406.8

Odor encoding in complex sequences: multielectrode array studies in the olfactory bulb

Georgia World Congress Center: Room B314

Monday, Oct 16, 2006, 2:45 PM - 3:00 PM

*A. NICOL¹, M. S. MAGNUSSON², K. M. KENDRICK¹;
¹Laboratory of Cognitive & Behavioural Neuroscience, Babraham Institute, Cambridge, UNITED KINGDOM, ²Human Behavior Laboratory, University of Iceland, Reykjavik, ICELAND.

Synfire chains, recurring sequences of spikes, can be detected in simultaneously sampled multiple neurons (Ikegaya et al, 2004). However, the facility for encoding sensory information in these sequences is equivocal; they may be similarly apparent in real and randomized spike data (Oram et al, 1999). We have studied sequences of spikes detected in multielectrode array (MEA) data using an alternative procedure, T-Pattern Analysis (TPA; Magnusson, 2000), in the context of odor encoding in the olfactory bulb (OB). In TPA, sequences are detected on the basis of significant critical interval relationships, rather than fixed or user-defined intervals between events. An MEA (5x6 or 6x8 electrodes, sampling area ~2.2mm²) was positioned in the mitral cell layer of the OB in 8 urethane anaesthetized rats. Recordings were made in 10s periods pre- and during odor delivery. Spikes were detected on >50% of electrodes. At each electrode spikes from 6.4±1.0 (sem) neurons were discriminated. Breathing was also recorded. Sequences of ≤17 elements were detected, often spanning the entire array. An average of 1412±73 (sem) sequences were detected in prestimulus (control) periods. Results were compared to those when the spike trains were randomized by (a) randomly redistributing each spike train over the sample period, or (b) more conservatively, by arranging each spike train on a continuous loop and randomly rotating each relative to the others, thereby randomizing temporal relationships between the trains whilst retaining the temporal structure of each. After either randomization procedure only shorter sequences were found and these were significantly fewer than in the real data. Some sequences (8.2% in control data) included breathing (inhale or exhale). Inhale-related, but not exhale-related sequences increased in number (+21.7%) and length (+38.4%) on odor presentation. The inhale-related sequences detected during odor delivery increased in number (+204%) and length (+142%) with odor concentration. Our findings show that sequences detected by TPA are a fundamental property of neural networks in the OB, and have the facility to encode odor information.

 A. Nicol, None; M.S. Magnusson, None; K.M. Kendrick, None.

Biotechnology & Biological Sciences Research Council, UK