Encoding of scents by the brain

I happen to be interested in smell right now -- it's a system in which chemical traces clearly function as iconic signs, but it's extremely ancient, and humans have this immense array of chemoreceptive genes, of which many have recently evolved into pseudogenes. So I found this commentary by Catherine Dulac very interesting. It gives some details about how mammals go from the many genes to the ability to pick out complex chemical combinations as odor signs. The abstract:

Sparse Encoding of Natural Scents
Natural scents are mixtures of tens to hundreds of individual chemical compounds. While previous studies have investigated how the olfactory bulb responds to simple odorants, how the nose and the brain respond to and interpret complex mixtures is less clear. In a paper in this issue of Neuron, Lin et al. combined gas chromatography and intrinsic signal imaging to examine the responses of individual olfactory bulb glomeruli in the mouse to natural odors and their component parts.

This passage from early in the article introduces the problem of encoding:

Remarkably, the neural response to complex and behaviorally relevant stimuli shows that the sensory processing of natural signals is both quantitatively and qualitatively different than the mere sum of their simple constituent features (Nelken, 2004 and Kayser et al., 2004). Simply put, millions of years of evolutionary pressure have enabled the brain to readily respond to objects commonly encountered in the environment and provided advantageous shortcuts to the systematic dissection of all individual features.
The discovery of a large gene family encoding olfactory receptors (OR) and its logic of expression, including the choice of a single OR gene by individual sensory neurons, and the convergence of axon terminals from all neurons expressing the same OR to one or few glomeruli have lead to the early prediction that odor quality is encoded by the activation of defined and restricted sets of glomeruli (Ressler et al., 1994 and Vassar et al., 1994)....

In other words, narrowing down individual odor signs would occur near the receptors, as receptors that contribute to a single odor would be coordinated into downstream signaling close to the source. These would add up by components into entire smells, or olfactory signs.

...Functional studies aiming at capturing the response of single olfactory neurons and individual glomeruli to simple chemical compounds largely supported this prediction....

So far so good; individual receptors actually do correspond to "simple" odors.

...However, one noticeable caveat became increasingly difficult to ignore: even simple chemical compounds, for example molecules containing a short hydrocarbon chain and few functional groups such as alcohol, ketone, or aldehyde motifs, generate specific but widespread responses that encompass surprisingly large subsets of receptor types and corresponding glomeruli (Malnic et al., 1999, Meister and Bonhoeffer, 2001, Rubin and Katz, 1999 and Uchida et al., 2000). A direct extrapolation to complex chemical cues, for example those encountered in food, plant, or animal odors, would therefore predict the recruitment of large and overlapping fractions of receptor types and corresponding glomeruli in the representation of natural scents, which are each composed of hundreds of constituent chemicals. Accordingly, the amazing specificity of olfactory recognition-humans are thought to be able to identify over 400,000 distinct odors (Mori et al., 2006)-requires the ability for the brain to discriminate large and complex combinations of glomerular activities, a process that could be facilitated by extensive inhibitory interactions between glomeruli.

Hmm...these smells seem to light up much of the entire olfactory sensing array, which would make it fairly hard to distinguish smells from each other.

The study by Lin and colleagues referred to by the commentary basically tested much lower concentrations of chemicals and found that experimental concentrations had just been too strong, causing activation of olfactory neurons that normally would not have been activated. This is a round about way of saying that everybody had been dousing these mouse neurons with too much Drakkar Noir, leaving them unable to distinguish smells from each other.

The result suggests that the simple additive model actually probably would work to explain odor reception:

In order to understand the relationship of the response to a mixture with the response to its parts, we compared the activation patterns seen for each stimulus to the cumulative response to its fractionated volatiles. One possibility is that extensive interactions among the responsive glomeruli prevent the prediction of the response to a stimulus from the responses to its components. On the contrary, however, we found that the response to a natural stimulus closely matches a linear summation of the responses to its monomolecular elements (Lin et al. 2006:941).

What about humans, and our many deactivated receptor genes?

Further, animals have selected to respond to few salient features among complex sources of odors, a theme already put forward by the late Larry Katz and his group describing the restricted response to mouse urine and body odor (Lin et al., 2005 and Luo and Katz, 2004) and here further explored with a large set of natural scents.

It seems to me that one option is that some of this processing of sparse cues may have been taken over by other neural processes. If odor sensing is already very selective in receiving certain cues, a need to be able to disinguish certain kinds of cues in more subtle concentrations might well involve higher level mental processes and leave basic olfactory reception in no need of improvement.

Or then there is this:

An amusing detail of the present study indicates how the ever resourceful humans may have found a way to escape olfactory poverty: in several instances the authors show that treatment of natural odorants by high temperature increases the number of olfactory salient features, a well known foundation of the very human specialty of cooking (McGee, 2004 and Wrangham and Conklin-Brittain, 2003).

A suggestion that is far more than amusing -- if odor concentrations are actually increased by cooking to levels beyond pleasant (always my sensation when I smell lamb cooking, for example), then lower expression or even absence of certain olfactory receptors might easily have been selected.


Dulac C. 2006. Sparse encoding of natural scents. Neuron 50:816-818. DOI link

Lin DY, Shea SD, Katz LD. 2006. Representation of Natural Stimuli in the Rodent Main Olfactory Bulb. Neuron 50:937-949. Article summary