Ronald R. Hoy
Department of Neurobiology & Behavior
W214 Mudd Hall, Cornell University
Ithaca, NY 14853 USA
phone :: (607) 254-4317
FAX :: (607) 254-1303
I am interested in elucidating the neural mechanisms underlying behaviorally-relevant computations. Put simply, the brain is a big computer and I want to reverse engineer it. I focus on two behaviors mediated by the auditory system in particular: sound localization and, to a lesser extent, speech recognition. Collectively, determining what the sound is and where it's coming from are the two main tasks the auditory system of any species must solve. My study species of choice are Ormia ochracea and barn owls. The former are flies which parasitize crickets by using them as hosts for larvae; the latter are nocturnal predators which rely on field mice for food. In both cases, a highly-developed and specialized auditory system is used to both recognize and localize their targets. What drew me to studying neuroscience, instead of continuing my undergraduate education in computer science and psychology, was a disappointment with the design and performance of the algorithms used by the artificial intelligence community. My hope is to facilitate better progress towards the creation of an intelligent machine through delineating how nervous systems solve similar computational problems.
A behavioral study of Ormia ochracea's ability to track and fixate both auditory and visual stimuli. Recent work has shown that Ormia can phonotax to sounds quite well in a vertically-oriented Y-maze walking task (Arthur and Hoy, 2006a). This experimental paradigm, however, does not preclude the use of so-called closed-loop cues (i.e. those created through the movement of the head and body). To address this issue, experiments are now underway in which flying flies are held in a fixed position with a rigid tether. The forces they apply to the tether are used to infer the direction of the percept: yaw torque corresponds to azimuth, while translational lift corresponds to elevation. The stimulus, either a speaker playing cricket calls or a piece of white card stock on a black background, is mounted on a motor-driven rotating boom. Preliminary data in the azimuthal plane show that gravid females readily track both the auditory and visual stimuli when the boom is slowly swept back and forth (Arthur and Hoy, 2006b). Moreover, they will fixate (keep the boom centered directly in front of them) both stimuli if a negative feedback loop is engaged between the force they apply to the tether and the direction the boom moves. Three important findings will likely result from this study: (1) a demonstration of auditory motion sensitivity; (2) a direct comparison of the acuity of the visual and auditory modalities in the same species using the same task; and (3) an assessment as to whether Ormia can localize in elevation.
A histological 3D reconstruction of whole Ormia ochracea used to create a predictive model of HRTFs. Head-related transfer functions (HRTFs) describe how the amplitude and phase of a sound change as it travels from the source to the eardrums. Changes can occur due to diffraction, reflection, interference, viscosity, and various other acoustic phenomena. Because Ormia's ears are on the prothorax in the narrow cleft between the head and the thorax, it is difficult to measure HRTFs directly without removing or retracting the head, both of which might have dramatic consequences on the HRTF itself. This study will use serial histological sections to make three dimensional reconstructions of entire Ormia females, and then use the dimensions of the cuticular surface to make a computational model of the sound field in the cleft. The model will be particularly informative about how rotations of the head, which Ormia normally make during saccades to a cricket, might effect the sound field. Moreover, the 3D reconstruction data set can be used to systematically look for asymmetries between the left and right ears, an important point to investigate for such asymmetries are used by barn owls to localize in elevation.
An extension of Lewicki's (1994) spike sorting algorithm. The ability to automatically separate single neuron activity from multi-neuron recordings can greatly enhance the quality of the analysis, as well as make experimental protocols vastly simpler. Lewicki's spike sorter has received widespread use since it's introduction due to it's ability to decompose overlapped spikes with an algorithm firmly based in bayesian statistics. In the elapsed time however computer technology has advanced to the point where certain practical limitations and assumptions made by the program for computational efficiency are no longer necessary. This project will update Lewicki's algorithm to (1) permit the analysis of multi-electrode recordings, (2) model the refractory period of neurons, (3) incorporate a time-dependent non-gaussian noise model, and (4) automatically adjust itself to changes in spike shape over time. In addition, the code will be re-worked to (a) handle more than two overlapped spikes, (b) take advantage of computers with multiple processors and vector processing subunits, and (c) interface with Matlab. The immediate application and impetus for such improvements is the analysis of whole auditory nerve recordings in the moth Cycnia tenera, which to date have been sorted by hand.
Arthur, BJ and Hoy RR (2006b). Tethered-flight sound-localization measurements in Ormia ochracea. Invertebrate Sound and Vibration, Toronto, Canada PDF
Arthur BJ and Hoy RR (2006a). Sound localization ability of the parasitoid fly Ormia ochracea in the elevational plane. JASA 120 (3): 1546-1549 PDF
Arthur BJ (2005). Distribution within the barn owl's inferior colliculus of neurons projecting to the optic tectum and thalamus. Journal of Comparative Neurology 492 (1): 110-121 PDF
Arthur BJ (2004). Sensitivity to spectral interaural intensity difference cues in space-specific neurons of the barn owl. Journal of Comparative Physiology A, 190 (2): 91-104 PDF
Saberi K, Takahashi Y, Konishi M, Albeck Y, Arthur BJ, Farahbod H (1998). Effects of interaural decorrelation on neural and behavioral detection of spatial cues. Neuron 21 (4): 789-798 PDF
Lewicki MS, Arthur BJ (1996). Hierarchical organization of auditory temporal context sensitivity. Journal of Neuroscience, 16 (21): 6987-6998 PDF
Arthur BJ (1996). A silicon model of the auditory neural perception of frequency modulated tones. International Symposium on Circuits and Systems, 4: 213-216. PDF