The Smell of Representations

ingful and are expected to elicit robust and reproducible The nose and palate never doubt / Their verdict on the physiological and behavioral responses. All of the above world without, / But instantaneously condemn / Or may facilitate the correlation of stimuli with the neuronal, praise each fact that reaches them: / ... Compared with physiological, and behavioral response. almost any brute, / Our savoring is less acute, / But, Codes Transformed subtly as they judge, no beast / Can solve the mystery It is fair to say that so far, the nose beats the mouth of a feast, / Where love is strengthened, hope restored, / in the competition for the attention of neuroscientists In hearts by chemical accord. (W. H. Auden, “Smelt (though the taste system has a lot to offer as well; e.g., and Tasted”). Odors and tastes dominate our life more Bures et al., 1998). Olfactory information is the outcome than we commonly tend to concede. Decisively, though of the interaction of volatile chemical compounds (comoften unconsciously, they have a lot to say about what monly of molecular weight , 0.5 kDa) with olfactory we eat, with whom we spend our life, and whether we receptors (ORs) in the sensory epithelium. In mammals, feel that the world smiles at us or pushes us under the there are millions of neurons in the olfactory epithelium rain. To love and hatred the nose contributes more than and about 1000 different types of G protein ORs, enthe cortex. It did so long before even the simplest cortex coded by a multigene family (Buck and Axel, 1991; Malhad emerged in evolution. But, primeval as they may nic et al., 1999). In vivo, the epithelium is immersed in be, the chemical senses are gaining popularity in the mucus (or sensillum lymph in invertebrates). The stimusearch for the golden bough of the neurosciences, lus–receptor interaction is therefore preceded by odornamely, the encoding of internal representations. This ant solubilization and probably transportation by odortrend is exemplified by the recent papers by Rubin and ant-binding proteins. This imposes constraints both on Katz (1999), who used intrinsic imaging to unravel what the experimental control of the on and off responses the rat’s nose tells the rat’s brain, and of Galizia et al. and on their kinetics. Indeed, this is one of the reasons (1999), who used calcium imaging to see what the bee why some brain physiologists, who fancy swift sensory smells. responses in the millisecond range, are not very enthusiBoth types of studies illustrate nicely how sophistiastic about smell and taste. The neurons in the olfactory cated imaging can be used to relate stimuli to brain epithelium transmit information to the olfactory bulb. activity maps that are construed as elements of internal The most characteristic structure in the bulb, located representations. Internal representations are neuronally below the olfactory nerve layer where the olfactory encoded, structured versions of the world that could nerves interweave, is composed of spherical regions potentially guide behavior (Dudai, 1992). Understanding called glomeruli (each 50–200 mm across, numbering the “syntax” and “semantics” of representations is esabout 1000 in the mouse and 2000 in the rat). It is here sential to understanding how the brain generates cogniwhere the individual olfactory axons branch and syntion, controls behavior, learns, and remembers. The curapse onto the dendrites of the relay neuron. The axons of rent majority view is that internal representations are the latter gather at the posterolateral part of the olfactory encoded in spatiotemporal patterns of activity of neubulb to form the lateral olfactory tract, which courses ronal populations. The search for representations deover the basolateral surface of the forebrain and conmands specific methodologies for recording and analyzveys the information to the olfactory cortex. In rodents, ing such activity. It also requires the identification of this paleocortical complex includes the anterior olfacappropriate biological Rosetta stones, which would tory nucleus, piriform cortex, olfactory tubercle, cortical present side-by-side spatiotemporal neuronal states, and medial nuclei of the amygdala, and transitional entophysiological states, sensory input, and behavioral outrhinal cortex. Olfactory information also reaches the put. In this respect, the chemical senses may provide thalamus and neocortex. some modest but useful advantages. First, chemical Attempts to analyze information processing in the olsenses, being “primitive,” may prove to be relatively factory system currently focus on the olfactory epithe“simple.” Second, chemical stimuli can be selected that lium, the olfactory bulb, and the piriform cortex. Molecuare very well defined and easily modified along a selar biology and neuroimaging have been especially lected molecular dimension. And third, the world of useful in the analysis of processing in the first two sta-