How do flies learn smells? The mushroom body and memory
Short answer: Fruit flies learn to pair a smell with sugar (a reward) or an electric shock (a punishment), and then approach or avoid that smell. The learning happens in the mushroom body: a smell activates a small group of Kenyon cells, and dopamine released at the moment of reward or punishment changes their connections to output neurons. Training repeated with breaks builds long-term memory that lasts for days.
The T-maze: the experiment that teaches flies smells
In 1974 Quinn, Harris and Benzer exposed groups of flies to two odours in turn; one of them came with an electric shock. Afterwards the flies avoided that odour, and the memory lasted a day. In 1985 Tully and Quinn standardised the method: about 100 flies receive one odour together with 12 electric shocks over one minute, then the other odour without shocks. They then choose between the two arms of a T-maze, and most stay away from the shock-paired odour. In the same way, flies also learn to approach an odour rewarded with sugar.
In 1976 the same test found the first "learning mutant": dunce. These flies sensed the odour and the shock normally but could not link the two. The broken gene encoded an enzyme that breaks down cAMP, a messenger molecule inside cells, which showed genetically, in the fly, the central role of cAMP signalling in learning.
A smell's path through the brain
Olfactory receptor neurons on the antennae converge on about fifty small balls of synapses (glomeruli) in the antennal lobe. From there projection neurons carry the smell to the mushroom body. Each side of the mushroom body has about 2,000 Kenyon cells, and every smell activates only a small fraction of them. This sparse code makes it easier to keep even very similar smells apart.
Dopamine as a teacher
The Kenyon cells' fibres run in parallel and are divided into 15 compartments. Each compartment has its own output neurons (34 cells of 21 types) and matching dopamine neurons (20 types). Reward activates dopamine neurons in the PAM cluster; punishment activates those in the PPL1 cluster. Dopamine weakens the connections from the Kenyon cells active at that moment to the output neuron of that compartment. The next time, the same smell shifts the balance between "approach" and "avoid" outputs.
Short-term and long-term memory
Memory from a single training session fades within hours and is barely detectable the next day. When training is repeated, two different memories can form:
- Spaced repetition (with rest between sessions): long-term memory that can last more than a week. It needs new protein synthesis: a drug that blocks protein synthesis (cycloheximide) prevents it, and a gene-regulating protein called CREB is required.
- Massed repetition (back to back): anaesthesia-resistant memory. It does not depend on new proteins and fades within a few days.
Spaced repetition is known to produce more lasting learning in people too; these fly studies are among the best-studied examples of the molecular basis of memory.
Sources
- Quinn WG, Harris WA, Benzer S (1974). Conditioned behavior in Drosophila melanogaster. PNAS 71: 708–712.
- Dudai Y, Jan YN, Byers D, Quinn WG, Benzer S (1976). dunce, a mutant of Drosophila deficient in learning. PNAS 73: 1684–1688.
- Tully T, Quinn WG (1985). Classical conditioning and retention in normal and mutant Drosophila melanogaster. J. Comp. Physiol. A 157: 263–277.
- Tully T. et al. (1994). Genetic dissection of consolidated memory in Drosophila. Cell 79: 35–47. https://repository.cshl.edu/id/eprint/31413/
- Yin JC et al. (1994). Induction of a dominant negative CREB transgene specifically blocks long-term memory in Drosophila. Cell 79: 49–58. https://repository.cshl.edu/id/eprint/31414/
- Aso Y. et al. (2014). The neuronal architecture of the mushroom body provides a logic for associative learning. eLife 3: e04577. https://elifesciences.org/articles/04577