Redwood Neuroscience
Title: “Communication and cortex: the computational
neuroethology of mouse vocalizations”
Robert
Liu
Abstract:
Ecological
theories hypothesize that an organism's
neural coding strategy reflects the statistics and regularities of the natural
world that it experiences. Evidence for this view is building in both vision
and audition. Further progress could be made by developing new animal models to
explore the neural representation of natural stimuli. Towards this goal, I am
pursuing a computational neuroethological approach to
auditory processing in the mouse, where genetic techniques may eventually
elucidate neural mechanisms. Coding is studied using a stimulus that naturally
elicits an observable behavioral response. After characterizing the statistical
properties of this natural stimulus (etho-
component), neurons are examined for their ability to represent these
characteristics (neuro-). Eventually, coding
algorithms can be evaluated (compu-) using behavior
as a constraint. This provides a systematic approach to the question of how the
neural code represents the statistics of behaviorally-relevant sensory stimuli.
I am carrying out this paradigm in the context of mouse ultrasound (> 25
kHz) communication. These vocalizations are produced in at least two behavioral
situations. First, the calls of isolated mouse pups elicit retrievals to the
nest by their mothers. Second, ultrasounds by adult males encountering females
seem to attract the latter and reduce its aggressiveness. Interestingly, the
calls from these two contexts fall into two acoustically distinct spectral and
temporal categories, laying a possible foundation for their categorical
identification and discrimination. Moreover, auditory cortical neurons in
mothers appear particularly well matched to a specific temporal statistic of
pup calls: the repetition period between successive vocalizations. This is
reflected first by the ability of the neural population to follow pup calls up
to (and not far beyond) the dominant vocal repetition rate; and second, by
oscillations in neural firing that have periods matching the distribution of
natural vocalization periods. Further studies will examine the possible
experience-dependent nature of the neural representation of communication
vocalizations.