Do cortical and thalamic bioelectric oscillations have a functional role? A brief survey and discussion

Do cortical and thalamic bioelectric oscillations have a functional role? A brief survey and discussion* P Buser, A Rougeul-Buser Imtitcrr cles N~wu~ciencrs.

CNRS and UPMC.

(Received

I I March

URA 1498. 9. quoi Sr-Bmmd,

75005 Pari.\. Ftmcc

1995; accepted 21 April 1995)

Summary - This paper first briefly describes rhythmic bioelectric oscillations that can be recorded from the neocortex (particularly in cats). This descriptive section is followed by a review of the mechanisms whereby these rhythms are generated, and a section in which WC try to go beyond the purely correlative aspect and discuss the possible functional role of synchronized oscillations in thalamo-neocortical channels. Based on the literature data, it seems that two distinct and opposite roles can be attributed to these oscillations: either to lower the level of awareness. such as in slow wave sleep, when these oscillations are widespread on the neocortex; or to contribute to perceptual processing. when these rhythms are more localized and possibly of a higher frequency. Electrocortical (ECoG) rhythmical activities have been known and described since the early days of electrophysiological explorations of the human and animal brain. Surprisingly though. little importance has generally been attached to these oscillatory activities, except as indicators of sleep stages in animals and humans and, in the case of the alpha rhythms (the tirst rhythm to be discovered in the thirties by Hans Berger), of a certain state of ‘relaxation’ in humans. Despite the well-accepted fact that the human alpha rhythm evidently occurs during waking (even if it is in a ‘relaxed’ state), the idea ha5 curiously prevailed that the waking (‘aroused’) state in general is mostly characterized by low voltage fast (‘desynchronized’) activity. It is only more recently that some studies have suggested a functional importance of a variety of regular oscillatory activities that can be recorded from the neocortex and/or from a variety of thalamic nuclei not only in the sleeping but also on the waking. behaving animal. This review considers possible reasons indicating that these rhythmical activities are more than just epiphenomena. Our analysis is essentially restricted to the cat. with only a very quick glance at other species bioelectric oscillations

/ neocortex

/ thalamo-neocortical

channels

From alertness to drowsiness and sleep. A list of rhythmic activities in the cat It is now well established that several subsets of rhythms can be recorded from the feline neocortex (no further mention will be made of rhythms recorded from the olfactory system (see eg Bressler and Freeman, 1980)).

During the waking state, at least six, if not more, types of rhythms have been described. According to their frequency. they can be categorized into three classes (40 Hz band, I &15 Hz band and 4-S Hz band) to which we recently added a fourth class: rhythms at about 20 Hz. Their cortical localizations and behavioural conditions for occurrence have been amply documented now and will therefore only be briefly surveyed (see eg Buser, 1987; Chatila et al, 1992; Rougeul-Buser, 1994; Rougeul-Buser and Buser, 1995) for some recent data and reviews) (table I).

The features of these rhythms thus disagree with the traditional belief. Indeed, the electrocortical activity in a waking animal is not simply of the desynchronized, low voltage fast activity type. There were some interesting early findings supporting this view: what we call the ‘mu’ rhythm was previously described by Sterman and Wyrwicka (1967) as sensorimotor rhythms. Lopes da Silva and Storm Van Leeuwen (1978) and Lopes da Silva er al (1973. 1980) described a true alpha rhythm in the posterior part of the dog cortex. Our findings have been summarized in several papers (see e!: Rougeul-Buser and Buser, 1995).

Table 1. The major feature!, of the frequency, localization possible behavioural concomitants of these waking rhythms. 40 Hz

band

III motor

cortex

Beta

parietal

cortex

focused

III visual

cortex

JO HzV

(‘gamma

rhythms’

to other

author.\):

develop

focused

nttentlw

and posterior 40

Hz

20 Hz

band

band

somatic

cortex

rhythms;

CC+

Develop

IO Hz

band

in visual

cortex

5 Hz

band

tended rior

I” >tate of actl\e

mu rhythms: Alpha

in fairly

anterior

cortical

foci

WC-

and poste-

develop

of

and

in quiet

dewlop

and quiet

Develop cwzs

accordmg in ate

waking

‘~3pectnncy’

rhythms:

waking

“‘This article was presented at the symposium ‘Oscillations and paroxysomal activity in the nervous system, held at Rennes, 9-10 June. 1994

(rt

expectancy

or quiet

tex

in \late

attention

intense

and visual cortex 14 Hz band in somatic

dwlop

and

in drowc~new

of wthdrawal

the environment.

m qwet

expectancy and/or

other

of attention

from

250

Slow wave sleep During slow wave sleep, things are in a way simpler in that only two patterns are classically distinguished, delta rhythms (2-3 Hz) and sleep spindles (14 Hz). As opposed to the above rhythms that develop during the waking state, the delta activity is widespread over the cortex. Sleep spindles are just as widely distributed, although they are most dense in the most mesial part of the cortex. Paradoxical

sleep

Paradoxical sleep (PS) is classically characterized by a desynchronized activity, devoid of any overt rhythmic pattern. Recently we incidentally observed some sustained episodes of mu 14 Hz and alpha 10 Hz rhythms during what has been recognized by several students of paradoxical sleep as ‘atypical phases’. These phases are stages where the animal looks quite asleep, but has no rapid eye movements, despite ECoG signs of PS with desynchronized pattern, except for these (unexpected) mu and alpha episodes.

What could be the mechanism cal cortical activity ?

for rhythmi-

Before analysing the possible significance of rhythmic activities, it would be valuable to understand the mechanism(s) by which they develop. Several classes of hypotheses have been proposed over the last two decades. According to the most commonly accepted hypothesis, the thalamus is the site of origin of the cortical rhythms, the cortex being merely passive in this case (see eg Steriade et al, 1990). The strongest support for this opinion is that the isolated cortex has since long been recognized as not developing more than very atypical, slow oscillations that have little to do with the typical rhythms in the intact animal (this observation goes back to the early days of cortical explorations). Moreover, the existence of thalamic rhythms developing even in the absence of the cortex strongly reinforces this hypothesis. However, this view that the thalamus acts as a command zone for the cortical rhythms, cannot explain a very curious and contradictory observation: while there are heavy projections, from each sensory cortical area down to the thalamic nucleus projecting upon it, there is little evidence so far for a physiological control back from the cortex down to the specific relay nucleus. Most electrophysiological studies have failed to illustrate such descending effects (see eg Richard et al, 1975). There are however some rare exceptions. Schmielau and Singer (1977) reported an action of the

visual cortex on direction selectivity of cells in the lateral geniculate; Beaux et al (1992) also noticed some changes in firing patterns in the same nucleus, following application of GABA on the visual cortex (see however below). Now turning to the thalamus itself, and asking how its neurons can oscillate, the data are very complex. No single hypothesis can be proposed and indeed there are several distinct possibilities. i) That the various rhythms described in the waking animal depend on distinct thalamocortical channels was shown by several groups including ourselves. Macroelectrode and microelectrode explorations, combined with coherence computations, suggested that at least three or four of the cortical rhythms involve a particular nucleus of the thalamus. The mu 14 Hz rhythms have thus been reported to be present in the nucleus ventralis posterior, mainly in the forelimb projection zone. The parietal beta rhythms at 40 Hz were identified in another nucleus, belonging to the posterior group (nucleus posterior, pars medialis, Porn). Finally the alpha activity is accompanied by rhythms in a part of the lateral geniculate body, comprising layers A et Al. Pending confirmation, one is tempted to generalize and suggest that each of these cortical rhythms represents an oscillatory pattern occurring in a given thalamocortical channel: one or another channel may thus start oscillating, depending on the behavioural condition(s). ii) If the thalamus is the source of oscillatory neuronal activity, what are the mechanisms underlying this rhythmicity? Currently, there are three possibilities and two further questions. a) The first hypothesis contends that all thalamic cells may be intrinsically capable of oscillating spontaneously. This was mainly demonstrated by the now classical findings of Jahnsen and Llinas (1984a, b) and confirmed by many others since, who enriched this basic idea with many more details, ending up on a mechanism based on the subtle interplay of a variety of ion channels, for Na’, or for K’ and still others for Ca”. However, these data were almost entirely obtained in vitro (see eg McCormick and Pape, 1990). b) Another hypothesis, on a different scale, results from our in vivo explorations of the nucleus ventralis posterior during the somatic cortex mu activities. While confirming the correlation between thalamic unit firing and cortical rhythms by microelectrode exploration, a curious finding came up. Most of the rhythmic thalamic cells that we isolated in the VP (10% of the total population) were identified as long-axoned thalamocortical cells (as assessed by the classical antidromic stimulation and collision test). However, we were unable to induce their firing through any of the usual tactile or

251

deep somatic stimuli. We therefore suggested that nucleus VP possibly hosts a subset of thalamo-cortical cells different from the classical thalamo-cortical relay cells which do not display rhythmicity. This putative dual system, with rhythmic cells and relay cells with different properties, awaits confirmation. It could be that we were dealing with cells that, for an unknown reason, had at the moment of the exploration, changed their level of polarization and passed from what Jahnsen and Llinas ( 1984a, b) called the ‘transfer mode’ to the ‘oscillatory mode’ and at least provisionally ceased to act as thalamo-cortical relay cells. The alternative hypothesis is of course that there are two subsets of cells with fundamentally different properties (also see the interesting overview by Sherman and Koch ( 1986) on transmission in the lateral geniculate nucleus). c) Both hypotheses (a and b above) consider that a nucleus - for example VP or LG - contains longaxoned neurons (all of them in hypothesis a, only some in hypothesis b) that ‘spontaneously oscillate’. A third way of explaining rhythmicity came up when it was suggested (see Steriade and Deschenes, 1984; Steriade and Llinas 1988 for the literature on this point) that the basis for oscillation was a loop formed through a very peculiar thalamic nucleus, the nucleus reticularis (RET). It is now well established that the RET does not project up to the cortex (unlike all other thalamic nuclei), but sends its axons back into all the thalamic nuclei, in an orderly way, except in the case of the antero-ventral nucleus (see Scheibel et al, 1973; Steriade and Llinas, 1988). Since each of these reticular neurons receives excitatory collaterals from the ascending thalamo-cortical axons, and since these neurons are now known to be GABAergic, the idea was raised that the oscillatory mode was due to the action of this feedback loop comprising an inhibitory link (see Steriade and Llinas 1988 for further details, and a more recent paper by Krosigk et al, 1993). This hypothesis has led to much debate. We tested the possible role of the n reticularis in two distinct situations: i) during the parietal beta rhythms; and ii) during sleep spindles (Canu and Rougeul, 1992; Canu et al, 1994). In accordance with the reticular hypothesis, we confirmed the participation of this nucleus in sleep spindles; however, we were unable to find reticular cells participating in the beta rhythms. We suggest therefore that localized rhythms, such as beta and maybe also mu or alpha. only involve particular channels, composed of a given thalamic nucleus and its cortical projection area, whereas more widespread activities such as sleep patterns may indeed involve the reticular nucleus. In their general view Crick (1984) and Crick and Koch (1990) suggested that sectors of the nucleus

reticularis could govern and modulate responsiveness and information processing in thalamo-cortical channels, providing a physiological substrate for the ancient and now rejuvenated ‘spot-light’ hypothesis of attention and consciousness. However, since we were able to show that localized rhythms such as beta or mu develop in given attentional situations, we cannot really agree with the suggestion of a role for the n reticularis in attentive processes. d) Oscillation vs synchronization: one or two distinct mechanisms? Most authors interested in this general problem considered the mechanisms for oscillations and there has been little analysis of the other aspect, that of synchronization. Evidently, if cells were oscillating individually, with no overall synchrony, no rhythmic local field potentials would be recorded at the thalamic or cortical level with gross electrodes. There must presumably be another mechanism. responsible for setting all individual firings in phase. No such mechanism has been identified yet. It may well be that there are collaterals between neurons, or that neurons that oscillate are interconnected through intemeurons. which are numerous at least in some nuclei (this is precisely the case for VP and LG in cat). The synchronization mechanism can however not be that suggested by Andersen et al (1964), based on the action of recurrent collaterals from the ascending axons, back to the inhibitory intemeuron(s). It provided at the same time a basis both for rhythmicity and for synchronization, through postulating that the inhibitory intemeurons were very few and heavily divergent, acting upon a large number of thalamo-cortical relay cells and thus automatically producing synchrony. The existence of such collaterals was investigated and finally rejected (particularly in case of the VP; some rare collaterals may exist in the GL). There is a need for another type of circuitry to account for coordination. In case of the sleep spindles, which display synchrony over widely separated areas in the cortex (and in the thalamus), the nucleus reticularis might well act both as a pacemaker and as a coordinator for many cell ensembles (Steriade and Llinas, 1988). As for the focal, discrete rhythms, for which we have no evidence, at the present time, of their dependence upon the reticular nucleus, there may presumably be other, as yet unidentified mechanisms. Whether local intemeurons (which are in most cases GABAergic, hence inhibitory) or other, less classical mechanisms are involved, such as gap connections (which do not seem to be numerous in the thalamus) remains a matter for future investigation. At any rate, a kind of ‘hyperlocal circuit-hypothesis’ should for the time being not be completely rejected (a more extensive discussion on this point may for instance be found in Buser, 1987).

252

e) Should we then definitely reject the idea that a corticothalamic feed-back contributes to the thalamoneocortical rhythmicity? A very long time ago, this became the ‘thalamo-cortico-thalamic’ reverberating circuit scheme, developed by the first generation of ‘neurocybemeticians’. It was very rapidly disproved and therefore consigned to oblivion. The situation has however recently swung back, due to some new and perhaps more consistent findings, indicating that the visual cortex appears to have a major action down onto the lateral geniculate nucleus, which may generate thalamic oscillations (Funke and Eyse, 1992; MC Cormick and Krosigk, 1992; Krosigk et al. 1993; Sillito et al, 1994).

Possible functional cal rhythms.

‘role’ of the thalamocorti-

Can we or should we attribute some functional role to this particular type of bioelectrical phenomenology? It could well be that the synchronized activities on the cortex and elsewhere are only epiphenomena with no influence on cognitive operations, such as perception, preparation to action, conscious behaviour or mental imagery (if any in the cat but why not?). The tendency to ‘deny importance to the brain waves’ is probably the most current attitude nowadays. Together with some other authors, we shall here adopt an alternative viewpoint, proposing that rhythmic neuronal firing and synchronization may on the contrary strongly influence the operational capabilities of the thalamo-cortical channel in which they develop. There are two extremely opposed situations to consider when trying to determine which ‘function’ may thus be fulfilled by synchronous oscillations. First, one in which there is a loss of awareness, which is the case during slow wave sleep; another one, in which a perceptual processing may, according to some authors, be dependent upon fast 40 Hz rhythms, at least in the visual cortex. Starting from these extreme cases, we can also speculate as to the role of other rhythms that also occur in the waking state. From drowsiness

to sleep

A certain class of synchronization clearly accompanies losses of awareness, when slow wave sleep sets in. This simultaneity may be purely correlative: delta and spindles although being symptomatologically very useful, may be purely epiphenomena. However, one may on the contrary hypothesize, as some authors have suggested, that ‘spindling’ and more generally regular

membrane potential oscillations in a neuron are not compatible with processing of the incoming information, or the organisation of an adequate output. Llinas developed this hypothesis, while showing that a given thalamic cell may undergo two distinct firing modes, depending on its level of polarization: it could either oscillate ‘spontaneously’, or transfer messages to the cortex. In other words, the first mode, that of oscillating, was not compatible with the information processing task which in this case, was to pass the peripheral sensory messages onto the cortical end-station. Drowsiness may be a similar, although less marked situation. Processing of the environmental information and messages could be impaired or at least diminished during this stage, while some sort of internal ‘selfthinking’ may still be operational, thus accounting for the ‘withdrawal of attention from the surrounding’. Of course, there is a strong argument against these views of slow sleep: paradoxical or REM sleep, is accompanied by fast low voltage activity with no sign of rhythmicity nor synchronization of individual rhythms (except for the mu and alpha episodes during ‘atypical’ phases, see above). The mystery remains complete: on the one hand, that loss of awareness when sleep sets in and that synchronized spindles simultaneously occur is undisputable. On the other hand, there is a very atypical sleep, ill defined in its conscious content except that we know that dreams have a high probability to occur during its course. The opposite reasoning can however be followed, to postulate that it is because there is no synchronization that this sleep stage represents a very peculiar stage of ‘awareness’, with a very high brain activity leading to dreaming, combined with motor inhibition (see eg Sakai, 1988). The other side of the coin: role of fast rhythms in the perceptual ‘binding’ process Recent studies have led to a quite different view of a particular category of rhythmic activities, at least at the level of the visual cortex. This category is in the 40 Hz frequency band (that some authors designate as belonging to the ‘gamma band’ and that we prefer to call 40 HzV). The problem is that of ‘binding’, a very old one indeed, with its roots in the Gestalttheorists (Kohler, 1930; Koffka, 1935). In their view, object recognition required that a visual scene be parsed and then coherence between elementary features be evaluated to permit discrimination of different objects within the scene. Feature detection thus first involves some segmentation of the viewed shape, followed by a perceptual grouping, currently known as ‘binding’ between the distinct parts to constitute the global form (Treisman, 1986).

253

More recent, now amply popularized data from several groups have suggested the existence of parallel processing of percepts in the primate visual system, with some areas devoted to form analysis, another to colour, another to movement (Z&i, 1978, 1989). How then could these various segments be reassembled into the original object with its diverse features? The problem once posed by the gestaltists now has a modem expression. E&horn et al (1988), Gray et al (1989) and Engel et al ( 1991), recently proposed that the physiological mechanism for reassembling the features could precisely be the cortical bursts at 40 Hz elicited by a visual stimulus. These activities would provide a time-space synchronization of the diverse unit activities, thus unified into a single feature. This contention currently has a very strong impact in the community. Many other data have now confirmed that oscillations in the 40 Hz band may accompany cognitive operations, such as voluntary movement in macaques (Murthy and Fetz, 1992) or auditory conscious perception in human (Pantev et al, 1991). Llinas er al (1994) have reported similar views, based upon MEG (magnetoencephalographic) explorations in man, illustrating the possible general role of the 40 Hz ‘thalamo-cortical resonance’ in generating cognition. A very interesting finding was that a significant 40 Hz rhythm also develops during REM sleep, but in this case. the ‘thalamo-neocortical system is not accessible to sensory input, but is hyperattentive to intrinsic activity, in which sensory input has no access to the machinery that generates conscious experience’. This sheds a new and original light on the difference between wakefulness and REM sleep (also see above). Quite recently, Bressler et a2 (1993) have shown that in the behaving monkey, there are increases in broad band coherences (mainly in the low-frequency range) between distant cortical structures which are not necessarily interconnected anatomically. Such coherences reveal functional connections. The enthusiasm for the ‘magic 40 Hz’ rhythm should however be viewed rigourously. Recently Fregnac et al (1994) very carefully analyzed the ‘binding’ hypothesis. They concluded from their intracellular explorations in kitten area 17, that “the conceptual frame according to which fast brain rhythms are involved in temporal coding and cognition is not supported by their observation. These activities which range between 7 and 20 Hz are propagated through diffuse reciprocal circuits involving cortical and/or thalamo-cortical networks”. The view that any electrobiological oscillation in the brain contributes to the perceptual and cognitive mechanisms should be tempered. The more so because in our opinion, neuronal synchronisation does not subtend

a single state and mechanism, but two opposite processes: loss of awareness and perceptual facilitation. In the first case, a large, diffuse synchronization involving much of the thalamo-neocortical areas may impede information processing; in the second case, pending confirmation, more localized rhythms may in contrast, be of fundamental importance to it.

Can we speculate further about the functional role of other rhythmic neocortical activities? The above data can be considered as extreme cases, with loss of consciousness in one case, and possibly perceptual processing in the other. Can we go a step further and try to propose ‘functions’ to other rhythms such as the anterior motor-parietal 40 Hz (beta), or the mu and ‘alpha’ or ‘alpha-lie’ expectancy activities ? The common factor being immobility, it cannot as such account for two distinct sets of frequencies. Other explanations must be found. The motor-parietal beta rhythms were shown to develop when the animal is in a state of motionless, focused attention. These rhythms may be similar to those presumably achieving temporospatial coding in the visual domain. Possibly, watching intensely a prey leads to a kind of mental preparation of a positive (motor) action toward it, with the task requiring organization and coordination of the successive visually oriented movements to be directed toward the target. The ‘mu’ and ‘alpha’ activities are certainly still more difficult to interpret. They were described as developing also while no movement was performed, but as the animal was ‘quietly’ expecting the appearance of a target and, in its presumed visual imagery, planning to move or not to move, depending on the environmental conditions. Conceivably, alpha and mu rhythms appear in response to the presence of a global motor programme which will have to be triggered at a given moment in the future: this programme may be ‘held in stand-by’ or ‘withheld’ for the duration of expectancy, a stage during which the animal may only have a global mental representation of the target which will have to be reached as soon as it will appear.

References Andersen P, Eccles J, Sears T (1964) The ventrobasal complex of the thalamus: types of cells, their responses and their functional organisation. J Physiol 174, 37S399 Beaux JC. Lestienne R, Imbert M, Grandjean B (1992) Modulations corticales de la structure temporelle fine des trains d’impulsions dans le corps genouillk lateral dorsal du chat. CR Acud Sci Paris 314, 31-36

254

Bressler S, Freeman W (1980) Frequency analysis of olfactory system EEG in cat, rabbit and rat. Electroenceph Clin Neurophvsiol 50, 19-24 Bressler S, Coppola R, Nakamura R. (1993) Episodic multiregional cortical coherence at multiple frequencies during visual task performance. Narure 366, 153- 156 Buser P (1987) Thalamocortical mechanisms underlying synchronized EEG activity. In: A Textbook of Clinical Neurophysiologv (Halliday A, Botler S. Paul R. eds) John Wiley & Sons Ltd, New York, 595-621 Canu MH, Rougeul A (I 992) Nucleus reticularis thalami participates in sleep spindles, not in beta rhythms concomitant with attention in cat. CR Acad Sci Paris 315, 513-520 Canu MH, Buser P, Rougeul A (1994) Relationship between posterior thalamic nucleus (Porn) and unit activity and parietal cortical rhythms (beta) in the waking cat. Neuroscience 60, 679688 Chatila M, Milleret C, Buser P. Rougeul A (1992) A 10 Hz ‘alphalike‘ rhythm in the visual cortex of the waking cat. Electroerrceph Clin Neurophpiol

83, 217-222

Crick F (1984) Function of the thalamic reticular complex: the search-light hypothesis. Prbc N&l Acnd Sci (ISA 81, 4586-4590 Crick E Koch C (1990) Towards a neurobiological theory of consciousness. Srtnin Neurosci 2, 263-275 Eckhom R, Bauer R. Jordan W, Brosch M, Kruse W, Munk M, Reiboek H (1988) Coherent oscillations: A mechanism for feature linking in the visual cortex? Eiol Cybem 60. 121-130 Engel A, Kiinig P, Singer W (1991) Direct physiological evidence for scene segmentation by temporal coding. Proc Not/ Acad Sci USA 88, 9136-9140 Fregnac Y, Bringuier V! Baranyi A (1994) Oscillatory neuronal activity in visual cortex: A critical reevaluation. hi: Temporal Coding in thr Brain (Buzsaki G. Llinas R, Singer W, Berthoc A, Christen Y, eds) Springer Verlag. Berlin, 81-102 Funke K, Eyse UT (1992) EEG-dependent modulation of response dynamics of cat dLGN relay cells and the contribution of corticogeniculate feedback. Brain Re.r 573, 2 17-227 Gray C, Konig P Engel A, Singer W ( 1989) Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Narirre 338. 334-337 Jahnsen H, Llinas R (1984a) Ionic basis for the electroresponsiveness and oscillatory properties of guinea-pig thalamic neurones bt virro. J Ph~siol 349. 227-247 Jahnsen H, Llinas R (1984b) Electrophysiological properties of guinea-pig thalamic neurones: an in vitro study. J PIIwIoI 349. 205226

Koffka K (1935) Principles of Gemtlt P,swho/o,q~. Routledge, Kegan Paul, London Kohler W ( 1930) Ge.stcr/t Ps~cl~olo~~. Bells and Sons. London Krosigk von M. Bal T. McCormick D ( 1993) Cellular mechanisms of a synchronized oscillation in the thalamus. Scirnce 261 1361364 Llinas R, Ribary U, Joliot M. Wang X ( 1994) Content and context in temporal thalamocortical binding. hi: Tempmd Coding in the Eroin (Buzsaki G. Llinas R. Singer W, Berthoz A. Christen Y. eds) Springer Verlag, Berlin, 251-272 Lopes da Silva F. Storm van Leeuwen W (1978) The cortical alpha rhythm in dog: the depth and surface profile of phase. In: Architectonics effhe Cerebral Corte~x. (Brazier MAB. Petsche H, eds) Raven Press, New York, 319-333 Lopes da Silva F, Van Lierop T. Schrijer C, Storm van Leeuwen W (1973) Organization of thalamic and cortical alpha rhythm:

spectra

and coherences.

Electroenceph

Cliu

Neurophysiol

35,

627439

Lopes da Silva F, Vos J, Mooibroek J, Van Rotterdam A (1980) Partial coherence analysis of thalamic and cortical alpha rhythms in dog. A contribution towards a general model of the cortical organization of rhythmic activity. In: Rhythmic EEG uctivities nnd Cortical Fmcrioning (Pfurtscheller G, Buser P, Lopes da Silva F. Petsche H, eds) Elsevier/North Holland, Amsterdam. 33-59 McCormick DA. Pape H (1990) Properties of a hyperpolarizationactivated cation current and its role in rhythmic oscillation in thalamic relay neurons. J Phvsiol 431. 291-318 McCormick DA. Krosigk M (1992) Corticothalamic activation modulates thalamic firing through glutamate metabotropic receptors, Pm Nuti Acad Sci USA 89, 2774-2778 Murthy V, Fetz E (1992) Coherent 25- to 35.Hz oscillations in the sensorimotor cortex of awake behaving monkeys. Pmc NatI Acad Sri USA 89, 5670-5674

Pantev C, Makeig S, Hoke M, Galambos R, Hampson S, Gallen C (1991) Human auditory evoked gamma-band magnetic fields. Pmc Nafl Acud Sci C/SA 88, 8996-9000 Richard D, Gioanni Y. Kitsikis A. Buser P (1975) A study of geniculate unit activity during cryogenic blockade of the primary visual cortex in the cat. Exp Brain Res 22, 235-242 Rougeul-Buser A, Buser P (1995) Electrocortical rhythms in the attentive cat: phenomenological data and theoretical issues, Irr: Oscillatoy events related brain dynamics (Pantev C, Elbert T, Lutkenhoner B, eds) Plenum Press, New York, in press Rouge&Buser A (I 994) Electrocortical rhythms in the 40 Hz band in cat. In search of their behavioural correlates. In: Trntporcrf Coding in the Brain (Buzsaki G, Llinas R, Singer W. Berthoz A, Christen Y, eds) Springer Verlag. Berlin, 103-l 14 Sakai K (1988) Executive mechanisms of paradoxical sleep. Arch 1tal Biol 479. 225-240

Scheibel M, Davies T, Scheibel A (1973) On thalamic substrates of cortical synchrony. Newo[ogy 23, 300-304 Schmielau F, Singer W C1977) The role of visual cortex for binocular interactions in the cat lateral geniculate nucleus. Brain Res IX. 354-36

I

Sherman S, Koch C (1986) The control of retinogeniculate transmission in the mammalian lateral geniculate nucleus. E.rp Brain Res 63. I-20 Sillito A. Jones H. Gerstein G, West D (1994) Feature-linked synchronisation of thalamic relay cell tiring induced by feedback from the visual cortex. Nature 369, 479482 Steriade M. DeschCnes M (1984) The thalamus as a neuronal ohcillator. Brain Res REV 8. l-63 Steriade M, Llinas R (1988)The functional states of the thalamus and the associated neuronal interplay. Phyiol Rev 68. 649-742 Steriade M, Gloor P, Llinas R, Lopes da Silva F, Mesulam M (1990) Basic mechanisms of cerebral rhythmic activities, E[rcfroenceph C/in Neurophysiol

76, 481-508

Sterman M. Wyrwicka W (1967) EEG correlates of sleep: evidence for separate substrates. Brain Res 6. 143-l 63 Treisman A (1986) In: Hundbook of Perception and Human performance (Boff K. Kaufman L. Thomas I, eds) Wiley, New York, 35. I-35.70 Zeki S (1978) Functional specialization in the visual cortex of the rhesus monkey. N&are 274, 423-428 Zeki S (1989) Colour vision and functional specialization in the visual cortex. Discussion Neurosci 6, 2. Elsevier Science Publisher. Amsterdam