Selective visual-spatial attention alters induced gamma band responses in the human EEG
Introduction
The visual system is regarded as a hierarchically organized processing device, i.e. information is analyzed in multiple stages (Felleman and van Essen, 1991). The retinal image is gradually transformed from localized, spatially restricted representations in the striate cortex (area V1) to more global areas such as the middle temporal area (MT), specialized for the analysis of motion or area V4 which has been reported to contain color-selective neurons (Livingstone and Hubel, 1988, Zeki, 1993, Tootell et al., 1996). It was suggested that distinct regions of information processing remain segregated throughout higher areas in the brain and can be divided into two streams of information processing (Ungerleider, 1995, Ungerleider and Haxby, 1994). The occipito-temporal or ventral pathway is thought to be specialized for the analysis of object features (color, shape etc. – the ‘what’ system), and the occipito-parietal or dorsal pathway seems to be specialized for the analysis of motion and the spatial relationship between objects (the ‘where’ system; Ungerleider and Haxby, 1994).
Due to the finite computational resources of the brain, only a limited number of stimuli will be processed at a given time (Desimone and Duncan, 1995). Therefore it was suggested that a selective attention mechanism focuses these resources to specific objects or locations in the visual field without moving the eyes to that location (Hillyard and Anllo-Vento, 1998, Hillyard et al., 1995, LaBerge, 1995). Several metaphors have been used to explain this mechanism. It has been likened to a ‘spotlight’ (Posner and Petersen, 1990, Posner et al., 1980) or a ‘zoom lens’ (Eriksen and St. James, 1986).
In her ‘feature integration theory’, Treisman (Treisman, 1993, Treisman and Gormican, 1988, Treisman and Gelade, 1980) proposed that the visual system decomposes the visual input into maps of simple features such as color, shape and orientation, whereby only features within the spotlight of attention are combined to a coherent representation of the object (Treisman and Gelade, 1980). Psychophysiological studies measuring visual event-related potentials (VEPs) have shown that focusing the spotlight on a certain location in space leads to an enhancement of early sensory evoked responses as compared to when stimuli were presented outside the beam of the spotlight (e.g. Hillyard and Anllo-Vento, 1998, Hillyard et al., 1998, Luck and Ford, 1998, Anllo-Vento and Hillyard, 1996, Mangun, 1995, Gomez-Gonzales et al., 1994, Mangun et al., 1993). Moreover, human positron emission tomography (PET) studies have suggested that attending to a particular stimulus enhances cerebral blood flow in those cortical areas which are specialized for processing the relevant features of that stimulus (Corbetta et al., 1990, Corbetta et al., 1993, Heinze et al., 1994). Using functional magnetic resonance imaging (fMRI), it has been shown that metabolic activity in the parieto-occipital stream was enhanced when subjects attended to moving stimuli as compared to when the identical stimulus was ignored (Haug et al., 1998, Beauchamp et al., 1997). Thus, attention seems to have the effect of boosting or facilitating the activity of neurons in cortical areas or pathways which are related to processing the attended stimulus (Hillyard et al., 1998, Duncan et al., 1997, Luck et al., 1997, LaBerge, 1995, Posner and Dehaene, 1994, Posner et al., 1980).
However, all these findings have difficulties in explaining how a visual object is represented in the brain. One suggestion was – based on theoretical considerations and findings in intracortical recordings in animals – that the object is coded in a Hebbian-like neuronal cell assembly which is distributed across different functional visual areas by means of synchronized bursts of action potentials in a frequency range above 20 Hz, i.e. the gamma band (Eckhorn et al., 1990, Eckhorn et al., 1992, Singer and Gray, 1995, Gray et al., 1990, Singer et al., 1990, Malsburg and Schneider, 1986, Milner, 1974). Contrary to the above mentioned VEPs, these synchronized oscillations are neither phase- nor time-locked to stimulus onset.
In human EEG, gamma band responses were induced by coherently moving lines giving the impression of a waterfall (Lutzenberger et al., 1995), by the short presentation of illusory Kanizsa and real triangles (Tallon-Baudry et al., 1997b, Tallon et al., 1995), by a visual search task (Tallon-Baudry et al., 1997a), by a coherently moving long bar (Müller et al., 1996, Müller et al., 1997, Müller et al., 1997) and by the perception of a Gestalt (Keil et al., 1999). In addition, induced gamma band responses were linked to cognitive processes, such as the processing of words (Pulvermüller et al., 1997, Pulvermüller, 1996, Pulvermüller et al., 1996), language and non-language stimuli (Eulitz et al., 1996) and, recently, to short-term memory processes during a delayed matching-to-sample task (Tallon-Baudry et al., 1998). All theses studies seem to support the idea of synchronized activity in wide spread cell assemblies linked to perceptual integration processes. However, in most of the studies, differences in induced gamma band activity were linked to the features of a stimulus, thus, stimuli which induced these activities were physically different from those which were linked to less power in the gamma band.
An alternative explanation for the findings in such experimental setups would be that the enhanced power in the gamma band is simply a by-product of stimulus processing (Kirschfeld, 1992). One method refuting this explanation is to employ an experimental setup which uses identical stimulus configurations. In one experimental condition, Tallon-Baudry et al. (1997a) presented their subjects an apparently meaningless picture that contained a hidden Dalmatian. The authors showed that gamma band power increased significantly when subjects perceived the dog as compared to when they viewed the picture without perceiving the dog. Here, it seems difficult to argue that the increase in gamma power is due to different features of the stimulus; in both cases the stimulus was identical with the exception that the dog was either perceived or unperceived. The limitation of this design is that once subjects perceive the dog, they will always perceive it. A method overcoming this limitations entails manipulating spatial selective attention. As mentioned above, early components of the VEP show a marked enhancement of amplitude when they were evoked by an attended stimulus as compared to when the identical stimulus was unattended (Hillyard and Anllo-Vento, 1998, Hillyard et al., 1998, Luck and Ford, 1998, Anllo-Vento and Hillyard, 1996, Mangun, 1995, Gomez-Gonzales et al., 1994, Mangun et al., 1993). This effect is linked to an amplification of neural responses in those visual pathways related to the processing of the stimulus (Hillyard et al., 1998). Consequently, one can hypothesize that if induced gamma band activity is a neural index of visual information processing, gamma power should be altered upon selective visual spatial attention by an amplification of the synchronized neural activity in the wide spread Hebbian cell assembly representing or coding the attended stimulus.
Previous studies on animals (Rougeul-Buser and Buser, 1994) and humans (Sheer, 1989) have suggested such an increase of gamma power is due to attentional processes. However, the experimental setup and the results of theses studies do not preclude that a shift of general arousal has caused the effects. Recently, we have demonstrated that power in a specific gamma band increased significantly when subjects attended to a long, coherently moving bar as compared to when that bar was ignored (Müller, 1998). Aside from behavioral data, we used objective physiological measures to control for subject's compliance and level of arousal. In particular, we were able to demonstrate that no differences between the experimental conditions were found with respect to alpha power, indicating no difference in the general level of arousal (Ota et al., 1996). We concluded that the increase in gamma power resulted from the amplification of synchronized neural spike activity in those cortical areas related to the processing of the moving bar (Müller, 1998). However, since we only recorded from a few occipital electrode sites, no information was obtained on the topography of that attention effect. In addition, aside from the standing bar serving as a baseline measure, no further standing control condition was introduced in order to compare the topographies for further strengthening of our conclusion of enhanced activity in the dorsal visual pathway when the bar was attended.
The present study was designed to overcome the limitations of our previous study by using a 128 channel EEG-montage and having not only a moving but also a standing control stimulus. Besides an enhancement in gamma power when the moving stimulus was attended, we expected an augmentation of gamma power on electrodes supposedly situated over cortical areas related to the dorsal stream in the moving as compared to the standing stimulus. In addition, we investigated whether a shift can be observed in the topographical distribution of the gamma band response to the hemisphere contralateral to the to-be-attended side after the onset of a attention direction cue. Thus, the present study goes beyond a simple replication of the effects found in our previous study.
Section snippets
Subjects
Fourteen healthy, right-handed university students (6 males, 8 females), 22–30 years of age (mean=25.8 years, SD 2.7 years), participated in the experiment. All had normal or corrected-to-normal visual acuity. They received class credit or a financial bonus for participation. Informed consent was obtained from each participant.
Stimuli and electrophysiological recordings
As in our previous study (Müller, 1998) we used a moving-bar-like paradigm to induce gamma band activity. However, the paradigm was modified in that a standing stimulus
Behavioral data
On average 708 targets were presented. Participants missed between 144 and 281 of the brown squares or rectangles (mean=232.5, SD=40.5), i.e. participants detected approximately 67% of target stimuli in each condition.
Attention effect
Fig. 3 depicts the normalized mean spectral power relative to baseline for all 5 frequency bands across subjects and the left and right parieto-occipital regional means.
Overall, when the rotating screen was attended, only the frequency band between 35–51 Hz exhibited a significant
Discussion
The purpose of the present study assisted in addressing the following questions: (1) can we replicate our previous finding of an enhanced spectral gamma band power when a moving stimulus was attended as compared to when the identical stimulus was ignored (Müller, 1998)? (2) Is the shift of visual spatial attention to either the left or right visual hemifield accompanied by a shift of gamma power towards the contralateral cortical hemisphere? (3) Is the attentional enhancement of spectral power
Acknowledgements
We are grateful to Ursula Lommen and Jürgen Wolf for help in data acquisition and to Lisa Green for editorial support. Research was supported by grants from the Deutsche Forschungsgemeinschaft and the Human Frontier Science Program.
References (85)
- et al.
Emulation of somatosensory evoked potential (SEP) components with the 3-shell head model and the problem of ‘ghost potential fields’ when using an average reference in brain mapping
Electroenceph clin Neurophysiol
(1990) - et al.
Competitive brain activity in visual attention
Curr Opin Neurobiol
(1997) - et al.
Oscillatory neuromagnetic activity induced by language and non-language stimuli
Cog Brain Res
(1996) - et al.
Improving spatial and temporal resolution in evoked EEG responses using surface Laplacians
Electroenceph clin Neurophysiol
(1993) - et al.
Visual stimulation alters local 40-Hz responses in humans: an EEG-study
Neurosci Lett
(1995) Auditory event-related dynamics of the EEG spectrum and effects of exposure to tones
Electroenceph clin Neurophysiol
(1993)- et al.
Scalp distributions of event-related potentials:an ambiguity associated with analysis of variance models
Electroenceph clin Neurophysiol
(1985) - et al.
Measurements by biphasic changes of the alpha band amplitude as indicators of arousal level
Int J Psychophysiol
(1996) - et al.
Cerebral asymmetry in visual attention
Brain and Cognition
(1990) - et al.
Spherical splines for scalp potential and current source density mapping
Electroenceph clin Neurophysiol
(1989)