Memories have both a “where” and a “when,” but how does the brain organize events according to space and time? A large body of research has shown that the hippocampus has “place cells” that recognize and fire when an animal is in a particular location, enabling the hippocampus to form spatial maps. In the August 25 Neuron, researchers led by Howard Eichenbaum at Boston University, Massachusetts, report that rat hippocampal neurons can also function as “time cells,” firing at specific moments over a stretch of time. Intriguingly, Eichenbaum and colleagues found that most of these neurons measure both time and space simultaneously. This implies that the cells are not dedicated to particular functions, but instead can map any stable feature of the environment, Eichenbaum speculated. If true, this would have implications for how memory works, and might eventually give researchers new ways to model memory loss and how it might be repaired in memory disorders. In agreement with Eichenbaum’s findings, Yuji Naya and Wendy Suzuki at New York University saw similar time-tracking behavior in the hippocampal neurons of monkeys performing a temporal memory task. Their results appeared in the August 5 Science.
“In effect, the hippocampus is a great organizer,” Eichenbaum proposed. “It uses stable features to organize specific events that occur. That is a much broader view of how the hippocampus operates [than the view that it simply maps space], and might eventually be a much more useful way to think about how memory works.”
An extensive literature going back 40 years describes hippocampal place cells, but few researchers have looked at how the hippocampus records time. Researchers led by Ray Kesner at the University of Utah, Salt Lake City, showed that rats with a lesion to the CA1 region of the hippocampus could not remember an object-odor paired association over time intervals of up to 10 seconds (Kesner et al., 2005). Eichenbaum and colleagues also demonstrated that CA1 is essential for temporal tasks of several seconds (see Farovik et al., 2010). It was not clear, however, what the hippocampus was doing during this time.
To answer that question, Eichenbaum and colleagues adapted the behavioral paradigm developed by Kesner. First author Christopher MacDonald trained four rats to perform a paired-association task with a temporal component. Researchers placed each rat in a modified T-maze containing one of two objects, which the rat had previously learned to associate with a particular smell. Then the rat entered an empty section of the maze for a delay period, which was typically 10 seconds. Finally, the animal entered the last section of the maze and sniffed an odor buried in sand in a flowerpot. If the rat smelled the odor associated with the object, the animal could dig in the pot for a food reward; if the odor did not match the object, the rat had to ignore the pot and move down the maze to get a reward. Once the animals were trained, MacDonald and colleagues surgically implanted electrodes into the CA1 region of the hippocampus, and recorded from pyramidal neurons throughout each trial. The researchers ran around 100 successive trials in each recording session, completing six sessions overall with the four animals.
The researchers found that those neurons active during the 10-second delay period fired at particular moments, much as place cells fire at selective locations. The overlapping firing periods of sets of neurons bridged the entire 10-second gap, suggesting that these neuronal ensembles were mapping time. “The hippocampus creates a framework of time much as it creates a framework of space—by taking an open period of time and parsing it into little pieces,” Eichenbaum said.
The researchers used complex statistical analysis to see whether the firing of these neurons corresponded to other variables in the environment as well. Surprisingly, they found that about three-fourths of the neurons seemed to be responding to both temporal and spatial cues in equal measures. For example, a particular neuron would only fire if the rat was standing in a specific spot at a particular time, i.e., both conditions had to be met for the neuron to activate. The remaining fourth were evenly split between pure time cells and pure place cells. Moreover, neuron firing was also influenced by what direction the animal was facing and how fast it was moving, an effect also seen in place cells. “Each cell really is encoding a complex set of features of the ongoing event,” Eichenbaum told ARF.
For the most part, different neurons fired during the object, delay, and odor periods, although some cells were active in more than one period, Eichenbaum said. For the neurons that fired when the animal was investigating the object or smelling the odor, the majority of active cells represented either time or space, but not both, perhaps suggesting that time and space are encoded differently during events than during less structured periods.
MacDonald and colleagues wondered if time cells were measuring absolute time or relative time. To investigate this, they changed the length of the delay to 20 seconds on several successive trials. They found that some of the neurons were measuring absolute time, continuing to fire at the same exact point in the delay as they had before, i.e., at two seconds into the delay, four seconds into the delay, etc. Other cells seemed to be measuring relative time, expanding or otherwise rescaling their firing activity to match the doubled delay length. About two-thirds of the active cells, however, did neither. Instead, they changed their behavior in unpredictable ways during the longer delay, for example, ceasing to fire altogether, changing their firing rate or pattern, or becoming active when they had previously been silent. Eichenbaum refers to this behavior as “retiming,” similar to the partial “remapping” seen in place cells when one feature of the environment is changed. “One implication is that the hippocampus is aware of both the similarities and differences in an altered environment,” Eichenbaum told ARF. That environmental complexity is captured by having some cells that continue to fire the same way, while others alter behavior to reflect the changes, he said.
The findings agree with Naya and Suzuki’s study, which used a different animal model and a slightly different temporal-order memory task. The researchers recorded from visual cortex and three medial temporal lobe regions—hippocampus, entorhinal cortex, and perirhinal cortex—as the monkeys performed the task. Like Eichenbaum and colleagues, Naya and Suzuki found that hippocampal activity evolved over time during the delay period. In addition, they saw evidence that the different brain structures cooperated to code distinct aspects of a sequential memory. Both the hippocampus and entorhinal cortex seemed to respond to time cues.
Eichenbaum said that together, his study and Suzuki’s “make a powerful argument that the hippocampus does some kind of mapping of time in an effort to organize memories of specific objects.” In a commentary published along with the Neuron paper, Matthew Shapiro at Mount Sinai School of Medicine, New York City, wrote, “The two studies agree that hippocampal representations evolve in time independent of other external variables, and that time cells could signal the unfolding history of experience.”
One of the next questions to answer, Eichenbaum said, is where this representation of time originates. Does the hippocampus generate it on its own? Or does a neighboring structure, such as the entorhinal cortex, produce the timing information and feed it to the hippocampus? In ongoing experiments, Eichenbaum plans to search for the circuitry that generates timing by recording from medial and lateral entorhinal cortex. The medial entorhinal cortex is known to process spatial information, while the lateral structure handles information about specific objects; either one could be a candidate for producing time signals.
These studies also have larger implications, Eichenbaum said. “We need to broaden our thoughts about what it is the hippocampus is doing. It is not dedicated to space.” Instead, sets of hippocampal neurons may bridge and organize any useful feature of remembered events. By better understanding how the hippocampus encodes episodic memories, perhaps researchers can ask more detailed questions about how memory breaks down in disorders such as Alzheimer’s disease, Eichenbaum suggested. “What is it about the framework in which information is deposited that is falling apart? Studies like this should give us better animal models so that we can study how the encoding of information by neurons is falling apart in aging or AD, and how it is altered or repaired by possible treatments.”—Madolyn Bowman Rogers
- Kesner RP, Hunsaker MR, Gilbert PE. The role of CA1 in the acquisition of an object-trace-odor paired associate task. Behav Neurosci. 2005 Jun;119(3):781-6. PubMed.
- Farovik A, Dupont LM, Eichenbaum H. Distinct roles for dorsal CA3 and CA1 in memory for sequential nonspatial events. Learn Mem. 2010 Jan;17(1):12-17. PubMed.
No Available Further Reading
- Naya Y, Suzuki WA. Integrating what and when across the primate medial temporal lobe. Science. 2011 Aug 5;333(6043):773-6. PubMed.
- Macdonald CJ, Lepage KQ, Eden UT, Eichenbaum H. Hippocampal "time cells" bridge the gap in memory for discontiguous events. Neuron. 2011 Aug 25;71(4):737-49. PubMed.
- Shapiro ML. Memory time. Neuron. 2011 Aug 25;71(4):571-3. PubMed.