thoughts on the meaning of brain electrical activity t-208

8/13/2019 Thoughts on the Meaning of Brain Electrical Activity T-208

1/13


2/13

j l9

? : 1 csh re l l b ch n 1 csh re l l t ch n c s h r e l l e chn


3/13

2a WR:116 00 2b WR: 109 31

Relaxation TOTAL Counting TOTAL

2c WR: 107 39

elaxation ALPHA

2d WR: 1 17 20

Counting ALPHA

28 WR: 134 33 2f WR: 156 42

Relaxation BETA Counting BETA

FIG. 2 a-f). Scalar represent tions of recrudescence In total alpha and betaEEC d u r ~ n gestlng and cou nt ~n g ackwards


4/13

216 KARL H PRIBRAM

Here is the dilemma. Nerve impulses are trans-mitted over definite, restricted paths in the sen-sory and motor nerves and in the central nervoussystem from cell to cell through definite inter-cellular connections. Yet all behavior seems tobe determined by masses of excitation, by theform or relations or proportions of excitationwithin general fields of activity, without regardto particular nerve cells. It is the pattern and notthe element that counts. What sort of nervousorganization might be capable of responding toa pattern of excitation without limited specia-lized paths of conduction? The problem is almostuniversal in the activities of the nervous systemand some hypothesis is needed to direct furtherresearch.

For almost half a century we have known that

the EEG recorded from the scalp does not somuch reflect an accumulation of nerve impulsesas it reflects the graded polarizations (hyper-and depolarizations) of synapses and the fine-fibred axonic and dendritic cortical connectionweb (e.g. Adey, 1967; Creutzfeld, 1961;Green, Maxwell, Petsche, 1961; GreenPetsche, 1961; Li, Cullen, Jasper, 1956; Ver-zeano Laufer, 1970; Verzeano Negishi,1960)'. It is in these graded polarizations thatpatterns of excitations (and inhibition) need to besought.

GRADED POLARIZATIONS IN THEBRAIN S CONNECTION WEB

However, it is nerve impulses that we can readilyrecord with microelectrodes, nerve impulses thatare generated in order to inform one part of thenervous system what is going on in another part.This has given rise to viewing the functions of thebrain and especially those of the cerebral cortex interms of its circuitry. Though circuitry is certainly

an important aspect of brain function, by them-selves circuits cannot account for the psychologi-cal processes emphasized by Lashley in the earlierquotation.

As a metaphor, the old vacuum tube servesadmirably. The tube's circuitry has interposedwithin it a plate. Minute changes in the chargeof the plate provide patterns in an otherwise stabletransmission of energy. These patterns, these

The situation is much the same as it is with the

electroretinogram, where the alpha and beta waves reflectretinal activity that is entirely devoid of nerve impulses.

designs, constitute the information processingcapabilities of early computers.

George Bishop (1 956), in a definitive essay thatdiscussed the natural history of the nerveimpulse , reviewed the evidence for attending

another aspect of neural activity, an aspect analo-gous to that served by the plate in the vacuumtube. Bishop indicated that graded slow poten-tials, hyper- and depolarizations, are more generalas well as more primitive than the impulses thatprobably developed when the early metazoansbecame too large. He cites the evidence that thecerebral cortex still operates largely by means ofconnections characteristic of primitive neuropil,the most appropriate mechanism for the produc-tion of a state, as contrasted to the transmissionabout such states. On the basis of such evidence, I

suggested (Pribram, 197 1, p. 105) that:

Nerve impulses and graded potentials are twokinds of processes that can function reciprocally.

simple hypothesis would state that the lessefficient the processing of synaptic arrival pat-terns into axonic departure patterns, the longerthe duration of the designs of the graded dendri-tic microprocess.-In short, nerve impulsesarriving at synapses generate a graded potentialdendritic microprocess. The design of this micro-process interacts with that already present by

virtue of the spontaneous activity of the nervoussystem and its previous 'experience.' The inter-action is modulated by inhibitory process and thewhole procedure produces effects akin to theinterference patterns resulting from the interac-tion of simultaneously occurring wave fronts.The graded potential processes thus act as crosscorrelation devices to produce new figures fromwhich the patterns of departure of nerve impulsesare initiated. The rapidly paced changes i nawareness could well reflect the duration of thecorrelation process.

One way of portraying the two different modesof operation of the brain cortex is to record, fromthe same microelectrode, with low and with highpass amplification. We find (Fig. 3) that the slowgraded activity recorded with low pass filteringprecedes the high pass filtered spiking recordedfrom the same electrode 75% of the time duringsensory stimulation (in 2369 recordings). Thisindicates that, just as for intracellular recordingsfrom axons, that depolarizations and hyperpolari-zations precede the generation of action potentials(spikes). Thus, as in the case of the vacuum tubeplate, the graded activity programmes the


5/13

TH O U G H TS O N T H E MEA N IN G O F BRA IN ELECTRICA L A CTIV ITY 217

Spikes , envelopeof spikes, an d slow potential

800 205 210 215 220 225 230 235 240 245 250

time ms)

FIG 3 An example of the relationship between spikes and multi-unit bursts to coincident local f ieldpotentials. Note that the ascending slope of the field potentials precedes that of the spikes and bursts. Ifthe field potentials were a consequence of the burst, the peak should coincide with or come later than themaximum number of bursts.

output of the axon from which recordings aremade. The graded activity intervenes betweenthe input and the output relation of the neuron.

The evidence available at that time, albeitindirect, was reviewed to support the hypothesisthat the graded processes recorded with low passfiltering, presumably occurring in dendrites, arecoordinated with awareness. This evidence didnot, however, include the displays of designs ofdendritic processes that are the critical underpin-nings of the hypothesis.

THE CONFIGURATION OFDENDRlBlC FIELDS A SPECTRAL

DOMAlN

Such displays are readily provided, thanks toKuffler (1953), who devised a technique thatmaps the configuration of dendritic activity frommicroelectrode recordings of nerve impulses fromthe axons connected to those dendrites. Kuffler

applied to microelectrode recordings from theoptic nerve the clinical technique of mapping

the visual field of a subject. Instead of obtaininga verbal o r instrumental respon se from the subject,Kuffler obtained the response from a single axon:A dramatic increase (or decrease) from baselinespontaneous activity in the number of nerveimpulses. The visual field of that axon isdescribed by the area in the visual environmentover which a stimulus is registered by theresponse of the axon. David Hubel, in a lectureat Stanford University, pointed out that such avisual receptive field actually represen ts thefunctional dendritic field of that axon under theconditions of' visual stimulation.

During the 1960s and 1970s many laboratories,including mine, mapped these functional dendriticfields in the visual system. Their shapes changefrom concentric (bull 's eye) to elongated as oneproceeds from optic nerve to cortex. Mathemati-cal descriptions and computer simulations can befitted to these shapes (for review see, for instancePribram, 1991, Lectu res 1,2, 4 5 . Here I present

similar shapes obtained from recordings madefrom axons in the somatosensory cortex of rats


6/13

0.4 0.6 0.8 1 1.2T F W

U

S. F. Flicks/R ) 0 T F RPS) 0 2 0.4 0.6 0.8 1T F R/S )

FIG. 4 a-f). Exam ples of receptive filled manifolds and their associated contour maps derived by an interpolation spline)procedure from recorded whisker stimulation. The contour map was abstracted from the manifold by plotting contou rs in termsof equ al numbers of bursts per recording interval IOOsec). Each figure sho ws baseline activity no whisker stimu lation ) at a givenelectrode location as a gr-plane located in terms of numb er of bursts per 100sec.


7/13

THOUGHTS ON THE MEANING O F BRAIN ELECTRICAL ACTIVITY 219

(see Fig. 4 . There are several characteristics ofthe shape of these maps that are important to thehypothesis under consideration. First, these mapsrepresent whisker stimulation in the spectral( pure frequency) domain, since variations ofthe stimulus are made by changing the spatialfrequency (spacings of grooves) of cylindricalgratings and the temporal frequency (speed ofrotation) with which the whiskers are engaged.The spectral representation emerges as a

density of whisker stimulation , which enfold sboth the spatial and the temporal aspects ofthe stimulation. As a corollary, in such arepresentation information about space andtime becomes distributed over the reach of themapping.

Second, the func tional dendritic field can thus

be mapped as a surface distribution. One way tolook at surface distributions is to consider themto be produced by nodes of interference amongwave forms with different origins. The dendriticconnection web receives inputs from arboursoriginating in a variety of axons. As suggestedby Eccles (1958), these inputs can be conceivedas forming wavefronts. But this conception isnot critical. When we additionally mapped theorientation of the spatial frequency grating (theorientation of the grooves), we were able toplot neuronal population vectors based on

regression to a sinusoid. However, these vec-tors capture only the most obvious aspects ofthe surface distributions; all of its subtleties areignored.

THE TEMPOR L WOLD

What d oes this concern with surface distributionsthat map a spectral domain gain us? Llinas, at thisconference, and several other laboratories (e.g.Singer, 1993) have tackled the so-called bindingproblem by indicating coherence among neural

oscillators (oscillation indicated by raster plots,see Fig. 5). Periodicity in the density distributionof nerve impulses (recorded from single axons)can be shown to be produced either by sensoryinput, or in Llinas' hands by input from the intra-laminar nuclei of the dorsal thalamus. Theseinvestigators have emphasized the temporaldimension of their findings, but, as came out inthe discussion of Llinas' results at this confer-ence , it is the temporal binding of se parate spatiallocations that is at stake.

Elsewhere (Pribram, 1966, p. 179), I suggestedthat such a temporal hold is necessary to:

the flexible rearrangement of memory pro-cesses. This temporal hold is assumed to beaccomplished through an operation similar tothat which gives rise to a temporary dominantfocus in the experiments of Zal'manson workingwith Ukhtomski (1927). (In these experiments

they applied a patty of filter paper soaked instrychnine to an appropriate location on themotor cortex which resulted in a shift of aconditional reflex from one leg to another.)Without regu lation by such a hold mechanism,the organism fluctuates inordinately among tem-poral signals and thus produces only a jumbleof arrival patterns.

The C ontingent Negative Variation (CNV ), theexpectancy wave of the EEG, was suggested torepresent su ch a temporal hold . Initially this nega-

tive variation was recorded deep to the frontallobe of the brain, which suggested a basal gang-lion or limbic origin. Of interest in the light ofLlinas' results obtained from stimulation of theintralaminar goup of nuclei is that, with the col-laboration of a medical student (McKegney,19-58 , I had traced, by way of the retrogradedegeneration technique, connections from theintralaminar complex to the perirhinal cortex (aswell as to the basal ganglia), an area currentlyheld responsible for many of the devastatingeffects on memory resulting from resections of

the medial portion of the temporal lobe2.Binding by way of a temporal hold that pro-duces coherence can readily be calculated in thespace-time domain provided there are onlya fewoscillators involved. When coherence must becalculated over many locations as is done in com -puterized tomography (CAT and PET scans andfMRI) and in the presentation at this conference

The in tralaminar complex becomes more prominentas one proceeds from rat to cat to monkey to man. Also,this complex divides the thalamus into two divisions: Oneprojects to the convexal cortex to include the posteriorpart of the frontal cortex and the parietal, occipital, andtemporal regions. This division maintains an anterior-posterior axis, which reflects the anterior-posterioraxisof the cortex.By contrast, the other, moremedial division,which proje ts to the far frontal (prefrontal)and cingulatecortex, losesthis anterior-posterior topological correspon-dence. This difference in thalamocortical organization isone reason for classifying the far frontal cortex with thelimbic formation. (The other is that resections of the farfrontal and l imbic cortices produce deficits in delayedalternation behaviour n monkeys whereas resectionsofthe convexal cortical regions do not; the reverse is true forsimple sensory discriminations.) For review seePribram(1954, 1958).


8/13

jun19, Stimulation Rate=4 Hz, Groove W idth= 0.8mm

006 0 degree1001 .

007 30 degree

50 100 150 200 250 50 100 150 200 250TIME ms) TIME ms)

012 60 degree 013 90 degree

50 100 150 200 250 50 100 150 200 250TIME ms) TIME ms)

018 120 degree 019 150degree

50 100 150 200 250TlME ms)

50 100 150 200 250TlME ms)

FIG. 5 Raster plots showing periodicity o f single-neuron firing as a function of a s wee p of a rat s whiskers across an 8m mgrating. The sw eep was initiated by electrical stimulation (stimulation n umbe r) of the 5th cranial nerve at the rate of 4Hz. Although

this is not readily seen in this figure, the neural periodicity isor identical with the sweep rate, especially when rates of8 or 12Hzwere used. Note that the periodic burst occurs only within the first 50msec of each sweep.


9/13

THOUGHTS ON THE MEANINGOF BRAIN ELECTRICAL ACTIVITY 221

by Petsche for determining different patterns ofthe EEG that characterize listening to differentmusical compositions, the computations arefacilitated considerably when they are carriedout in the spectral domain (by using anFFT or

similar algorithm). My suggestion, in keepingwith my hypothesis, is that if it is easier for thecomputer to perform the algorithm in this domain,i t is likely that it is easier for the brain to do it thatway.

SCAN PATHS

The analogy with tomography raises an issue thatneeds to be addressed: We do not perceive in thespectral domain. How is the inverse transformimplemented, so that processing can shift back

from the spectral to the space-time domain? Thespectral domain is time symmetric. What isneeded is a mechanism that breaks time symme-try. Prigogine (e.g. 1994) has noted that undercertain conditions, the time symmetry inherentin the spectral domain can be broken. He hassuggested that under these conditions scattering(dissipation of energy) is persistent rather thantransitory; that is, the system is an open, not aclosed one. Open systems provide opportunitiesfor a range of possibilities for different futures3.Whether Prigogine's formulation is the correctone or not, the important point of his programmeis that open systems are open to possibilities.When a path is taken, a choice is made amongpossibilities (e.g. by the formation of an attractor)such that the path not taken is forever lost, andtime symmetry is broken.

This emphasis on paths brings Prigogine'sformulation regarding the breaking of time sym-

Mathematically, Prigogine's discussion concernscertain generalized quantum andlor classical systems dri-ven by (non-self-adjoined) Hamiltonian operators (forquantum systems) andlor Liouville operators (for classi-cal systems) which are chosen so that their t imedevelopments are kept contractive (i .e. lose information)and dissipative (i .e. lose energy). My interpretation (Pri-bram, 1994) of his equations is that damping term seliminate the imaginary component of non-square integr-able Eigen functions due to their evolution operator (e.g.a Hamiltonian). Thus, undertaking a path , choosing ,making implicit or e xplicit directed movement-whetheras attention to input, as intending an action or as thought(i .e. rumm aging through memory)-breaks time symm e-try. The path taken loses information or better stated,loses uncertainty in Shannon's formulation, and, in the

terminology of nonlinear dynamics, forms an attractor ina dissipative process.

metry into register with the issue addressed hereregarding the mechanism whereby an inversetransform accomplishes a return from the spectralto the space-time domain. Effron (1989) hasshown the importance of scan-paths to visual

and other sensory processing. Bolster andI (Bol-ster Pribram, 1993) have shown, by recordingsof the electrical activity of the parietal, temporal,and far frontal intrinsic association cortices ofmonkeys while they were choosing among singleor conjoined features, that these brain s ystem s areinvolved in the construction of scan-paths. Otherevidence (reviewed by Pribram, 1960, 1974,

99 ) indicates that these parts of the cortex oper-ate back onto the primary sensory pathwaystohelp organize the sensory input.

Individual scan-p aths, when externalized as

eye movements (which they need not be), donot look directed any more than do the indivi-dual paths taken by the maxim um squaredamplitude recordings shown in Fig. 1. Butwhen scans are tracked over time, patterns tendto centre on the more informative parts of afigure (Bags haw , Mackw orth, Pribram, 1970).Thus, on the basis of the evidence that isdemonstrated in Fig.], the temporal hold (asdemonstrated in Llinas' experim ents) can resultin brief mo men ts iterated until a pattern ofscan-paths accomplishes a percept. On the basisof our experiments and those of Petche's pre-sented at this conference, the duration of such aneural moment to approximate 100th of a secondon average (in the range from5 to 20msec) andthat of the temporal hold would be coordinatewith the span of attention.

A MODEL

The following exam ple shows how a system o per-ating in the spectral domain can be constructedfrom scan-paths; specifically, how hippocampal(limbic) function can interact with operations inthe cortical convexity to facilitate learning.

The example takes into consideration JeffreyGray's and J.N.P. Rawlins' (1975) proposal thatthe hippocampal formation acts as a comparator;D.S. Olton's 1983) and my own (Pribram, 197 1)emphasis on its function as a memory buffer;Abraham Amsel's (in press) experiments regard-ing the effects of hippocampal manipulations onvicarious trial (and error) performances; andE.Roy John's (1967) demonstration that electrical

responses are evoked in limbic structures duringand only during early phases of learning.


10/13

222 K A R L H PRIBRAM

Specifically, the example attempts to resolvesome apparently discrepant findings regardingthe results of microelectrode recordings of theactivity of single neurons or small groups ofsuch neurons during performances of rats in

mazes. The discrepancy is that under some con-ditions, a path in space is outlined by the record-ings; under other conditions, spatial cues appearto be represented in a distributed fashion.

The current formulation was instigated byanother presented by J McClelland and BruceMcNaughton (McClelland, 1996). The McClelland-McNaughton model takes into account only thelatter's finding of a representation in hippocampalneurons of a path in space. Thus their modeldirectly matches hippocampal activity with theactivity of the cortical convexity (as would be

expected of a comparator). On the input sidesuch a model is plausible. However, their modelalso demands such a comparative process on theoutput side. This is implausible in view of resultsobtained by Paul MacLean and myself (PribramMacLean, 1953) when mapping cortical connec-tivity by strychnine neuronography. Although wewere readily able to show multiple inputs to thehippocampal formation, we were totally unable toactivate ny isocortical region by stimulating thehippocampal cortex. The finding was so strikingthat MacLean (1990) developed the theme of aschizophysiology of cortical function.

On the other hand, such outputs are plentiful tothe amygdala, to the n. accumbens septi, and toother subcortical structures via the fornix. Con-firmation of the difference between input (encod-ing) and output (decoding) operations involvingthe hippocampal formation has recently comefrom studies in humans using fMRI (Gabrieli,Brewer, Desmond, Glover, 1997 . Encodinginto memory was found to activate the parahip-pocampal cortex, which includes the entorhinal

cortex (which receives input from the remainderof the isocortex), whereas decoding (retrieval)was found to activate the subiculum which (Gab-rieli, et al., 1997, p. 265) provides the majorsubcortic l output of the hippocampal region viathe fornix .

The subcortical nuclei do not have the laminarstructure of cortex and so are poor candidates forthe point-to-point match a computer would ordi-narily be conceived to implement. On the otherhand, a match could readily be achieved if thecomparison would involve a stage during which

processing involved a distributed stage, much aswhen a holographic memory is used to store and

retrieve information (for instance with holofishe).It is the evidence that such a distributed store is, infact, built up in the hippocampal formation duringlearning that makes such a model plausible.

Landfield (1976) and O'Keefe (1986) have

developed such a model. O'Keefe (1986, pp.82-84) reviews the evidence and describes themodel as follows:

Attempts to gain a n idea of the way in which anenvironment is represented i n the hippocampusstrongly suggest the absence of any topographicisomorphism between the map and the environ-ment. Furthermore, it appears t h a t a small clusterof neighboring pyramidal cells would map, albeitcrudely, the entire environment. This observa-tion, taken together with the ease that manyexperimenters have had i n finding place cellswith arbitrarily located electrodes in the hippo-campus, suggests t h a t each environment is repre-sented many times over i n the hippocampus, i n amanner similar to a holographic plate. n bothrepresentation systems the effect of increasingthe area of the storage which is activated is toincrease the definition of the representation.

A second major similarity between the way inwhich information can be stored on a holo-graphic plate and the way environments canbe represented in the hippocampus is that thesame hippocampal cells can participate i n therepresentation of several environments (O'Keefe

Conway, 1978; Kubie Ranck, 1983). I n theKubie and Ranck study the same place cell wasrecorded from the hippocampus of female rats i nthree different environments: All of the 28 non-theta cells had a place field in at least one of theenvironments, and 12 had a field in all threeenvironments. There was no systematic relation-ship amongst the fields of the same neurone inthe different environments. One can concludet h a t each hippocampal place cell can enter intothe representation of a large number of environ-ments, and conversely, that the representation of

any given environment is dependent on theactivity of a reasonably large group of placeneurones.

The third major similarity between the holo-graphic recording technique and the constructionof environmental maps in the hippocampus is theuse of interference patterns between sinusoidalwaves to determine the pattern of activity in therecording substrate (see Landfield, 1976). I noptical holography this is done by splitting abeam of monochromatic light into two, reflectingone beam off the scene to be encoded and theninteracting the two beams a t the plane of the

substrate. In the hippocampus something similarmight be happening. The beams are formed


11/13


12/13

224 K A R L H. PRIBRAM

take a single wire and stick it into the guts of thecom puter (and hope you won t short anything out)to find out in machine language what is going on.Despite such odds, by hard work and, what isessential, by synthesizing the results with thoseobtained with other techniques, meaning has beenharvested from recordings of brain electricalactivity. Have we got it totally right as yet? Prob-ably not. But when I think back to what we knewhalf a century ago about how the brain operates toorganize our perceptions and our memory and ourbehaviour, I can only be optimistic:24 000 neuro-scientists are in the trenches ready to seize thenext vantage.

Adey, W.R. (1967). Intrinsic organization of cerebraltissue in alerting, orienting, and discriminative re-sponses. In G.C . Quarton, T. Melnechu k, F.O.Schmitt (Eds.), The neurosciences (pp. 615-633).New York: The Rockefellar Press.

Amsel, A. (in press). On cognitive maps, vicarioustrial-and-error, and impulsivity. In K.H. Pribram(Ed.), Brain and values. Hillsdale, NJ: LawrenceErlbaum Associates Inc.

Bagshaw, M.H., Macworth, N.H., Pribram,K.H. (1970). The effect of inferotemporal cortexablations on eye movements of monkeys duringdiscrimination training. International Journal ofNeuroscience 1 153-1 58.

Bishop, G. (1956). Natural history of the nerve im-pulse. Physiologicrrl Review 36 376-399.

Bolster, B., Pribram, K.H. (1993). Cortical involve-ment in visual scan in the monkey.Perception andPsychophysics 53(5), 505-5 18. Austin, TX: ThePsychonomic Society.

Creu tzfeld, O.D . (1961). G eneral physiology of corti-cal neurons and neuronal information in the visualsystem. In M.A.B. Brazier (Ed.),Brain and beha-vior (pp. 299-358). Washington, DC : American In-stitute of Biological Sciences.

Eccles, J.C. (1958). The physiology of imagination.Scien tijc American 199 135-1 46.

Effron, R. (1989). The decline and fall of hemisphericspecialization. Hillsdale, NJ: Lawrence ErlbaumAssociates Inc.

Gabrieli, J.D.E., Brewer, J.B., Desmond, J.E., Glo-ver, G.H. (1997). Separate neural bases of two fun-damental memory processes in the human medialtemporal lobe. Science 276 264-266.

Gray, J., Rawlins, J.N.P. (1975). Comparator andbuffer memory: An attempt to integrate two modelsof hippocamp al function. In R.L. Isaacson K.H.Pribram (Eds), The hippocampus Vol. I and I1 (pp.159-1 96). New York: Plenum.

Gree n, J.D., Maxw ell, D.S., Petsche, H. (1961 ).Hippocampal electrical activity 111. Unitary eventsand genesis of slow waves. EEG Clinical Neuro-

physiology 13 854-867.

Green, J.D., Petsche, H. (1961). Hippocampal elec-tral activity 11. Virtual generators. EEG CliniccrlNeurophysiologj 13 847-853.

Harth, E., Unnikrishnan, P., Pandya, A.S. (1987 ).The inversion of sensory processing by feedbackpathways: Model of visual cognitive functions.Science 237 84- 87.

John, E.R. (1967). Mechanisms of memory. NewYork: Academic Press.

Kubie, J.L., Ranck, J.B. (1983). Sensory-beha-vioural correlates in individual hippocampus neu-rones in three situations: Space and context. In W.Seifert (Ed.),Neurobiology of the hippocarnpus (pp.433-447). London: Academic Press.

Kuffler, S.W . (1953 ). Discharge patterns and func-tional organization of mammalian retina.Jounzalof Neurophysiology 16 37-69.

Landfield, P.W. (1976). Synchronous EEG rhythms:Their nature and their possible functions in memory,information transmission and behaviour. In E.H.Gispen (Ed.),Molecular and functional rzeurobiol-ogy. Amsterdam: Elsevier.

Lashley, K.S. (1942). The problem of cerebral orga-nization in vision. In Biological symposia Vol. \ / I /Visual mechanisms (pp. 301-322). Lanc aster, UK:Jacques Cattell Press.

Li, C.L., Cullen, C., Jasper, H.H. (1956). Laminarmicroelectrode analysis of cortical unspecific re-cruiting responses and spontaneous rhythms.Jour-nal of Neurophysiology 19 13 1-143.

MacLean, P. (1990) The Triune brain in evolution:Role in paleocerebral functions. New York: PlenumPress.

McClelland, J. (1996). Role of the hippocampus inlearning and memory: A computational analysis.In T. Ono, B.L. McNaughton, S. Molotchnikoff,E.T. Rolls, H. Nichijo (Eds.),Perception memoryand emotion: Frontier in neuroscience (pp. 601-6 13). Oxford: Elsevier Scien ce.

McKegney, F.P. (1958). Telencephalic projections ofthe midline and intralaminar nuclei in the cat.YaleJournal of Biological Medicine 30 4 15-428.

O Keefe, J. (1986). Is consciousness the gateway tothe hippocampal cognitive map? A speculativeessay on the neural basis of mind.Bruin and Mind59-98.

O Ke efe, J., Conwa y, D.H. (1978). Hippocampalplace units in the freely moving rat: Why they firewhere they fire. Experimental Brain Research 31

573-590.Olton, D.S. (1983). Memory functions and the hippo-

campus. In W. Seifert (Ed.),Neurobiology of thehippocumpus. New York: Academlc Press.

Petsche, H., Gogolak, G., Van Zwieten,X. (1965). Rhythmicity of septa1 cell dischargesat various levels of reticular excitation.EEG Clin-ical Neurophysiology 19 25-33.

Pribram, K.H. (1954). Toward a science of neuropsy-chology (method and data). In R.A. Patton (Ed.),Current trends in psychology and the behavioralsciences (pp. 1 15-142). Pittsburgh, PA : U niversityof Pittsburgh Press.

Pribram, K.H. (1958). Neocortical function in beha-

vior. In H.F. Harlow C.N. Woolsey (Eds.), Bio-


13/13

THOUGHTS ON THE MEANING OF BRAIN ELECTRICAL ACTIVITY 225

logical and biochemical bases of behavior pp. 15 1-172). Madison, W1: University of W isconsin Press.

Pribram, K.H. 1960 ). The intrinsic systems of theforebrain. In J. Field, H.W . Magoun , V.E. HallEds.), Handbo ok of physiology neurophysiologypp. 1323-1324). W ashington, DC: American Phy-

siological Society.Pribram, K.H. 1966 ). Som e dimensions of remem-

bering: Steps toward a neuropsychological modelof memory. In J. Gaito Ed.), Macromolecules andbehavior pp. 165-187). New York: AcademicPress.

Pribram, K.H. 1971). Languages of the brain: Ex-perimental paradoxes and principles in neuropsy-chology. Englewood Cliffs, NJ: Prentice-Hall.Previous editions: Monterey, CA: BrooksICole,

1977; New York: Brandon House, 1982.)Pribram, K.H. 1974). How is it that sensing so much

we can do so little? In K.H. Pribram Contrib. Ed.),Central processing of sensory input. InF.O. Schmitt

F.G. Worden Eds.), The neurosciences thirdstudy program pp. 249-261). Cambridge, MA:MIT Press.

Pribram, K.H. 1986). The role of cortico-corticalconnections. In F Lepore, M. Ptito, H. JasperEds.), Two hemispheres-one brain; functions of

the corpus callosum. New York: Alan R. Liss.Pribram, K.H. 1 991 ) . Brain and perception: Holon-

omy and structure in jgu ra l processing. Hillsdale,NJ: Lawrence Erlbaum Associates Inc.

Pribram, K.H. 1994 ). Afterword to Origins: Brainand self organization pp. 707-708). Hillsd ale, NJ:Lawrence Erlbaum Associates Inc.

Pribram, K.H., Mac lean, P.D. 1953). Neurono-graphic analysis of medial and basal cerebral cortex.11. Monkey. Journal of Neurophysiology 16 324-340.

Prigogine, I. 1994 ). Mind and matter: Beyond theCatesian dualism. In K.H. Pribram Ed.),Origins:Brain and self organization pp. 2-15). Mahwah ,NJ: Lawrence Erlbaum Associates Inc.

Psaltis, D, Mok , F 1995). Holographic memories.Scientific American 273 5), 70-76.

Singer , W. 1993 ). Synchronization of cortical activ-ity and its putative role in information processingand learning. Annual Review of Physiology 55349-374.

Stum pf, C. 1965). Drug action on the electrical ac-tivity of the hippocampus.International Review ofNeurobiology 8 77-138.

Ukhtomski, A.A. 1927 ). Concerning the condition ofexcitation in dominance. Novoe y refteksologie Ijiziologii nervnoisysternry 2, 3-15.

Verzeano, M., Laufer, M. 1970). The activity ofneuronal networks in the thalamus of the monkey. InK.H. Pribrani D. Broadbent E ds.),The biology ofmemory. New York: Academic Press.

Verzeano, M., Negishi, K. 1960 ). Activity in corticalthalamic networks: study with multiple microelec-trodes. Jourrzal of G en. Physiolog y 43, 177-195.

thoughts on the meaning of brain electrical activity t-208

Documents