Human Olfaction, Crossmodal Perception, and Consciousness

Abstract

When philosophers attempt to provide a theory of perception they usually focus exclusively on vision, assuming, without argument, that it can serve as a model for perception in general, extendible, with minor modifications, to all other senses. This is a hopeless strat-egy and Andreas Keller's monograph is a useful corrective to it. The concluding chapter, Comparing Olfaction and Vision, would be an excellent self-standing essay. In fact, it is Keller's desire to distance himself from visuocentrism in the philosophy, psychology, and neu-roscience of perception that led him to pursue a second doctorate on philosophy and write this monograph on olfactory perception. As Keller reminds us, " visual perception in humans is fundamen-tally different from human olfactory perception " (p.193). Olfaction does not maintain a permanent olfactory scene in the way vision sustains our view of our immediate surroundings. There is no agreed way to organize perceptual smell space or identify its funda-mental categories as there is for color or sound space. Olfaction has close ties to the emotions and only weak links to language; odors are difficult to name. The grouping of odors in olfactory perception reflects the behavioral needs and responses of individuals, not physi-cal similarities among the stimuli. These distinguishing features give olfaction a different function and character from vision; in particu-lar, olfactory perception does not involve, for Keller, the perception of objects. A key reason for this conclusion is Keller's conviction that the perceptual space of olfaction is not spatially or temporally struc-tured. The arguments for this claim come in Chapter 3 where he discounts different attempts to establish candidates for olfactory objects. His strategy is to consider criteria for objecthood in a per-ceived scene and to see whether these criteria are met in the case of smells. These include a distinction between figure and ground in perception, and the so-called Many Properties problem of deciding, for any perceived scene, which properties go with which objects; an instance of the binding problem. Keller works through cases and finds them wanting. Smells may be distinct to the perceiver, the smell of baked goods and fresh coffee, say, but which is the figure and which is the ground? We seem unable to provide an answer. The trouble with this strategy is that is takes the figure-ground distinc-tion in olfaction to be modeled on the visual case, and Keller even tells us at one point: To decide if any of these situations [the examples given] should be considered as evidence for figure-ground seg-regation, it is informative to compare them to analogous situations in visual perception. (p.74) At this point, readers may be wondering what happened to the insistence that we do not model olfactory perception on vision. The worry is that Keller has not exercised sufficient imagination in the range of cases he considers when attempting to distance him-self from the visuocentric outlook. The point is that we can find temporal and spatial aspects of olfactory experiences if we look hard enough. For example, if I am in a restaurant and someone lights up a ciga-rette, I have a pretty accurate idea of how near or far away they are from me. Although the smell of cigarette smoke cannot convey the direction from which the smoke emanates, it conveys an idea of the distance of the source. Similarly, I may enter a room and learn just from the character of the perceived odor that someone was smoking recently, or had been smoking a little while ago. And although most of us will have habituated to the smell of our own homes, this is not the same as not smelling it anymore. People with acquired anosmia speak of feeling alienated and cut off from their own home. By con-trast, normosmics will have the (perhaps unconsciously) perceived smell of their home as a baseline against which to detect when some-one has been there, or has been smoking, or left the garbage out. These perceptible changes to the baseline odor are foregrounded for the perceiver in a figure-ground way. For the Many Properties Problem Keller provides a nice exam-ple—the smell of pizza—but he draws entirely the wrong morals and it is illustrative to see why.