| The Theta Rhythm
generalized theta activity appeared" (Matsuoka, 1990)
The theta rhythm has been observed during a wide variety of behaviors in many different mammalian species. In particular, it seems to be involved with memory consolidation, contextual representation, and (among humans) the performance of tasks involving focused attention. Various correlations in animals and humans are outlined and two theories regarding the functional role of the theta rhythm in the brain are presented.
1.1. The Theta Rhythm The ìTheta Rhythmî refers to the firing frequency of between roughly 4 and 7 Hz. Theta is just one of a number of commonly recorded frequencies in the brain. A typically recorded frequency band includes: delta (1-3.4 Hz), theta-1 (4.8-5.6 Hz), theta-2 (5.8-6.6 Hz), theta-3 (6.8-7.6 Hz), alpha-1 (8-10 Hz), alpha-2 (10.2-12.4 Hz), and beta (12.6-20 Hz) (Matsuoka, 1990). Of note here is that what is judged to be the theta frequency varies between species and laboratories. For the purposes of this paper, when referring to ìthetaî, I will mean the generally accepted human range of from 4-7 Hz.
The theta rhythm originates in the brainstem as non rhythmic neuronal impulses. These signals are relayed to the septum which converts them into rhythmic bursts at the theta frequency. These bursts are sent from the medial septum and the diagonal band of the septum to all parts of the hippocampus (Miller, 1991). The Hippocampus then relays these rhythmic firing bursts to the cortex. Although there are connections to the entire cortex, the majority seem to go to the prefrontal areas with some connections to temporal and limbic areas, and less to the rest of cortex. There seems to be cases in which theta is generated by the cortex itself, skipping this pathway. This type of theta activity probably involves different types of information processing.
Of particular interest is the connectional structure of the hippocampus in comparison to the cortex. The hippocampal pathway begins in the entorhinal cortex, passes first to the dentate gyrus via the perforant pathway, then along the mossy fibers to area CA3, after that it goes to area CA1 via the Schaffer collaterals, then to the subiculum, and finally back out to the entorhinal cortex which forms the majority of connections to and from the cortex. The thing that is unique about the hippocampus is that, unlike the rest of the brain, there are very few reciprocal connections, for the most part signals only flow in one direction. The significance of that fact, and its relationship to the theta rhythm unfortunately will not be addressed in this paper.
2.1. Context Representation In Cortico-Hippocampal Interplay and the Representation of Contexts in the Brain, Robert Miller proposes a theory describing how the interaction of the hippocampus, the theta rhythm, and the isocortex allow the brain to represent contexts. In the brain, as in a book or a sentence, the context is what gives individual inputs meaning. The brain receives a virtually infinite amount of information from the world, and in order to make sense of it, it must form contexts within which it can gain understanding. In a relatively simple brain, contexts might refer only to different environments that the animal explores; however in the human brain, contexts could represent anything from the meaning of a word in a sentence to the meaning of a look from a girl across the room. Without contexts, there is no meaning. The study of how the brain represents contexts is at the heart of how the brain understands, and perhaps becomes conscious of, everything it perceives. In order for the brain to integrate all of the available data into a context, it has been theorized that groups of cells responding to different inputs must somehow join together to form ìglobal cell assembliesî. Miller argues against the idea that these cell groupings are formed using a simple ìcombinatorial matrixî, citing that there arenít enough connections between spatially distant neurons. This is where the theta rhythm comes into play.
Before Millerís proposed biological mechanism for context representation can be discussed, it is important to note the behavioral correlations that led him to his theory. The theta rhythm or RSA (rhythmic slow activity) as it is also referred to, is present in ìrodents, carnivores, and primatesî (Buzsaki, 1996). Miller notes that, Those periods of an animalís activities when information important to that species requires to be gathered from the environment, are the times when RSA is most likely to be generated by the hippocampus. (Miller, 1991) That is, when animals need to learn from their environment to survive, they typically exhibit theta. For example, in the wild, rats maintain relatively large home territories from which they often make forays as much as a mile away. Rats need to keep track of where they find food, cats, and also where they have recently visited. Therefore, rats need to gain information while they are moving. This correlates with the fact that in rats, hippocampal RSA is most prominent during locomotion (Miller, 1991). Similar correlations have been found with other animals such as rabbits, cats, and dogs. Cats, like rats, are territorial but they are also hunters. While hunting, the most important time for learning is during the relatively immobile period of stalking. As the cat stalks its prey, it must learn how not to scare it away before commencing the fatal pounce. If it doesnít do that, it will have to survive on Purina alone, a sad thing for the proud cat. As expected, hippocampal theta is prominent during immobility as well as locomotion in the cat. The behavioral correlates of theta in humans are not as simple as in the cat or the rat, however it is theorized that the same basic mechanism is being utilized across all species exhibiting theta.
Given the observed correlation between exploratory-learning and theta activity, Miller seeks to give a biological explanation of their relationship. This is ìthe theory of resonant self-organizing phase locked loopsî. He hypothesizes that the theta rhythm acts to strengthen the connection weights of certain, already present loops between the hippocampus and the cortex. 2.3.1. Self Organization The organization of these loops is based on the principle of Hebbian learning. That is, correlation between the firing of pre and post synaptic cells, within about 20 milliseconds, leads to an increased connection weight between the two neurons, possibly caused by LTP. To see how the oscillation of the hippocampus could entrain these loops, let us imagine an extremely simple loop, including one cortical neuron ìN1î and one hippocampal neuron ìH1î. Imagine that neuron N1 is activated, (perhaps by some environmental stimuli), it sends an impulse to H1 in the hippocampus. Now, the hippocampus is firing between 4 and 7 Hz, about once every 250 to 142 milliseconds. Recall the Hebbian learning rule, if the input to a hippocampal neuron from N1 arrives within 20ms of that cellís firing, synaptic facilitation can occur. Now imagine that the time it takes for the neural signal to travel around the loop and back to the hippocampus is approximately equal to the theta period (between 142 and 250ms). In this case, Hebbian learning could occur every time the signal reached the hippocampus. The loop could now be termed ìresonant.î The actual loops are necessarily more complicated than this simple circuit, but the underlying principle remains the same: loops selected for resonance ìare those whose total conduction time approximates to the theta period (or a simple fraction or multiple of it).î(Miller, 1991) Conduction time refers to the total time it takes a signal to travel out from the hippocampus and back again. Thus connections between the hippocampus and the cortex are theorized to organize themselves into resonant loops locked in phase to the theta rhythm. 2.3.2. Context Representation through Resonant Loops Each resonant loop, or pattern of loops, activates neurons widely dispersed across the cortex. According to Miller, these neurons do not have enough interconnections to form global cell assemblies according to simple Hebbian learning. Resonant loops allow for this interconnection of distant parts of the brain. This resultant interconnectivity, Miller theorizes, is what allows for the representation of contexts by the brain.
One advantage of the cortico-hippocampal information storage network over a purely cortical network is that the number of physical connections needed can be greatly reduced. According to Millerís theory, a projection neuron from the cortex might need to form only one connection to the hippocampus in order to participate in a resonant loop. On the other hand, a single hippocampal neuron could project to thousands of different cortical sites and participate in many different loops. Thus vastly fewer connections are needed than if each cortical neuron had to form a direct connection with each of its partners in crime. In this way, the hippocampus acts as the index to the book of the cortex, indicating what is stored where, and letting the much bigger cortex actually store the information. Another way in which the hippocampus aids in the economy of information storage is that it doesnít seem to contain any of the redundancy of representation common to cortical areas such as the visual system.
Keeping in mind that this still only a theory, it is to be expected that some evidence is lacking. The primary missing piece is the evidence for resonant activity in the cortex itself, apart from the hippocampal formation. This could be because people havenít paid attention to it or because of a lack of the proper recording technology. Despite this, other lines of evidence seem to support the theory of resonant loops. Among these is the discovery of ìplace cellsî in the rat hippocampus. These are cells whose firing rate is maximized when the rat is in a particular location within its environment. These are predicted by the convergence of various cortical inputs onto small areas of the hippocampus: these various inputs linking together in the formation of a cortical representation of that location . Buzsáki lends support to Millerís theory by saying that ìin essence, the map of the environment will be ëencodedí as a temporally ordered pattern in the recurrent collaterals of the CA3 regionî(Buzsáki, 1996). It seems clear that the mechanism described above, or on similar to it, is involved in the representation of environments and, very likely, contexts in the brain. However there seems to be a missing piece to this theory.
The joint firing of hippocampal neurons that occurs during theta activity is theorized to provide the mechanism by which memories are encoded in the hippocampus and the cortex. A question that arises when considering this theory is, ìwhy doesnít the joint firing of hippocampal neurons during sleep and other non-learning states disrupt the network representation of information encoded during exploratory theta activity?î In order to answer that question, Buzsáki proposes a two stage process of memory consolidation in the feature article of the Mar/Apr 1996 issue of ìCerebral Cortexî. These two stages are termed ìopen loopî and ìclosed loopî.
During the open-loop state, the aroused animal gathers information from its environment, like the rat exploring its territory. This state is linked to theta and gamma (40-100 Hz) oscillations in the hippocampus. As has been described above, theta is related to information storage in the hippocampus. The role of gamma waves is still unclear, however they are certainly related to theta waves as shown by Buzsáki. First, changes in theta frequency (6-10 Hz -see they keep changing it!-) are coupled to changes in gamma frequency (40-100 Hz.). Second, the frequency and power of gamma activity is modulated by the phase of the theta rhythm. Third, most interneurons oscillate at both theta and gamma field frequencies. (Buzsáki 1996)
The open loop state is characterized by the steady firing of hippocampal neurons carried out by sustained sodium-spike activity. This mode of neural activity is thought to provide the ìhigh fidelity throughputî necessary for the continuous recording of information thought to occur during the open loop state. During this state, information is theorized to be recorded in area CA3 of the hippocampus. From there, it can later be moved on to area CA1 and eventually to the rest of cortex during the closed loop state. This transfer is made possible by the high level of sub-cortical neurotransmitters, such as acetylcholine, histamine, serotonin, and noradrenaline, that are present in the open loop system.
When the animal enters a closed loop state, such as ìslow wave sleep, awake immobility, drinking, eating, face washing, and groomingî, the amount of available neurotransmitters (NTs) decreases. This decreased level of NTs inhibits the steady potassium current necessary to maintain sustained sodium spike activity. This results in the conversion of hippocampal and cortical cells to more of a ìburst firingî mode, characterized by the appearance of sharp waves (SPWs). An SPW is a burst of firing that lasts between 40 and 100 msec and occurs irregularly (0.02-3/sec). During these powerful blasts, more calcium is released than during typical steady firing. This increase in calcium release is theorized to have a role in LTP which in turn is theorized to play a key role in memory consolidation. In addition to increased calcium levels, SPWs result in the joint firing of many hippocampal, entorhinal, and presubicular neurons. This correlation is also necessary for LTP to occur.
During the closed loop state, memories, previously recorded in area CA3, are theoretically shot out to area CA1 and then to the cortex on the aforementioned SPWs. This theory provides a solution to the problem outlined at the beginning of this section by assigning different functional purposes to the different observed states. Information is gathered during exploratory (open loop) behavior and stored in area CA3 via theta and gamma activity. From there it is relayed to CA1 and then out to the cortex during the non-aroused (closed loop) state using SPWs.
The primary line of evidence linking SPWs to delayed memory transfer involves the observation that those cells that are the most active during exploration are the most likely to participate in SPWs during the next occurrence of slow-wave sleep. This implies that information acquired during the open loop state is re-expressed in the closed loop state as SPWs. Although this two stage hypothesis remains unproven, the implications are there that it is an important theory for future research.
It is well established that theta is involved in rodent spatial memory and environmental representation. Experiments and theories have even shed some light on the theta-related mechanisms that underlie this type of cognition. However, perhaps due to the lack of volunteers for internal electrode recording experiments, precise information about the functioning of theta in humans is lacking (although it is probably closely related). What we are left with in human research is primarily EEG recordings during different behaviors, without much explanation as to the actual internal processes. So, until more precise, non-invasive, spatial and temporal brain imaging techniques are developed, we are left guessing at the actual functioning of theta in the human brain.
Theta has been observed to be correlated with different behaviors and different states of consciousness: including learning (Seppo, 1995), problem solving (Crawford, 1994 Laukka, 1995 Matsuoka, 1991), meditation (zen, transcendental etc.)(Wallace, 1972), REM sleep, mental imagery, verbal and spatial reasoning tasks, hypnosis, laughter, tobacco withdrawal (Herning, 1983) and during drug induced euphoria (alcohol, nitrous oxide, helium during diving related compression)(Matsuoka, 1991).
In one study, subjects were engaged in a simulated driving task in which they had to learn how to navigate through a series of streets represented as animation on a computer screen (Laukka et al. 1995). They found that the percent of frontal-midline theta activity increased as the subjects grew better at the task, making more correct decisions. The experimenters concluded that theta correlated to the relaxed concentration that was needed to complete the task, once it had been learned, as well as during learning - successful behavior producing more theta than unsuccessful.
Excluding those experiments involving drugs, the activities in which theta is most prevalent are those in which the subject is engaged in focused, but also relaxed, concentration. Be it a meditative mantra, the voice of a hypnotist or a problem that they are solving, theta continues to occur. This observation has lead some scientists to link theta to attentional and disattentional processes in the brain.
One such process which quite specifically involves attention is hypnosis. In order to be hypnotized, people have to focus their attention on themselves and the person that is hypnotizing them. If they are distracted, the trance is broken. Highly hypnotizable people exhibit significantly more theta before and during hypnosis then low hypnotizables (Crawford,1994). It is theorized that theta activity in the posterior areas of the brain is involved with selective attention, and in the frontal areas with the cognitive filtering, ìdisattending to extraneous stimuliî, neccesary for entry into the hypnotic state (Crawford, 1994). Thus theta activity is involved in the selection and maintenance of attention during hypnosis. 4.3.2. Zen Meditation Another uniquely human activity linking theta activity to attention is meditation. During zen meditation, experienced practitioners have been recorded using EEGs. As they reach deeper and deeper meditative states or trances, the typical patterns of EEG activity begin with alpha activity (8-10Hz) and gradually move into the theta range as the trance progresses. (It is important to point out some descrepancies between what different scientists call ìthetaî. Matsuoka puts the upper bound at 7.6 Hz., while Buzsaki puts it at 10. So when Matsuoka sites alpha activity, Buzsaki would refer to high frequency theta.)
Wallace and Benson studied the physiological effects of transcendental meditation in 1974. Transcendental meditation is a form of yoga in which subjects choose a ìsuitableî sound or thought and attend to it, allowing their mind to ìexperience it freelyî. During this process, subjects displayed alpha and theta activity similar to that exhibited by the zen practioners in Matsuokaís study. Once again, theta is related to relaxed concentration, or well focused attention. Interestingly, the amount of experience in meditation makes fairly little difference in the EEG activity, relative beginners exhibiting nearly the same patterns as masters. Also, the physiological effects of meditation are fairly standard across most disciplines, indicating that the different forms are merely different paths to the same thing.
The meditative state of consciousness seems to have many benefits. Wallace and Benson observed what Benson would later term ìthe relaxation responseî during transcendental meditation (Benson, 1996). This refers to the physiological state basically opposite to what is known as the fight or flight reflex (3 fís). During fight or flight, heart rate, blood pressure, and the levels of adrenaline (norepinephrin) released into the blood are increased. During meditation, the opposite effects are noted, in addition to a slower metabolism and lower blood lactate level. Bensonís various studies have indicated a direct correlation between eliciting ìthe relaxation responseî (basically through a form of meditation) and better mental and physical well-being. Slow waves, theta and alpha, have been linked to this ìrelaxation responseî and are theorized to have a functional role in its creation.
Increases in theta activity have been correlated with drug induced euphoria and tobacco withdrawal. Matsuoka theorizes a possible link between theta and ìdysfunction in the frontal cortex.î His research was prompted by the observation of a correlation between forced laughing attacks, caused by a tumor ìin the right frontal falx,î and the occurence of frontal midline theta activity. He found theta activity in a number of other laughter inducing situations such as the inhalation of nitrous oxide, the consumption of ethyl alcohol in a social environment, and during the induction of high pressure nervous syndrome during simulated dives in a hyperbaric chamber. Matsuoka proposes that theta is correlated to a dysfunction in the frontal cortex, but does not indicate what type of dysfunction. To formulate an explanation, the phenomenon of laughter must be investigated. In a normal environment, the predecessors of laughter are some of the most complex processes of the human brain. The presence of these processes that allow people to ìgetî jokes, or to see the humor in a situation, is even theorized to be a sort of ìlitmus testî for consciousness. During drug or tumor induced laughter, these processes are skipped; the physiological response of laughter coming about without these complex processes to inititate it. A possible explanation of the occurrence of theta during laughter, drug induced or not, is that theta is related to the feeling of pleasure that laughter produces. Seppo et al. (1995) theorized that theta had to do with the feeling of satisfaction after successfully completing a task. An interesting study would be to see if theta occured during non-drug related laughter. If the same pattern of activity was seen, more evidence for the linking of theta to pleasure production would be accumulated, however if the pattern changed, theta would have to be related to some other concurrent phenomena. Pleasure could be an important part of task-learning, the feeling of pleasure reinforcing the correct response. However, seeing that theta is present in rats during exploration, and it is doubtful that they get any pleasure from running around a maze, it would make more sense for theta to be involved in cognitive processing beyond the production of pleasure. This would lend support to the idea that theta serves similar purposes across species, although the possibility remains that theta plays an entirely different role in humans than in other species.
Theta rhythms have been observed in humans and animals during a wide variety of activities; thus understanding the role of theta is integral to the understanding of the functioning brain. Observing correlations between theta and behavior is the first step, the next step is understanding how the theta rhythm actually allows the brain to enact the processes that lead to these behaviors. Theories, such as those outlined in this paper, have suggested ways to explain thetaís function in the consolidation of memory and the representation of contexts. In order to test these theories, better computer models of the brain must be developed. This would allow more experiments to be performed that deal with the how of brain processes rather than just the when and where. Eventually these computer simulations could lead to a complete understanding and representation of the human brain. If this can ever be done, not only would cognitive neuroscience complete its goals, but a means would be provided for the birth of HAL. Either as an appropriately enhanced chimp or a computer program, humans would have the power to create a new form of intelligent life. But lest we get to far ahead ahead of ourselves, it must be kept in mind that, for now, people donít even understand the interaction of individual neurons. The goal of understanding the brain is far off, perhaps even unnattainable, but the study of the theta rhythm provides a step in the right direction.
|