100 000 dollars for decoding memory from neural connections: Aspirational Neuroscience proposes five experiments that could win the prize

100 000 dollars for decoding memory from neural connections: Aspirational Neuroscience proposes five experiments that could win the prize

On November 6, a link to a working document appeared on the Hacker News forum. The prize will go to the first team to use a map of brain connections to identify what an animal experienced, for example, which of 32 odors a fly learned to associate with an electric shock.

The organization is led by neuroscientist Kenneth Hayworth, president of the Brain Preservation Foundation. Its starting point is the consensus that the vast majority of long-term memory is stored in the strength of synaptic connections, although “most experimental evidence is indirect.” To test this directly, all five experiments follow the same design: an animal is given a known experience, its brain is imaged under a microscope, the images are used to map its connections, and a team blinded to the answer uses that map to identify the experience.

The document defines memory as a trace of an individual animal’s experience; innate reflexes and traces of injury do not qualify. It proposes measuring decoding difficulty in bits: one bit means choosing between two possibilities, and five means choosing among 32. One bit is insufficient: in a 2015 experiment, synaptic strength revealed which of two directions each of 14 rats had been trained to take in response to a low-pitched tone.

The first experiment uses flies, in which learning takes place in the mushroom body. There, an odor is encoded by a group of Kenyon cells. During learning, dopamine neurons weaken the synapses that these cells make onto output neurons controlling approach and avoidance. The same odor activates different groups of Kenyon cells in different flies because the wiring from the antennal lobe glomeruli to these cells is nearly random. The glomeruli, however, are the same across flies and respond to the same odors. The team therefore traces which glomeruli supply input to each Kenyon cell, predicts the group of cells activated by each odor, and looks for the group whose synapses have weakened.

Some traces of learning are too easy to read: a 2021 study found that long-term memory of an odor paired with sugar was accompanied by an increase in the number of synaptic complexes associated with the input neurons for that odor, and those neurons can be identified by their glomerulus. The document therefore proposes rewarding some flies with sugar and giving others electric shocks, so that the trace must be sought at the output.

The remaining experiments require teams to identify a tone paired with an electric shock, the orientation of stripes shown to a mouse, and an environment a rodent has visited. The final experiment requires reconstructing a zebra finch’s song. The document proposes identifying the environment from the hippocampus: if cells active in the same locations are more strongly connected to one another, the pattern of their connections will reproduce the environment’s shape, forming a ring for a circular track or a sheet for a square arena. According to the leading model, the sequence of a song is determined by a chain of cells in which each activates the next, so that sequence should be readable from the connections. The document defines an unambiguous win as a synthesized recording that resembles the individual bird’s song more closely than the songs of other birds.

All five experiments rest on the assumption that a synapse strengthened by learning is larger than usual and that this difference is visible in the images. For the hippocampus, the document cites a finding that challenges the proposal: a 2025 study in Science found that mouse cells active during fear learning showed no strong preference for connecting to one another.

The foundation’s executive director, Ariel Zeleznikow-Johnston, described the prize as the spiritual successor to the foundation’s 2018 prize for preserving a pig brain: that prize tested whether connections between neurons had survived, while the new prize tests whether memories can be read from those connections. According to Zeleznikow-Johnston, success would strengthen the case for brain preservation, while failure after sustained efforts would weaken it.