Manipulating Synthetic Optogenetic Odors Reveals the Coding Logic of Olfactory Perception
How are things that we experience in the world represented in the brain? Multiple advances in monitoring neural activity, such as with electrophysiology and calcium imaging, have revealed tremendous complexity in how the brain represents what we see, hear, or smell. While many features of neural activity patterns (the identity and timing of neuronal activity) are found to correlate with changes in the external sensory world, it is unclear which of these activity features are directly causal for perception and how individual activity features may combine to generate perceptions. For example, the activity of some cells in a pattern may be more important than others, or the sequential activation of multiple cells or the latency of their activation relative to brain rhythms may carry important information about the encoded perception.
In order to understand the link between neural activity and perception, we used mouse olfaction as our model system, where we measured perceptual responses to precise manipulation of neural activity. We chose mouse olfaction for its experimental tractability: the relevant olfactory circuits are known and, importantly, accessible for direct manipulation. Axons of olfactory sensory neurons in the nose, expressing the same receptor genes and having similar odor selectivity, converge to discrete neuropil, or glomeruli, in the olfactory bulb. In addition, odor information is correlated with the precise activation of glomeruli: each odor can be uniquely encoded in the identity and timing of glomeruli activated.
We used genetically engineered mice where channelrhodopsin is expressed in all olfactory glomeruli, training mice to recognize synthetic patterns of glomerular activation, or “synthetic odors.” Subsequently, we measured how recognition changed as we systematically manipulated learned synthetic patterns. Some manipulations lead to larger changes in recognition than others, and the degree of change reflects the importance of each manipulated feature to perception. Because we manipulated features both individually and in concert, we could precisely quantify which combinations of features were important for perception.
We found that the perceptual responses of mice not only depended on which glomeruli were activated, but also when they were activated. In other words, temporal sequences are akin to timed notes in a melody. Surprisingly, we found that the most perceptually relevant activation latencies were defined relative to other glomeruli in a sequence, and not to brain or body rhythms such as animal sniffing, as previously postulated from observational studies. Furthermore, we found a “primacy effect” where earlier-activated glomeruli in the sequence had a larger effect on behavioral responses. Modifying later glomeruli in the sequence had small effects—this confirmed previously published work from our lab.
To account for all results, we formulated a simple computational model based on template matching: new activity sequences are compared to learned sequences, or templates. The model weighs relative timing within each sequence, and also accounts for the greater importance of earlier-activated glomeruli. Based on our model, the degree of mismatch between the new sequence and learned template predicts the extent to which recognition should degrade as glomerular activity changes, across many different manipulations.
In summary, we developed an experimental and theoretical framework to map a broad space of precisely and systematically manipulated brain activity patterns to behavioral responses. Under this framework, we found that we were able to summarize the key computations performed by the olfactory system into a simple model which accounted for animals’ perceptual responses. Finally, our framework might have relevancy for questions beyond olfaction: for example, the approach can also be used for testing the causal links between complex, spatio-temporal neural activity as well as perception or behavior.
—Edmund Chong, MS, and Monica Moroni, PhD
Read the paper ”Manipulating synthetic optogenetic odors reveals the coding logic of olfactory perception” in Science, published June 19, 2020.