His graduate studies in 1963 of the
electrophysiology of the
olfactory bulb produced one of the first diagrams of a
brain microcircuit. Building on this work he collaborated with
Wilfrid Rall, just then founding the new field of
computational neuroscience, at the
National Institute of Health (NIH) to construct the first computational models of brain
neurons: the
mitral and
granule cell. This predicted previously unknown
dendrodendritic interactions between the mitral and granule cells, subsequently confirmed by
electron microscopy. These interactions were hypothesized to mediate lateral inhibition in the processing of the sensory input as well as generate oscillatory activity involved in odor processing. The model suggested active properties in the dendrites, which was subsequently confirmed, through which the model accounts for non-topographic interactions throughout the
olfactory bulb. This paper was included in the "Essays on
APS Classic Papers" series. The culmination of this line of research within the domain of computational neuroscience can be identified in a seminal paper published in 1968 by Wilfrid Rall and Gordon M. Shepherd. Distinguishing itself from previous works that offered conceptual frameworks, this paper delved into the intricate architecture of a particular system, namely the
olfactory bulb. The next problem Shepherd addressed was how odors are represented in the brain. Using the
brain imaging technique of the time, 2-deoxy-D-glucose (2DG)
autoradiography, it was shown for the first time that odors are encoded by different spatial activity patterns in the
glomeruli of the olfactory bulb, a tangle of nerves formed by connections between mainly olfactory sensory cells and mitral cells. This demonstrated that the neuronal basis of the sense of smell in vertebrates consists of the representation of odors by glomerular activity patterns, or "odor images," which are then processed by the widely distributed microcircuits of the olfactory bulb. The odor-induced patterns included a "modified glomerular complex," which was the first of a subsystem of specific glomeruli in the olfactory bulb. In their research, Shepherd's laboratory employed the olfactory bulb as a foundational model to explore the integrative functions of neuronal dendrites. Their findings revealed that dendrites possess the capacity to house multiple computational units, that
action potentials propagating in a retrograde direction within dendrites can execute specific functional operations, and that dendritic spines (small membrane protrusion from a neuron's dendrite that typically receives input from a single axon at the synapse) can function as autonomous input-output units. Furthermore, the
laboratory contributed a fundamental circuit for the
olfactory cortex of the brain. Concurrent with this research, novel concepts were developed to replace the classical '
neuron doctrine', and the term 'microcircuit' was introduced to characterize specific patterns of synaptic interactions in the nervous system. ==Later work==