e., fine structure and formant transitions (Rosen, 1992). Theoretical work recently demonstrated that the shape of a prototypical 50 ms diphone-like stimulus can be represented by a three-bit code corresponding to three ≈40 Hz gamma cycles (Shamir et al., 2009). Such a binary encoding by the low-gamma rhythm represents a critical and

buy SCH 900776 necessary downsampling step in the process of transforming acoustic into phonological representation after which many spectrotemporal details of speech are lost. While others have put forward the hypothesis that syllabic sampling at theta rate might be altered in dyslexia (Abrams et al., 2009, Giraud et al., 2005 and Goswami, 2011), we focus here on the complementary idea that an anomaly in phonemic sampling at low-gamma rate could have direct consequences for phonological processing. We hypothesize that the oscillatory behavior in the low-gamma band observed in typical control participants is the optimal

phonemic sampling rate, and that too slow or too fast sampling would affect the format of phonemic representations. More specifically, gamma oscillations downshifted relative to controls would result in diminished phonemic discrimination (Tallal et al., 1993), whereas too fast gamma sampling by oscillations shifted upward might flood the auditory system with overdetailed spectrotemporal information and thereby saturate theta-based auditory buffer capacity (Hsieh et al., 2011) and phonological working memory. To assess GS-1101 these hypotheses, we compared auditory cortex gamma

oscillations in 23 dyslexic and 21 control participants. We used a frequency tagging magnetoencephalography (MEG) experiment with source reconstruction, in which auditory steady state cortical responses (ASSR) were evoked by a white noise with a range of amplitude Thymidine kinase modulations (10–80 Hz; Figure 1) that broadly covered the phonemic sampling domain. We predicted that phonological performance in dyslexics should reflect a deficit in low-gamma oscillations within a 25–35 Hz frequency window centered on the dominant 30 Hz phonemic rate. Consistent with the asymmetric sampling theory (AST) that postulates stronger low-gamma sampling in left than right auditory cortex (Giraud et al., 2007, Morillon et al., 2010, Poeppel, 2003 and Telkemeyer et al., 2009), we further assumed that the low-gamma deficit in dyslexics should be more pronounced in the left than in the right hemisphere. In both dyslexics and fluent readers, oscillatory responses were observed for acoustic amplitude modulations presented at the same rate (Figure S1 available online, maximal responses on the diagonal). We identified two regions within each hemisphere where entrainment by the modulated sound was maximal.