Does hippocampal theta exists in the human brain? (Comment to the paper of Uchida et al. 2001)

The hidden surface of the human medial temporal lobe (MTL) and its inaccessibility by traditional EEG recording procedures produced a large gap between human and animal data regarding hippocampal activity patterns. A wide-range of state-specific activity patterns of the hippocampal formation were established in animal studies, but few of them were unambiguously detected in humans. Two papers of Uchida et al. (Brain Research, 2001, 891:7-19 and Neurobiology of Sleep-Wakefulness Cycle, 2001, 1(1):1-8) present data with the explicit intention of reducing the gap between human and animal studies and "establishing human MTL activity patterns", which we agree "is of utmost importance" in "unveiling the enigma of human brain function" (Uchida et al 2001a, p. 8.). These and the previously reported data of the same group (Hirai et al., 1999a, b) are major advances in the field. They exceed previous studies in relatively high number of subjects, high frequency sampling and quantitative analysis of the electrocorticogram (ECoG), as well as careful avoidance of epileptic activity in the analyzed ECoG segments. In spite of these major advances, we disagree with the recording technique and the interpretation of the data, presented in their papers in Neurobiology of Sleep-Wakefulness Cycle and Brain Research. The authors invoke the phenomenon of hippocampal theta (also known as hippocampal rhythmic slow activity or hippocampal RSA) as one of the major starting point of their study and the basis of the interpretation of their results. Hippocampal RSA is a basic neurophysiologic feature of waking- and REM sleep related arousal (and perhaps cognition) in animal models. It is unclear whether similar theta activity exists in primates including humans. However, recording of hippocampal RSA has its indispensable conditions, which were established on the basis of a large number of animal experiments (Robinson, 1980). Perhaps the most important condition is related to the reference point. As hippocampal RSA is synchronous over large mediotemporal areas, it can be recorded in monopolar derivation, provided that the reference point is distant from the hippocampal formation. Bipolar recordings detect hippocampal RSA only when the referred points are below and above the pyramidal layer respectively. It is hard to imagine, how this latter criterion can be met in human studies, but clearly the subdural electrodes presented in the study of Uchida et al. (2001a) are inappropriate for bipolar recording of hippocampal RSA. Given this fact and the intention of the authors to detect hippocampal RSA, it is surprising that they used bipolar recordings in their study. This could lead them to erroneous assumptions regarding the frequency of human hippocampal RSA. Presenting the logaritmized spectral plots of a single subject's monopolarly referred ECoG is (Uchida et al. 2001a, Fig. 9.) not convincing because of three reasons: (1) the distance between the reference point (ipsilateral basal temporal lobe) and the parahippocampal contact point is probably not sufficient (theta activity can be still detected in the basal temporal cortex of rats); (2) log-transformed spectral plots up to the 200 Hz value reported in figure 9. (p. 17.) did not allow visual inspection in the 1-5 Hz frequency range, which is most frequently found to behave like hippocampal RSA in human studies cited by Uchida et al. (2001a, b); (3) statistical analysis is based only on bipolar signals, which are unsuitable for detecting hippocampal RSA, so we do not know if the similar statistical analysis of the monopolar signals would produce significant state-specific differences in other frequency bands too. Otherwise it is quite improbable that a synchronized rhythmic activity would appear in a similar way in bipolar and monopolar derivations. It is more probable that the 10-20 Hz activity reported by Uchida et al. (2001a, b) is not synchronous over the parahippocampal gyrus, which is not the case for hippocampal RSA. When used in animal studies, bipolar recordings produced signals which were rich in fast activity (Robinson 1980). This is exactly what we see in the records presented by Uchida et al. (2001a). In a similar study (Bodizs et al. 2001), using mediotemporal corticography with foramen oval electrodes in 12 epilepsy patients, we also found relatively low amplitude, fast activity in bipolar recordings during REM sleep. But when the same records were reanalyzed in extracranial reference they showed synchronized, rhythmic slow activity of 1.50-3.00 Hz. Our present database contain multiple all-night recordings from 14 patients, who all had this 1.50-3.00 Hz monopolarly recorded rhythmic activity during REM sleep. Other studies reporting hippocampal RSA-like patterns in human subjects always found frequencies below 10 Hz. Some of these papers were cited by Uchida et al. (2001a, b; but see also Wieser 1984 and Mann et al. 1997). Given these reports and the huge data coming from animal experiments we feel it is premature to consider beta-1 activity as the "functional equivalent of the animal hippocampal theta rhythm" (Uchida et al. 2001b).