Essays on Science and SocietyEppendorf

Play It Again, Brain

See allHide authors and affiliations

Science  01 Nov 2013:
Vol. 342, Issue 6158, pp. 574
DOI: 10.1126/science.1245966

You did it last night, and you will do it again tonight. And you're not alone, because every living organism with a brain needs to do it: sleep. And, for something so critical to our well-being, it's remarkable that we don't understand why we need to sleep. We may feel our conscious mind shutting off for much of the night and think that our brain is resting, but our brain actually remains very active. And what our brain is doing when we sleep provides a hint about what many neuroscientists believe to be a fundamental role of sleep.

While we are awake, our sensory experience of the world creates a unique pattern of neural activity throughout our brain. One region of the brain, the hippocampus, is thought to play a central role in storing the memory of these neural activity patterns that represent the day's events. Much like a computer that gets backed up periodically, when we go to sleep, the hippocampus is believed to initiate a “backing-up” process referred to as systems-level memory consolidation. Memories stored in the hippocampus get reactivated, in turn, driving the re-encoding of these memories in neocortical brain regions where long-term memories are thought to eventually reside.

We can observe this in a rat implanted with tiny microwires that record neural activity from its hippocampus. When the rat is awake, its hippocampus acts as its global positioning system, encoding the animal's position in space with the activity of place cells (1). Each place cell has a different preferred location, and as the rat moves along a spatial trajectory, a particular sequence of place cells is activated [see the figure (A)]. When the rat goes to sleep, something interesting happens. These same place cells are spontaneously reactivated in the same order [see the figure (B)] and recreate the sequential pattern that had just occurred while the rat was awake and running (2, 3). These “replay” events are a neural memory trace of the rat's recent experience of running along a spatial trajectory and have been speculated to be the driving mechanism that is responsible for systems-level memory consolidation.

Eppendorf and Science are pleased to present the essay by Daniel Bendor, a 2013 finalist for the Eppendorf and Science Prize for Neurobiology.

If our long-term memories are the consequence of hippocampal replay while we sleep, then making the hippocampus selectively reactivate a particular memory in our sleep should reinforce that memory. While Hollywood has focused on more nefarious “while-you-were-sleeping”–based technologies for memory manipulation, such as memory deletion in “Eternal Sunshine of the Spotless Mind” and false-memory creation in “Inception,” reinforcing and strengthening existing memories could have helpful therapeutic effects, from improving our memory capacity before taking an exam to slowing down dementia-related memory degradation. Recent human data suggest that pairing a sensory cue with a behavioral task can enhance memory consolidation if this cue is also present during a postbehavior nap (4, 5). In other words, if you study for an exam in a rose garden, you can improve your test score by placing a bouquet of roses next to your bed before you go to sleep.

Examples of place cells, sleep replay, and cue-biased replay.

(A) Place fields (top) and place cell activity (bottom) as the rat runs along a linear trajectory. (B) Sequential firing patterns of place cells during sleep replay events. (C) In this experiment, rats were trained to run to the right after hearing sound R and to run to the left after hearing sound L. During non-REM sleep, if sound L was played, replay events were more likely to involve place cells with place fields on the left side of the track. If sound R was played, replay events were more likely to involve place cells with place fields on the right side of the track.

How does a cue associated with a previous experience, presented to you in your sleep, improve your memory of this experience? As a postdoctoral fellow in the laboratory of Matthew Wilson at Massachusetts Institute of Technology (MIT), I investigated this question by studying the hippocampal activity of rats during an auditory-spatial discrimination task and a post-training sleep session. Rats were placed on a linear track and, after their nose poke initiated a trial, they heard one of two sounds. If they heard the first sound (sound R) they had to run to the right side of the track to receive a reward, whereas for the second sound (sound L), they had to run to the left side of the track. After performing the task, rats slept in a remote location while these two sounds were repeatedly played quietly in the background. While the rat was in the non-REM stage of sleep, reactivated memory traces repeated spontaneously, however, after either sound R or sound L was played, replay events were more likely to reflect the track location associated with that sound (6). In other words, after sound R was played, place cells on the right side of the track were more likely to be active during replay events than if sound L had played (the converse was true for sound L) [see the figure (C)]. This cue-dependent bias was not observed in awake rats, which suggested that cue-biasing of replay events is a phenomenon specific to sleep.

But biasing the activity of neurons does not necessarily mean that I was biasing what (or where) the rat was thinking about. To interpret what the rat was thinking, I next used a probabilistic (Bayesian) neural decoding algorithm. If you know the probability that a neuron will be active when the rat is at a certain location, then for a population of neurons you can “invert” this probability to predict the rat's location from the neuronal population's activity at any given moment (7, 8). Now, instead of analyzing neural activity, one can analyze what this neural activity means (i.e., where the rat thinks he is during a replay event). Using this method, I observed a similar cue-induced positional bias in the replay. Thus, playing sound R increased the chance that the sleeping rat thought about the right side of the track during successive replay events (and the converse was also true for sound L). These results help explain how a sound associated with a previous experience, when played while you are sleeping, can be used to improve your memory consolidation.

So, next time you go to sleep, remember that you have a choice regarding what your brain is going to replay. Listen wisely.


Daniel Bendor is a lecturer in the Department of Cognitive, Perceptual, and Brain Sciences and the Institute of Behavioral Neuroscience at University College London. Dr. Bendor received his Ph.D. from Johns Hopkins University under the mentorship of Dr. Xiaoqin Wang, studying temporal processing in auditory cortex and the neural correlate of pitch and flutter perception. For his postdoctoral research, he investigated the role of the hippocampus in memory encoding and consolidation, while working with Dr. Matthew Wilson at the Massachusetts Institute of Technology. He has recently started his own lab at University College London, where his research focuses on how neural ensembles encode perceptual and memory-related information.


For the full text of finalist essays and for information about applying for next year's awards, see Science Online at

References and Notes

  1. Acknowledgments: The research discussed in this essay was supported by a Merck Award–Helen Hay Whitney Postdoctoral Fellowship (D.B.), a Charles King Trust Postdoctoral Fellowship (D.B.), and NIH grants 1-K99-DC012321-01 (D.B.) and 5R01MH061976 (M. Wilson).

Navigate This Article