Brain oscillations in perception and memory - CiteSeerX

International Journal of Psychophysiology 35 Ž2000. 95]124

Brain oscillations in perception and memory a E. Bas¸ar a,b,U , C. Bas¸ar-Eroglu ˘ c , S. Karakas¸b,d, M. Schurmann ¨ a

Institute of Physiology, Medical Uni¨ ersity Lubeck, 23538, Lubeck, Germany ¨ ¨ b ¨ TUBITAK Brain Dynamics Research Unit, Ankara, Turkey c Institute of Psychology and Cognition Research, 28334, Bremen, Germany d Institute of Experimental Psychology, Hacettepe Uni¨ ersity, Beytepe, Ankara, Turkey Received 23 March 1999; accepted 23 March 1999

Abstract Gamma oscillations, now widely regarded as functionally relevant signals of the brain, illustrate that the concept of event-related oscillations bridges the gap between single neurons and neural assemblies. Taking this concept further, we review experiments showing that oscillatory phenomena such as alpha, theta, or delta responses to events are strongly interwoven with sensory and cognitive functions. This review argues that selecti¨ ely distributed delta, theta, alpha, and gamma oscillatory systems act as resonant communication networks through large populations of neurons. Thus, oscillatory processes might play a major role in relation with memory and integrati¨ e functions. A new ‘neurons-brain’ doctrine is also proposed to extend the neuron doctrine of Sherrington. Q 2000 Elsevier Science B.V. All rights reserved. Keywords: Memory; Brain oscillations Ždelta, theta, alpha, gamma.; Evoked potentials; Event-related potentials; Sensory processing; Cognitive processing; Distributed networks

1. Introduction 1.1. Aim of the report ‘During the ‘Decade of the Brain’ brain science is coming to terms with its ultimate problem: understanding the mechanisms by which the im-

mense number of neurons in the human brain interact to produce the higher cognitive functions’ ŽFreeman, 1998.. As one of the candidate mechanisms, oscillatory neuroelectric acti¨ ity has recently attracted much interest. In particular, this holds for synchronous gamma activity in spatially distributed cells. In this framework, the present review has several aims, namely:

U

Corresponding author. Tel.: q49-451-500-4170; fax: q49451-500-4171. E-mail address: [email protected] ŽE. Bas¸ar.

1. To survey functionally-related findings in oscillatory brain acti¨ ity in the frequency range

0167-8760r00r$ - see front matter Q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 8 7 6 0 Ž 9 9 . 0 0 0 4 7 - 1

96

2.

3.

4.

5.

E. Bas ¸ar et al. r International Journal of Psychophysiology 35 (2000) 95]124

between 0.5 and 100 Hz, i.e. by surpassing approaches centered on the gamma band. A particular aim is to demonstrate that the alpha band } so far mostly neglected } deserves more interest. To emphasize that oscillatory networks are selectively distributed and that oscillatory activities are related to sensory as well as cognitive processes. This integrative view might help to reconsider several contro¨ ersies. To propose a new approach to a fundamental problem } searching for general communication properties in the brain. In particular, it is suggested that complex stimuli elicit superimposed alpha, gamma, theta responses Žto be combined like letters in an alphabet, Bas¸ar, 1998.. To discuss a possible role of selecti¨ ely distributed oscillatory systems in working memory processes. Support for this view will be derived from recent results about distributed memory networks. It will be argued that sensory perception Ževen of simple stimuli. is closely interwoven with cognition and memory. This makes us expect that the theoretical issues covered here might be addressed in future ERP-based experiments. To propose Žaccording to what Mountcastle Ž1992. called a ‘paradigm change in neuroscience’. a ‘neurons-brain doctrine’ extending the classical view of Sherrington.

1.2. E¨ ent-related oscillations The functional significance of oscillatory neural activity begins to emerge from the analysis of responses to well-defined events Ž e¨ ent-related oscillations, phase- or time-locked to a sensory or cogniti¨ e e¨ ent .. Among other approaches, it is possible to investigate such oscillations by frequency domain analysis of event-related potential ŽERP., basing on the following hypothesis ŽBas¸ar, 1980, 1992.: 1. The EEG consists of the activity of an ensemble of generators producing rhythmic activity in several frequency ranges. These oscillators

are active usually in a random way. However, by application of sensory stimulation these generators are coupled and act together in a coherent way. This synchronization and enhancement of EEG activity gives rise to ‘evoked’ or ‘induced rhythms’. 2. Evoked potentials representing ensembles of neural population responses were considered as a result of transition from a disordered to an ordered state. 3. The compound ERP manifests a superposition of evoked oscillations in the EEG frequencies ranging from delta to gamma Ž‘natural frequencies of the brain’ such as alpha: 8]13 Hz, theta: 3.5]7 Hz, delta: 0.5]3.5 Hz and gamma: 30]70 Hz.. Time-locked responses of specific frequency after stimulation can be identified by computing the amplitude frequency characteristics ŽAFCs. of the averaged ERPs ŽBas¸ar, 1980; Roschke et al., 1995.. ¨ The AFC describes the brain system’s transfer properties, e.g. excitability and susceptibility, by revealing resonant as well as salient frequencies. It therefore does not simply represent the spectral power density characterizing the transient signal in the frequency domain but the predicted behavior of the system Žbrain. if sinusoidally modulated input signals of defined frequencies were applied as stimulation. As reflecting the amplification in a given frequency channel, the AFC is expressed in relative units. Hence, the presence of a peak in the AFC reveals the resonant frequencies interpreted as the most preferred oscillations of the system during responding to stimulus. To calculate the AFCs, auditory ERPs were first averaged and then transformed to the frequency domain by means of one sided Fourier Transform ŽLaplace transform, see Solodovnikov, 1960; Bas¸ar, 1980. as shown in Fig. 1. The AFCs serve also to define filter limits for response-adaptive digital filtering of the averaged ERPs. The filtered curves obtained in this way show the time course of oscillatory activity in a certain frequency range ŽBas¸ar, 1980.. More recently, a new technique called ‘wavelet analysis’ has been applied to ERP analysis. Wavelet analysis confirms the results obtained by


97

Fig. 1. Combined time and frequency domain analysis of EEG-EP epochs. FFT, Fast Fourier Transform; AFC, amplitude frequency characteristics Žfrom Schurmann and Bas¸ar, 1994.. ¨

using the AFCs and digital filtering. In addition, wavelet analysis can be used for signal retrie¨ al and selection among a large number of sweeps

recorded in a given physiological or psychological experiment ŽDemiralp et al., 1999.. As will become clear below, the combination of

98


these methods yields results leading to the conclusion that alpha-, theta-, delta-, and gammaresponses are functionally relevant brain responses-related to psychophysiological functions, in short, ‘real signals’ ŽBas¸ar, 1998, 1999.. We intend to show that these oscillations have multifold functions and may act as uni¨ ersal operators or codes of brain activity. Besides frequency and site of oscillations, several other parameters are dependent on specific functions, namely enhancement, time locking, phase locking, delay and duration of oscillations Žfor methods to assess these parameters, see e.g. Kolev and Yordanova, 1997.. 1.3. Selecti¨ ely distributed oscillatory systems in the brain } a general concept The idea of ‘distributed system’ plays an important role in the statements or theories of all scientists working on general aspects of the integrative brain activity. In Mountcastle’s words ‘ prominent among them is the concept that the brain is a complex of widely and reciprocally interconnected systems and that the dynamic interplay of neural acti¨ ity within and between these systems is the ¨ ery essence of brain function’. Again coming back to those statements the large entities of the brain are themselves composed of replicated modules. The linked sets of modules of the various brain entities are defined as a ‘distributed system’. Freeman Ž1975. has named the theory of using dynamics of neural masses as ‘the new Sherrington doctrine’, in which neural populations play the significant functional role. John et al. Ž1988. described a ‘hyperneuron’ consisting, again, of neural populations as a functional important entity in the brain Žfor details see below.. In order to facilitate the comprehension of signal transfer of the brain and in order to avoid several controversies related to localisation and functional correlates of brain oscillatory responses the concept of selecti¨ ely distributed oscillatory systems in the brain has been introduced: by means of the application of combined analysis procedure of EEG and EPs we recently emphasized the functional importance of oscillatory responses Žin

the framework of brain dynamics. related to association and ‘long distance’ communication in the brain. According to a great amount of results alpha networks, theta networks and gamma networks Ž or systems . are selectively distributed in the brain Žfor reviews see Bas¸ar et al., 1992; 1997a,b; Bas¸ar, 1998.. The synchronous occurrence of such responses in multiple brain areas hints at the existence of distributed oscillatory systems and parallel processing in the brain. Such diffuse networks would facilitate the information transfer in the brain according to the general theory of resonance phenomena. The term ‘diffuse’ was used in order to describe the distributed nature of the frequency response in the brain. It is not yet not possible to define connections between the elements of these systems neuron by neuron tracking, or to define the directions of signal flow and exact boundaries of neuronal populations involved. However, this description is necessary to emphasize that oscillatory phenomena in these frequency ranges are not unique features of the observed single subsystem of the brain, and that their simultaneous existence in distant brain structures may be a relevant and important point in the description of an integrative neurophysiology. Further details about the possible functional significance of such distributed oscillatory systems will be given below.

2. Gamma oscillations As for e¨ ent-related gamma oscillations, the most prominent examples nowadays are oscillatory responses in the frequency range of 40]60 Hz occurring in synchrony within a functional column in the cat visual cortex ŽEckhorn et al., 1988; Gray and Singer, 1989.. This has been suggested as a possible mechanism of feature linking in the visual cortex, being related to the ‘binding problem’. This theory, however, does not fully explain the ‘ubiquity of gamma rhythms’ ŽDesmedt and Tomberg, 1994, Schurmann et al., 1997a.. It this ¨ respect, it may be helpful to consider further


studies of gamma oscillations, partly going back to Lord Adrian Ž1942.. While the interpretations are heterogeneous, the empirical findings may be roughly classified into sensory Žor obligatory. vs. cogniti¨ e gamma responses. 2.1. Sensory function Some examples of sensory functions follow here: 1.

2.

Auditory and ¨ isual gamma responses are selectively distributed in different cortical and subcortical structures ŽFig. 2.. They are phase-locked stable components of EPs in cortex, hippocampus, brain stem, and cerebellum of cats occurring 100 ms after the sensory stimulation with a second window of approximately 300 ms latency ŽBas¸ar, 1980, 1999; Bas¸ar et al., 1997b. A phase-locked gamma oscillation is also a component of the human auditory and ¨ isual

Fig. 2. Responses to auditory stimuli in single cats, filtered in gamma band Ž30]70 Hz. filtered auditory responses of single cats in GEA, RF, HI and CE. The gamma band filtered grand averages of responses are shown in the bottom row Žfrom Bas¸ar et al., 1995..

99

response ŽFig. 3; Bas¸ar et al., 1987.. A new strategy by application of six cognitive paradigms showed that the 40-Hz response in the 100-ms after stimulations has a sensory origin, being independent of cognitive tasks ŽKarakas¸ and Bas¸ar, 1999. The auditory MEG gamma response is similar to human EEG responses with a close relationship to the middle latency auditory evoked response ŽPantev et al., 1991.. 3. An early phase locked 40-Hz response was recorded in visceral ganglion of Helix pomatia using electrical stimulation ŽSchutt ¨ and Bas¸ar, 1992.. In arthropods also, light-induced gamma responses have been observed ŽKirschfeld, 1992. 2.2. Cogniti¨ e processes Several investigations dealt with cogniti¨ e processes related to gamma responses, some of them based on measuring the P300 wave. This positive deflection typically occurs in human ERPs in response to ‘oddball’ stimuli or omitted stimuli interspersed as ‘targets’ into a series of standard stimuli: 1. A P300-40 Hz component has been recorded in the cat hippocampus, reticular formation and cortex Žwith omitted auditory stimuli as targets.. This response occurs approximately 300 ms after stimulation, being superimposed with a slow wave of 4 Hz ŽFig. 4; Bas¸ar-Eroglu ˘ and Bas¸ar, 1991.. Preliminary data indicate similar P300-40 Hz responses to oddball stimuli in humans ŽBas¸ar et al., 1993; Yordanova et al., 1997.. However, a suppression of 40-Hz activity after target stimuli has also been reported ŽFell et al., 1997.. 2. Attention-related 40-Hz responses were observed in humans, especially over the frontal and central areas ŽTiitinen et al., 1993.. 3. During visual perception of re¨ ersible or ambiguous figures a significant increase Žalmost 50%. in human frontal gamma EEG activity has been recorded ŽBas¸ar-Eroglu ˘ et al., 1996.. 4. The spatiotemporal magnetic field pattern of

100


Fig. 3. ŽA. Ten randomly selected single EEG-EP trials filtered with digital filters of 30]50 Hz. U , average of these trials. Stimulation is applied at time ‘0 ms’. ŽB. Ten single EEG-EP trials, digitally filtered Ž30]50 Hz. and selected for high enhancement, i.e. high amplitude increase after stimulus Žin comparison with prestimulus EEG amplitude.. UU , average of these trials. ŽC. Selectively averaged evoked potentials ŽEPs.; averages of 40 single trials each, with different selection of trials Žhuman vertex recordings.. Ža. EP averaged from randomly selected single trials Ž‘conventional’ averaged EP.. Žb. Same EP, filtered 30]50 Hz. Žc. EP averaged from single trials selected for specific criteria Ž marked amplitude enhancement in the 40-Hz range after stimulation.. Žd. Same EP, filtered with band limits 30]50 Hz. Že. EP averaged from single trials with low amplitude enhancement. Žf. Same EP, filtered with band limits 30]50 Hz. Žg. Identical with Že. } average of low enhancement trials. Žh. Result of applying a 30]50-Hz stop-band filter Žwhich theoretically rejects the 40-Hz response. to the conventional averaged EP of Ža.. Note the similarity of Žh., obtained with stop band filtering, and Žg., obtained with low enhancement trials ŽModified from Bas¸ar et al., 1987..

gamma band acti¨ ity has been interpreted as a coherent rostrocaudal ‘sweep of activity’ ŽLlinas and Ribary, 1992..

2.3. Selecti¨ ely distributed gamma system This wide spectrum of experimental data is in accordance with a hypothetical ‘selecti¨ ely distributed parallel processing gamma system’ with multiple functions. Rather than being highly specific correlates of a single process, gamma oscillations might be important building blocks of elec-

trical activity of the brain. Being related to multiple functions, they may: Ži. occur in different and distant structures; Žii. act in parallel; and Žiii. show phase locking, time locking or weak time locking. Notably, simple electrical stimulation of isolated invertebrate ganglia evokes gamma oscillations Žin the absence of perceptual binding or higher cognitive processes. ŽSchutt ¨ and Bas¸ar, 1992.. In conclusion, gamma oscillations possibly represent a universal code of CNS communication ŽBas¸ar, 1998, 1999.. This view might also serve as a synthesis overcoming controversies in earlier reports.


101

3. Alpha oscillations 3.1. Functions of 10-Hz oscillations As to the alpha range, a ‘renaissance of functional alphas’ is under way. The interpretation of alpha rhythms as an ‘idling rhythm’ rests on observations such as blocking of ‘spontaneous’ occipital alpha oscillations upon opening of the eyes or blocking of central mu rhythm upon movement onset ŽKuhlman, 1978. Ž‘event-related desynchronization’ ŽPfurtscheller et al., 1997.. A reverse effect Žincrease of mu rhythms during visual information processing, ‘event-related synchronization’ ŽKoshino and Niedermeyer, 1975; Pfurtscheller et al., 1997. has also been reported. However, co-existing with these well-known phenomena and in relationship with Adrian’s ‘evoked alpha’ ŽAdrian, 1942., several forms of ‘functional alpha’ have been observed during sensory and cognitive processes ŽBas¸ar et al., 1997a,b,c; Bas¸ar, 1998, 1999; Schurmann et al., ¨ 1997b.:

Fig. 4. Event-related potentials of lower pyramidal layer ŽCA3. of hippocampus Žone cat.. Top: single ERP sweeps Žepochs. filtered at 30]50 Hz. Middle: averaged ERP filtered at 30]50 Hz. Bottom: unfiltered ERP, average of 50 artefact-free epochs. ŽModified from Bas¸ar-Eroglu ˘ and Bas¸ar, 1991..

1. In the auditory and visual pathways in cats, adequate stimuli elicit alpha responses Ždamped 10-Hz oscillations of approx. 300 ms., which are visible without filtering ŽBas¸ar, 1980, 1998, 1999. Žfor confirmation by wavelet analysis, see Bas¸ar Ž1998.. Human alpha responses similar to those in the cat brain were also described ŽBas¸ar et al., 1997a,b,c.. Examples of such function-related alpha responses are given in the next paragraph. Multiple sclerosis patients with opticus neuritis show reduced alpha responses to visual stimuli, in consistence with a sensory function to the alpha response ŽBas¸ar, 1998.. 2. Thalamo-cortical circuits are not unique in generating alpha responses. Hippocampal and reticular 10-Hz responses are relatively modality-independent, hinting at possible supra-modal functions. 3. Cognitive targets significantly influence the alpha responses in P300: Using an oddball paradigm, prolonged event-related alpha os-

102


cillations up to 400 ms were observed ŽBas¸ar, 1998, 1999.. 4. Memory-related e¨ ent-related alpha oscillations can be observed in well-trained subjects one second before an expected target. New results ŽKlimesch et al., 1994; Bas¸ar et al., 1997a,b. demonstrate that alpha activity is strongly correlated with working memory and probably with long term memory engrams. 5. Alpha activity is not unique to mammals: spontaneous and electrically evoked 10-Hz oscillations in isolated ganglia of Helix pomatia and Aplysia ŽBullock and Bas¸ar, 1988; Schutt ¨ and Bas¸ar, 1992; Bas¸ar, 1999. serve as an example. In parallel to the gamma band, these results are consistent with a hypothetical selecti¨ ely distributed alpha system. Event-related alpha oscillations may facilitate association mechanisms in the following way: When a sensory or cognitive input elicits ‘10 Hz wave-trains’ in several brain structures then it can be expected that this general activity can serve as a resonating signal. The co-existence of evoked alpha oscillations with alpha blocking and event-related desynchronization ŽPfurtscheller et al., 1997. hints at multiple processes being reflected in alpha oscillations. An example of such co-existence are earlier measurements where high amplitude spontaneous alpha activity coincided with alpha blocking while low amplitude alpha preceded EPs of high amplitude ŽBas¸ar, 1998.. Again, parallel obser¨ ations at the cellular le¨ el are noteworthy: Evoked oscillations in the 8]10Hz frequency range in visual cortex neurons upon visual stimulation suggest a relation to scalp-recordable alpha responses ŽSilva et al., 1991; Dinse et al., 1997.. The sum of these observations permits a tentative interpretation of alpha as a functional and communicati¨ e signal with multiple functions. This interpretation of 10-Hz oscillations Žat the cellular level, or in populations. might be comparable to the putative universal role of gamma responses in brain signaling. For a more complete descriptions of functionrelated alpha the reader is referred to ŽBas¸ar et

al., 1997c.; some examples follow in the next paragraph. 3.2. E¨ idence of function-related alpha response with some examples 3.2.1. The alpha response in cross-modality measurements As an example of the function-related alpha responses mentioned in the previous paragraph, we will deal with topographic differences of frequency components. In particular, we will summarize results of measurements from auditory and visual areas. As auditory and visual stimuli were used, the condition were either ‘adequate stimulation’ Žauditory cortex recording of auditory EP; visual cortex recording of visual EP. or ‘inadequate stimulation Žvisual cortex recording of auditory EP and vice versa.. Such experiments are referred to as ‘cross-modality’ measurements Žsee Hartline, 1987; Bas¸ar and Schurmann, 1994. ¨ Panel A in Fig. 5 shows single-trial EPs filtered in the 8]15-Hz range. The left column refers to auditory stimulation with visual cortex recordings; the right column to visual stimulation with visual cortex recordings; i.e. inadequate vs. adequate stimulation. Responses to visual } adequate } stimulation show amplitude increase and timeand phase-locking. A distinct response is also seen in the filtered averaged EP in panel B. The unfiltered averaged EP also shows an alpha-like waveform. In contrast, responses to auditory stimulation } inadequate } neither show amplitude increase or phase locking, nor can we see an alpha response in the filtered average. There is a type of response in the unfiltered EP in panel C, but this is not an alpha response. Fig. 6 gives another example of time-locking and amplitude increase in single trial responses to adequate stimulation: Single sweeps, filtered in the 8]15-Hz range } visual stimuli, visual cortex recordings } are superimposed. These superimposed single sweeps filtered in the 8]15-Hz range Župper curves. are very similar in waveform to the wide-band filtered curves Žlower curves.. It is thus not only by filtering that the alpha response can be illustrated in these sweeps. Alpha responses


103

Fig. 5. EPs recorded from the cat brain by using intracranial electrodes. Ža. Single EEG-EP trials, filtered 8]15 Hz. Žb. Averaged EP, filtered 8]15 Hz. Žc. Averaged EP, wide-band filtered. Left column, inadequate stimulation Žvisual cortex recording with auditory stimulation.. Right column, adequate stimulation Žvisual cortex recording with visual stimulation. Žfrom Schurmann et al. ¨ Ž1997b...

are even visible in the broad-band filtered sweeps which are alpha-type responses. By contrast, Fig. 7 refers to a circumstance under which alpha responses cannot be recorded, i.e. to a measurement with inadequate stimulation. Note the lack of time-locking and the lack of amplitude increase. Thus, alpha responses were recorded with adequate stimuli in primary sensory areas. Adequate vs. inadequate differences were larger for alpha responses than for theta responses, demonstrat-

ing the functional relevance of frequency components. As an aside, in ‘cross-modality’ recordings from the auditory cortex Žgyrus ectosylvianus anterior. of the cat brain we observed a complementary effect: large alpha enhancements were present in auditory EP recordings. In visual EP recordings from the auditory cortex such alpha enhancements were not observed. Critics might argue that the filtering procedure gives rise to this type of enhancements. However, even the averaged visual EP without filtering Žsee

104


Fig. 7. Superimposed single trial EEG-EP epochs recorded from the cat brain with inadequate stimulation Žauditory cortex recordings with visual stimulus.. Upper panel, filter 8]15 Hz. Lower panel, wide-band filter Ž1]45 Hz. Žfrom Schurmann ¨ et al. Ž1997b... Fig. 6. Superimposed single trial EEG-EP epochs recorded from the cat brain with adequate stimulation Žvisual cortex recordings with visual stimulus.. Upper panel, filter 8]15 Hz. Lower panel, wide-band filter Ž1]45 Hz. wfrom Schurmann et ¨ al. Ž1996. In: Bas¸ar et al., 1997ax.

Fig. 6, above. shows a 10-Hz oscillatory waveform. For an extended discussion of such criticism see ŽBas¸ar, 1998.. We learned from these experiments that such damped alpha activity is not present in all parts of the brain or elicited by all types of stimuli: it is only by the combination of EP frequency analysis, of adequate stimuli and of appropriate electrode positions that such activities can be recorded. The results underline the following properties of the neural tissues under study: In the 10-Hz frequency range Žfilter limits: 8]14 Hz. we recorded large enhancements of single visual EPs in the visual cortex Žalso reflected in the amplitude frequency characteristics in the shape of a dominant 12-Hz peak.. In the language of systems theory significant Žsharp. peaks in the amplitude charac-

teristics of the transfer function characterize resonant behavior of the studied system Žsee chapter 7 in Bas¸ar, 1999.. One may also express this behavior as tuning of the ‘device’, or one might express the resonant frequency channels as the ‘natural frequencies’ of the system. In our case we may say that neural tissues in the occipital cortex are tuned to respond with 12 Hz and 1]5 Hz to adequate Žvisual. stimuli and with 1]5 Hz to inadequate Žauditory. stimuli. The response magnitudes to both visual and auditory stimulation are similar in the low 1]5-Hz frequency range. It is important to note that the 10]12-Hz response peak has almost disappeared in the case of inadequate Žauditory. stimuli, which, in turn, did not evoke alpha enhancements in single EEG-EP epochs of Fig. 7. 3.2.2. Alpha responses in human EEG and MEG in cross-modality experiments It is useful to compare the cat data to EEG and MEG recordings in humans. EEG measurements


were performed in N s 11 subjects ŽBas¸ar and Schurmann, 1994.. Fig. 8 shows filtered curves ¨ computed from grand averages of occipital recordings ŽO1.. The upper half of the figure shows theta responses whereas the lower half shows alpha responses. The alpha response to auditory stimulation Žinadequate for the visual cortex, occipitally located. is on the left, where the response is of low amplitude. The response to visual stimulation, however, is on the right, with a distinct alpha response. Note that the adequate]inadequate difference is less for the theta response. The hypothesis as given previously is thus supported: as observed in cats it is mainly the alpha response which is dependent on whether or not a stimulus is adequate. A correlation between the alpha response and primary sensory processing is thus plausible both for human and for cat EEG-EP data. MEG measurements were performed both with a BTI 7 channel MEG system ŽSaermark et al.,

105

1992. and with a PHILIPS 19-channel MEG system ŽBas¸ar et al., 1992; Schurmann et al., 1992a,b.. ¨ The methods used were similar to those used for EEG recordings where possible. We used auditory stimuli Ž2000 Hz; 80 dB sound pressure level. and selected sensor positions close to the auditory cortex and close to the visual cortex. The data shown in Fig. 9 were obtained with the seven-channel system where the different positions required two experimental sessions. Panel A shows temporal recordings, Panel B occipital recordings, in both cases with auditory stimuli. The underlying cortical areas being the primary auditory cortex and the primary visual cortex, auditory stimuli are regarded as adequate in the first case ŽFig. 9, panel A. and as inadequate in the second case ŽFig. 9, panel B.. High amplitude alpha responses are visible in panel A with adequate stimulation. In contrast, panel B with inadequate stimulation does not show such alpha responses.

Fig. 8. Frequency components of grands average EPs Ž N s 11.. Top, filter limits: 4]7 Hz, ‘theta response’. Bottom, filter limits: 8]15 Hz, ‘alpha response’. Left, acoustical stimulation. Right, visual stimulation. Žfrom Schurmann et al. Ž1997b... ¨

106


point out, that the amplitude, time course, and frequency responses in EPs strongly depend on the amplitude of the prestimulus alpha activity. Again, parallel obser¨ ations at the cellular le¨ el are noteworthy: Evoked oscillations in the 8]10Hz frequency range in visual cortex neurons upon visual stimulation suggest a relation to scalp-recordable alpha responses ŽSilva et al., 1991; Dinse et al., 1997.. The sum of these observations permits a tentative interpretation of alpha as a functional and communicati¨ e signal with multiple functions. This interpretation of 10 Hz oscillations Žat the cellular level, or in populations. might be comparable to the putative universal role of gamma responses in brain signaling.

4. Theta oscillations 4.1. Theta response oscillations in cogniti¨ e processes

Fig. 9. Human MEG responses to auditory stimulation: averaged evoked fields recorded in a typical subject Žfilter limits: 8]15 Hz.. Ža. Seven channels with ‘pure temporal’ location. Žb. Seven channels with ‘pure occipital’ location Žfrom Schurmann et al. Ž1997b... ¨

3.3. Brain’s selecti¨ ely distributed alpha system with multiple functions Similar to the gamma band the selecti¨ ely distributed alpha system in the brain is interwoven with multiple functions and control functions: 1. The 10-Hz processes may facilitate, association mechanisms in the brain: when a sensory or cognitive input elicits ‘10 Hz wave-trains’ in several brain structures then it can be expected that this general activity can serve as a resonating signal ‘ par excellence’ ŽBas¸ar, 1980.. 2. Alpha activity controls EPs. Experiments of several authors Žsee Bas¸ar et al., 1997a, 1998.

Experimental data suggests that e¨ ent-related theta oscillations are related to cognitive processing and cortico-hippocampal interaction ŽMiller, 1991; Klimesch et al., 1994; Bas¸ar, 1999.. Some examples follow: 1. Theta is the most stable component of the cat P300-like response ŽBas¸ar, 1999.. 2. Bimodal sensory stimulation induces large increases in frontal theta response thus demonstrating that complex events require frontal processing ŽBas¸ar, 1999.. 3. Event-related theta oscillations are prolonged andror have a second time window approximately 300 ms after target stimuli in oddball experiments. Prolongation of theta is interpreted as being correlated with selecti¨ e attention ŽBas¸ar-Eroglu ˘ et al., 1992.. 4. Event-related theta oscillations are also observed after an inadequate stimulation whereas event-related alpha oscillations are not existent if the stimulation is an inadequate one. Accordingly the associative character for event-related theta oscillations is


more pronounced than for higher frequency event-related oscillations ŽBas¸ar-Eroglu ˘ et al., 1992.. 5. ‘Orienting’ } a coordinated response indicating alertness, arousal or readiness to process information } is related to theta oscillations and manifested in cat experiments during exploration and searching and motor behavior ŽBas¸ar, 1998, 1999.. 6. Results of Miller Ž1991. on cortico-hippocampal signal processing support the functional role of theta transmission in all cognitive states related to association. According to the statements above it is clear that event-related theta oscillations can be con-

107

sidered as important building blocks of functional signaling in the brain. 4.2. E¨ idence of theta response: an example Demiralp and Bas¸ar Ž1992. have measured significant theta responses following expected visual and auditory targets. Their results help to understand the functioning of the diffusely distributed theta system of the brain. EPs as well as ERPs were recorded from 10 healthy subjects in auditory and visual modalities Žonly visual EPs will be shown in this chapter.. For ERP recordings ‘the omitted stimulus paradigm’ was employed, in which the subjects were expected to mark mentally the onset time Žtime prediction task. of the omitted stimulus Žtarget..

Fig. 10. Superimposed standard visual EPs ŽVEP. and responses to third attended light stimuli in the visual omitted stimulus paradigm Ž3 ATT. of 10 subjects obtained from frontal ŽF3., parietal ŽP3. and occipital ŽO1. regions. Wide-band filtered recordings Žfrom Demiralp and Bas¸ar, 1992..

108


The bottom row of Fig. 10 shows the unfiltered averaged responses, superimposed, of 10 subjects recorded from F3, P3 and O1 leads upon application of the standard visual EP paradigm ŽVEP.. The upper row of Fig. 10 illustrates the responses of the same subjects to the third attended visual stimuli Ž3.ATT. in the visual omitted stimulus paradigm. In this paradigm, theta responses are visible even without filtering. Especially the frontal EP response looks like an ‘almost pure theta oscillation’ visible without filtering. The averaged responses were filtered in various frequency bands by means of digital filters with zero phase-shift Žband limits selected according to

maxima in amplitude frequency characteristics .. Fig. 11 shows the visual EPs ŽVEP, bottom. and the responses to the third attended light stimuli Ž3.ATT, top. in the omitted stimulus paradigm Žsuperimposed. and the grand averages obtained in both conditions filtered in the theta frequency band Ž3]6 Hz.. Note that the similarity between wide-band filtered curves ŽFig. 10, above. and theta-filtered curves ŽFig. 11, above. is highest for frontal recordings Ž‘pure theta’ responses.. Further details of the results will be presented in the companion volume. The highest, statistically significant, theta increases during cognitive performance were ob-

Fig. 11. Superimposed standard visual EPs ŽVEP. and responses to third attended light stimuli in the visual omitted stimulus paradigm Ž3 ATT. of 10 subjects and their grand averages filtered in theta frequency band Ž3]6 Hz. Žfrom Demiralp and Bas¸ar, 1992..


tained in frontal and parietal recording sites. In the visual modality the theta response increase in the frontal recording site was slightly higher than that in the parietal recording site Ž48% vs. 45%.. Since the cognitive task in this study was also mainly based on anticipation to an expected stimulus, it is not surprising that the greatest changes are in frontal regions.

5. Delta oscillations 5.1. Delta responses in cogniti¨ e processes As to delta oscillations, experimental data hint at functional correlates roughly similar to those mentioned for theta oscillations, i.e. mainly in cognitive processing: 1. The response to visual oddball targets have their highest response amplitude in parietal locations, whereas for auditory target stimuli the highest delta response amplitudes are observed in central and frontal areas ŽSchutt ¨ and Bas¸ar, 1992; Bas¸ar, 1999.. 2. Cogniti¨ e functions: The amplitude of the delta response is considerably increased during oddball experiments. Accordingly, it was concluded that the delta response is related to signal detection and decision making ŽBas¸arEroglu ˘ et al., 1992.. 3. In response to stimuli at the hearing threshold delta oscillations are observed in human subjects, in consistence with the hypothetical relation to signal detection and decision making ŽBas¸ar, 1999.. 4. A waveform observed in response to deviant stimuli not attended by the subject, the mismatch negati¨ ity ŽNaatanen, 1992. is shaped ¨¨ ¨ by a delayed delta response superimposed with a significant theta response. 5. Phase-locked delta responses are probably the major processing signals in the sleeping cat and human brain ŽBas¸ar, 1980.. The topographic distribution of the results is again consistent with a distributed response system. The delta response obtained during a typical P300

109

experiment will be described in the following to give a good support about the cognitive nature of the delta responses. Details are given in Bas¸ar Ž1999.; some examples follow in the next paragraph. 5.2. Functional significance of the delta response } examples from experiments with ‘cogniti¨ e’ paradigms The P300 human response to a special type of auditory stimuli shows that delta responses can be considered as ‘real brain responses’ with precise functional correlates. This was demonstrated in a study using an auditory oddball paradigm ŽBas¸arEroglu ˘ et al., 1992.. Standard auditory EPs Ždelta response amplitude set to 100%. were compared with responses to oddball stimuli where the normalized delta amplitude was approximately 600% Žsee Table 1.. This remarkable increase is an example of a major change in the frequency contents of an EP as mentioned in the beginning of this chapter. Taking into account the psychophysiological foundation of the P300 paradigm this hints at cognitive processing as a functional correlate of the delta response. The same conclusion was drawn from a study employing a visual oddball paradigm with standard vs. target checkerboard stimuli ŽSchurmann et al., 1995; see below.. ¨ In a new series of experiments the first group of voluntary healthy subjects Ž21]29 years; six male; four female. underwent visual evoked potential Ž VEP . measurements with reversal of a 509 checkerboard pattern. In the second group Ž17]33 years; seven male; three female. two stimuli were applied in pseudorandom order: NON-TARGET Ž75% occurrence. was checkerboard reversal. Subjects were instructed to pay attention to TARGET Ž25%. stimuli, i.e. checkerboard reversal with horizontal and vertical displacement by 259. Fig. 12 shows amplitude frequency characteristics computed from a¨ eraged ERPs. Fig. 13 shows single trials of target responses clearly demonstrating that pure delta responses are visible in such target responses even without filtering. Maxima in the 10-Hz range were common to VEP and TARGET and largest in occipital positions. Prominent maxima in the 0.5]3.5-Hz range were only


110

Table 1 The medians of maximum amplitudes of delta, theta and alpha frequency components of auditory evoked potentials ŽAEP. responses to third attended stimuli in the omitted stimulus paradigm Ž3.ATT. and responses to non-frequent target tones in the oddball paradigm ŽODDBALL. obtained in frontal ŽF3., vertex ŽCz., parietal ŽP3. and occipital ŽO1. recording sites Delta Ž1]3 Hz.

AEP 3.ATT

F3

P300 AEP 3.ATT

Cz

P300 AEP 3.ATT

P3

P300 AEP 3.ATT

O1

P300

1.8 3.0 Ž66%.a 10.6 UU Ž489%.a, 5.3 7.3 Ž37%.a 10.9 UU Ž106%.a, 2.4 3.2 Ž33%.a 9.7 UU Ž304%.a, 1.6 1.9 Ž19%.a 8.2 UU Ž413%.a,

Theta Ž3]6 Hz. Window 1 Ž0]250 ms.

Window 2 Ž250]500 ms.

4.7 6.7 Ž43%.aUU 4.5 Žy4%.a 8.2 9.5 Ž16%.a 7.2 Žy12%.a 4.0 4.4 U Ž10%.a, 2.7 Žy33%.a 2.3 2.3 Ž0%.a 2.8 Ž22%.a

2.0 2.0 Ž0%.a 6.5 UU Ž225%.a, 3.0 2.9 Žy3%.a 9.7 UU Ž223%.a, 1.6 1.8 Ž13%.a 5.8 UU Ž263%.a, 1.3 1.5 Ž15%.a 3.9 UU Ž200%.a,

a

The percent changes of amplitudes in Ž3.ATT. and ŽODD-BALL. conditions as the percent of the standard AEP amplitudes are given in parentheses. U P- 0.05. UU P- 0.01.

observed after TARGET stimuli. Filtered a¨ eraged ERPs Ždelta. in Fig. 2c show a prominent positive deflection in TARGET responses at approximately 400 ms Žamplitude: up to 244% in comparison to VEP.. Amplitude differences of VEPs vs. responses to TARGET were significant for delta Žd.f.s 2, 27; F s 4.60; P- 0.05. but not for alpha Ž8]15 Hz. responses Žd.f.s 2, 27; F s 1.55; MANOVA test; factor TASK.. Thus, the delta response is clearly more dependent on the P300 task than the alpha response ŽSchurmann et al., 1995.. ¨ The slow positive wave in TARGET responses belongs to the family of the P300-waves Žcf. Bas¸ar et al., 1987, 1993; Bas¸ar-Eroglu, ˘ and Bas¸ar, 1991.

which are widely accepted to be related to the processing of task-rele¨ ant, surprising e¨ ents and to reflect a manifold of cognitive processes. Polich and Kok Ž1995. emphasize that, although the explanations of the P300 center around the basic information processing mechanisms of attention allocation and immediate memory, a substantial portion of P300 variation appears to be caused by factors not only related to alterations of the task structure but also fluctuations in the arousal state of the subject. The ‘uni¨ ersal’, i.e. general, modality-independent character of the ‘P300-delta response’ is underlined by similar responses in auditory P300 experiments ŽBullock and Bas¸ar, 1988; Bas¸ar et al., 1997a,b; Bas¸ar, 1999.. Furthermore, detecting auditory stimuli close to the hearing threshold produced slow induced delta rhythms, possibly correlates of signal detection and decision making ŽBas¸ar, 1980..

6. Communication networks in the brain and oscillatory responses 6.1. Most general transfer functions in the brain Fessard Ž1961. emphasized that the brain must not be considered simply as a juxtaposition of private lines, leading to a mosaic of independent cortical territories, one for each sense modality, with internal subdivisions corresponding to topical differentiations. What are principles dominating the operations of heterosensory communications in the brain? This knowledge needs an extensive use of multiple microelectrode recordings, together with a systematic treatment of data by computers Žcf. Eckhorn et al., 1988; Gray and Singer, 1989.. Fessard indicated the necessity of discovering principles that govern the most general } or transfer functions of multiunit homogeneous messages through neuronal networks. The transfer function describes the ability of a network to increase or impede transmission of signals in given frequency channels. The transfer function, represented mathematically by frequency characteristics or wavelets, ŽBas¸ar, 1980; Bas¸ar-Eroglu ˘ et al., 1992. constitute the main


111

Fig. 12. Responses to visual evoked potential ŽVEP. stimuli Ž left column. and responses to TARGET stimuli Ž right column.. Ža. Time domain: wide-band filtered averaged event-related potentials ŽERPs. in subject J.A. Žb. Frequency domain: amplitude frequency characteristics computed from ERPs shown in Ža.. Along the x-axis, frequency in logarithmic scale; along the y-axis, amplitude in relative units ŽdB.. Amplitudes are normalized in such a way that the amplitude at 1 Hz is equal to 0 dB. Žc. Time domain: 0.5]3.5-Hz filtered ERPs in typical subject J.A. Žfrom Schurmann et al., 1995.. ¨

112


Fig. 12. Ž Continued.

framework for signal processing and communication. The existence of general transfer functions would then be interpreted as the existence of networks distributed in the brain having similar frequency characteristics facilitating or optimizing the signal transmission in resonant frequency channels ŽBas¸ar, 1999.. In an electric system an optimal transmission of signals is reached when subsystems are tuned to the same frequency range. Does the brain have such subsystems tuned in similar frequency ranges, or do there exist common frequency modes in the brain? The empirical results reviewed here imply a positive answer and provide a satisfactory framework to Fessard’s question. Frequency selectivities in all brain tissues containing selectively distributed oscillatory networks Ž delta, theta, alpha, beta, gamma. constitute and govern mathematically the general transfer functions of the brain. To fulfil Fessard’s prediction all brain tissues, both mammalian and invertebrates would have to react to sensitive and cognitive inputs with oscillatory activity or with similar transfer functions.

The degree of synchrony, amplitudes, locations and durations or phase lags are continuously varying, but similar oscillations are most often present in the activated brain tissues ŽBas¸ar, 1999.. 6.2. E¨ ent processing in distributed systems The synchrony of selectivities described earlier by our group could have a conceptual parallel in ‘selectively distributed processing’ in neurocognitive networks ŽMesulam, 1990, 1994.. In Mesulam’s neurological model of cognition, the unimodal areas of cortex provide the most veridical building blocks of experience Žfor functional anatomy see also chapter 3 in volume I.. Transmodal nodes bind information in a way that introduces temporal and contextual coherence. The formation of specific templates belonging to objects and memories occurs in distributed form but with considerable specialization. This arrangement leads to a highly flexible and powerful computational system which underlies the selectively distributed processing. In our earlier work we often


113

This means that the oscillatory 10 Hz response or delta response should also show selective behavior in various anatomical structures distinguished by their physiological functions Žsee above for a discussion about such ‘selectively distributed oscillatory systems’.. This idea will be detailed in the following section.

7. Distributed percepts and memory networks 7.1. Memory as distributed property of cortical systems

Fig. 13. Top: single trial ERPs Žresponses to TARGET. in subject C.G. Bottom: averaged ERP Žresponses to TARGET. in subject C.G. Right parietal recordings ŽP4. Žfrom Schurmann ¨ et al., 1995..

used the expression or concept of distributed oscillatory systems and their resonance as selective activities. According to Mesulam, functional selectivities exist in distributed functions that are based on the anatomy. The electrophysiological activity of selectively distributed systems must be also of selective behavior. Accordingly, oscillatory response susceptibility of the sensory cortices, of the hippocampus, thalamus or cerebellum should also be differentiated, depicting selective behavior to stimulation from the milieu interior or exterior. Goldman-Rakic Ž1988, 1997., in search of a topography of cognition, concludes: ‘If subdivisions of limbic, motor, sensory, and associative cortex exist in developmentally linked and functionally unified networks, as the anatomical, physiological, and behavioral evidence suggests, it may be more useful to study the cortex in terms of information processing functions and systems rather than traditional but artificially segregated sensory, motor, or limbic components and individual neurons within only one of these components’.

In this paper we intend to describe some hints or remarks related to working memory and event related oscillations. This is suggested as an example of the relationship between the concepts of distributed systems and e¨ ent-related oscillations, which probably will attract considerable attention in memory research in future. According to Fuster Ž1997. our thinking on the cortical organization of primate memory is undergoing a Copernican change, from a neurophysiology that localizes different memories in different areas as to one that views memory as a distributed property of cortical systems as stated. According to Fuster’s empirically founded hypothesis, the same cortical systems that serve us to perceive the world serve us to remember it. Fuster Ž1997. states that memory reflects a distributed property of cortical systems. An important part of higher nervous function, as perception, recognition, language, planing, problem solving and decision making, is interwoven with memory. Memory is a property of the neurobiological systems it serves and inseparable from their other functions. Perceiving is the classification of objects by activation of the associative nets that represent them in memory. It is reasonable to assume, as Hayek Ž1952. did, that memory and perception share, to a large extent, the same cortical networks, neurons and connections. To understand the formation and topography of memory, it is useful to think of the primary and sensory motor areas of the cortex that we may call phyletic memory or memory of the species. The structure of

114


primary sensory and motor cortices may be considered a fund of memory that the species has acquired in evolution. We can call it memory because, like personal memory, it is information that has been acquired and stored, and can be retrieved Žrecalled. by sensory stimuli or the need to act. Perceptual memory is memory acquired through the senses. It comprises all that is commonly understood as personal memory and knowledge, i.e. the representation of events, objects, persons, animals, facts, names and concepts. From the hierarchical viewpoint, at the bottom are memories of elementary sensations; at the top, the abstract concepts that, although originally acquired by sensory experience, have become independent from it in cognitive operations. According to Fuster Ž1997. the classic terms representation, retrieval, recall, recognition, short term memory and long-term memory are still valid for current discourse, but need to be neurobiologically redefined. Arguably, the smallest memory network Žnetlet . is the cortical cell group or module representing a simple sensory or motor feature in the interface between the organism and its environment. Single neuron recordings in monkeys that are trained to perform working memory tasks have identified components of a working memory circuit in the prefrontal cortex. In these studies, the neuronal processes that are related to task performance can be dissociated, on the scale of milliseconds to seconds. During a working memory task, as the stimulus is sequentially registered, stored over a period of a second and then translated into a motor response, specific neural populations respond in a characteristic ways. One class of prefrontal neuron responds to a visual stimulus as long as the stimulus is in view. In contrast, other prefrontal neurons are activated at the offset of the stimulus, and they remain active all the time that the monkey has to remember the location or features of an object ŽGoldman-Rakic, 1988, 1997; Fuster, 1995.. For investigations in humans, Cohen et al. Ž1997., and Courtney et al. Ž1997. used functional MRI ŽfMRI., finding parallels to the knowledge gained from single-cell recordings in animals.

Courtney et al. Ž1997. presented subjects with pictures of human faces, and asked them to recall whether the picture being shown was the same, or different, from one that had been presented 8 s earlier. The authors found that activations in the prefrontal areas correlated most strongly with delay periods, compared with activations in the visual areas, which were more strongly correlated with sensory stimulation. Cohen et al. Ž1997. presented subjects with written consonants, one at a time every 10 s, and asked them to judge whether each consonant was the same as a letter presented one, two or three trials back in the sequence. This task requires that subjects remember the order of consonants, as well as their identity. The farther back in the sequence the consonant to be recalled occurs, the greater the ‘load’ on working memory. These authors showed that activations in the prefrontal cortex are maintained throughout the 10-s interstimulus interval and, importantly, that the degree of prefrontal activation is higher for the conditions with the greatest memory load. By contrast activations by the primary visual, somatosensory and motor cortices, as well as in several secondary regions, are not sustained across the 10-s interval, and they are not related to memory demand. They are probably responsive to the sensory or perceptual, but not memory-aiding, aspects of working memory tasks. Further, according to the fMRI results of Courtney et al. Ž1997. early extrastriate visual areas demonstrate transient, relatively non-selective responses to complex visual stimuli and later extrastriate visual areas demonstrate transient, selective responses to faces, indicating a more specialized role in the processing of meaningful images, and both extrastriate visual and prefrontal cortical areas demonstrate sustained activity during memory delays, indicating a role indicating a role in maintaining an active representation of the face in working memory. 7.2. What is the role of brain oscillations in memory processes? 7.2.1. General remarks According to the view of Fuster Ž1997. stating


that memory reflects a distributed property of cortical systems and to our view outlined in the present paper it can be hypothesized that the selectively distributed oscillatory systems Žor networks. may provide a general communication framework can be a useful concept for functional mapping of the brain ŽMesulam, 1990, 1994.. Communication in these networks might contribute to the formation of specific templates belonging to objects and memories. According to a model of cognition, this formation occurs as selecti¨ ely distributed processing with considerable specialization and in anatomically differentiated localizations ŽMesulam, 1990, 1994; for details about memory as a distributed property of cortical systems see also Fuster, 1997.. In particular, analysis of the hypothetical distributed oscillatory systems may lead to fundamental functional mapping of the brain, complementary to morphological studies. As one can extract from Mesulam’s and Fuster’s work, there must be common codes for perpetual signal transfer between neural networks for parallel and serial processing and also for possible reverberation circuits and loops between neural networks. Oscillations might serve as adequate codes for this general communication by putting the networks to resonate. A more general view is that functional or oscillatory networks modules are distributed not only in the cortex but in the whole brain ŽBas¸ar, 1998.. In the following we will try to discuss an electrophysiological parallel between Fuster’s ‘memory network’ and the distributed oscillatory systems mentioned earlier. When analyzing the field potentials it is difficult to define boundaries of brain nuclei and their electrical activity. Nevertheless, this approach is useful, since a great amount of data can be collected and interpreted from several electrodes distributed in the brain. Furthermore, it is possible to perform measurements during continuously changing cognitive states. This way, EPs or EEG segments are recorded in the cortex, limbic system, and thalamus, and cerebellum. They can be compared in waking and freely behaving animals. This type of recording during behavioral states

115

can not possibly be managed with recordings of single cell electrodes. Studies on functional correlates of structures like sensory cortices, hippocampus, and thalamic relay nuclei are mostly based on experiments using unit recordings. A major difficulty concerning the interpretation of experiments with single unit recordings Žfor example experiments on cortico-thalamic information transfer . is that the results are limited to a few neurons. Accordingly, the author assumes that every hypothesis on the localisation of ‘thalamo-cortical circuit as 10 Hz generator’ is restricted and not acceptable with regard to the results of experiments described in this book: The alpha, theta and gamma generators are selectively distributed in the brain. Bas¸ar Ž1999. reviewed several classifications of memories given by distinguished memory investigators in order to find a strategy to explore the electrophsyiology of distributed memories. He concluded that the brain oscillations in EEG and ERPs could provide a good foundation to attack the problem both at the neurophysioloical and psychological levels since brain oscillations have similar frequency codes in the whole brain. Remembering and memory are manifestations of various and multiple functional processes depending on the complexity of the input to the CNS. Already the electrical response to a simple light flash bases on simple memory processes at the lowest hierachical order. When we talk about a memory process-either a short } or long-term one } then we have in our mind the perception of a sensory input which is matched with information already stored in the neural tissue. If a simple light evokes alpha and gamma responses than it is almost obligatory to assume that elementary oscillatory responses are also manifestations of several memory processes at different hierarchical levels. The topology of the memories depending on the modality of the input must be different Žsee examples given above: cross modality experiments; measurements in cortical and subcortical structures .. So far, such studies have rarely been performed, as Bas¸ar et al. Ž1999. point out. Therefore, results and their interpretations are to be considered as preliminary. Accordingly, the multi-

116


ple distributed memories can not be treated in details and a classification on all levels of distributed memories can not be yet provided. In performing many complex tasks, it is necessary to hold information in temporary storage to complete the task. The system used for this is referred to as working memory ŽBaddeley, 1996.. Working memory is the temporary, ad hoc, activation of an extensive network of short- or long-term perceptual component of that network would be, as any other perceptual memory, retrievable and expandable by a new stimulus or experience. Fuster states that working memory has the same cortical substrate as the kind of short-term memory traditionally considered the gateway to long-term memory. According to the functional descriptions above, a simple light stimulation, or a complex light stimulation or a light stimulation with some tasks Ževent. should evoke oscillatory responses with different time hierarchies. Our view is that functional or oscillatory networks modules are distributed not only in the cortex but in the whole brain ŽBas¸ar, 1999.. The detection of a target signal during a P300 type of experiment requires also a type of working memory. The subject has to sustain knowledge related to the target Žnature of the target, frequency, shape, color, etc., depending on the type of the experiment. during the experiment. The linkage between P300 amplitude and latency measures and working memory processes has been already established ŽSanquist et al., 1980; Howard and Polich, 1985; Pratt et al., 1989; Fabiani et al., 1990; Scheffers and Johnson, 1994.. The matching process after the detection of the target should rely in this case on working memory or the success of this type of memory P300 vs. N100. The ERP is a compound neuroelectric signal which is rich in functional information ŽBullock, 1993. and related to a large spectrum ranging from single percepts to complicated memory processes. Furthermore, in the analysis of integrative brain functions it is indispensable to consider not only one specific ERP in a given brain structure, but to take into account that distributed ERPs are interrelated due to the evident strong parallel processing in the whole brain. Accord-

ingly, it is necessary to analyze the entire brain in order to understand even a specific function manifested by neurolectric activity of a given structures: For example, when we consider or analyze cognitive processes usually the most marked ERPs are recorded in fronto-parietal areas or in various association cortices. However, it is necessary to take into account recordings from other areas as well, e.g. from sensory cortices Žpossibly indicating parallel processing; Bas¸ar and Schurmann, ¨ 1994; Bas¸ar, 1998, 1999.. Several types of analysis categories are crucial in the functional interpretation of ERPs: 1. The analysis of stimulus itself: what can a stimulus evoke in the brain? It can evoke simple sensory percepts, complex sensory percepts, bimodal percepts or memory-related functions, etc. 2. The analysis of ERPs should be performed in related our unrelated function-dependent areas. For example, if a complex semantic event or memory demanding task is presented as stimulation is presented to the brain, usually frontal recordings and or parietal recordings are considered to carry the most important information. In this case it is very important to analyze ERPs recorded in the occipital cortex Žan area thought to be less involved in high level cognitive processing.. This shows what is missing in occipital ERPs in comparison to association ares, or what is recorded additionally. These steps are analogous to the fMRI analysis mentioned above. 3. The component analysis by means of eventrelated oscillations provide a real advantage over conventional ERP analysis as, for example, the results of cross modality measurements demonstrate: In occipital areas auditory stimulation does not evoke 10 Hz responses, although an ERP is measured upon visual stimulation. This demonstrates the dependence of the 10-Hz response on visual perception. Accordingly, the spatial resolution of ERPs is highly increased. 4. Studies with single-cell recordings and with fMRI point out that memory networks are distributed. Although the ERPs and the


event-related oscillations do not have the excellent spatial resolution of fMRI or the exact one-to-one location of single cell recordings, they have several outstanding advantages in memory research. 5. When compared with fMRI the time resolution of ERPs Žand of event-related oscillations. is excellent, since it is possible to measure function-related neuroelectric changes within a few milliseconds. 6. In ERP studies the neuroelectric or Žneuromagnetic recordings. can be also performed by human studies, almost impossible to perform with single cell recordings. Moreover it is possible to apply simultaneous measurements with several recording electrodes in distant locations, this in turn allowing the dynamic comparisons between various structures of the human cortex Žalso diverse subcortical structures during experiments with the animal brain.. For example, immediate comparison of frontal theta or alpha activity with the occipital ones is possible. 7. Similarly, increased focused attention evokes ample theta response in frontal recording, but almost no alpha response, whereas in occipital recordings alpha and theta responses are superimposed. During oddball recordings a 600% increase of delta responses can be recorded Žsee Table 1.. Such an enhancement cannot be obtained by conventional repetitive stimuli. As Fuster Ž1997. underlines there are in the brain as much memory types as the number of percepts. Not only for the analysis of working memory but for a simultaneous analysis of perceptual memory the applications of event-related oscillations is very useful as a complement to fMRI and single cell studies. These remarks clearly show that the analysis of event-related oscillations fills an important gap for the analysis of selectively distributed percepts and memories. 7.2.2. Examples of selecti¨ ely distributed oscillatory responses in search of memory To see something, even the simplest light signal, is already a memory process-related to a

117

fundamental inborn retrieval process: a baby perceives the light and shows reflex responses to light before going through learning processes. This is probably a basic decoding process. In Fig. 14 ŽSchurmann and Bas¸ar, unpublished ¨ results . responses to target and non-target stimulation Župon checkerboard stimulation. in alpha and delta frequencies are shown. The occipital 10 Hz response is large in posterior areas Žrelated to vision.. The delta response, however, is distributed, being most marked in posterior areas upon target stimuli. As explained above, the target signal also requires working memory processes. So the fact that delta responses are most marked in posterior areas hints at selective distribution of ‘memory oscillations’. Ample occipital 10 Hz responses are not recorded in frontal locations indicating that the frontal lobes are not involved in primary visual processing. This response is a sign of perceptual memory: When no visual perception occurs, then there are no 10 Hz responses. As to the delta response in the auditory P300 paradigm, a distributed highly enhanced response in the whole cortex is observed ŽBas¸ar-Eroglu ˘ et al., 1992., the maxima being in frontal and parietal areas. The auditory 10-Hz signal to auditory signal is missing. These findings again indicate a functional selective distribution 10 Hz responses are recorded in primary sensory areas. Frontal and parietal delta areas. As to the theta response the findings are yet more complicated to interpret: in the auditory P300 paradigm only target signals have a prolonged theta oscillation Žsecond window. marked in parietal and frontal recording indicating a correlation to working memory. In experiments where subjects pay attention to the third applied signal in an evoked potential experiment, the third attended signal show ample theta increases again especially in frontal location. In this report it is not our aim to differentiate the functional roles of theta and delta responses in working memory processes; we solely indicate that this oscillatory responses are selectively distributed depending on the memory load required during the experiments. A P300-40 Hz response as visible in Fig. 4

118


Fig. 14. Grand average ERPs Ž509 checkerboards; N s 9 subjects., filtered 1]5 Hz Ža. and 8]15 Hz Žb., respectively. Left: Visual evoked potentials ŽVEP., middle: responses to non-target stimuli; right: responses to target stimuli.

Žabove. illustrates the superposition of theta and gamma response following omitted stimuli as targets. The combination of gamma and delta response oscillations are here again involved with working memory. These findings have remarkable parallelities to

fMRI experiments showing topographical functional selectivity. Moreover, it is possible to define the working and perceptual memory components in a time window already 100 ms upon a memory load. At least two types of memory responses with a time hierarchy are observed: Ž1. sensory Žper-


ceptual. memory oscillatory response; and Ž2. late window responses related to working memory. This time hierarchy occurring in the 1-s period following stimuli, are not reflected in fMRI studies, and not found in conventional ERPs that, in the best case, allow analysis of two relevant peaks. 7.3. Grandmother cell or grandmother cell assemblies Sherrington introduced the notion of ‘one ultimate pontifical ner¨ e-cell’ which integrates CNS function. The grandmother neuron is a neuron that responds to nothing else but the face of one’s grandmother. Barlow Ž1972. stated that the sensory system is organized to achieve as completely as possible a representation of the sensory stimulus with a minimum of active neurons. According to his view, perception consists of achieving a selection from the very numerous high-level neurons corresponding to a pattern of external events of the order of complexity of the event that is symbolized by the word ‘grandmother’. According to what Mountcastle Ž1992. called a ‘paradigm change in neuroscience’, the role of neural assemblies gains an important rank in functional studies. In this place it is therefore appropriate to summarize some of relevant viewpoints about the organisation of neural masses, before dealing with the ‘problem of the grandmother cell’. Szenthagothai’s well known illustration of a 300-mm diameter cortical module is one of the important examples of neural modules and local nerve circuits, an ensemble that plays a significant functional role. According to Szenthagothai, modular organization allows a higher degree of specific connectivity to be achieved with a minimum of genetic instructions. Mountcastle defined the basic function unit as a ‘minicolumn’ approximately 30 mm in diameter containing 100]300 neurons. Larger processing units called ‘macrocolumns’ contain up to several hundred ‘minicolumns’. Concerning the functional level, Damasio and Damasio Ž1994. state that our brains use dynamic records, rather than static, immutable memory traces. For example the record of the face of a person you know is a set of neuron circuit changes

119

which can be reactivated, rather than the ‘picture’ that is stored somewhere in the brain. In our recent publications we developed, a step by step, a theory on the existence of the selectively distributed alpha and theta and 40 Hz systems of the brain. Furthermore, the existence of EEG-modules Žactive in parallel and selectively } or diffusely } distributed the brain. was proposed. EEG oscillation-modules, in this context, are neural populations giving rise to coherent EEG activities in different frequency channels. We assume that the EEG oscillations which can attain coherent states belong to functional repertoires of the higher brain function. The superposition of several of these signals makes it possible to connect behavioral events with an ensemble of event-related oscillations. Since the ERPs are now very popular for scientists of psychophysiology who use a diversity of psychological paradigms, the consideration of EEG dynamics can bring a deeper understanding to the correlation between cognitive performance and physiological states. According to Szentagothai connections are established between vertical columns and modules. These connections can link also frontal areas to occipital ones. Now the question is whether the hypotethized EEG generating modules in the frontal and occipital areas are also be connected. If event-related 10 Hz oscillations are measured in occipital and frontal areas, then it is certainly legitimate to define such modules and thus connections between these modules. The same possibility also exists in theta-, delta-, and gamma frequency bands. With the knowledge presented in Bas¸ar Ž1999. it is not possible to clearly indicate that the modules described above also serve for processing information in the EEG frequency channels. Volume conduction between modules as distant as occipital and frontal cortices is unlikely. Accordingly, the high amplitude delta responses following target signals in P300 experiments ŽFig. 12. must have different local generators. Furthermore, it is probable that neural impulses can reach several parts of the cortex following an event and generate in cortical areas oscillatory responses in various frequency chan-

120


nels depending on the sensory information or task. According to experiments described in Bas¸ar Ž1998, 1999. gamma activity is not recorded only in a small given brain population but in widely distributed neural populations. The 10-Hz resonances are also widely distributed and depend on functions processed in the brain. Depending on ‘regimes’ or ‘states’ of the brain, the limbic system, brainstem, thalamus, and cortex are all involved with 2, 4, 10, and 40 Hz firing or with all of them. The distribution of coherently working EEG modules does not take place only in the cortex; we assume that they are selectively distributed in the entire system. One of the examples is the frontal-hippocampal]parietal-system which gains a highly organized theta state during attentive processes. By combining examples and concepts arising from the analysis of EEG and evoked potential responses, we aim to extend from modular concepts to modular frequency systems operating in the entire brain. This is a new hypothesis emerging from a recent book ŽBas¸ar, 1999.. This hypothesis is based on experiments where brain functions are explained on the basis of measurement of field potentials which reflect the activity of large neural populations. These neural populations can be also be interpreted as functionally active modules. These explanations are based on the different frequency systems of the brain which are not defined in space and time precisely. But even then, such explanations seem more efficient in explaining the dynamic aspect of the percepts. In this report we have described numerous types of oscillatory activities with definitive or tentative explanations of their functional relations. The results from several laboratories clearly demonstrated that it is not possible to assign a single function to a given type of oscillatory activity. These oscillations have multifold functions and may act as uni¨ ersal operators or codes of brain activity. Besides frequency and site of oscillations, several other parameters are dependent on specific functions, namely enhancement, time locking, phase locking, delay and duration of oscillations. Concerning the question at the beginning of this section Ž‘Does a grandmother neuron exist?’.,

our hypothesis about the functional role of event-related brain oscillations is as follows: complex and integrative brain functions are manifested in the superposition of several oscillations. Stryker Ž1989. and Bas¸ar et al. Ž1997a,b,c. described results of cellular gamma activity ŽBullock and Bas¸ar, 1988; Desmedt and Tomberg, 1994. by commenting that neurons in the visual cortex activated by the same object in the world tend to discharge rhythmically and in unison. He raised the question ‘Is grandmother an oscillation?’ According to the studies reviewed above, the observation of the grandmother picture would activate oscillations, not only in the visual cortices, but in all parts of the brain Žprobably including frontoparietal delta, occipital alpha, theta and gamma oscillations.: every simple input trigger diverse oscillations in selectively distributed areas of the brain. Accordingly, distributed neural groups of all frequencies have to be involved in the processing of this complicated percept ŽDesmedt and Tomberg, 1994; Dinse et al., 1997..

8. A ‘neurons-brain’ doctrine: new thoughts With the neuron doctrine alone, as it was originally proposed by Sherrington, it is not possible to interpret the functional contributions of alpha, theta, delta and gamma responses. The generators giving rise to these frequency responses are extremely sensitive to the modality of sensory and cognitive inputs. For now, such generators can only be explored with macroelectrodes that are placed with an adequate physical separation. Tracking of properties of functionally-related distant single neurons is not yet possible because of technical limitation. At the first step approaching the higher functions of the brain is easier when we assume the existence of modules that generate EEG-like signals. At this point, it should again be remembered that potentials measured from human subjects reflect the properties of a large number of neurons and possibly, multiple neural populations. During an experiment of ‘P300 type’ the brain goes over to a regime in which distributed theta and delta neural networks are mostly in play.


During pure sensory stimulations several structures of the brain go over to a 10-Hz regime. During slow sleep stage the brain does not only show slow wave activity in the delta frequency range, but all EPs from intracranial and cortical structure of the cat brain and cortex of the human brain depict a dominance of delta responsiveness. There are several trends of extending or renewing Sherrington’s ‘neuron doctrine’. Freeman proposed a ‘Neo-Sherringtonian view’ of integrative brain function whereas Barlow and Mountcastle proposed modern views on the neuron doctrine. Here, an attempt at extending the Sherringtonian neuron doctrine to neural populations with oscillatory firing properties is proposed. It relies on experiments as reviewed above Žfor details, see Bas¸ar, 1998, 1999. and serves to describe a basic framework for signaling of communication and functional operating in the brain. This new doctrine cannot be perfect and cannot cover all results and principles accumulated in the last decades. However, it provides a plausible and progressive framework, which might replace the old doctrine. The latter one should be considered as a special case of the more general new ‘neurons-brain doctrine’. 1. The neuron is the basic signaling element of the brain. 2. Oscillatory neural acti¨ ity is considered as a basic signal reflecting natural frequencies of the brain Žthis thesis relies on works of Eckhorn et al., 1988; Eckhorn, 1994; Gray and Singer, 1989; Silva et al., 1991; Dinse et al., 1997.. 3. Neural assemblies replace the neuron in the description of complex brain functions. This view diverges from Sherrington’s ‘neuron doctrine’. As a metaphor to physical sciences, neurons can be considered as atoms, neural assemblies, that contribute to a function, as molecules. Accordingly the metaphor is similar to statistical mechanics and gas laws Žsee chapter 4.2 in Bas¸ar, 1998.. 4. Oscillatory acti¨ ities Ž e¨ ent-related, induced or spontaneous. govern the most general transfer

121

functions in the brain Žfrequency characteristics and power spectra are governed with alpha, gamma, theta, delta, etc. oscillations } which is confirmed by the wavelet approach.. Furthermore, as stated in Section 6.1 the general transfer functions provide a framework for the electrical information processing in the brain. 5. Oscillations in different frequency ranges are a property of the neurons Žsee no. 2 above.. Selecti¨ ely distributed oscillatory neural populations, however, behave with ‘molecular properties’ Žalpha, beta, gamma, delta, theta being ‘atomar properties’ .. These oscillatory networks are activated upon sensory stimulation or event-related tasks by manifestation of synchronization of neural activity; partial synchrony; enhancements; or blocking or desynchronization of oscillations; depending on the nature of the sensation or event, and accordingly depending on the function performed. These selectively distributed networks are operators of general brain functions including communication and association and data retrieval Žmolecular properties, see previous item.. 6. Major operating rhythms play a key role in association and communication. Topological distribution of oscillations is heterogeneous and their functions are multifold. Accordingly, parallel processing is not perfect between distributed populations since the major operating rhythms are selectively distributed. Examples: Alpha responses do not appear in the medial geniculate nucleus and in the auditory cortex to light stimulation, whereas the lateral geniculate nucleus and the primary visual cortex respond with large alpha enhancements Žmore examples are summarized in chapter 30 in Bas¸ar, 1999.. 7. Types of neurons do not play a major role for frequency tuning of oscillatory networks. The neural architectonics of the cerebellar cortex, cerebellum and hippocampus are completely different. In spite of this, all these structures behave with almost similar frequency responses.

122


8. Distributed oscillatory networks react selectively upon application of pharmacological agents. Examples: Caerulein causes a great change in hippocampal evoked response which takes the shape of a homogeneous 3]4-Hz response, whereas the frequency response of the cerebellum remained completely unchanged since this agent does not have any action in the cerebellum. Acetylcholine activates the 4-Hz response of the hippocampus enormously, whereas other structures are less influenced Žsee chapter 8 in Bas¸ar, 1999.. Functions in the brain are manifested by varied degrees of superpositions of oscillations in EEG frequency ranges. There are varied degrees of responsiveness depending on the strength of the stimulation or the event presented to the CNS. Accordingly, neuronal assemblies do not react with a type of all-or-none behavior as in the single neuron doctrine. There exists a strong inverse relation between prestimulus oscillations and brain responses. Spontaneous oscillations control the amplitude and shape of population responses Žsee chapters 12 and 14 in Arieli et al., 1996; Bas¸ar, 1998..

References

Adrian, E.D. Ž1942. Olfactory reactions in the brain of the hedgehog. J. Physiol., 459]473. Arieli, A., Sterkin, A., Grinvald, A., Aertsen, A., 1996. Dynamics of ongoing activity: explanation of the large variability in evoked cortical responses. Science 273, 1868]1871. Bas¸ar, E., 1980. EEG-brain dynamics. Relation between EEG and evoked potentials. Elsevier, Amsterdam. Bas¸ar, E., Rosen, B., Bas¸ar-Eroglu, ˘ C., Greitschus, F., 1987. The associations between 40-Hz EEG and the middle latency response of the auditory evoked potential. Int. J. Neurosci. 33, 103]117. Bas¸ar, E., 1992. Brain natural frequencies are causal factors for resonances and induced rhythms wEpiloguex. In: Bas¸ar, E., Bullock, T.H. ŽEds.., Induced Rhythms in the Brain. Birkhauser, Boston, MA, pp. 425]467. ¨ Bas¸ar, E., Schurmann, M., 1994. Functional aspects of evoked ¨ alpha and theta responses in humans and cats. Occipital

recordings in ‘cross modality’ experiments. Biol. Cybern. 72, 175]183. Bas¸ar, E., Bas¸ar-Eroglu, C., Parnefjord, R., Rahn, E., ˘ Schurmann, M., 1992. Evoked potentials: Ensembles of ¨ brain induced rhythmicities in the alpha, theta and gamma ranges. In: Bas¸ar, E., Bullock, T.H. ŽEds.., Induced Rhythms in the Brain. Birkhauser, Boston, pp. 155]181. ¨ Bas¸ar, E., Bas¸ar-Eroglu, M., ˘ C., Demiralp, T., Schurmann, ¨ 1993. The compound P300-40 Hz response of the human brain. Electroencephalogr. Clin. Neurophysiol. 87, 14P. M., Bas¸ar, E., Bas¸ar-Eroglu, ˘ C., Demiralp, T., Schurmann, ¨ 1995. Time and frequency analysis of the brain’s distributed gamma-band system. IEEE. Eng. Med. Biol. 14, 400]410. M., Bas¸ar, E., Hari, R., Lopes da Silva, F.H., Schurmann, ¨ ŽEds.., 1997a. Brain Alpha activity-new aspects and functional correlates Žspecial issue.. Int. J. Psychophysiol. 26: 1]482. M., Bas¸ar-Eroglu, Bas¸ar, E., Schurmann, ¨ ˘ C., Karakas¸, S., 1997b. Alpha oscillations in brain functioning: an integrative theory. Int. J. Psychophysiol. 26, 5]29. Bas¸ar, E., Yordanova, J., Kolev, V., Bas¸ar-Eroglu, ˘ C., 1997c. Is the alpha rhythm a control parameter for brain responses? Biol. Cybern. 76, 471]480. Bas¸ar, E., 1998. Brain Function and Oscillations. I. Brain Oscillations: Principles and Approaches. Springer, Berlin, Heidelberg. Bas¸ar, E., 1999. Brain Function and Oscillations. II. Integrative Brain Function. Neurophysiology and Cognitive Processes. Springer, Berlin, Heidelberg. M., 1998. Bas¸ar, E., Rahn, E., Demiralp, T., Schurmann, ¨ Spontaneous EEG theta activity controls frontal visual evoked visual evoked potential amplitudes. Electroencephalogr. Clin. Neurophysiol. 108, 101]109. M., Bas¸ar-Eroglu Bas¸ar E., Demiralp T., Schurmann ¨ ˘ C., Ademoglu A., 1999. Oscillatory brain dynamics, wavelet analysis, and cognition. Brain Lang. 66, 146]183. Bas¸ar-Eroglu, ˘ C., Bas¸ar, E., 1991. A compound P300-40 Hz response of the cat hippocampus. Int. J. Neurosci. 60, 227]237. M., Bas¸ar-Eroglu, ˘ C., Bas¸ar, E., Demiralp, T., Schurmann, ¨ 1992. P300-response: possible psychophysiological correlates in delta and theta frequency channels, a review. Int. J. Psychophysiol. 13, 161]179. D., Kruse, P., Bas¸ar, E., Stadler, M., Bas¸ar-Eroglu, ˘ C., Struber, ¨ 1996. Frontal gamma-band enhancement during multistable visual perception. Int. J. Psychophysiol. 24, 113]125. Baddeley, A., 1996. The fractionation of working memory. Proc. Natl. Acad. Sci. USA 93, 13468]13472. Barlow, H.B., 1972. Single Units and Sensation: a neuron doctrine for perceptual psychology. Perception 1, 371]394. Bullock, T.H., 1993. How do brains work? Birkhauser, Boston. ¨ Bullock, T.H., Bas¸ar, E., 1988. Comparison of ongoing compound field potentials in the brain of invertebrates and vertebrates. Brain Res. Rev. 13, 57]75. Cohen, J.D., Perlstein, W.M., Braver, T.S., Nystrom, L.E.,

E. Bas ¸ar et al. r International Journal of Psychophysiology 35 (2000) 95]124 Noll, D.C., Smith, E.E., 1997. Temporal dynamics of brain activation during a working memory task. Nature 386, 604]608. Courtney, S.M., Ungerleider, L.G., Keil, K., Haxby, J.V., 1997. Transient and sustained activity in a distributed neural system for human working memory. Nature 386, 608]611. Damasio, A.R., Damasio, H., 1994. Cortical Systems for retrieval of concrete knowledge: the convergence zone framework. In: Koch, C., Davis, J.L. ŽEds.., Large scale Neuronal theories of the brain. MIT Press, Cambridge, MA, pp. 61]74. Demiralp, T., Bas¸ar, E., 1992. Theta rhythmicities following expected visual and auditory targets. Int. J. Psychophysiol. 13, 147]160. Demiralp T., Ademoglu A., Schurmann M., Bas¸ar-Eroglu ¨ ˘ C., Bas¸ar E., 1999. Detection of P300 waves in single trials by the wavelet transform. Brain Lang. 66, 108]128. Desmedt, J.E., Tomberg, C., 1994. Transient phaselocking of 40 Hz electrical oscillations in prefrontal and parietal human cortex reflects the process of conscious somatic perception. Neurosci. lett. 168, 126]129. Dinse, H.R., Kruger, K., Akhavan, A.C., Spengler, F., Schoner, ¨ ¨ G., Schreiner, C.E., 1997. Low-frequency oscillations of visual, auditory and somatosensory cortical neurons evoked by sensory stimulation. Int. J. Psychophysiol. 26, 205]227. Eckhorn, R., 1994. Oscillatory and non-oscillatory synchronizations in the visual cortex and their possible roles in associations of visual features. Prog. Brain Res. 102, 405]426. Eckhorn, R., Bauer, R., Jordan, W. et al., 1988. Coherent oscillations: a mechanism of feature linking in the visual cortex? Biol. Cybern. 60, 121]130. Fabiani, M., Karis, D., Donchin, E., 1990. Effects of mnemonic strategy manipulation in a van restorff paradigm. Electroencephalogr. Clin. Neurophysiol. 75, 22]35. Fell, J., Hinrichs, H., Roschke, J., 1997. Time course of 40 Hz ¨ EEG activity accompanying P3 responses in an auditory oddball paradigm. Neuro-Sci. Lett. 235, 121]124. Fessard, A., 1961. The role of neuronal networks in sensory communications within the brain. In: Rosenblith, W.A. ŽEd.., Sensory Communication. MIT press, Boston, MA. Freeman, W.J., 1975. Mass Action in the Nervous System. Academic press, New York. Freeman, W.J., 1998. Preface to Brain Function and Oscillations. Springer, Berlin, Heidelberg. Fuster, J.M., 1995. Memory in the cortex of the primate. Biol. Res. 28, 59]72. Fuster, J.M., 1997. Network memory. Trends Neurosci. 20, 451]459. Goldman-Rakic, P., 1988. Topography of cognition: ‘Parallel distributed networks in primate association cortex. Ann. Rev. Neurosci. 11, 137]156. Goldman-Rakic, P., 1997. Space and time in the mental universe. Nature 386, 559]560. Gray, C.M., Singer, W., 1989. Stimulus-specific neuronal oscil-

123

lations in orientation columns of cat visual cortex. Proc. Natl. Acad. Sci. USA 86, 1698]1702. Hartline, P.H., 1987. Multisensory convergence. In: Adelman, G. ŽEd.., Encyclopedia of Neuroscience. Birkhauser, Bos¨ ton, Basel, Stuttgart, pp. 706]709. Hayek, F.A., 1952. The Sensory Order. University of Chicago Press Howard, L., Polich, J., 1985. P300 latency and memory span development. Dev. Psychol. 21, 283]289. John, E.R., Prichep, L.S., Fridman, J., Easton, P., 1988. Neurometrics: computer-assisted differential diagnosis of brain dysfunctions. Science 239, 162]169. Karakas¸, S., Bas¸ar, E., 1999. Early gamma response is sensory in origin. A conclusion based on cross comparison of results from multiple experimental paradigms. Int. J. Psychophysiol. 31, 13]31 Kirschfeld, K., 1992. Oscillations in the insect brain: do they correspond to the cortical gamma waves of vertebrates? Proc. Natl. Acad. Sci. USA 89, 4764]4768. Klimesch, W., Schimke, H., Schwaiger, J., 1994. Episodic and semantic memory: an analysis in the EEG theta and alpha band. Electroencephalogr. Clin. Neurophysiol. 91, 428]441. Kolev, V., Yordanova, J., 1997. Analysis of phase-locking is informative for studying event-related EEG activity. Biol. Cybern. 76, 229]235. Koshino, Y., Niedermeyer, E., 1975. Enhancement of Rolandic mu rhythm by pattern vision. Electroencephalogr. Clin. Neurophysiol. 38, 535]538. Kuhlman, W.N., 1978. Functional topography of the human mu rhythm. Electroencephalogr. Clin. Neurophysiol. 44, 83]93. Llinas, R.R., Ribary, U., 1992. Rostrocaudal scan in the human brain: a global characteristic of the 40-Hz response during sensory input. In: Bas¸ar, E., Bullock, T.H. ŽEds.., Induced Rhythms in the Brain. Birkhauser, Boston, p. ¨ 1992. Mesulam, M.M., 1990. Large scale neurocognitive networks and distributed processing for attention, language, and memory. Ann. Neurol. 28, 597]613. Mesulam, M.M., 1994. Neurocognitive networks and selectively distributed processing. Rev. Neurol. 150, 564]569. Miller, R., 1991. Cortico-hippocampal interplay and the representation of contexts in the brain. Springer, Berlin, Heidelberg. Mountcastle, V.B., 1992. Preface. In: Bas¸ar, E., Bullock, T.H. ŽEds.., Induced Rhythms in the Brain. Birkhauser, Boston, ¨ MA, pp. 217]231. Naatanen, R., 1992. Attention and Brain Function. Erlbaum, ¨¨ ¨ Hillsdale, NJ. Pantev, C., Makeig, S., Hoke, M., Galambos, R., Hampson, S., Gallen, C., 1991. Human auditory evoked gamma-band magnetic fields. Proc. Natl. Acad. Sci. USA 88, 8996]9000. Pfurtscheller, G., Neuper, C., Andrew, C., Edlinger, G., 1997. Hand and foot area mu rhythms. Int. J. Psychophysiol. 26, 121]135.

124


Polich, J., Kok, A., 1995. Cognitive and biological determinants of P300: an integrative review. Biol. Psychol. 41, 103]146. Pratt, H., Michalewski, H.J., Barrett, G., Starr, A., 1989. Brain potentials in a memory-scanning task. I. Modality and task effects on potentials to the probes. Electroencephalogr. Clin. Neurophysiol. 72, 407]421. Roschke, J., Mann, K., Riemann, D., Frank, C., Fell, J., 1995. ¨ Sequential analysis of the brain’s transfer properties during consecutive REM episodes. Electroencephalogr. Clin. Neurophysiol. 96, 390]397. Saermark, K., Mikkelsen, K.B., Bas¸ar, E., 1992. Magnetoencephalographic evidence for induced rhythms. In: Bas¸ar, E., Bullock, T.H. ŽEds.., Induced Rhythms in the Brain. Birkhauser, Boston, MA, pp. 129]145. ¨ Sanquist, Th.F., Rohrbaugh, J.W., Syndulko, K., Lindsley, D.B., 1980. Electrocortical signs of levels of processing: perceptual analysis and recognition memory. Psychophysiology 17, 568]576. Scheffers, M.K., Johnson Jr., R., 1994. Recognition memory and search for attended letters: An event-related potential analysis. J. Psychophysiol. 8, 328]347. Schurmann, M., Bas¸ar, E., 1994. Topography of alpha and ¨ theta oscillatory responses upon auditory and visual stimuli in humans. Biol. Cybern. 72, 161]174. Schurmann M., Bas¸ar-Eroglu ¨ ˘ C., Rahn E. et al., 1992a. A comparative study of alpha responses in human MEG temporo-parietal and occipital recordings and cat intracranial EEG recordings. Proceedings of the IEEE Symposium on Neuroscience and Technology, Lyon, pp. 132]137 Schurmann, M., Bas¸ar-Eroglu, ¨ ˘ C., Rahn, E. et al., 1992b. A comparative study of alpha responses in human MEG

temporo-parietal and occipital recordings and cat intracranial EEG recordings. Proceedings of IEEE Satellite Symposium on Neuroscience and Technology, pp. 132]137 Schurmann, M., Bas¸ar-Eroglu, ¨ ˘ C., Kolev, V., Bas¸ar, E., 1995. A new metric for analyzing single-trial event-related potentials ŽERPs. application to human visual P300 delta response. Neuro-Sci. Lett. 197, 167]170. Schurmann, M., Bas¸ar-Eroglu, ¨ ˘ C., Bas¸ar, E., 1997a. Gamma responses in the EEG: elementary signals with multiple functional correlates. Neuroreport 8, 531]534. Schurmann, M., Bas¸ar-Eroglu, ¨ ˘ C., Bas¸ar, E., 1997b. A possible role of evoked alpha in primary sensory processing: common properties of cat intracranial recordings and human EEG and MEG. Int. J. Psychophysiol. 26, 149]170. Schutt, ¨ A., Bas¸ar, E., 1992. The effects of acetylcholine dopamine and noradrenaline on the visceral ganglion of Helix Pomatia II: Stimulus evoked field potentials. Comp. Biochem. Physiol. 102C, 169]176. Silva, L.R., Amitai, Y., Connors, B.G., 1991. Intrinsic oscillations of neocortex generated by layer 5 pyramidal neurons. Science 251, 432]435. Solodovnikov, V.V., 1960. Introduction to the Statistical Dynamics of Automatic Control Systems. Dover, New York. Stryker, M.P., 1989. Is grandmother an oscillation? Nature 338, 297]298. Tiitinen, H., Sinkkonen, J., Reinikainen, K., Alho, K., ¨ Lavikainen, J., Naatanen, R., 1993. Selective attention en¨ ¨ ¨¨ ¨ hances the auditory 40-Hz transient response in humans. Nature 364, 59]60. Yordanova, Y., Kolev, V., Demiralp, T., 1997. The phase-locking of auditory gamma band responses in humans is sensitive to task processing. Neuroreport 8, 3999]4004.

Brain oscillations in perception and memory - CiteSeerX

Brain oscillations in perception and memory - CiteSeerX

Suggest Documents

Emotional Perception and Memory in Amnesia - CiteSeerX

Brain oscillations and predictive processing

Memory and Learning in Figure-Ground Perception - CiteSeerX

Neuronal oscillations and speech perception: critical ... - BioMedSearch

visual memory and perception - Tel

Memory Encoding and Retrieval in the Aging Brain - CiteSeerX

Memory systems in the brain and localization of a memory

Brain Oscillations and Cognitive Processes - CSJ Archive

Cortical oscillations in auditory perception and ... - Semantic Scholar

Brain oscillations and functional connectivity during ... - BioMedSearch

Working memory and the brain

Perception, memory and aesthetics of indeterminate art - CiteSeerX

Similarity in Perception: A Window to Brain Organization - CiteSeerX

In sync: gamma oscillations and emotional memory - ScienceOpen

Oscillations in working memory and neural binding: A ... - PLOS

Perception of Music and Dimensional Complexity of Brain ... - CiteSeerX

Probabilistic Models of the Brain: Perception and Neural ... - CiteSeerX

Composition of Brain Oscillations and Their Functions in ... - BM-Science

Spontaneous neural oscillations bias perception by modulating

NREM sleep oscillations and brain plasticity in aging

Object Perception, Attention, and Memory (OPAM)

Oscillations in Monterey Bay - CiteSeerX

Supplementary Information Memory and Perception-based Facial ...

PITCH PERCEPTION, WORKING MEMORY, AND ...