REINVESTIGATION OF LONG-LASTING MEMORY-TRACE REPLAY

M. Tatsuno¹*; P. Lipa¹; K.C. Stengel²; B. L. McNaughton¹

1. NSMA , Univ Arizona, Tucson, AZ, USA
2. Neuralynx, Tucson , AZ , USA

Neuronal traces of recent experiences are reactivated during sleep, but, in hippocampus, this activity typically decays to levels that are statistically undetectable with current methods within about 30 minutes (Kudrimoti et al., 1999). Using a different analysis method, however, Ribeiro et al. (2004) recently reported that rats 'reverberate' spatio-temporal neuronal patterns from novel experiences for up to 48 hours. If this is correct, it has very important implications for the trace-reactivation theory of memory consolidation.

We critically reexamined the Ribeiro et al. results and studied neuronal interactions among various brain regions during off-line processing periods using a high-density recording array, developed in collaboration with Neuralynx Inc. This array allows independent manipulation of 240 electrodes on a 12x20 array with 0.675 mm spacing and was implanted over the rat's neocortex.

Two sets of 50 hour continuous recordings were conducted, following the essential design of Ribeiro et al., and more than 60 neurons and 200 EEGs were recorded simultaneously. The data were analyzed by several template matching methods, including the Louie-Wilson measure (Louie & Wilson, 2001), which was used in the Ribeiro et al. study, and the explained variance measure (Kudrimoti et al., 1999). We did not observe significant long-lasting reverberation, but did find that the level of correlation was strongly dependent on the choice of measures in the template matching algorithm and that most of the template-match variance was due to long-term changes in mean firing rates, not moment-to-moment firing patterns. Because of this, we believe that, to establish long-lasting reverberation, the target templates need to be compared over the whole recording trace, rather than comparing 'control' templates over the first half of the recording and 'exposure' templates over the latter half. This control procedure was not reported by Riberio et al., whose conclusion, we believe, is weakened by this omission.

Supported by MH46823

Key words: high-density recording, reverberation, template matching, explained variance