In the experiment, rats learned which lever to press to receive water, where the correct lever depended on which lever they had pressed previously (the levers were retractable; there was a variable delay between the first and second presentation of the levers). Microelectrodes in the rats’ brains provided data that enabled researchers to work out the firing patterns of neurons in CA1 that resulted from particular firing patterns in CA3 (previous research had established that long-term memory involves CA3 outputs being received in CA1).

Normal neural communication between these two subregions of the hippocampus was then chemically inhibited. While the rats still remembered the general rule, and still remembered that pressing the levers would gain them water, they could only remember which lever they had pressed for 5-10 seconds.

An artificial hippocampal system that could reproduce effective firing patterns (established in earlier training) was then implanted in the rats’ brains and long-term memory function was restored. Furthermore, when the ‘memory prosthetic’ was implanted in animals whose hippocampus was functioning normally, their memory improved.

The findings open up amazing possibilities for ameliorating brain damage. There is of course the greatly limiting factor that effective memory traces (spatiotemporal firing patterns) need to be recorded for each activity. This will be particularly problematic for individuals with significant damage. Perhaps one day we will all ‘record’ ourselves as a matter of course, in the same way that we might put by blood or genetic material ‘in case’! Still, it’s an exciting development.

The next step will be to repeat these results in monkeys.