Skip to Main Content

Project List


Full Shepherd bibliography with chapters, reviews and books available here.

The nature of the interaction between an odor molecule and an olfactory receptor protein defines the way that information about the identity of the odor molecule is encoded by the brain.  Rather than the traditional idea of key and lock, the more likely mechanism involves a broad graded reactivity between many receptors with different specificities of binding regions. 

For an interest in integrative neuroscience, this molecular mechanism is critical in giving insight into the primitives that are the basis for encoding olfactory information.  The first modeling studies were carried out by a talented undergraduate, then graduate student, Michael Singer (Singer et al, 1995; Singer, 2000) and by the Firestein lab (2000).  The modeling studies provide support for binding regions that are optimal for certain odor molecules but can interact to varying degrees with a range of others, showing close resemblance to the tuning curves for the first receptors identified and their odor molecule preferences.

Imaging of odor induced activity in the olfactory glomeruli was introduced using 2deoxyglucose (Sharp et al, 1975) followed by high-resolution fMRI (Xu et al, 2003). These are only two of the many methods that have been used to show patterns of activation of the olfactory glomeruli by different odor molecules.  They are consistent with the fact that subsets of similar olfactory receptor cells converge on single glomeruli (Mombaerts et al, 1996), giving rise therefore to differential activation of the glomerular sheet.  The patterns of activation are likely more extensive than are revealed by these methods, accounting for more widespread encoding of odor molecules, which means more resistance to losing the discrimination of odors to infection and other damage in the lifetime of animals dependent on the sense of smell for behavior.  See Shepherd et al (2021) for details.
An early electrophysiological study with Charles Phillips and Tom Powell of the olfactory bulb postulated that long-lasting inhibition of mitral (M) cells is due to the granule (G) cell acting as an interneuron (Shepherd, 1963). The first computer models of brain neurons with Wilfrid Rall predicted that the inhibition is mediated by unprecedented dendrodendritic interactions between mitral and granule cells (Rall and Shepherd, 1968), which subserve both lateral inhibition and oscillatory activity underlying olfactory processing. The prediction was supported by electronmicroscpic evidence for dendrodendritic synapses between the cells (Rall et al, 1966). This collaboration is fully explained in the Rall Archive (see below). In the intervening 50 years multiple studies have supported this model (Shepherd et al, 2021). The illustration from the study of Xiong and Chen (2002) shows propagating impulses in the lateral dendrites of mitral cells, which enable the lateral inhibition to process the patterns of odor activation widely throughout the olfactory bulb. Current studies are giving a deeper understanding of the central role that mitral-granule inhibition plays in olfactory processing underlying odor perception.

Video: Computational Modeling

Computational Modeling Video
Recognition of odor images, and the importance of retronasal smell, are giving us a new understanding of how the brain creates the perception of food flavor.

Wilfrid Rall research notebooks (1962-1972), correspondence and scientific instruments

We stand on the shoulders of giants. I’ve explored this perspective in studies of the founding of the neuron doctrine (two eds.), the role of Wilfrid Rall in creating computational neuroscience, the rise of modern neuroscience in the 1950s, Angelo Mosso and the origins of brain imaging and cognitive neuroscience, and a biography in preparation of John Farquhar Fulton.

The Spanish TV dramatization of the breakthrough of Ramon y Cajal that launched the beginning of modern neuroscience at the cellular level.