A state bearing strong similarities to Rapid Eye Movement (REM) or active sleep is dominant during the perinatal period in mammals and is characterized by episodes of variability, bursting and intermittent activity in the brainstem and in spontaneous tonic and phasic behavioral events such as REMs, periods of nuchal atonia and myoclonic twitching^l. In the past these spontaneous active sleep­associated behaviors in the fetus have not been distinguished from independent Poisson random processes.² We observed that episodes of nuchal atonia recorded from fetal sheep during gestational days E121­E133 have burst within burst pattern suggestive of statistical self­similarity which is indicative of long­run correlations between events (see Figure 1).

Figure. 1. On E123 of gestation, erratic intermittent bursting is apparent in nuchal activity with intervening periods of atonia. Nuchal EMG activity along the abscissa (2 volt range normalized to 250 standard units), plotted over 100 minutes on the ordinate (top trace). Rescaling the amplitude, abscissa, and time, the ordinate, over a 600 second subset of the original series (middle trace) reveals smaller clusters within larger clusters of both muscle discharges and atonia. Rescaling both dimensions again over 65 seconds (bottom trace) demonstrates similar qualitative "self­affinity" which is bounded from below by the sampling rate. Correlations hidden within natural time series with short­ and long­term erratic fluctuations (e.g., annual patterns of riverflow, rainfall, tree rings, etc.) can be uncovered using Hurst's Rescaled Range Analysis (Range normalized by S.D. or R/S).³ Hurst's analysis examines how the R/S for the entire time series is related to the R/S for a number of smaller time windows. A log­log plot of the average R/S for non­overlapping windows of 4, 8, 16....2048 points vs the size of the windows yields a line with a slope (H) that ranges from 0 to 1. Time series of independent random statistical processes have H = 0.5 indicating no correlations. Time series with positive correlations result in H > 0.5, indicating the existence of long­run correlations (i.e., fractals in time). Processes with H > 0.5 have long clusters of events in time that are more likely to be followed by long clusters of events, than short clusters, in a pattern termed "persistence".

To determine if time series of nuchal atonia (NA) durations had non­random fluctuations, episodes of NA were estimated from nuchal EMG sampled at 1Hz from fetal sheep (five subjects). Analysis of 70 24hr records containing at least 2048 NA episodes yielded a mean H = 0.70. H values derived from the nuchal EMG of neonatal rats P2 to P10 (collected for 2hr @ 300 Hz) ranged from 0.65 to 0.87 and were very similar to H values reported for fetal sheep. H also demonstrated a significant increase with 2 hours of maternal deprivation, indicating that nuchal atonia periods became more clustered. Probability distributions of spontaneous nuchal atonia events in both species were also found to be well described by convolutionally stable Levy distributions, suggesting deviations from Central Limit Theorem assumptions and non­convergent 2nd moments. ruling out a Poisson process.

From these findings we infer that this global structure of REM or active sleep in the fetus or neonate may be providing convolutionally stable patterns of asynchronous stimulation over many time scales to the developing CNS. Also RE associated behaviors in developing mammals lack a characteristic scale of measurement in time, requiring fractal methods analysis which provide measures of global structure independent of sampling rate.

1 Corner, M. A. (1990). Brainstem control of behavior: Ontogenetic aspects In: W.R. Klemm and Robert P. Vertes (Eds.) Brainstem Mechanisms of Behavior, (pp. 239­268). John Wiley and Sons, Inc., New York.

2 Narayanan, C.H., Fox, M.W. & Hamburger, V. (1971) Prenatal development of spontaneous and evoked activity in the rat. Behavior, 40, 100­134.

3 Bassingthwaighte, J.B., Liebovitch, L.S. & West B.J. (1994). Fractal Physiology. Oxford University Press, New York.

Research supported by NIMH­MH19116, NICHD HD­28014 and ONR Biological Intelligence Div.