Encoding and Decoding Mechanosensory Information on Multiple Timescales
MetadataShow full item record
CitationBaker, Allison Eva. 2015. Encoding and Decoding Mechanosensory Information on Multiple Timescales. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractNatural stimuli contain information on multiple timescales, ranging from milliseconds to seconds, or longer. Therefore, sensory circuits must be able to encode a wide range of frequencies. A general strategy is to dedicate different neurons to different timescales, with some neurons encoding fast timescales and others encoding slow timescales. Another strategy neural circuits use to increase information transfer is to establish opponent channels, where some neurons detect one type of change in a stimulus (e.g., an increase in intensity) while others detect the opposite change (e.g., a decrease in intensity). Opponency and temporal diversity are known to coexist in neural circuits, but it remains unclear how the two might interact to extract specific features of natural stimuli. We investigated this interaction from the perspective of Johnston’s organ, the largest mechanosensory organ in the fruit fly Drosophila. Johnston’s organ neurons (JONs) encode a wide range of antennal movements by using the two aforementioned strategies: temporal diversity and opponency (i.e., some JONs are excited by pushing the antenna backward, and others are excited by pulling the antenna forward). In this study, we identify a type of neuron in the brain that combines input from a pair of opponent JON subgroups. As a population, these central neurons represent a broadband temporal filter that transmits a wide range of low frequencies. Interestingly, these neurons are relatively uniform in their responses to DC stimuli, such as those produced by gravity, but show opponent responses to amplitude modulated sound (some neurons are excited by increases in sound energy, while others are inhibited). We show that this “frequency-specific opponency” can arise from the interaction between temporal filtering and a rectifying nonlinearity in JONs, followed by mixing of these peripheral signals in the brain. Moreover, this frequency-specific opponency allows central neurons to encode sounds over a wider range of antennal resting positions. We also show that these peripheral signals combine to produce temporal diversity in the central neuron population. Overall, our results illustrate how temporal filters and opponency can interact to extract temporal features of a stimulus in a flexible and robust manner.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:17467466
- FAS Theses and Dissertations