Joan M. Sinnott - US grants

Although a current view of human speech sound perception argues for the use of "general mammalian auditory mechanisms", there is at present no extensive source of precise, comprehensive data comparing either the psychoacoustic or speech discrimination capacities of humans and other animals. As a start towards a more systematic study of the contribution of sensory vs. linguistic processes to speech perception, we propose a series of experiments designed to directly compare humans and monkeys in the discrimination and identification of speech and non-speech acoustic signals. Monkeys will be trained with operant conditioning techniques using positive reinforcement. A go, no-go procedure will be used for discrimination and a go-right, go-left procedure for identification. Humans and monkeys will be tested with identical stimuli and procedures. We already have preliminary data indicating that monkeys are much closer to humans in their capacity for intensity discrimination than for frequency discrimination. Accordingly, we have made predictions concerning the relative salience for humans and monkeys of speech sounds based on spectral vs. intensity cues. The ultimate objective is to precisely define both the quantitative and qualitative limitations of the monkey auditory-sensory system in relation to the human, in order to determine to what extent the monkey can provide an adequate model of human speech sound processing at the phonetic level. This work will have implications for current models of speech perception and the sensory encoding of speech signals at the earliest stages of perceptual analysis.

Although a current view of human speech sound perception argues for the use of "general mammalian auditory mechanisms", there is at present no extensive source of precise, comprehensive data comparing either the psychoacoustic or speech discrimination capacities of humans and other animals. As a start towards a more systematic study of the contribution of sensory vs. linguistic processes to speech perception, we propose a series of experiments designed to directly compare humans and monkeys in the discrimination and identification of speech and non-speech acoustic signals. Monkeys will be trained with operant conditioning techniques using positive reinforcement. A go, no-go procedure will be used for discrimination and a go-right, go-left procedure for identification. Humans and monkeys will be tested with identical stimuli and procedures. We already have preliminary data indicating that monkeys are much closer to humans in their capacity for intensity discrimination than for frequency discrimination. Accordingly, we have made predictions concerning the relative salience for humans and monkeys of speech sounds based on spectral vs. intensity cues. The ultimate objective is to precisely define both the quantitative and qualitative limitations of the monkey auditory-sensory system in relation to the human, in order to determine to what extent the monkey can provide an adequate model of human speech sound processing at the phonetic level. This work will have implications for current models of speech perception and the sensory encoding of speech signals at the earliest stages of perceptual analysis.

Animals perform much like humans in many psychoacoustic capacities. However, several tasks reveal striking differences between animals and humans. Animals are markedly inferior to humans in pure tone frequency discrimination. Animals are deficient in discriminating intensity and duration decrements. Animals are deficient in certain tasks involving octave generalization. On the other hand, parakeets perform better than humans in forward and backward masking tasks. The proposed research will compare humans, macaque monkeys and chinchillas in processing simple tones, musical tones, animal sounds, and speech sounds, with emphasis on those tasks for which animals have been shown to be markedly inferior (or superior) to humans. All species will be tested using the same stimuli, procedures, and threshold criteria. Both go, no-go discrimination and left/right identification procedures will be used. Monkeys and chinchillas will be trained with operant conditioning techniques and food reinforcement. The specific objective is to precisely define similarities and differences in human, non-human primate, and non-primate mammalian hearing, in order to determine which species are adequate models for human auditory capacities. The ultimate goal is to integrate existing psychoacoustic data from humans, monkeys, chinchillas and birds in order to more clearly delineate which aspects of human hearing are inherited from vertebrate, mammalian or primate ancestors, and which are species- specific to humans.

I joined the USA at the invitation of Dr. Charles Brown, Associate Professor of Psychology, in order to collaborate with him in building a new laboratory to study primate hearing and communication. The USA has provided us with an ideal environment totally supportive of primate research. This RCDA will allow me to develop my own research program independent of Dr. Brown and facilitate my promotion to a senior-level research rank. During the tenure of this award, I aim to obtain the required comparative data to write a theoretical article designed to evaluate evidence for differential use of the two primary auditory coding mechanisms (place vs. periodicity) in humans and other animals. I plan to become proficient in the technical aspects of managing a high-tech, state-of-the-art acoustic laboratory and to learn new techniques involved in the measurement and analysis of auditory evoked potentials. I will dedicate time for supervision and training of individual students who engage in research projects in our laboratory. My research compares the psychoacoustic capacities of humans, monkeys, and gerbils. Emphasis is on those tasks for which humans have been shown to perform markedly different from animals; e.g. pure tone frequency discrimination and forward and backward masking tasks. All species will be tested using the same stimuli, discrimination procedures, and threshold criteria. Animals will be trained with operant conditioning techniques and food reinforcement. The psychoacoustic performance of gerbils will be followed through their lifespans in order to develop an animal model for aging of the auditory system. Other experiments will compare humans' and monkeys' abilities to categorize musical tunes and monkey sounds in order to develop animal models for the perceptual processing of complex stimuli. The ultimate goal is to integrate data from humans, monkeys and other animals in order to more clearly delineate which aspects of human hearing are inherited from generalized mammalian vs. primate ancestors, and which are species-specific to humans.

Through the application of digital audio technology, a new microcomputer-based laboratory will expose psychology students and speech pathology and audiology students to the fundamentals of acoustics, sound propagation, distortion, time-domain and frequency-domain representation of sound, sound spectrograms, resonance and the format structure of the human voice, the source-filter theory of speech production, categorical perception of speech and non-speech stimuli, fused and segregated perceptual images, multidimensional scaling, and associated topics. The laboratory will employ an active hands- on approach, so that students will acquire a firm understanding of the relationship between the physical properties of sound and sound production on one hand, and the perceptual consequence of changes in stimulation on the other. Both major- and non-major upper-division undergraduate students will utilize this laboratory .

The long-term objective is to investigate to what extent "special" versus "general" processing mechanisms contribute to human speech perception. The specific aims are to study the perception of trading relations in English phonology by humans and monkeys. A trading relation is one in which multiple, and seemingly unrelated, acoustic cues contribute to the same phonetic percept. Trading relations are studied by synthesizing both normal and altered speech sound continua in which the value of one acoustic cue is reduced. Listeners perceive a shift in the identified phoneme boundary to compensate for the reduced cue. Both general mammalian auditory and human-specific, articulation-mediated processes have been proposed to account for trading relations. The comparative approach is a powerful tool to help distinguish between these two alternatives. We propose to train two different monkey species to identify speech sounds using both go/no-go and go-left/go-right procedures in order to test their sensitivity to trading relations. Monkeys will be trained with operant conditioning techniques and positive (food) reinforcement. Two human subject populations consisting of both native American and non-native (Hispanic) speakers of English will be tested with the same basic procedures. Comparisons between these subject groups will help to clarify if trading relations are native- language-specific, human-specific, or reflect general processing strategies of the primate auditory system. Results will be related to a theoretical framework put forth by developmental psycho-linguists that distinguishes between "maintenance", "facilitation" and "induction" as different processes by which linguistic experience affects speech perception.