Coding of perceived odor intensity

One of the aims of systems neuroscience is to link neural activity in different stages of brain processing with specific sensations or percepts. This has been accomplished for our senses of touch and vision, but not for our sense of smell, or olfaction. To better understand olfaction, we started with a basic sensation common to many sensory systems: intensity. Based on prior work, we knew that mammals (rats and humans) compare quantities (concentrations) of different odors based on their perceived intensity.

Previous work showed that neuronal activity in different parts of the brain (the piriform cortex, entorhinal cortex, and amygdala) correlates with odor concentration.  However, this is not enough evidence for consistency with perception. In rats and humans, the perceived intensity of odors grows systematically with their concentration but rapidly decreases after prolonged sampling (or sniffing) due to adaptation. Thus, perceived intensity for a given odor is a function of at least two variables: the physical concentration of an odor outside the nose and the sampling duration. Therefore, in order for a neuronal response to represent intensity, it must change consistently with concentration and adaptation.

We hypothesized that a neural representation, or code, for odor intensity may already exist in the activity patterns of mitral and tufted cells (MTC) in the olfactory bulb. To test this hypothesis, we characterized changes in electrophysiological activity of MTCs as a function of odor concentration and adaptation and examined them for consistency with perception. MTCs respond to odors not only by changing the total number of spikes they fire (rate code) but also by changing how spikes are distributed over time or their temporal pattern. During every sniff cycle, MTCs fire a characteristic pattern of spikes.  When a new odor is presented, this pattern can change even while the total number of spikes remains the same (temporal code).

Left: Schematics of the effect of each of the transformation on sniff cycles.
Right: Raster plots of the response of MT cell aligned according to different schematics.

We first looked at how MTC odor responses change as a function of odor concentration (Fig. 1).  We next examined changes in responses due to adaptation by comparing the response on the first sniff of odor with consecutive sniffs; Fig. 3).  We compared how MTC responses change with concentration and adaptation based on total spike count (rate code; Fig. 5) and as well as spike temporal pattern (temporal code; Fig. 4). We found that changes in the total number of spikes fired by MTCs were not similar to odor dilution and adaptation (Fig. 5).  However, the temporal pattern of the spikes within each sniff changed in suggestively similar ways, in terms of both the latency and the amplitude of the response.

Each odorant activates multiple different receptor types and thus multiple MTCs in the olfactory bulb. Therefore, odors are likely represented by the activity of multiple cells, so we looked at how does odor concentration encoded by a population of MTCs. To do this, we grouped the measured responses of MTCs together into population response vectors (PRVs).  We build a PRV for each sniff of each odor concentration on each odor presentation (or trial). We ended up with many PRVs.  To see how these PRVs change across concentration and sniff number we applied Principal Component Analysis (PCA), which helps identify the most meaningful basis to re-express a noisy, data set. The assumption underlying PCA is that this new basis (or principal components, PCs) will filter out the noise and reveal the most prominent features of the data set. We visualized PRV responses on each sniff and concentration by plotting the responses in the space of the first three PCs, which accounted for 70% of the total variance in our data set (Fig. 6).  This showed that changes in PRVs with concentration and adaptation were similar to each other and therefore consistent with a representation of odor intensity. From this, we formed the following hypothesis: adaptation should make it harder to distinguish PRVs for different concentrations.

We tested this hypothesis quantitatively. Animals can gauge odor intensity from a single sniff of odor, so we checked how well a simple computer algorithm, called a classifier, could identify odor concentration based on the Euclidean distance between a single PRV and the average PRVs observed for different concentrations. Using PRVs from the first sniff of odor the classifier was able to identify the coded concentration with 92% accuracy. But performance using subsequent sniffs fell below 78% (Fig. 7B). Thus, although concentration information was still largely intact after adaptation, odor concentrations were harder to distinguish, consistent with our first hypothesis.

Our neural results suggested that the change in the coded concentration was abrupt, reduced significantly after even one sniff of odor.  However, prior work in olfactory perception measured changes over longer time intervals, not across individual sniffs.  To test whether perceived intensity changes abruptly after just one sniff of odor, we performed psychophysical experiments with human subjects. We taught human volunteers to rate the intensity of odors on specific sniffs after we turned on a constant odor source (Fig. 9). Volunteers rated the intensity of different odor concentrations presented either on the first sniff or after several adapting sniffs. Intensity ratings on the first sniff were well described using a Hill function (a well-known nonlinear relationship with odor concentration; Fig. 9A). However, a single adapting sniff decreased intensity ratings. We converted intensity to concentration units using the Hill function.  Ratings after adaptation corresponded to a roughly 2-fold dilution of the odor. Thus, consistent with our MTC measurements, perceived intensity fell abruptly after one adapting sniff and by a similar amount.

To summarize, we compared neural activity in the mouse olfactory bulb with odor intensity perception in humans. We found the two to be consistent.  Odor responses changed with decreasing concentration and repeated sampling of a constant odor source in similar ways. Using a classifier, we found that the odor concentration coded on later sniffs was sharply lower than on the first sniff. These neural results were consistent with the sharp sniff to sniff drop in odor intensity reported by human volunteers. Our data suggest that responses of neurons in the olfactory bulb are consistent with odor intensity perception.

Link to the article

7 comments

  1. Great content! Super high-quality! Keep it up! 🙂

  2. Like!! I blog frequently and I really thank you for your content. The article has truly peaked my interest.

  3. As in other sensory systems, the sampling of olfactory stimuli is tightly controlled by the animal, with important consequences for information coding, processing, and perception. Indeed, considering olfaction as a system in which stimulus sampling, behavioral state, motor system function, and information processing strategies are closely coordinated is fundamental to understanding olfaction in the behaving animal. This chapter touched on how active sensing is important and integrated at each of these levels. For a more detailed review of the relationship between odor sampling and nervous system function at a particular level, the reader is referred to several excellent reviews ( Schoenfeld and Cleland 2005, 2006 ; Buonviso et al. 2006 ; Mainland and Sobel 2006 ; Scott 2006 ; Wachowiak and Shipley 2006 ).

  4. My brother recommended I might like this web site. He was totally right. This post actually made my day.
    You cann at imagine just how much time I had spent for this info! Thanks!

  5. Wohh exactly what I was searching for, appreciate it for posting
    Actually when someone doesn’t understand then its up to other viewers that they will assist, so here it takes place.

  6. Gosh This was an extremely wonderful post. Thank you for supplying this information.

  7. I was able to find good information from your articles.

Leave a Reply