For decades, scientists have believed that our memory for events works by first creating a neural code for that event, and then later reinstating that code when remembering that event. However, recent work based on a powerful new technique called ‘calcium imaging’ has challenged this view, showing that over weeks and months many parts of the brain actually use a surprisingly different neural code to represent their content. This includes the hippocampus, a key hub for memory for events, where mental maps of the same space continue to change and change with time. These sorts of changing or ‘drifting’ memory representations can mean a lot of different things, and right now scientists across the world are trying to figure out just how random or organized these changes are.

In this work, Dr. Mark Brandon and his team address this question directly. In work led by Alexandra Keinath, and published this week in Nature Communications, the hippocampus of mice was recorded with calcium imaging as these mice repeatedly explored a bunch of specially-chosen similar Lego environments. Over more than a month of experiments, they tracked their neural codes and found that their mental maps of these places changed – not randomly – but in a very particular way. This particular organization allows the brain to keep track of where the mouse is without having to do any extra work, despite all the ongoing changes to their neural codes. Moreover, their data suggests that these ongoing changes might themselves be representing something interesting, such as time or intervening experience, beyond the mental map itself in an unexpected way, raising new questions for future work.

Zooming out a bit, this work shows that seeing changes in a neural code alone is not enough to draw conclusions about what that neural code represents. Thus it is especially important that we use carefully-designed experiments like this one to characterize the structure of drift in neural codes all throughout the brain when trying to understand how this drift affects memory and perception.

Like all scientific work, this study also has its limitations. One big limitation is that it can only speak to the organization of neural drift for the environments that the mice explored. Therefore, it is possible that the sort of structure we observed only occurs for representations of similar environments. In this case, the researchers needed to record from these similar environments for technical reasons, but perhaps this limitation can be overcome in future work.

Another more general limitation is currently confronting all work on neural drift. Right now, we know of some examples of neural drift and some examples of extremely stable long-term representations in different brain regions in a variety of species, such as mice, birds, and monkeys. However, it remains to be seen whether we observe similar phenomena in human brains, and if so whether the structure of neural drift looks anything like it does in our rodent friends.


During this work ATK was supported by a McGill University Healthy Brains for Healthy Lives CFREF postdoctoral fellowship and a Natural Sciences and Engineering Research Council (NSERC) Banting postdoctoral fellowship. CAM was supported by a Fonds de Recherche du Québec – Santé (FRQS) postdoctoral fellowship. Funding was provided by the Canadian Institutes for Health Research (Project grants #367017 and #377074), the Natural Sciences and Engineering Research Council of Canada (Discovery grant #74105), the Canada Research Chairs Program, and the Brain Canada Foundation (Future Leaders in Canadian Brain Science) to MPB.