Brain oscillations bind smell to memory
As a foodie, one scene in Pixar’s Ratatouille particularly resonated with me: Anton Ego, the acerbic food critic, takes one bite out of the protagonist’s ratatouille and is instantly reduced to tears by a childhood memory of his mother’s cooking.
Smell, our oldest sense, is particularly powerful at unlocking our deepest memories. Scientists have long wondered why this is the case. One reason may be the way our brain processes olfactory information: unique among our senses, smell does not require relay by the thalamus before it enters its main processing hub – the olfactory bulb. Instead, smell directly stimulates the olfactory bulb, and simultaneously activates the medial amygdala and the entorhinal cortex – areas important for learning emotionally powerful events. The entorhinal cortex is known to transfer information from the cortex to the hippocampus, a main learning hub in the brain. Thus, smell is directly linked to memory.
But how exactly are these distributed brain regions communicating with each other? Previous studies have revealed a mode of neural chatter called oscillations, in which distributed neurons with specialized functions activate rhythmically to “synch up” different brain areas. We already know that these brain oscillations are important for retrieving memory, but their role in learning is still a mystery. A team of neuroscientists from the Kavli Institute for Systems Neuroscience decided to see if these synchronized brain waves drive the learning of smell-memory associations.
Researchers trained a group of rats to associate two distinct odours with two cups placed at different spots in a box – a task that depends on the hippocampus and its associated cortical areas. For example, the scent of bananas meant that the food reward was hidden in the left cup, while a whiff of pine trees pointed to the right cup as the jackpot. After 3 weeks of training, the rats could successfully locate the reward 85% of the time in two consecutive trials. Researchers then implanted 16 electrodes into their hippocampus and entorhinal cortex to peek into their brain activity.
Since smell rapidly triggers a linked memory, researchers reasoned that oscillations - required to “fish out” a related memory - should be most prominent as the animals sampled the odour cues. Just as they thought, immediately after exposing the rats to the odour pair, the distal CA1 (dCA1) area of the hippocampus erupted in activity. Researchers observed oscillatory waves between 20-40Hz, which gradually increased in magnitude as the rats sniffed the cue odour, but dropped back to baseline once the rats made up their minds and started running towards the cups. A similar pattern of activity occurred during odour sampling in the lateral entorhinal cortex (LEC), which projects to the hippocampus. The oscillations of dCA1 and LEC were tightly coupled to each other, while another area of the entorhinal cortex (MEC) did not exhibit similar oscillations. These results suggest that in well-trained rats, increased synchronization between dCA1 and LEC are the result of direct neuron-neuron connections between the two brain areas.
Top: Synchrony at 20Hz (dashed line) between LEC and dCA1 increases (red patches) as the rats start sniffing the odour cues. Bottom: MEC and dCA1 on the other hand don't seem to talk to each other.
But how do these oscillations change during learning, if at all? Researchers next repeated the experiment with a new group of rats, but this time they monitored brain waves during which rats mastered the odour-place associations. Once again, researchers found 20-40Hz brain waves in the dCA1 and LEC, which increased in coherence, but not magnitude, as learning progressed. So it seems that learning synchronizes pre-existing brain waves between these connected areas, as opposed to generating them anew.
So what do these oscillations do? Since they spring up during the odour-sampling phase, researchers reasoned that these brain waves might help rats “fish out” the correct odour-associated memory. To test this, researchers separately looked at data from successful trials and misremembered trials, in which rats went after the wrong cup. Compared to successful trials, the coherence between dCA1 and LEC was dramatically lower in error trials, suggesting the two brain areas weren’t effectively talking to each other.
When researchers zoomed in on each brain area, they found that as learning progressed, a subset of neurons gradually developed specificity for a particular odour. In other words, some neurons only activated in response to one odour but not the other, hence forming a neural “representation” for the selected odour. During the initial sniffing phase, odour selectivity sometimes increased, which strongly correlated with the emergence of the 20-40Hz oscillations. What's especially interesting is that this only occurred in successful trials. In failed ones, both selectivity and oscillations were pretty much absent. Taken together, it seems that odour-specific representations in the dCA1 and LEC, along with coupling between the two brain areas, are necessary for successful odour-based navigation.
According to Edyard Moser, director of the Kavli Institute and lead author of the paper, this is the first study that correlates the emergence of a specific band of brain waves to hippocampus-associated learning and recall. Perhaps this is why smell potently triggers memory: when you encounter a new fragrance, increased oscillations spanning the olfactory bulb, the entorhinal cortex and the hippocampus binds the scent to the where and what. A whiff of the same scent – even years or decades later – reactivates these previously synchronized networks and pulls the associated memory trace from the depth of your mind.
In this study, the authors used individual fragrances; humans have the ability to detect up to a trillion different scents individually and in countless combinations. Do complex scents - the smell of forests, of perfume, of food, of home – also have their each unique representation? If so, are these memories also retrieved by oscillating brain waves?
Igarashi KM, Lu L, Colgin LL, Moser MB, & Moser EI (2014). Coordination of entorhinal-hippocampal ensemble activity during associative learning. Nature PMID: 24739966
