This proposal aims to explore the effect of speech processing strategies that may improve speech perception by users of pulsatile cochlear implants. In addition, the proposed research aims to probe the underlying reasons for the perceptual improvements we may observe. Stimulation strategies currently used by speech processors for cochlear implants will be modified in two ways: 1) frequency-to-electrode maps will be designed to ensure tonotopicity, i.e., that electrical pulses encoding higher frequency acoustic energy are sent to electrode pairs closer to the cochlear base, and 2) stimulation rate will be substantially higher than the FO rate that is currently used. The experimental hypotheses behind these strategies are: 1) Natural spectral representation- tonotopical stimulation results in more 'natural' percepts, making it easier for users of pulsatile cochlear implants to interpret spectral acoustic cues. (2) Adequate representation of temporal cues and dynamic spectral cues- high stimulation rates (400 to 800 pulses per second as opposed to 80 to 200) will provide a better representation of fast spectral transitions, resulting in improved perception of different stop sounds in particular and improved perception of the place feature in general. In addition, these high stimulation rates will provide better representation of temporal cues, some of which (such as VOT) underlie certain voicing and manner distinctions. (3) Sufficient perceptual contrast. The ability to discriminate along acoustic continua known to be relevant in making phonetic decisions is an essential underpinning to perceptual performance. We will attempt to identify what acoustic cues underlie perceptual differences under different stimulation strategies. The proposed research includes two within-subject studies. Both studies will employ vowel and consonant confusion tests, and a words-in-sentence test. In the first study, subjects will be tested with their current maps to establish a baseline. Then they will be fitted with the new (tonotopic) maps and tested at the beginning and at the end of 2-3 weeks of tonotopic map use. Finally, they will be reprogrammed with the original map and re-tested to ensure baseline replicability. The second study will compare the best map for each patient (either their original map or the new tonotopic map) against a high-rate version of the map. Since currently available speech processors cannot implement the proposed high-rate strategies, all testing and training for this second study will be done in the laboratory, using computer-stored stimuli delivered through a special purpose interface. Confusion matrices will be analyzed using Sequential Information Transfer Analysis and log-linear modeling. In addition, stimuli will be analyzed to determine what specific electrical stimulation cues are employed by different subjects to identify speech sounds.
