scale and represents a dB value relative to the maximum modulation level for each subject (see Fig. 2d,e). Although the events are embedded in the stimulus, they are still distinguishable from the ongoing amplitude modulations. This is due to the fact that the events are generally somewhat louder than the ongoing modulation. The reason for this is because of the exponential mapping outlined above that restricts the modulating waveform to spend ~90% of its time below –12 dBfs. Furthermore, due to the random nature of the modulating waveform, nonevent-related amplitudes exceeding –12 dBfs generally have a short duration compared with the 67 ms duration of events. Subjects were instructed to respond only when a target in the attended stream was heard. Each trial contained a total of 24 events (i.e., both targets and distracters). The proportion of targets and distracters in each trial was randomly assigned ranging from 8 targets (and therefore 16 distracters) to 16 targets (and therefore 8 distracters). On average, 48.75% of events were targets and 51.25% were distracters. No event, either within or between streams, could occur within 1 s of another and the maximum separation between events within streams could not be more than 9 s. EEG was recorded for later analysis where both the responses to the HIGH-RIGHT and LOW-LEFT streams for each condition were extracted. The stimulation paradigm and the stimuli are outlined in Figure 2. this in mind, the time axis for the AESPA carries a different meaning than the time axes in traditional ERP studies. Each point on the time axis can be interpreted as being the relative time between the continuous EEG and the continuous input intensity signal. Therefore, the AESPA at –100 ms, for example, indexes the relationship between the input intensity signal and the EEG 100 ms earlier; obviously this should be 0. As another example, the AESPA at 100 ms indexes how the input intensity signal affects the EEG 100 ms later. Results Behavioral Results In order to keep subjects highly engaged in the attentional task, discriminating between targets and distracters in the attended stream was deliberately made difficult. The difficulty of the task was evidenced by the relatively high number of distracters to which subjects responded (see Table 1). However, the fact that subjects responded to a significantly greater percentage of targets than distracters (main effect of event, F1,16 = 28.088, P < 0.001) shows that subjects were capable of performing the task. Furthermore, the very low percentage of events responded to in the ignored stream indicates that attention 1226 Endogenous Auditory Spatial Attention d Power et al. to the intended stream was taking place. Despite the fact that we found a significant main effect of stimulus (F1,16 = 14.203, P = 0.002), due to higher number of event responses when the right ear was attended than when the left ear was attended, there was no stimulus 3 event interaction (P > 0.05). This suggests the ability to carry out the task (i.e., to distinguish between targets and distracters) did not differ depending on which ear was attended. Electrophysiology Results Based on the running t-tests carried out on the GFPs, we identified the Nc and Pd components as being of particular interest for further investigation. Nc and Pd were defined as the Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 Figure 3. Responses for indicated regions and GFPs for the stimuli presented at the left and right ears when attended and unattended (upper panels). Time intervals of interest identified by running t-tests are shaded gray in the GFPs. Topographic maps of the Nc component when attended and unattended as well as difference topographies are shown in lower panel. stimulus type (P > 0.05 for both). This tells us that there was no hemispheric bias due to the spatial nature of the stimuli. The possibility that increased activity in the Nc component window may be due to an engagement of additional nonobligatory generators and not increased activity of the sensory activity was investigated using the TANOVA method (Murray et al. 2008). Topographies in attended and unattended conditions for the LOW-LEFT response were not found to be statistically different (LOW-LEFT attended vs. unattended: P = 0.5855). This suggested that the same generators were involved in both attended and unattended conditions for the responses to the LOW-LEFT stimulus. In the case of the HIGH-RIGHT responses, however, the TANOVA did suggest that the topographies were statistically different between conditions (P = 0.0147). That said, the TANOVAs did not indicate statistically dissimilar topographies between the LOW-LEFT unattended and HIGH-RIGHT unattended responses (P = 0.3365) or between the LOW-LEFT attended and HIGH-RIGHT attended responses (P = 0.0874) suggesting that regardless of stimulus similar generators were engaged. To further investigate the location and strength of the Nc generators and the possible attention affects, a dipole analysis was carried out on the Nc component. Starting with the Nc component of the unattended LOW-LEFT responses, we attempted to fit 2 symmetrical regional sources (a regional source in BESA consists of 3 orthogonal dipoles at the same location). Fitting the sources to a window encompassing the whole Nc component (111--154 ms: identified from the GFPs) placed the sources at Talairach coordinates x = –32.6, y = –16.8, z = 13.6. This configuration accounted for 98.88% of the variance in the fitted window, and the sources are located within 1 cm of Heschl’s gyrus (HG). In fact fixing the source locations to the center of the auditory core (at talairach x = –46, y = –24, z = 12) only slightly reduced the variance explained to 98.12%. Applying this same model to the attended LOW-LEFT responses accounted for 96.76% of the variance. Thus, bilateral sources in AC provided an excellent model of the Nc component in both the attended and unattended conditions. This coupled with the insignificant dissimilarity in topographies shown by the TANOVA indicates that the same sources were involved when the stimulus was attended and unattended. Following the same procedure for the HIGH-RIGHT responses resulted in an initial localization of the unattended Nc to talairach coordinates x = –37.7, y = –24.9, z = 18.5 with 97.73% of the variance explained. Again this is within 1 cm of HG and fixing the sources to the center of the auditory core as before only slightly reduced the variance explained to 97.24%. Applying this model to the attended HIGH-RIGHT responses accounted for 93.36% of the variance. Although the model accounted for slightly lower variance than in the case of the LOW-LEFT stimuli bilateral sources in AC again provided an excellent model of the Nc component. Employing the same model for the LOW-LEFT and HIGH-RIGHT responses is backed up by the TANOVA results, which suggest that both the LOWLEFT and HIGH-RIGHT stimuli employ similar generators when unattended as well as when attended. Indeed, it is possible that the lower signal power of the HIGH-RIGHT responses (indicated by the main effect of stimulus mentioned above) may have played a part in TANOVA results that suggested topographical differences between HIGH-RIGHT conditions. These noisier responses may also account for the slightly lower explained variance in the BESA model. Thus, the fact that Cerebral Cortex June 2011, V 21 N 6 1227 Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 RMS amplitude in a ~10 ms interval around the mean latency of each component of the grand averages of the LOW-LEFT and HIGH-RIGHT responses (i.e., Nc: 136 ms and Pd: 208 ms, see Fig. 3). In order to test the effects of attention for each stimulation condition, we performed a 4-way 2 3 2 3 4 3 2 repeated measures ANOVA using factors of stimulation (LOWLEFT stream vs. HIGH-RIGHT stream), attention (Attended vs. Unattended), electrode region (left, right, central, and frontal), and component (Nc vs. Pd). Greenhouse--Geisser corrections were applied to the repeated measures factors where the sphericity assumption was violated with the corrected degrees of freedom reported. First and most importantly, there was a main affect of attention (F1,16 = 27.75, P < 0.001), indicating that components of the responses to the attended stream were enhanced. There was also a significant effect of stimulus (F1,16 = 54.004, P < 0.001). This is due to the fact that the later cortical responses (i.e., Nc and Pd) to the LOW-LEFT stream are greater in amplitude than responses to the HIGH-RIGHT stream. This is likely due to the logarithmic nature of frequency representation in auditory cortex, that is, the higher the frequency the smaller the amount of cortex devoted to it (Romani et al. 1982). Using a wider carrier stimulus bandwidth for the higher frequency stream may result in more similar HIGH and LOW responses. There was no stimulus 3 attention interaction (P > 0.05), suggesting that both the HIGH-RIGHT and LOW-LEFT streams were similarly affected by attention. A significant attention 3 region 3 component interaction (F1.98,31.71 = 7.636, P = 0.002) was found. To further interrogate the components driving this interaction, we performed separate 2-way 2 3 4 repeated measures ANOVAs on Nc and Pd with factors of attention (attended vs. unattended) and region (left, right, frontal, and central). In the case of Nc, we found a significant effect of attention (F1,16 = 20.28, P < 0.001) as well as a significant attention 3 region interaction (F3,48 = 3.44, P = 0.024). In the case of Pd, the effect of attention was not significant (P > 0.05), although there was an attention 3 region interaction (F3,48 = 7.41, P < 0.001). In order to ascertain the regions driving these interactions, post hoc t-test were carried out and identified the Nc effect to be driven by attention in left (t16 = 5.21, P < 0.001), right (t16 = 4.1, P = 0.001), central (t16 = 4.6, P < 0.001), and frontal (t16 = 2.4, P = 0.027) regions. The Pd interaction was driven by an effect in the frontal region (t16 = 2.4, P = 0.027). Employing bonferonni correction, however, resulted in a significant Nc effect only in left, right, and central regions but no Pd effect. This suggests that the Pd effect indicated by the running t-test performed on the GFP is marginal, whereas the Nc effect is robust. This is further indicated by an attention 3 component interaction in the initial 4-way ANOVA that approached significance (F1,16 = 3.138, P = 0.096). Just because both left and right regions are similarly affected by attention does not rule out the possibility that responses may be biased to one hemisphere over the other. Since it has been suggested that spatial processing may by lateralized (Zatorre and Penhune 2001; Spierer et al. 2009), we sought investigate whether this was the case here. The initial 4-way ANOVA resulted in a stimulus 3 region interaction (F3,48 = 10.02, P < 0.001). This allowed us to perform post hoc t-tests on the LEFT-LOW and HIGH-RIGHT responses to inspect the regional differences driving this interaction. We found no difference between the right and left regions for either the responses with the higher signal-to-noise ratio (i.e., the responses to the LOW-LEFT stimulus) did not result in statistically significant topographical differences combined with the fact that the Nc component in all stimulus conditions is well explained by sources located in AC suggests that the Nc modulation results from an enhancement of the obligatory sensory activity in AC and not the engagement of supplementary nonobligatory activity. Discussion 1228 Endogenous Auditory Spatial Attention d Power et al. Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 In this study, we used continuous and simultaneous competing stimuli to assess endogenous auditory attention in a cocktail party-like environment. The use of continuous stimuli has eliminated possible confounding effects associated with exogenous attention and discrete stimulation. We found a strong attentional effect of the Nc component in left and right hemispheres as well as central areas. Since the AESPA response primarily represents obligatory sensory processing and since Nc generators have been localized to AC, we have shown that sensory processing in AC is modulated by endogenous top-down attention. Our results are at odds with Näätänen’s attentional trace hypothesis (Näätänen 1982), which suggests that obligatory sensory components are not affected by attention. He proposes that attention acts by way of an endogenous processing negativity (PN), which is due to a comparison process between the neural representation of the stimulus and the relevant attentional trace. This PN overlaps and is superimposed on true sensory processes giving the impression of modulation of sensory activity in many cases. Näätänen et al. (1992) do concede that in some instances simultaneous effects on sensory activity cannot be ruled out entirely due to the inability of current methods to distinguish between obligatory sensory activity and overlapping voluntary activity. They remain skeptical, however, of the involvement of sensory processes in attention effects (Näätänen et al. 1992). Although we have isolated an endogenous attention effect on obligatory sensory processes, due to the nature of AESPA responses we are precluded from investigating the undeniably important voluntary components, such as the PN, which are not well synchronized to stimulus fluctuations. Our results do agree with Hillyard et al. (1973), however, who suggests that sensory processes are affected by attention. The results also agree with the findings of Woldorff et al. (1993), who found attentional modulation of activity localized to AC in the ranges 20--50 ms and 80--130 ms and argue that these effects are due to attentional modulation of sensory processes. Recently, evidence has been emerging that the majority of sound feature processing is achieved subcortically and that AC represents sounds in terms of auditory objects (Nelken 2004). Furthermore, it has been suggested that AC is involved in sensory memory (Näätänen and Winkler 1999; Näätänen 2001; Ulanovsky et al. 2003). A suggested mechanism for sensory memory is stimulus-specific adaptation (SSA; Ulanovsky et al. 2003), which has been posited as a possible neuronal correlate both of the decreased N1 to repeated stimuli and of the mismatch negativity. SSA is the process by which neurons decrease their responses to sequences of identical stimuli, i.e., the activity of neurons is affected by stimulus history. Furthermore, activity of sustained responses has been shown to be significantly affected by SSA (Ulanovsky et al. 2003) and thus, due to the continuous nature of the AESPA stimulus, it is likely that our responses primarily represent activity of the subset of neurons that are least susceptible to SSA. That is to say those cells most susceptible to SSA would contribute minimal activity in response to a continuous stimulus, whereas those least susceptible to SSA, that is, the neurons least involved in sensory memory and most involved in feature processing, would contribute most to the response. This would suggest that the attention effect found here is due to enhanced feature processing and not related to sensory memory representation. Also, the fact that SSA has been shown to be more prominent for sustained responses than for onset responses of neurons in primary AC (Ulanovsky et al. 2003) may account for the smaller size of the Nc components relative to the AEP-N1 component (Lalor et al. 2009). Although we only found attentional effects around 136 ms, it is possible that earlier AESPA components, especially the Nb and Pc components, may be affected by attention and but these components are somewhat ill defined. This may be due to the frequency content of the carrier stimuli: Previously, these components were seen to be ill defined when a 1 kHz tone was used as the carrier stimulus as opposed to broadband noise (Lalor et al. 2009). A wider carrier stimulus bandwidth may allow for a more detailed investigation of these components. The lack of an early effect on well-defined components such as Pa may be due to the nature of the task employed here. It has been shown that the locus of attention is flexible and varies depending on the processing stage most heavily loaded by the task in question (Vogel et al. 2005; Kelly et al. 2008). We employed a difficult event discrimination task (i.e., discrimination between targets and distracters), which is likely to load later processes as opposed to simpler frequency deviant identification tasks (e.g., Woldorff et al. 1993). Recent efforts at assessing attention to simultaneously presented stimuli have looked at transient responses to the onset of amplitude-modulated sounds (Ross et al. 2010). Attentional modulation was found as early as 143 ms and was localized to AC. However, whether this is due to increased sensory activity or a separate endogenous effect is not clear. Furthermore, whether this effect has been diluted by unaffected or inhibited responses to the simultaneous unattended stimulus is unclear. Thus, not only has employing the AESPA method allowed for the investigation of endogenous attention effects on obligatory sensory processes that are unaffected by simultaneous voluntary components, it has also allowed for the isolation of truly separate responses to concurrent stimuli. That said, however, we can only assert that sensory processes are affected by endogenous attention and cannot investigate the certain effects of endogenous attention on nonsensory cognitive processes. A recent attempt was made to examine attention in a competing stimulus environment using an intricate paradigm (Bidet-Caulet et al. 2007). The authors of that study assessed both transient responses and SSRs from depth electrode recordings in patients with epilepsy. While the results from this study were encouraging, they exhibited a number of inconsistencies: Responses to certain stimuli were enhanced when attended in some conditions and reduced when attended in other conditions, and no attention effects were found in a significant number of subjects. Furthermore, the use of onsets and SSRs precluded the assessment of the timing of attentional enhancement during sustained attention. This study also Funding Irish Research Council for Science, Engineering and Technology. Notes We thank Dr Simon P. Kelly for useful comments on the manuscript and Dr Robert Whelan for assistance with the statistical analysis. Conflict of Interest: None declared. References Ahveninen J, Jääskeläinen IP, Raij T, Bonmassar G, Devore S, Hämäläinen M, Levänen S, Lin F-H, Sams M, ShinnCunningham BG, et al. 2006. Task-modulated ‘‘what’’ and ‘‘where’’ pathways in human auditory cortex. Proc Natl Acad Sci. 103:14608--14613. Alho K, Teder W, Lavikainen J, Näätänen R. 1994. Strongly focused attention and auditory event-related potentials. Biol Psychiatry. 38:73--90. Bidet-Caulet A, Fischer C, Besle J, Aguera PE, Giard MH, Bertrand O. 2007. Effects of selective attention on the electrophysiological representation of concurrent sounds in the human auditory cortex. J Neurosci. 27:9252--9261. Cohen YE, Knudsen EI. 1999. Maps versus clusters: different representations of auditory space in the midbrain and forebrain. Trends Neurosci. 22:128--135. De Sanctis P, Ritter W, Molholm S, Kelly SP, Foxe JJ. 2008. Auditory scene analysis: the interaction of stimulation rate and frequency separation on pre-attentive grouping. Eur J Neurosci. 27: 1271--1276. Escera C, Alho K, Schröger E, Winkler I. 2000. Involuntary attention and distractability as evaluated with event-related potentials. Audiol Neurootol. 5:151--166. Hillyard SA, Hink RF, Schwent VL, Picton TW. 1973. Electrial signs of selective attention in the human brain. Science. 182:177--180. Hopfinger JB, West VM. 2006. Interactions between endogenous and exogenous attention on cortical visual processing. Neuroimage. 31:774--789. Jonides J. 1981. Voluntary versus automatic control over the mind’s eye movement. Atten Perform. 9:187--203. Kelly SP, Gomez-Ramirez M, Foxe JJ. 2008. Spatial attention modulates initial afferent activity in human primary visual cortex. Cereb Cortex. 18:2629--2636. Kerlin JR, Shahin AJ, Miller LM. 2010. Attentional gain control of ongoing cortical speech representations in a ‘‘Cocktail Party’’. J Neurosci. 30:620--628. Lalor EC, Foxe JJ. 2010. Neural responses to uninterrupted natural speech can be extracted with precise temporal resolution. Eur J Neurosci. 31:189--193. Lalor EC, Power AJ, Reilly RB, Foxe JJ. 2009. Resolving precise temporal processing properties of the auditory system using continuous stimuli. J Neurophysiol. 102:349--359. Lehmann D, Skrandies W. 1980. Reference-free identification of components of checkerboard-evoked multichannel potential fields. Electroencephalogr Clin Neurophysiol. 48:609--621. Linden RD, Picton TW, Hamel G, Campbell KB. 1987. Human auditory steady-state evoked potentials during selective attention. Electroencephalogr Clin Neurophysiol. 66:145--159. Murray MM, Brunet D, Michel CM. 2008. Topographic ERP analyses: a step-by-step tutorial review. Brain Topogr. 20:249--264. Näätänen R. 1982. Processing negativity: an evoked-potential reflection of selective attention. Psychol Bull. 92:605--640. Näätänen R. 2001. ‘‘Primitive Intelligence’’ in the auditory cortex. Trends Neurosci. 24:283--288. Näätänen R, Picton T. 1987. The N1 wave of human electric and magnetic response to sound: a review and analysis of component structure. Psychophysiology. 24:375--425. Näätänen R, Teder W, Alho K, Lavikainen J. 1992. Auditory attention and selective input modulation: a topographical ERP study. Neuroreport. 3:493--496. Näätänen R, Winkler I. 1999. The concept of auditory stimulus representation in cognitive neuroscience. Psychol Bull. 125:826--859. Nelken I. 2004. Processing of complex stimuli and natural scenes in the auditory cortex. Curr Opin Neurobiol. 14:474--480. Pinek B, Duhamel JR, Cavé C, Brouchon M. 1989. Audio-spatial deficits in human: differential effects associated with left versus right hemisphere parietal damage. Cortex. 25:175--186. Rauschecker JP, Tian B. 2000. Mechanisms and streams for processing of ‘‘what’’ and ‘‘where’’ in auditory cortex. Proc Natl Acad Sci U S A. 97:11800--11806. Cerebral Cortex June 2011, V 21 N 6 1229 Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 posited a dominant role for left hemisphere in attentional selection with right hemisphere being inhibited as a function of attentional load. This is at odds with our results that show voluntary attentional enhancement effects over both left and right hemispheres and no lateralization of responses. However, further inspection of their results reveals attentional enhancement in right hemisphere for a number of attentional conditions, suggesting that it may be premature to make strong conclusions about the hemispheric specialization of auditory attention in sensory areas. There is much debate relating to lateralization of attentional and spatial processing with some studies favoring a right hemisphere dominance (Tanaka et al. 1999; Spierer et al. 2009), others a left lateralization (Pinek et al. 1989), others a bias to the hemisphere contralateral to the stimulus (Zatorre et al. 1995), others whole field neglect following unilateral lesions, that is, no difference between left and right lesion subjects (Zatorre and Penhune 2001). Indeed, in their study that encountered mixed lateralization results when studying patients with unilateral temporal lobe excisions either encroaching on or sparing HG Zatorre and Penhune (2001) summed up the observed variation succinctly: ‘‘the existence of individual differences likely illustrates differential patterns of functional lateralization.’’ Thus, the possibility of a predominant lateralization of spatial processing in auditory cortex is still an open question. Our results suggest that intensity processing is not affected by the spatial properties of a stimulus. This does not rule out the possibility of the involvement of spatially sensitive centers, not represented in the intensity processing characterized by the AESPA responses, which may be lateralized: Frequency-specific interaural time differences, interaural level differences as well as monaural amplitude spectrum cues are thought to integrate in a nonlinear fashion to create a map of space represented by location-specific neurons (Cohen and Knudsen 1999). Thus, since the activity of these location-specific neurons is not linearly related to the intensity of the stimulus then it is unlikely to be accounted for in the AESPA response. This would further back up our assertion that the AESPA response is due to the activity of intensity processing neurons only (Lalor et al. 2009). In light of this, the current results, although enlightening as to the effect of endogenous attention on intensity processing, are not directly comparable with previous studies on the lateralization of spatial processing. Furthermore, there has been much interest in the purported separation of ‘‘what’’ and ‘‘where’’ streams in the auditory system (Rauschecker and Tian 2000; Tian et al. 2001; Ahveninen et al. 2006). Given that amplitude modulation is the driving feature of the stimulus employed here (i.e., a likely what feature), it may be that the what stream is preferentially driven by the current implementation of the AESPA method and thus activity of the location relevant where stream is not represented. Implementation of a paradigm whereby the intensity of a stimulus is kept constant but the location of the stimulus is modulated would shed further light on this possibility. 1230 Endogenous Auditory Spatial Attention d Power et al. unilateral visuospatial neglect. J Neurol Neurosurg Psychiatry. 67:481--486. Tian B, Reser D, Durham A, Kustov A, Rauschecker JP. 2001. Functional specialization in rhesus monkey auditory cortex. Science. 292:290--293. Ulanovsky N, Las L, Nelken I. 2003. Processing of low-probability sounds by cortical neurons. Nat Neurosci. 6:391--398. Vogel EK, Woodman GF, Luck SJ. 2005. Pushing around the locus of selection: evidence for the flexible-selection hypothesis. J Cogn Neurosci. 17:1907--1922. Woldorff M, Gallen CC, Hampson SA, Hillyard SA, Pantev C, Sobel D, Bloom FE. 1993. Modulation of early sensory processing in human auditory cortex during selective attention. Proc Natl Acad Sci U S A. 90:8722--8726. Zatorre RJ, Penhune VB. 2001. Spatial localization after excision of human auditory cortex. J Neurosci. 21:6321--6328. Zatorre RJ, Ptito A, Villemure JG. 1995. Preserved auditory spatial localization following cerebral hemispherectomy. Brain. 118: 879--889. Downloaded from https://academic.oup.com/cercor/article-abstract/21/6/1223/349063 by guest on 30 October 2018 Ritter W, De Sanctis P, Molholm S, Javitt DC, Foxe JJ. 2006. Preattentively grouped tones do not elicit MMN with respect to each other. Psychophysiology. 43:423--430. Romani GL, Williamson SJ, Kaufman L. 1982. Tonotopic organization of the human auditory cortex. Science. 216:1339--1340. Ross B, Hillyard SA, Picton TW. 2010. Temporal dynamics of selective attention during dichotic listening. Cereb Cortex. 20:1360--1371. Ross B, Picton TW, Herdman AT, Hillyard SA, Pantev C. 2004. The effect of attention on the auditory steady-state response. Neurol Clin Neurophysiol. 22:1--4. Spierer L, Bellmann-Thiran A, Maeder P, Murray MM, Clarke S. 2009. Hemispheric competence for auditory spatial representation. Brain. 132:1953--1966. Sussman E, Ritter W, Vaughan HG, Jr. 1999. An investigation of the auditory streaming effect using event-related brain potentials. Psychophysiology. 36:22--34. Tanaka H, Hachisuka K, Ogata H. 1999. Sound lateralisation in patients with left or right cerebral hemispheric lesions: relation with