We also found no effect of learning on song coding or auditory scene processing in the higher-level AC, in contrast with previous reports that used the European Starling (e.g., Gentner and Margoliash, 2003 and Meliza and Margoliash, 2012), which may suggest differences in cortical plasticity between PI3K inhibitor species with open-ended (European Starling) and close-ended (zebra finch) learning periods.

We propose and model a cortical circuit based on feedforward inhibition that recapitulates salient aspects of the neural coding transformations observed between the primary and higher-level AC. Although the results of the simulation are in close agreement with our physiologic and pharmacologic findings, the model makes assumptions regarding the identity and connectivity of excitatory and inhibitory neurons, and the relative timing of excitatory and inhibitory inputs. The model also assumes that excitatory and inhibitory inputs to BS neurons are perfectly cotuned in frequency,

because in the model excitation is directly supplied and inhibition is indirectly supplied by the same neuron in the primary AC. Although we do not explicitly verify these assumptions, they are supported by previous studies showing that the higher-level AC receives direct Gefitinib ic50 synaptic input from the primary AC and is richly interconnected by local interneurons (Vates et al., 1996), and that neurons

in the songbird (Mooney and Prather, 2005) and mammalian (Atencio and Schreiner, 2008) cortex can be segregated based on action potential width into excitatory (broad) and inhibitory (narrow) populations. Our data show that primary AC and NS neurons in the higher-level AC have similar spike train patterns, firing rates, selectivity, and STRFs, in support of NS neurons receiving direct excitatory very input from the primary AC. Spectrally cotuned but temporally offset excitation and inhibition have been demonstrated in the mammalian auditory cortex (Wehr and Zador, 2003). Our proposed model captures our experimental findings and makes testable hypotheses about how the auditory cortex is organized to transform behaviorally relevant information. Across organisms and sensory modalities, examples of sparse coding (Crochet et al., 2011, DeWeese et al., 2003, Stopfer et al., 2003 and Weliky et al., 2003), contextual sparsification (Haider et al., 2010 and Vinje and Gallant, 2000), and feedforward inhibition (Tiesinga et al., 2008, Vogels et al., 2011 and Wehr and Zador, 2003) are common.