, 2004). Most models of synchrony are indeed based on random connectivity (Pfeuty et al., 2007 and Wang Dasatinib supplier and Buzsáki, 1996). In contrast, recent work has highlighted the emergence of highly spatially heterogeneous activity states when local clustering of electrical and chemical synapses is considered

(Lau et al., 2010). The enhanced clustering of both electrical and chemical synaptic connections among MLIs, as well as their structured overlap, may therefore form the substrate for complex spatial patterns of network activity underlying computations in the cerebellar cortex. A complementary way to examine the effect of different network topologies on network function is to study how different network motifs change network dynamics. Zhao et al. (2011) showed that deviations from random

networks caused by overrepresenting different network motifs involving two connections in either a divergent, convergent, or chain configuration can have opposing effects on synchrony. What could be the functional consequences of the overrepresentation of transitive chemical motifs we find among MLIs? The “synaptic chain model” is an example of such a transitive network architecture containing feedforward motifs and Androgen Receptor Antagonist is known to generate highly structured temporal dynamics (Abeles, 1991 and Seung, 2009). Loops, on the other hand, are examples of intransitive network motifs and can generate oscillations and self-maintaining rhythms (Wang and Rinzel, 1992). Although some circuits may exploit such dynamics (Manor et al., 1999 and Wang and Rinzel, 1992), the reverberating effects of loops between brain regions have been proposed to cause instability (Crick and Koch, isothipendyl 1998); this may also occur at the local circuit

level where oscillations may lead to tremor. Thus, structured connectivity containing feedforward motifs may be beneficial for network stability. In signal processing, finite impulse response filters implemented by a feedforward motif are more stable and reliable than infinite impulse response filters implemented by a loop motif (Rabiner and Gold, 1975). It remains to be determined if such features are also exhibited by neural networks with transitive connectivity. In the cerebellum, synchrony between MLIs (Mann-Metzer and Yarom, 1999) may be restricted to sagittal bands where electrical clustering is high. There, electrical coupling allows improved spatial averaging of the activity levels in the input population (Alcami and Marty, 2013), by sampling from a large number of parallel fibers. Inhibitory connections across sagittal planes may help synchronize successive planes with each other.