Pattern of BOLD signal in auditory cortex relates acoustic response to perceptual streaming
© Hill et al; licensee BioMed Central Ltd. 2011
Received: 2 July 2011
Accepted: 17 August 2011
Published: 17 August 2011
Segregating auditory scenes into distinct objects or streams is one of our brain's greatest perceptual challenges. Streaming has classically been studied with bistable sound stimuli, perceived alternately as a single group or two separate groups. Throughout the last decade different methodologies have yielded inconsistent evidence about the role of auditory cortex in the maintenance of streams. In particular, studies using functional magnetic resonance imaging (fMRI) have been unable to show persistent activity within auditory cortex (AC) that distinguishes between perceptual states.
We use bistable stimuli, an explicit perceptual categorization task, and a focused region of interest (ROI) analysis to demonstrate an effect of perceptual state within AC. We find that AC has more activity when listeners perceive the split percept rather than the grouped percept. In addition, within this ROI the pattern of acoustic response across voxels is significantly correlated with the pattern of perceptual modulation. In a whole-brain exploratory test, we corroborate previous work showing an effect of perceptual state in the intraparietal sulcus.
Our results show that the maintenance of auditory streams is reflected in AC activity, directly relating sound responses to perception, and that perceptual state is further represented in multiple, higher level cortical regions.
The natural world presents a rich mixture of auditory events that overlap in frequency and time. One of the brain's greatest perceptual challenges is to segregate this mixture into distinct "streams", so that it can attribute acoustic energy to discrete sources in the environment. This analysis of an auditory scene is essential for much of our daily acoustic experience, notably for communication where it is posed as the 'cocktail party problem' . In addition to its importance for healthy listeners, stream segregation may be impaired in various neurological disorders such as dyslexia , schizophrenia  and Asperger syndrome , and the inability to segment and selectively attend to sounds is a major problem with hearing impairment [5, 6].
Decades of psychoacoustic studies have characterized the basic phenomenology of streaming with sequences of sounds. The classic paradigm uses alternation between two sounds that differ along one stimulus dimension [7–9], such as spatial location [10, 11]. The sounds (usually referred to as A and B) typically alternate along with silent gaps (-) in an ABA- pattern. When these stimuli are close in the relevant dimension they are grouped into a single stream and perceived as triplets with a galloping rhythm. At larger separations the streams segment, and subjects perceive a repeating stream of A sounds (A-A-A-) and a separate, more slowly repeating B stream (B---B---). At intermediate frequency separations the single and two stream percepts are bistable, where listeners switch between perceptual states after an initial buildup [12, 13]. However, despite its perceptual importance, the neural mechanisms of streaming remain unclear.
A central area of contention is the role of early auditory cortex in forming and maintaining streams . Evidence from different methodologies has failed to converge on a single answer. Animal studies have relied mainly on recordings from early auditory cortex that characterize the changing neural representation of tones during the buildup of streaming  or physical changes to the stimulus that correlate with perceptual state [16, 17]. Theories based on this data posit that auditory cortex (AC) plays a key role in both the formation and maintenance of auditory streams through modulation of the receptive fields of auditory neurons . However these conclusions are practically limited since it is difficult to record extracellularly in many regions of cortex simultaneously and since animals cannot signal their perceptual state unambiguously. Meanwhile, human studies using both electroencephalography (EEG) [19, 20] and magnetoencephalography (MEG) [21, 22] have also supported the importance of AC in streaming. These studies found correlates of segregation in electrical and magnetic waveforms believed to be generated in AC and time locked to the individual tones within a sequence. However the stimulus-locked nature of waveform analysis could not characterize non-AC signals which occur on the time scale of percepts rather than individual sounds.
In contrast, an influential fMRI study by Cusack in 2005 challenged the importance of AC by showing a single area in right posterior IPS where activity was greater during the split percept relative to the grouped percept , and failing to find any effect of percept in AC. These findings led Cusack to propose a model of stream segregation that relied on top-down control of auditory information for the maintenance of streams rather than automatic segregation in early sensory cortex. He argued that IPS is a multimodal region sensitive to object number and provides the key neural mechanism for the segmentation of auditory sources. Finally, recent fMRI experiments have found effects related to streaming in AC, either as stimulus properties change in a way that correlates with streaming [24, 25] or during the momentary switches from one percept to another [26, 27]. However, it is unclear how these stimulus driven effects or switch events are related to the persistent neural activity that maintains a single percept over an extended period of time.
Taken as a whole, these findings from multiple methodologies present an inconsistent picture of the neural mechanisms of auditory streaming. Animal researchers have clear theories for the neural mechanisms in AC that could sustain streaming, but have thus far not recorded from cortical regions outside of auditory cortex. EEG and MEG evidence suggests an involvement of AC in the continuous maintenance of auditory streams yet fMRI experiments have failed to find corroborating evidence. Therefore, in this study, we used fMRI to provide some continuity between disparate lines of evidence: specifically, to test if the same networks that strongly represent incoming auditory information are sensitive to perceptual state by showing an overall activity difference between group and split percepts. We observe an effect in early AC which is sustained through the length of the percept.
Fifteen subjects participated in the study (mean age 22.1 years +/- 1.8 years SD). All subjects had no history of neurological disorders or hearing loss. Participants gave written informed consent in accordance with procedures approved by the University of California Institutional Review Board and were paid for their participation. Two of the subjects were removed based on behavioral performance inside the scanner (see results section).
In addition to the basic streaming stimuli, we included increased intensity deviants for both A and B sounds to ensure that subjects were actively attending to the sound sequences, and so we could confirm that detection - as a proxy measure of attention - does not explain any perceptual streaming effects. Deviants occurred 10 times for each sound at pseudo-random times throughout the 100 second block. The deviants were also calibrated for each subject to target a d' of 2 (group mean 2.77 dB, +/- 1.00 SD).
Task and Calibration
Subjects were given a short training on how to distinguish the two percepts, which typically alternated spontaneously and categorically between a single stream containing both the A and B tones (referred to as grouped) and two separate streams of the A tones alone and the B tones alone (referred to as split). Once they were comfortable with the distinction, subjects began the full calibration. First, a rough psychometric function for streaming as a function of ILD separation was calculated using three 100 second long blocks. Subjects were instructed to press and release one of two buttons with their right middle and right index fingers to indicate a switch to a grouped and split percept, respectively. All subjects were assumed to begin each block in the grouped percept . Subjects were given a self-timed break between each 100 sec block. Streaming thresholds, defined as the spatial disparity necessary for the subjects to spend 50% of a block in the split percept, were estimated from each subject's approximate three-point psychometric function through linear interpolation. Then, with spatial disparity held at this threshold, detection thresholds for the intensity deviants were estimated in a similar fashion. Subjects pressed a button with their left index finger each time they detected a deviant. Responses occurring from 200-1200 msec after the onset of the deviant were scored as hits; responses outside of this window were counted as false alarms. To further refine our estimate of each subject's ILD and deviant detection thresholds, subjects then began an adaptive 1-up 1-down staircase procedure , with the initial values set at the previously estimated thresholds. This algorithm targeted 50% of the time spent in the split percept during each 100 sec block, and a d' of 2 for deviant detection across both deviant types. Subjects performed both tasks simultaneously, and after each block the values for spatial separation and deviant disparity were adapted independently. The spatial separation had a step size of 1 dB and the detection task had a step size of 0.5 dB. Subjects proceeded until the direction of change reversed at least 8 times for each of the two metrics, a procedure that usually lasted 10-14 blocks. Then, the values for each reversal were averaged to find a threshold for streaming and deviant detection. This threshold was finally confirmed with a 6 block run, and adjusted by hand if the average streaming differed from 50% by more than 10%, and if the d' was less than 1.5 or more than 3. This calibration procedure occurred within two weeks before a participant's scanning session, and lasted approximately 1.5 hours.
Scanning was separated into six, 8.47 minute sessions. Sessions began and ended with a 30 second fixation period that served as a baseline. Subjects performed the previously described streaming and deviant detection tasks during four, 100 sec blocks each separated by a 16 sec fixation period. Each session consisted of 2 LRL and 2 RLR stimulus blocks in pseudo-random order.
MRI data were collected in a Siemens MAGNETOM Tim Trio System 3 Tesla scanner with a 32-channel RF headcoil and a whole body gradient system. Foam padding was used to minimize head motion. Each session began with a series of images to determine regional anatomy, including a sagittal localizer (Repetition Time (TR) = 250 ms, Echo Time (TE) = 3.2 ms). Single-shot gradient-echo echoplanar images (EPI) were acquired for thirty-six near-axial slices. The functional scans had the parameters: TR of 2 s, TE 25 ms, 64 × 64 acquisition matrix, 3.4 mm slice thickness, a 220 mm field of view, 3.45 mm in plane resolution, bandwidth of 2604 Hz/Px and a flip angle of 90°. A high resolution three-dimensional MPRAGE image for use in intersubject coregistration was taken at the end of the session with a voxel size of 0.45 × 0.45 × 0.95 mm. Auditory stimuli were presented with a piezoelectric audio system customized for use in high magnetic fields (Sensimetrics Corp, model S14 http://www.sens.com/s14/). The earbuds of the audio system passively attenuated the scanner noise to 60 dB (attenuation level based on the manufacturer's specifications), and stimuli were played at 80 dB. All sounds were filtered to account for known frequency response of the earbuds, ensuring that stimuli were perceived as intended.
Behavioral data was analyzed using custom in-house scripts written in Matlab 7.4 (Mathworks, http://www.mathworks.com/). fMRI data was analyzed using a combination of in-house scripts and the modified general linear model (GLM) in SPM 8 http://www.fil.ion.ucl.ac.uk/spm/software/spm8/. EPIs were slice time corrected, realigned to the first scan, coregistered to the subject's MPRAGE, normalized to the Montreal Neurological Institute (MNI) template , and smoothed with an 8 mm Gaussian smoothing kernel unless otherwise noted. The following covariates were added to the design matrix: a block regressor for the 100 second sound sequences (one for each type of stimulus, LRL and RLR), a perceptual regressor which had the value of 1 when subjects grouped sounds and -1 when subjects split sounds, and separate impulse regressors for deviant onsets, hits and false alarms. The standard approach in neuroimaging studies would be to model the key conditions (grouped and split) separately, and then contrast their parameter estimates or betas. For bistable perceptions this standard method poses significant problems. The categorical nature of the perceptual phenomenon requires that when one of the two possible states ends, the other begins. Thus after standard high-pass filtering, perceptual states are strongly collinear (with a correlation coefficient more extreme than -0.90 in our tests). Models with highly collinear regressors are mathematically unstable and can lead to unreliable results . The inclusion of a bimodal covariate to model perception surmounts this limitation by combining the strongly-anticorrelated regressors into a single covariate. This bimodal regressor is functionally equivalent to a grouped > split contrast between the parameter estimates of independent regressors. Therefore, positive regression coefficients associated with the perceptual regressor signals regions which have higher activity levels during grouped percept relative to split percept, while negative parameter estimates indicate the inverse. All these regressors were convolved with the standard SPM8 hemodynamic response function (HRF). Motion parameters and session covariates were also included as nuisance regressors. All group level statistical tests were t-tests on beta parameter estimates against the null hypothesis that they equal zero.
In order to maintain statistical independence between regressors used to code sound onset and perceptual state, only subjects who streamed between 35% and 65% during the scanning session were used. Subjects with larger or smaller streaming percentage would by definition have large portions of sound blocks spent in a single percept, which would cause the sound onset and percept to have similar time courses, leading to collinear regressors. Thirteen of the 15 subjects met this criterion, and represent the group referred to in all subsequent analyses. Within this group, the mean proportion of streaming was 49.5% (+/- 8.0% SD). There was no significant difference in streaming percentage between the two stimulus types, LRL and RLR and all subsequent analyses are collapsed across both stimulus types. The group mean of each subject's median inter-switch interval was 9.3 seconds (+/- 4.3 seconds SD), putting it in a range which is amenable to detection by a GLM after filtering with a HRF.
The mean d' for deviant detection was 1.84 (+/- 0.9 SD), indicating that subjects actively attended to the stimuli. To further analyze the effect of perceptual state on deviant detection, we performed an analysis of variance (ANOVA) on the target hit rate in a repeated measures 2 × 2 design, with the factors of deviant side and perceptual state. Hit rate was used for the ANOVA because false alarms cannot be attributed to a particular side. This ANOVA revealed that there was a main effect of target location on hit rate (F(1,12) = 8.23, p = 0.014), with left targets being detected with more frequency than right targets. There was no effect of perceptual state (F(1,12) = 0.37, p = 0.55) or interaction between location and perceptual state (F(1,12) = 1.31, p = 0.27).
Clusters of voxels significantly modulated by percept
Cluster Size (Voxel)
Peak MNI Coordinates
26 -32 67
-1 -59 46
Our results show that early AC reflects the sustained perceptual phenomenon of streaming, and that the spatial patterns of AC subregions most sensitive to sound also show the greatest perceptual effect. This broadly supports the theory that sound segregation modulates the same neural circuits that process basic sound object features . The direction of our perceptual effect, that AC activity is greater for separate streams than a grouped stream, and the existence of buildup in streaming supports the notion that the grouped percept is the default perception and that additional metabolic effort occurs when perceiving two or more streams. Animal models of streaming propose that this effort manifests as narrowing receptive fields of neurons sensitive to the stimulus dimension along which the putative streams are separated . While there are examples where the local circuits of a cortical region can promote sharpening of receptive fields through short-range reciprocal inhibition from audition , vision , and olfaction , an alternative explanation is that other cortical regions could direct or otherwise interact with local streaming related sharpening. Theories of auditory streaming have so far not made specific claims about whether this modulation arises from local network processes or interaction between AC and higher level areas.
Our data suggests that AC is only one of several regions involved in streaming. Higher cortices may play a cooperative role by interpreting a segregated scene or modulating the streaming itself. For instance, consistent with Cusack's previous work , rIPS may track the number of distinct objects after they have been segregated by auditory cortex or it may allow broad behavioral goals to influence streaming mechanisms. Such high level control of auditory streaming mechanisms is evident behaviorally as listeners can consciously influence the number of perceived objects . Further studies that are optimized to detect functional connectivity between multiple cortical regions and explicitly modulate top down signals such as task demands or expectation may shed light on the interaction between higher level areas and AC.
Technical challenges may have played a key role in the prior lack of fMRI based evidence for AC's involvement in the continuous maintenance of auditory streaming. Cusack's study used sparse scanning techniques, and a number of stimuli that spent a large amount of time in a single percept . Both of these factors would drastically reduce statistical power, possibly giving a false negative for the effects of perceptual state. In addition, general linear models (the usual method employed for analyzing fMRI data sets) can have difficulty dissociating the activity for bi-stable percepts. If regressors are included for each perceptual state independently, the alternating nature of bi-stable percepts can cause the regressors to become collinear after standard filtering needed to remove known low frequency noise in fMRI data. We instead used a single bimodal perceptual regressor that, although precluding analyses on each percept alone, offers a powerful measure of the activity distinguishing between percepts.
It is also worth considering why EEG and MEG studies have not found effects outside of auditory cortex. Clearly, models of streaming would benefit from data on non-AC (e.g. rIPS) neural signals with the excellent temporal resolution of EEG and MEG. One possibility is that higher level processes lack consistent phase-locking with the stimuli. So while previous studies have focused on stimulus-locked event related potentials/fields, induced activity (loosely time-locked but not precisely phase-locked) may have remained undetected. In addition, the MEG studies cited used source-filtered waveforms that ignored currents generated outside of AC [21, 22]. Designs and analyses that detect non-stimulus locked activity and integrate data from multiple current sources may improve the correspondence between findings in these different modalities.
Surprisingly in our attentional control task, subjects were no worse (or better) at detecting targets while streaming separate objects. This has direct relevance for theories of how attention operates within, or spreads across, object representations [35–37]. Our results suggest that there is no cost associated with small numbers of objects. In general, auditory streaming appears to be a relatively untapped paradigm for the study of object-based attention considering the advantages of having a single controlled stimulus which fluctuates between two different object schema every few seconds. Some investigations have suggested that certain task sets, such as deviant detection in a single stream, may promote perceptual segregation . Even though the recorded d' values between our two perceptual states is equivalent, it may be that performing a secondary task designed to spread attention across multiple streams may have impacted participants' overall streaming percentage. However, this contextual effect should not influence the interpretation of our results because our calibration procedure ensures that we find a streaming threshold in the presence of both simultaneous tasks and such task demands are equally present during both split and grouped percepts. Thus any comparison between the two percepts controls for task related factors.
Using carefully calibrated stimuli and a focused approach, we are the first group to show sustained activity in AC that distinguishes between perceptual streaming states. Auditory cortex showed higher levels of activity for split percept compared to the grouped percept. These results strengthen the continuity between multiple lines of evidence supporting a role for AC in the formation and maintenance of auditory streams. At the same time, our work is consistent with previous fMRI experiments to suggest that AC does not function alone in this task. Future studies will address the interaction between multiple cortical regions and improve our understanding of this important perceptual phenomenon.
Acknowledgements and Funding
This work was supported by the National Institutes of Health: National Institute on Deafness and other Communication Disorders [R01-DC8171 to L.M., T32-DC8072-01A2 to K.H., F31-DC011429 to C.B.].
- Cherry EC: Some Experiments on the Recognition of Speech, with One and with Two Ears. J Acoust Soc Am. 1953, 25: 975-979. 10.1121/1.1907229.View Article
- Petkov CI, O'Connor KN, Benmoshe G, Baynes K, Sutter ML: Auditory perceptual grouping and attention in dyslexia. Brain Res Cogn Brain Res. 2005, 24: 343-354. 10.1016/j.cogbrainres.2005.02.021.View ArticlePubMed
- Nielzen S, Olsson O: Perceptual grouping due to pitch and amplitude in hallucinating schizophrenics. Psychopathology. 1997, 30: 140-148. 10.1159/000285040.View ArticlePubMed
- Lepisto T, Kuitunen A, Sussman E, Saalasti S, Jansson-Verkasalo E, Nieminen-von Wendt T, Kujala T: Auditory stream segregation in children with Asperger syndrome. Biological psychology. 2009, 82: 301-307. 10.1016/j.biopsycho.2009.09.004.PubMed CentralView ArticlePubMed
- Festen JM, Plomp R: Effects of fluctuating noise and interfering speech on the speech-reception threshold for impaired and normal hearing. J Acoust Soc Am. 1990, 88: 1725-1736. 10.1121/1.400247.View ArticlePubMed
- Middelweerd MJ, Festen JM, Plomp R: Difficulties with speech intelligibility in noise in spite of a normal pure-tone audiogram. Audiology. 1990, 29: 1-7. 10.3109/00206099009081640.View ArticlePubMed
- Bregman AS: Auditory Scene Analysis: The Perceptual Organization of Sound. 1994, MIT Press
- van Noorden LP: Minimun differences of level and frequency for perceptual fission of tone sequences ABAB. J Acoust Soc Am. 1977, 61: 1041-1045. 10.1121/1.381388.View ArticlePubMed
- Carlyon RP: How the brain separates sounds. Trends in cognitive sciences. 2004, 8: 465-471. 10.1016/j.tics.2004.08.008.View ArticlePubMed
- Boehnke SE, Phillips DP: The relation between auditory temporal interval processing and sequential stream segregation examined with stimulus laterality differences. Perception & psychophysics. 2005, 67: 1088-1101. 10.3758/BF03193634.View Article
- Hartmann WM, Johnson D: Stream Segregation and Peripheral Channeling. Music Perception: An Interdisciplinary Journal. 1991, 9: 155-183.View Article
- Pressnitzer D, Hupe JM: Temporal dynamics of auditory and visual bistability reveal common principles of perceptual organization. Curr Biol. 2006, 16: 1351-1357. 10.1016/j.cub.2006.05.054.View ArticlePubMed
- Denham SL, Winkler I: The role of predictive models in the formation of auditory streams. Journal of physiology, Paris. 2006, 100: 154-170. 10.1016/j.jphysparis.2006.09.012.View ArticlePubMed
- Shamma SA, Micheyl C: Behind the scenes of auditory perception. Current opinion in neurobiology. 2010, 20: 361-366. 10.1016/j.conb.2010.03.009.PubMed CentralView ArticlePubMed
- Micheyl C, Tian B, Carlyon RP, Rauschecker JP: Perceptual organization of tone sequences in the auditory cortex of awake macaques. Neuron. 2005, 48: 139-148. 10.1016/j.neuron.2005.08.039.View ArticlePubMed
- Fishman YI, Arezzo JC, Steinschneider M: Auditory stream segregation in monkey auditory cortex: effects of frequency separation, presentation rate, and tone duration. J Acoust Soc Am. 2004, 116: 1656-1670. 10.1121/1.1778903.View ArticlePubMed
- Fishman YI, Reser DH, Arezzo JC, Steinschneider M: Neural correlates of auditory stream segregation in primary auditory cortex of the awake monkey. Hear Res. 2001, 151: 167-187. 10.1016/S0378-5955(00)00224-0.View ArticlePubMed
- Micheyl C, Carlyon RP, Gutschalk A, Melcher JR, Oxenham AJ, Rauschecker JP, Tian B, Courtenay Wilson E: The role of auditory cortex in the formation of auditory streams. Hear Res. 2007, 229: 116-131. 10.1016/j.heares.2007.01.007.PubMed CentralView ArticlePubMed
- Snyder JS, Alain C, Picton TW: Effects of attention on neuroelectric correlates of auditory stream segregation. Journal of cognitive neuroscience. 2006, 18: 1-13. 10.1162/089892906775250021.View ArticlePubMed
- Snyder JS, Holder WT, Weintraub DM, Carter OL, Alain C: Effects of prior stimulus and prior perception on neural correlates of auditory stream segregation. Psychophysiology. 2009, 46: 1208-1215. 10.1111/j.1469-8986.2009.00870.x.View ArticlePubMed
- Gutschalk A, Micheyl C, Melcher JR, Rupp A, Scherg M, Oxenham AJ: Neuromagnetic correlates of streaming in human auditory cortex. J Neurosci. 2005, 25: 5382-5388. 10.1523/JNEUROSCI.0347-05.2005.PubMed CentralView ArticlePubMed
- Gutschalk A, Oxenham AJ, Micheyl C, Wilson EC, Melcher JR: Human cortical activity during streaming without spectral cues suggests a general neural substrate for auditory stream segregation. J Neurosci. 2007, 27: 13074-13081. 10.1523/JNEUROSCI.2299-07.2007.View ArticlePubMed
- Cusack R: The intraparietal sulcus and perceptual organization. Journal of cognitive neuroscience. 2005, 17: 641-651. 10.1162/0898929053467541.View ArticlePubMed
- Wilson EC, Melcher JR, Micheyl C, Gutschalk A, Oxenham AJ: Cortical FMRI activation to sequences of tones alternating in frequency: relationship to perceived rate and streaming. Journal of neurophysiology. 2007, 97: 2230-2238. 10.1152/jn.00788.2006.PubMed CentralView ArticlePubMed
- Deike S, Gaschler-Markefski B, Brechmann A, Scheich H: Auditory stream segregation relying on timbre involves left auditory cortex. Neuroreport. 2004, 15: 1511-1514. 10.1097/01.wnr.0000132919.12990.34.View ArticlePubMed
- Kondo HM, Kashino M: Involvement of the thalamocortical loop in the spontaneous switching of percepts in auditory streaming. J Neurosci. 2009, 29: 12695-12701. 10.1523/JNEUROSCI.1549-09.2009.View ArticlePubMed
- Schadwinkel S, Gutschalk A: Transient BOLD activity locked to perceptual reversals of auditory streaming in human auditory cortex and inferior colliculus. Journal of neurophysiology. 2011, 105: 1977-1983. 10.1152/jn.00461.2010.View ArticlePubMed
- Leek M: Adaptive procedures in psychophysical research. Attention, Perception & Psychophysics. 2001, 63: 1279-1292.View Article
- Evans AC, Collins DL, Mills SR, Brown ED, Kelly RL, Peters TM: 3D statistical neuroanatomical models from 305 MRI volumes. Proc IEEE-Nuclear Science Symposium and Medical Imaging Conference. 1993, 1813-1817.
- Hocking RR: A Biometrics Invited Paper. The Analysis and Selection of Variables in Linear Regression. Biometrics. 1976, 32: 1-49. 10.2307/2529336.View Article
- Guimaraes AR, Melcher JR, Talavage TM, Baker JR, Ledden P, Rosen BR, Kiang NY, Fullerton BC, Weisskoff RM: Imaging subcortical auditory activity in humans. Hum Brain Mapp. 1998, 6: 33-41. 10.1002/(SICI)1097-0193(1998)6:1<33::AID-HBM3>3.0.CO;2-M.PubMed CentralView ArticlePubMed
- Dasika VK, White JA, Carney LH, Colburn HS: Effects of inhibitory feedback in a network model of avian brain stem. Journal of neurophysiology. 2005, 94: 400-414. 10.1152/jn.01065.2004.View ArticlePubMed
- Katz LC, Callaway EM: Development of local circuits in mammalian visual cortex. Annu Rev Neurosci. 1992, 15: 31-56. 10.1146/annurev.ne.15.030192.000335.View ArticlePubMed
- Urban NN: Lateral inhibition in the olfactory bulb and in olfaction. Physiol Behav. 2002, 77: 607-612. 10.1016/S0031-9384(02)00895-8.View ArticlePubMed
- Ho MC, Yeh SL: Effects of instantaneous object input and past experience on object-based attention. Acta Psychol (Amst). 2009, 132: 31-39. 10.1016/j.actpsy.2009.02.004.View Article
- Duncan J: Selective attention and the organization of visual information. J Exp Psychol Gen. 1984, 113: 501-517.View ArticlePubMed
- Shomstein S, Yantis S: Object-based attention: sensory modulation or priority setting?. Perception & psychophysics. 2002, 64: 41-51. 10.3758/BF03194556.View Article
- Deike S, Scheich H, Brechmann A: Active stream segregation specifically involves the left human auditory cortex. Hear Res. 2010, 265: 30-37. 10.1016/j.heares.2010.03.005.View ArticlePubMed
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.