Figure 6.

Profound impairment in speech token classification in the MGB in ouabain treated mice, even after correcting for threshold shift and increasing ensemble size. (A, left) Spectrograms of the speech token “teed” in original form (top) and resynthesized for mouse hearing (bottom). Formant frequencies are indicated by thick yellow lines. (A, right). Schematic of speech tokens arranged according to phonetic similarity. Acoustic properties that vary across speech tokens are indicated in insets. (B) Mean probability of correct classification of speech tokens using a 10 ms bin width across ensembles ranging from 1 to 20 multi-unit sites. Ouabain-treated animals show significant impairment, even though speech tokens were presented 20 dB above threshold for all sites. The probability of assigning the correct speech token identity by chance is indicated by the dashed gray line. (C) Mean confusion matrices from decoding single multi-unit PSTH patterns in sham (left) and ouabain (right) conditions. Axes are arranged according to schematic in (B).