Neurogastronomy: How the Brain Creates Flavor and Why It Matters

In all mammals, the output fibers from the olfactory bulb gather in a bundle called the lateral olfactory tract, which connects to the next stage, the olfactory cortex. The tract is relatively short in most animals, but very long in humans, an inch or so (up to 30 mm), in order to reach from the olfactory bulbs, sitting in front over the nasal cavity, to the olfactory cortex on the underside of the brain. The length reflects the expansion of the brain as the neocortex grew in size during mammalian and especially primate evolution. As in the case of the olfactory nerves, all our sensations of smell depend on this connection.

The olfactory cortex is little noticed when the many higher cognitive functions of the cerebral cortex are discussed. What is the olfactory cortex for? Why is it that the coordinated, multidimensional smell image from the olfactory bulb cannot be sent straight to the highest cortical level—the neocortex—to serve as the basis of perception?

Modern research reveals the olfactory cortex to have remarkable properties that make it an essential player in the human brain flavor system. Most significantly, it represents the transition from the steps of extracting features in the smell stimulus to the steps of creating the perceptual qualities of smell. It is where the external features of the outside world meet the internal features of our perceptual world.

If you are able to smell something, it is because of the olfactory cortex. Similarly, if you sense the flavor of something, it is because of the olfactory cortex. Understanding how the olfactory cortex functions has to be an important subject in neurogastronomy.

Out of the extracted odor image of the stimulus in the olfactory bulb the olfactory cortex creates the basis for the human perception of a unified smell perception, what is called an odor object. How does it do it?

To answer this question, we look inside the olfactory cortex to see what happens to the odor image. These experiments can’t be carried out in the human or the monkey, so the laboratory rat and mouse are necessary for studying them.

The main nerve cell in the olfactory cortex is a pyramidal cell. As shown in figure 7.1, its cell body is shaped like a small pyramid, giving rise to a large “apical” dendrite that ascends to the surface and several “basal” dendrites from the base of the pyramid. There must be something very useful in this arrangement of dendrites, because the pyramidal cell is the main kind of cell in all types of cerebral cortex. One could call it the CEO, the Chief Executive Officer, of the cortex. It is worth knowing something about this kind of cell. Your mind depends on it for normal thinking, and it is the target for degeneration in Alzheimer’s disease. You can say that your mind is what the pyramidal cell does.

To pursue our analogy, this CEO has two possible actions, which it exerts through branches of its axon. One is to send impulses to excite its co-workers, the interneurons. As in the olfactory bulb, and in fact as in most parts of the brain, the interneurons feed back inhibition onto an excited pyramidal cell, controlling its output, and onto neighboring pyramidal cells, to sharpen contrast. So far this seems similar to the way a mitral cell is organized.

However, there is a big difference that accounts for much of what goes on in all areas of the cortex. The axon collaterals also feed back excitation onto an excited pyramidal cell and its neighbors. These re-excitatory feedback collaterals in the olfactory cortex were discovered in 1973 by Lewis Haberly, using physiological recordings when he was a graduate student in my laboratory at Yale. They were also independently revealed, using anatomical methods, by Joseph Price at Washington University in St. Louis.

Feeding back excitation onto an already excited cell seems like a recipe for runaway excitation, but in fact it is a fundamental element in not only the olfactory cortex but in all types of cerebral cortex. In normal function this feedback excitation is counterbalanced by the feedback inhibition through inhibitory interneurons. We suggested that together with the feedback inhibition this forms a “basic circuit” for all cortical regions, and that the basic circuit is modified in the different cortical regions for the specific functions in each. This basic circuit is similar to the “canonical circuit” for the cerebral cortex, suggested by Rodney Douglas and Kevan Martin, then at Oxford University.

The fibers in the lateral olfactory tract carrying the odor image pass along the surface of the olfactory cortex and send many branches to the underlying layer, where they transfer their activity by means of synapses onto the most distal apical branches of pyramidal cells (see figure 7.1). There is thus a radical difference from the way input comes into the olfactory bulb, where receptor cells with the same response sensitivities—that is, the same “molecular receptive range”—converge onto one glomerular module. The mitral cells, the output of which still reflects the glomerular module to which they are connected, thus distribute their output across the olfactory cortex to many pyramidal cells. In this way, the information is changed from a mosaic image to a distributed representation of that image.

What is the nature of this distributed image? An answer was suggested by Haberly. After finishing his dissertation, he took postdoctoral training with Price in St. Louis, bringing together the two people who had put the re-excitatory collaterals on the map. Starting with the evidence for the smell patterns being sent to the olfactory cortex from the olfactory bulb, Haberly began to study the literature that was emerging on pattern recognition devices. He compared the anatomical and functional organization of the olfactory cortex that he knew so well with other parts of the brain doing similar kinds of pattern recognition.

In 1985, Haberly wrote a classic paper on the olfactory cortex in which he suggested that the “olfactory cortex serves as a content-addressable memory for association of odor stimuli with memory traces of previous odor stimuli.” He noted the properties necessary for the cortex to function in this manner: a large number of integrative units (pyramidal neurons and synapses) relative to the number of memory traces; a highly distributed, converging-diverging, input (from the olfactory bulb fibers); and positive feedback (the re-excitatory axon collaterals) via highly distributed interconnections between units. All these properties are found in the basic olfactory cortical microcircuit. Not surprisingly, this basic microcircuit is similar to that in the hippocampus, which is well known for its role in long-term memory (chapter 21).

The final property necessary for a cortex to mediate learning and memory consists of the synapses that are reinforced by coincident action of presynaptic and postsynaptic activity. This is the so-called Hebb rule, named after the psychologist Donald Hebb, who in 1949 suggested that this coincident action would build memory into brain circuits. This too is a property of the synapses in both olfactory and hippocampal cortices.

Haberly was particularly intrigued by a comparison of the recognition of odor images with recognition of faces in the visual system. He noted that processing of the highly distributed and complex patterns of activity in the olfactory cortex is different from the initial feature extraction known to be carried out in the primary visual cortex, resembling more closely the discrimination of complex visual patterns such as faces carried out by higher-order visual association areas. We have already seen the usefulness of a comparison with the visual system. As discussed in chapter 8, there is thus an analogy between recognition of the complex patterns laid down by odors in the olfactory cortex and recognition of the complex patterns of faces in visual association areas. By studying the microcircuits, we can now see that in both cases, the extensive horizontal connections through re-excitatory collaterals are essential for the storage and recognition mechanisms.

This has been a fertile hypothesis for the field of smell. An elegant summary of the experiments supporting it is contained in Learning to Smell: Olfactory Perception from Neurobiology to Behavior by Donald Wilson of New York University and Richard Stevenson of McQuarrie University in Sydney, Australia. The main point is that, whereas the representation of smells in the olfactory bulb is driven by stimulus properties, the representation in the olfactory cortex is memory based. Wilson and Stevenson identify several defining characteristics of how this works, which may be summarized as follows:

1. The olfactory cortex responds especially to changes in its input signals from the olfactory bulb; it adapts (reduces its responses) to continued stimulation with the same smells. This, in fact, was the reason that first attempts using functional imaging in the human failed to identify responses to smell in the olfactory cortex. Noam Sobel and his colleagues realized that the responses were adapting out, and used changing smells to reveal the olfactory cortex activity.

2. This system learns. The basic cortical circuit has the ability to improve its performance with repeated exposure to different smells. The recurrent excitation strengthens the cells activated by input from the olfactory bulb. The lateral inhibition enhances the contrast between activated and less activated cells. Finally, the synaptic strengths change so that the system can store these changes as a memory and match them to the input.

3. These changes with learning enable the system to improve its ability to match an input pattern to a stored pattern, so that finer discrimination between more similar smell molecules can occur.

4. These changes also enable the system to improve its signal-to-noise ratio, so that detection and discrimination of a particular smell can be enhanced against a background of many smells.

5. The olfactory cortex microcircuit functions to take an input reflecting many diverse stimuli and construct out of it a coherent odor object. This is an analogy with how the visual system takes an input of different shapes and sizes and constructs a “visual object.” An important aspect of such an object is that seeing only a small part of it still enables the system to “fill in” the missing parts. This is a feature of all sensory experience. As we have noted, we catch a glimpse of a distant figure, a few notes of a melody, a whiff of a smell, and can instantly “fill in” the rest. This especially applies to perceiving a face or just a part of a face, as discussed in chapter 8. It also means that the system “degrades gracefully” if damaged.

6. The odor object is in a form that can be combined with other sensory inputs to produce the sensation of flavor. This occurs at the final level in the smell pathway, the orbitofrontal cortex.

Given these impressive properties of the olfactory cortex, a key question is whether conscious perception of smell arises there. I have asked many behavioral psychologists this question, but apparently the crucial experiment of interfering with the next stage has never been done, at least not in primates. (For evidence from trauma to that level in humans, see chapter 25.)

The argument in favor of conscious perception arising in the olfactory cortex is that, although many call this the primary olfactory cortex, it is not really equivalent to the primary cortex in other sensory systems, where that term is reserved for the receiving area in the neocortex. The olfactory cortex, as Haberly suggested, is more equivalent to a higher-association area in other sensory systems, even though it is not yet at the neocortical level.

The argument against conscious perception arising in the olfactory cortex is that the smell information has not yet passed through the thalamus or reached the level of the neocortex, which is necessary for all other sensory systems.

So we need to go to the next level—the orbitofrontal area of the neocortex—for possible answers to where conscious smell arises.

In addition to the role of the olfactory cortex in smell perception, another function began to emerge some two decades ago with the evidence that it contains an area sensitive to amino acids in the diet. Of the 20 amino acids required for building proteins, 10 are essential; if one of them is missing from the diet, an animal’s health begins to fail and it will die if the deficit is uncorrected. Rats will cease feeding within 30 minutes if their chow lacks just one of these. What is the mechanism?

A series of studies over the past 20 years has shown that the sensor is in the brain and, surprisingly, has narrowed it to the olfactory cortex. Evidence from Dorothy Gietzen at the University of California, Davis, and her colleagues Shuzhen Hao and Tracy Anthony published in 2007 indicates that the pyramidal cells contain a molecular mechanism that senses the lack of an essential amino acid by its inability to “charge” its appropriate transfer ribonucleic acid (tRNA) molecule with that amino acid. How this is communicated to the cell membrane to change the cell’s activity, and what is the pathway for communicating this message to the rest of the brain, is still under study. One can speculate that the exquisite balance between excitation and inhibition in this region enables the cells to be sensitive detectors of slight changes in the presence of the amino acid. This is an unexpected hidden function of the mammalian olfactory system, one that is apparently present in humans. It may be crucial to the nutrition of people in conditions of poverty and starvation around the world. It may also be a critical element in the need for vegetarian diets to include foods with all the essential amino acids. The relation to flavor is only indirect, but it presumably means that for an omnivore such as ourselves vegetable flavors must be learned in order to supply the needed amino acids.